Olena-patches

olena-patches@lrde.epita.fr

9625 discussions

last-svn-commit-768-g8f266e9 mln/util/array.hh: Add last() method.

by Guillaume Lazzara

--- milena/ChangeLog | 4 ++++ milena/mln/util/array.hh | 25 +++++++++++++++++++++++-- 2 files changed, 27 insertions(+), 2 deletions(-) diff --git a/milena/ChangeLog b/milena/ChangeLog index e03d35c..c4bed2c 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,5 +1,9 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + * mln/util/array.hh: Add last() method. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Add an optional base index for debug::iota. * mln/debug/iota.hh, diff --git a/milena/mln/util/array.hh b/milena/mln/util/array.hh index 0fe34e7..c7febfe 100644 --- a/milena/mln/util/array.hh +++ b/milena/mln/util/array.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2008, 2009 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2008, 2009, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -187,6 +187,11 @@ namespace mln /// \pre i < nelements() mutable_result operator[](unsigned i); + /// \brief Return the last element. + ro_result last() const; + + /// \brief Return the last element. + mutable_result last(); /// Empty the array. All elements contained in the array are /// destroyed. \post is_empty() == true @@ -553,6 +558,22 @@ namespace mln template <typename T> inline + typename array<T>::ro_result + array<T>::last() const + { + return v_[nelements() - 1]; + } + + template <typename T> + inline + typename array<T>::mutable_result + array<T>::last() + { + return v_[nelements() - 1]; + } + + template <typename T> + inline bool array<T>::is_empty() const { -- 1.5.6.5

13 years, 10 months

last-svn-commit-769-g77429c3 Rename files in Scribo.

by Guillaume Lazzara

* scribo/primitive/extract/elements.hh, * src/primitive/extract/extract_discontinued_hlines.cc, * src/primitive/extract/extract_discontinued_lines.cc, * src/primitive/extract/extract_discontinued_vlines.cc, * src/primitive/extract/extract_thick_hlines.cc, * src/primitive/extract/extract_thick_vlines.cc: Rename to... * scribo/primitive/extract/non_text.hh, * src/primitive/extract/discontinued_hlines.cc, * src/primitive/extract/discontinued_lines.cc, * src/primitive/extract/discontinued_vlines.cc, * src/primitive/extract/thick_hlines.cc, * src/primitive/extract/thick_vlines.cc: ... this. * scribo/toolchain/internal/content_in_doc_functor.hh: Rename function call from extract::elements to extract_non_text. * src/primitive/extract/Makefile.am: Update target names. --- scribo/ChangeLog | 23 +++++++ .../primitive/extract/{elements.hh => non_text.hh} | 64 +++++++++++--------- .../toolchain/internal/content_in_doc_functor.hh | 2 +- scribo/src/primitive/extract/Makefile.am | 26 ++++---- ...scontinued_hlines.cc => discontinued_hlines.cc} | 0 ...discontinued_lines.cc => discontinued_lines.cc} | 0 ...scontinued_vlines.cc => discontinued_vlines.cc} | 0 .../{extract_thick_hlines.cc => thick_hlines.cc} | 0 .../{extract_thick_vlines.cc => thick_vlines.cc} | 0 9 files changed, 73 insertions(+), 42 deletions(-) rename scribo/scribo/primitive/extract/{elements.hh => non_text.hh} (82%) rename scribo/src/primitive/extract/{extract_discontinued_hlines.cc => discontinued_hlines.cc} (100%) rename scribo/src/primitive/extract/{extract_discontinued_lines.cc => discontinued_lines.cc} (100%) rename scribo/src/primitive/extract/{extract_discontinued_vlines.cc => discontinued_vlines.cc} (100%) rename scribo/src/primitive/extract/{extract_thick_hlines.cc => thick_hlines.cc} (100%) rename scribo/src/primitive/extract/{extract_thick_vlines.cc => thick_vlines.cc} (100%) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 9cbfca1..e08f2fd 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,28 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Rename files in Scribo. + + * scribo/primitive/extract/elements.hh, + * src/primitive/extract/extract_discontinued_hlines.cc, + * src/primitive/extract/extract_discontinued_lines.cc, + * src/primitive/extract/extract_discontinued_vlines.cc, + * src/primitive/extract/extract_thick_hlines.cc, + * src/primitive/extract/extract_thick_vlines.cc: Rename to... + + * scribo/primitive/extract/non_text.hh, + * src/primitive/extract/discontinued_hlines.cc, + * src/primitive/extract/discontinued_lines.cc, + * src/primitive/extract/discontinued_vlines.cc, + * src/primitive/extract/thick_hlines.cc, + * src/primitive/extract/thick_vlines.cc: ... this. + + * scribo/toolchain/internal/content_in_doc_functor.hh: Rename + function call from extract::elements to extract_non_text. + + * src/primitive/extract/Makefile.am: Update target names. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Add Base64 conversion routines. * scribo/scribo/convert/from_base64.hh, diff --git a/scribo/scribo/primitive/extract/elements.hh b/scribo/scribo/primitive/extract/non_text.hh similarity index 82% rename from scribo/scribo/primitive/extract/elements.hh rename to scribo/scribo/primitive/extract/non_text.hh index ddf2c92..a017f1f 100644 --- a/scribo/scribo/primitive/extract/elements.hh +++ b/scribo/scribo/primitive/extract/non_text.hh @@ -26,12 +26,12 @@ /// \file /// -/// \brief Find in a document elements which are not text. +/// \brief Find in a document non text which are not text. /// /// \fixme To be optimized! -#ifndef SCRIBO_PRIMITIVE_EXTRACT_ELEMENTS_HH -# define SCRIBO_PRIMITIVE_EXTRACT_ELEMENTS_HH +#ifndef SCRIBO_PRIMITIVE_EXTRACT_NON_TEXT_HH +# define SCRIBO_PRIMITIVE_EXTRACT_NON_TEXT_HH # include <mln/core/image/image2d.hh> # include <mln/core/alias/neighb2d.hh> @@ -58,6 +58,10 @@ # include <mln/clustering/kmean_rgb.hh> # include <mln/fun/v2v/rgb8_to_rgbn.hh> +# include <mln/util/timer.hh> + +# include <mln/io/pbm/save.hh> + namespace scribo { @@ -72,7 +76,7 @@ namespace scribo template <typename L, typename I> component_set<L> - elements(const document<L>& doc, const Image& input); + non_text(const document<L>& doc, const Image& input); # ifndef MLN_INCLUDE_ONLY @@ -107,9 +111,9 @@ namespace scribo template <typename L, typename I> component_set<L> - elements(const document<L>& doc, const Image& input_) + non_text(const document<L>& doc, const Image& input_) { - trace::entering("scribo::primitive::extract::elements"); + trace::entering("scribo::primitive::extract::non_text"); const I& input = exact(input_); mln_precondition(doc.is_valid()); @@ -135,9 +139,16 @@ namespace scribo image2d<t_rgb5> img_rgb5 = mln::data::transform(doc.image(), t_rgb8_to_rgb5()); + // DEBUG + io::pbm::save(content, "text_area.pbm"); + + mln::util::timer t; + t.start(); img_lbl8 = mln::clustering::kmean_rgb<double,5>((img_rgb5 | pw::value(content)), 3, 10, 10).unmorph_(); data::fill((img_lbl8 | !pw::value(content)).rw(), 0u); + t.stop(); + std::cout << t << std::endl; mln::util::array<unsigned> card = mln::labeling::compute(accu::math::count<value::label_8>(), @@ -162,15 +173,25 @@ namespace scribo std::cout << "Removing small elements" << std::endl; { - image2d<bool> elts; - initialize(elts, img_lbl8); - data::fill(elts, false); - data::fill((elts | (pw::value(img_lbl8) != pw::cst(0))).rw(), true); + image2d<bool> elts; + initialize(elts, img_lbl8); + data::fill(elts, false); + data::fill((elts | (pw::value(img_lbl8) != pw::cst(0))).rw(), true); + + // DEBUG + data::fill((elts | (pw::value(doc.line_seps()) != pw::cst(0))).rw(), + false); + // END OF DEBUG + - scribo::def::lbl_type nlabels; - elts = filter::components_small(elts, c8(), nlabels, 40); + scribo::def::lbl_type nlabels; + elts = filter::components_small(elts, c8(), nlabels, 40); - output = primitive::extract::components(elts, c8(), nlabels); + // DEBUG + io::pbm::save(elts, "elements.pbm"); + // END OF DEBUG + + output = primitive::extract::components(elts, c8(), nlabels); } @@ -207,20 +228,7 @@ namespace scribo } } -// mln::io::pbm::save(merged_elts, "merged_elts.pbm"); - -// mln::util::array<image2d<value::rgb8> > elt_ima; -// unsigned i = 0; -// for_all_comps(c, elt_comp) -// if (elt_comp(c).is_valid()) -// { -// elt_ima.append(preprocessing::crop(doc.image(), elt_comp(c).bbox())); -// mln::io::ppm::save(elt_ima(i), mln::debug::filename("elt.ppm", i)); -// ++i; -// } - - - trace::exiting("scribo::primitive::extract::elements"); + trace::exiting("scribo::primitive::extract::non_text"); return output; } @@ -233,4 +241,4 @@ namespace scribo } // end of namespace scribo -#endif // ! SCRIBO_PRIMITIVE_EXTRACT_ELEMENTS_HH +#endif // ! SCRIBO_PRIMITIVE_EXTRACT_NON_TEXT_HH diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index b8aa56d..ed691e8 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -463,7 +463,7 @@ namespace scribo // Extract other Elements on_new_progress_label("Extracting Elements"); component_set<L> - elements = scribo::primitive::extract::elements(doc, original_image); + elements = scribo::primitive::extract::non_text(doc, original_image); on_progress(); diff --git a/scribo/src/primitive/extract/Makefile.am b/scribo/src/primitive/extract/Makefile.am index 2c8188e..a46cd68 100644 --- a/scribo/src/primitive/extract/Makefile.am +++ b/scribo/src/primitive/extract/Makefile.am @@ -1,5 +1,5 @@ -# Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -# (LRDE). +# Copyright (C) 2009, 2010, 2011 EPITA Research and Development +# Laboratory (LRDE). # # This file is part of Olena. # @@ -18,17 +18,17 @@ include $(top_srcdir)/scribo/scribo.mk -noinst_PROGRAMS = \ - extract_discontinued_lines \ - extract_discontinued_vlines \ - extract_discontinued_hlines \ - extract_thick_vlines \ - extract_thick_hlines \ +noinst_PROGRAMS = \ + discontinued_lines \ + discontinued_vlines \ + discontinued_hlines \ + thick_vlines \ + thick_hlines \ lines_pattern -extract_discontinued_lines_SOURCES = extract_discontinued_lines.cc -extract_discontinued_vlines_SOURCES = extract_discontinued_vlines.cc -extract_discontinued_hlines_SOURCES = extract_discontinued_hlines.cc -extract_thick_vlines_SOURCES = extract_thick_vlines.cc -extract_thick_hlines_SOURCES = extract_thick_hlines.cc +discontinued_lines_SOURCES = discontinued_lines.cc +discontinued_vlines_SOURCES = discontinued_vlines.cc +discontinued_hlines_SOURCES = discontinued_hlines.cc +thick_vlines_SOURCES = thick_vlines.cc +thick_hlines_SOURCES = thick_hlines.cc lines_pattern_SOURCES = lines_pattern.cc diff --git a/scribo/src/primitive/extract/extract_discontinued_hlines.cc b/scribo/src/primitive/extract/discontinued_hlines.cc similarity index 100% rename from scribo/src/primitive/extract/extract_discontinued_hlines.cc rename to scribo/src/primitive/extract/discontinued_hlines.cc diff --git a/scribo/src/primitive/extract/extract_discontinued_lines.cc b/scribo/src/primitive/extract/discontinued_lines.cc similarity index 100% rename from scribo/src/primitive/extract/extract_discontinued_lines.cc rename to scribo/src/primitive/extract/discontinued_lines.cc diff --git a/scribo/src/primitive/extract/extract_discontinued_vlines.cc b/scribo/src/primitive/extract/discontinued_vlines.cc similarity index 100% rename from scribo/src/primitive/extract/extract_discontinued_vlines.cc rename to scribo/src/primitive/extract/discontinued_vlines.cc diff --git a/scribo/src/primitive/extract/extract_thick_hlines.cc b/scribo/src/primitive/extract/thick_hlines.cc similarity index 100% rename from scribo/src/primitive/extract/extract_thick_hlines.cc rename to scribo/src/primitive/extract/thick_hlines.cc diff --git a/scribo/src/primitive/extract/extract_thick_vlines.cc b/scribo/src/primitive/extract/thick_vlines.cc similarity index 100% rename from scribo/src/primitive/extract/extract_thick_vlines.cc rename to scribo/src/primitive/extract/thick_vlines.cc -- 1.5.6.5

13 years, 10 months

last-svn-commit-770-g2a17f6d Various changes in scribo core structures.

by Guillaume Lazzara

* scribo/scribo/core/component_info.hh: Remove useless line_id. * scribo/scribo/core/line_set.hh, * scribo/scribo/core/component_set.hh: Add new constructors. * scribo/scribo/core/document.hh: Store line separators. * scribo/scribo/core/line_info.hh: Share attributes and add new constructors. (is_textline): New method. * scribo/scribo/core/line_links.hh: Do not allocate useless data. * scribo/scribo/core/paragraph_info.hh (operator<<): New. * scribo/scribo/core/paragraph_set.hh: Share attributes and add new constructors. --- scribo/ChangeLog | 22 ++ scribo/scribo/core/component_info.hh | 17 +-- scribo/scribo/core/component_set.hh | 65 +++++-- scribo/scribo/core/document.hh | 48 +++++- scribo/scribo/core/line_info.hh | 367 ++++++++++++++++++---------------- scribo/scribo/core/line_links.hh | 2 +- scribo/scribo/core/line_set.hh | 29 +++- scribo/scribo/core/paragraph_info.hh | 10 + scribo/scribo/core/paragraph_set.hh | 84 +++++++- 9 files changed, 425 insertions(+), 219 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index e08f2fd..ab028c3 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,27 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Various changes in scribo core structures. + + * scribo/scribo/core/component_info.hh: Remove useless line_id. + + * scribo/scribo/core/line_set.hh, + * scribo/scribo/core/component_set.hh: Add new constructors. + + * scribo/scribo/core/document.hh: Store line separators. + + * scribo/scribo/core/line_info.hh: Share attributes and add new + constructors. + (is_textline): New method. + + * scribo/scribo/core/line_links.hh: Do not allocate useless data. + + * scribo/scribo/core/paragraph_info.hh (operator<<): New. + + * scribo/scribo/core/paragraph_set.hh: Share attributes and add new + constructors. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Rename files in Scribo. * scribo/primitive/extract/elements.hh, diff --git a/scribo/scribo/core/component_info.hh b/scribo/scribo/core/component_info.hh index 4ed6db7..1b03318 100644 --- a/scribo/scribo/core/component_info.hh +++ b/scribo/scribo/core/component_info.hh @@ -47,7 +47,6 @@ namespace scribo class component_info { typedef mln::util::object_id<scribo::ComponentId, unsigned> component_id_t; - typedef mln::util::object_id<scribo::LineId, unsigned> line_id_t; public: component_info(); @@ -69,9 +68,6 @@ namespace scribo component::Type type() const; void update_type(component::Type type); - // The line it is rattached to. 0 means an invalid line. - line_id_t line_id() const; - bool is_valid() const; private: @@ -82,8 +78,6 @@ namespace scribo component::Tag tag_; component::Type type_; - - line_id_t line_id_; }; @@ -109,7 +103,7 @@ namespace scribo const mln::point2d& mass_center, unsigned card) : id_(id), bbox_(bbox), mass_center_(mass_center), card_(card), - tag_(component::None), type_(component::Undefined), line_id_(0) + tag_(component::None), type_(component::Undefined) { } @@ -179,14 +173,6 @@ namespace scribo inline - component_info::line_id_t - component_info::line_id() const - { - return line_id_; - } - - - inline bool component_info::is_valid() const { @@ -204,7 +190,6 @@ namespace scribo << ", mass_center=" << info.mass_center() << ", card=" << info.card() << ", tag=" << info.tag() - << ", line_id=" << info.line_id() << ")" << std::endl; } diff --git a/scribo/scribo/core/component_set.hh b/scribo/scribo/core/component_set.hh index 103826f..7ddcf16 100644 --- a/scribo/scribo/core/component_set.hh +++ b/scribo/scribo/core/component_set.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -55,6 +55,8 @@ # include <mln/core/routine/duplicate.hh> +# include <mln/value/next.hh> + # include <scribo/core/macros.hh> # include <scribo/core/component_info.hh> @@ -95,6 +97,8 @@ namespace scribo void fill_infos(const mln::util::array<pair_data_t>& attribs); + // Useful while constructing incrementaly (XML loading). + void soft_init(const mln_value(L) ncomps); L ima_; mln_value(L) ncomps_; @@ -128,6 +132,9 @@ namespace scribo /// Constructor without argument. component_set(); + // Constructor from internal data. + component_set(const mln::util::tracked_ptr<data_t>& data); + /// Constructor from an image \p ima and the number of labels \p ncomps. component_set(const L& ima, const mln_value(L)& ncomps); @@ -348,6 +355,19 @@ namespace scribo } + template <typename L> + inline + void + component_set_data<L>::soft_init(const mln_value(L) ncomps) + { + mln_precondition(infos_.nelements() == 0); + + ncomps_ = ncomps; + infos_.reserve(ncomps_); + infos_.append(component_info()); // Component 0, i.e. the background. + } + + } // end of namespace mln::internal @@ -361,6 +381,13 @@ namespace scribo template <typename L> inline + component_set<L>::component_set(const mln::util::tracked_ptr<data_t>& data) + { + data_ = data; + } + + template <typename L> + inline component_set<L>::component_set(const L& ima, const mln_value(L)& ncomps) { data_ = new internal::component_set_data<L>(ima, ncomps); @@ -400,7 +427,7 @@ namespace scribo const component_info& component_set<L>::info(const mln_value(L)& id) const { - return this->data_->infos_[id]; + return data_->infos_[id]; } template <typename L> @@ -408,7 +435,7 @@ namespace scribo component_info& component_set<L>::info(const mln_value(L)& id) { - return this->data_->infos_[id]; + return data_->infos_[id]; } // template <typename L> @@ -416,7 +443,7 @@ namespace scribo // const component_info& // component_set<L>::operator()(const mln_value(L)& id) const // { -// return this->data_->infos_[id]; +// return data_->infos_[id]; // } // template <typename L> @@ -424,7 +451,7 @@ namespace scribo // component_info& // component_set<L>::operator()(const mln_value(L)& id) // { -// return this->data_->infos_[id]; +// return data_->infos_[id]; // } template <typename L> @@ -432,7 +459,7 @@ namespace scribo const component_info& component_set<L>::operator()(const component_id_t& id) const { - return this->data_->infos_[id]; + return data_->infos_[id]; } template <typename L> @@ -440,7 +467,7 @@ namespace scribo component_info& component_set<L>::operator()(const component_id_t& id) { - return this->data_->infos_[id]; + return data_->infos_[id]; } @@ -476,7 +503,7 @@ namespace scribo const L& component_set<L>::labeled_image() const { - return this->data_->ima_; + return data_->ima_; } @@ -485,7 +512,7 @@ namespace scribo bool component_set<L>::is_valid() const { - return this->data_->ima_.is_valid(); + return data_ != 0 && data_->ima_.is_valid(); } @@ -503,7 +530,7 @@ namespace scribo L& component_set<L>::labeled_image_() { - return this->data_->ima_; + return data_->ima_; } @@ -512,7 +539,7 @@ namespace scribo mln_concrete(L) component_set<L>::valid_comps_image_() const { - mln::util::array<bool> f(mln::value::next(this->data_->ncomps_)); + mln::util::array<bool> f(mln::value::next(data_->ncomps_)); f(0) = true; for_all_comps(c, (*this)) @@ -520,8 +547,8 @@ namespace scribo mln_value(L) new_ncomps; mln_concrete(L) - output = mln::labeling::relabel(this->data_->ima_, - this->data_->ncomps_, + output = mln::labeling::relabel(data_->ima_, + data_->ncomps_, new_ncomps, f); @@ -534,7 +561,7 @@ namespace scribo bool component_set<L>::has_separators() const { - return this->data_->separators_.is_valid(); + return data_->separators_.is_valid(); } @@ -544,9 +571,9 @@ namespace scribo component_set<L>::add_separators(const mln_ch_value(L, bool)& ima) { if (! has_separators()) - this->data_->separators_ = ima; + data_->separators_ = ima; else - mln::logical::or_inplace(this->data_->separators_, ima); + mln::logical::or_inplace(data_->separators_, ima); } @@ -555,7 +582,7 @@ namespace scribo const mln_ch_value(L, bool)& component_set<L>::separators() const { - return this->data_->separators_; + return data_->separators_; } @@ -564,7 +591,7 @@ namespace scribo void component_set<L>::clear_separators() { - this->data_->separators_.destroy(); + data_->separators_.destroy(); } diff --git a/scribo/scribo/core/document.hh b/scribo/scribo/core/document.hh index f112410..e5ac825 100644 --- a/scribo/scribo/core/document.hh +++ b/scribo/scribo/core/document.hh @@ -91,16 +91,24 @@ namespace scribo const component_set<L>& whitespace_seps_comps() const; void set_whitespace_separators(const image2d<bool>& whitespace_seps); + bool has_line_seps() const; + const mln::image2d<bool>& line_seps() const; + const component_set<L>& line_seps_comps() const; + void set_line_separators(const image2d<bool>& line_seps); + + private: const char *filename_; mln::image2d<mln::value::rgb8> image_; - line_set<L> lines_; paragraph_set<L> parset_; component_set<L> elements_; mln::image2d<bool> whitespace_seps_; component_set<L> whitespace_seps_comps_; + + mln::image2d<bool> line_seps_; + component_set<L> line_seps_comps_; }; @@ -201,7 +209,6 @@ namespace scribo return parset_.lines(); } - template <typename L> const paragraph_set<L>& document<L>::paragraphs() const @@ -294,6 +301,43 @@ namespace scribo } + template <typename L> + bool + document<L>::has_line_seps() const + { + return line_seps_.is_valid(); + } + + + template <typename L> + const mln::image2d<bool>& + document<L>::line_seps() const + { + return line_seps_; + } + + + template <typename L> + const component_set<L>& + document<L>::line_seps_comps() const + { + return line_seps_comps_; + } + + + template <typename L> + void + document<L>::set_line_separators(const image2d<bool>& line_seps) + { + line_seps_ = line_seps; + + mln_value(L) ncomps; + line_seps_comps_ = primitive::extract::components(line_seps, + mln::c8(), ncomps); + } + + + # endif // ! MLN_INCLUDE_ONLY diff --git a/scribo/scribo/core/line_info.hh b/scribo/scribo/core/line_info.hh index 54a5094..c82160a 100644 --- a/scribo/scribo/core/line_info.hh +++ b/scribo/scribo/core/line_info.hh @@ -58,12 +58,76 @@ namespace scribo // Forward declarations. template <typename L> class line_set; + template <typename L> class line_info; typedef mln::util::object_id<scribo::LineId, unsigned> line_id_t; + + namespace internal + { + /// Data structure for \c scribo::line_info. + template <typename L> + struct line_info_data + { + line_info_data(); + line_info_data(const line_set<L>& holder, + const mln::util::array<component_id_t>& comps); + + + bool hidden_; + line::Tag tag_; + mln::box2d bbox_; + mln::box2d ebbox_; + mln::util::array<component_id_t> components_; + + // Values relative to the line bbox. + int baseline_; + int meanline_; + + // Values relative to the baseline. + unsigned x_height_; + int d_height_; + int a_height_; + + // Character related stats. + unsigned char_space_; + unsigned char_width_; + + // Words related stats. + unsigned word_space_; + + // Reading direction + line::ReadingDirection reading_direction_; + + // Line type + line::Type type_; + + // Is this line in reverse video? + bool reverse_video_; + + // Text orientation + float orientation_; + + // Text reading orientation + float reading_orientation_; + + bool indented_; + + std::string text_; + + // Line set holding this element. + line_set<L> holder_; + + }; + + } // end of namespace scribo::internal + + + template <typename L> class line_info { + typedef internal::line_info_data<L> data_t; typedef mln::util::object_id<scribo::ComponentId, unsigned> component_id_t; typedef mln::util::object_id<scribo::LineId, unsigned> line_id_t; @@ -74,6 +138,8 @@ namespace scribo line_info(); + line_info(const line_id_t& id, data_t* data); + line_info(const line_set<L>& holder, const line_id_t& id, const mln::util::array<component_id_t>& comps); @@ -134,9 +200,10 @@ namespace scribo const std::string& text() const; void update_text(const std::string& str); - bool is_valid() const; + bool is_textline() const; + /// Hidden status. /// /// When a line is hidden, it should not be used in routines @@ -180,6 +247,9 @@ namespace scribo /// Returns the delta used to compute the extended bbox. int delta_of_line() const; + /// Update the extended bbox. + void update_ebbox(); + private: // Members void copy_data(const line_info<L>& other); @@ -189,81 +259,59 @@ namespace scribo /// Update bbox and ebbox_ attributes. void update_bbox_and_ebox(line_info<L>& other); - /// Update the extended bbox. - void update_ebbox(); - mln::box2d merged_ebbox(const scribo::line_info<L>& info_l, const scribo::line_info<L>& info); void update_components_type(component::Type type); private: // Attributes - // WARNING: NEVER FORGET TO UPDATE COPY CONSTRUCTOR REDEFINITION!!!! - line_id_t id_; - bool hidden_; - line::Tag tag_; - mln::box2d bbox_; - mln::box2d ebbox_; - mln::util::array<component_id_t> components_; - - // Values relative to the line bbox. - int baseline_; - int meanline_; - - // Values relative to the baseline. - unsigned x_height_; - int d_height_; - int a_height_; - - // WARNING: NEVER FORGET TO UPDATE COPY CONSTRUCTOR REDEFINITION!!!! - - // Character related stats. - unsigned char_space_; - unsigned char_width_; - - // Words related stats. - unsigned word_space_; - - // Reading direction - line::ReadingDirection reading_direction_; - - // Line type - line::Type type_; + mln::util::tracked_ptr<data_t> data_; + }; - // Is this line in reverse video? - bool reverse_video_; - // Text orientation - float orientation_; + template <typename L> + std::ostream& + operator<<(std::ostream& ostr, const line_info<L>& info); - // Text reading orientation - float reading_orientation_; - bool indented_; +# ifndef MLN_INCLUDE_ONLY - std::string text_; + namespace internal + { - // Line set holding this element. - line_set<L> holder_; + template <typename L> + line_info_data<L>::line_info_data() + { + hidden_ = false; + } - // WARNING: NEVER FORGET TO UPDATE COPY CONSTRUCTOR REDEFINITION!!!! - }; + template <typename L> + line_info_data<L>::line_info_data(const line_set<L>& holder, + const mln::util::array<component_id_t>& comps) + : hidden_(false), tag_(line::None), components_(comps), + type_(line::Undefined), holder_(holder) + { + // FIXME: set valid information for these attributes in + // force_stats_update. + word_space_ = 0; + reading_direction_ = line::LeftToRight; + reverse_video_ = false; - template <typename L> - std::ostream& - operator<<(std::ostream& ostr, const line_info<L>& info); + orientation_ = 0.; + reading_orientation_ = 0.; + indented_ = false; + } -# ifndef MLN_INCLUDE_ONLY + } // end of namespace scribo::internal template <typename L> line_info<L>::line_info() - : id_(0), hidden_(false) + : id_(0) { - } template <typename L> @@ -275,47 +323,24 @@ namespace scribo if (! is_valid()) id_ = other.id(); - hidden_ = other.hidden_; - - tag_ = other.tag(); - bbox_ = other.bbox(); - ebbox_ = other.ebbox(); - components_ = other.components(); - - baseline_ = other.baseline(); - meanline_ = other.meanline(); - - x_height_ = other.x_height(); - d_height_ = other.d_height(); - a_height_ = other.a_height(); - - char_space_ = other.char_space(); - char_width_ = other.char_width(); - - word_space_ = other.word_space(); - - reading_direction_ = other.reading_direction(); - - type_ = other.type(); - - reverse_video_ = other.reverse_video(); - - orientation_ = other.orientation(); - - reading_orientation_ = other.reading_orientation(); - - indented_ = other.indented(); + data_ = other.data_; + } - text_ = other.text(); - holder_ = other.holder(); + template <typename L> + inline + line_info<L>::line_info(const line_id_t& id, data_t *data) + : id_(id), data_(data) + { } + template <typename L> inline line_info<L>::line_info(const line_info<L>& other) - : id_(0), hidden_(false) + : id_(0) { + //data_->hidden_ = false; copy_data(other); } @@ -357,22 +382,10 @@ namespace scribo line_info<L>::line_info(const line_set<L>& holder, const line_id_t& id, const mln::util::array<component_id_t>& comps) - : id_(id), hidden_(false), tag_(line::None), components_(comps), - type_(line::Undefined), holder_(holder) + : id_(id) { - + data_ = new data_t(holder, comps); force_stats_update(); - - // FIXME: set valid information for these attributes in - // force_stats_update. - word_space_ = 0; - reading_direction_ = line::LeftToRight; - reverse_video_ = false; - - orientation_ = 0.; - reading_orientation_ = 0.; - - indented_ = false; } @@ -397,7 +410,7 @@ namespace scribo line::Tag line_info<L>::tag() const { - return tag_; + return data_->tag_; } @@ -405,7 +418,7 @@ namespace scribo void line_info<L>::update_tag(line::Tag tag) { - tag_ = tag; + data_->tag_ = tag; } @@ -413,14 +426,14 @@ namespace scribo const mln::box2d& line_info<L>::bbox() const { - return bbox_; + return data_->bbox_; } template <typename L> const mln::box2d& line_info<L>::ebbox() const { - return ebbox_; + return data_->ebbox_; } @@ -428,14 +441,14 @@ namespace scribo const mln::util::array<typename line_info<L>::component_id_t>& line_info<L>::components() const { - return components_; + return data_->components_; } template <typename L> unsigned line_info<L>::card() const { - return components_.size(); + return data_->components_.size(); } @@ -443,7 +456,7 @@ namespace scribo int line_info<L>::baseline() const { - return baseline_; + return data_->baseline_; } @@ -451,7 +464,7 @@ namespace scribo int line_info<L>::meanline() const { - return meanline_; + return data_->meanline_; } @@ -459,7 +472,7 @@ namespace scribo int line_info<L>::ascent() const { - return baseline_ - a_height() + 1; + return data_->baseline_ - a_height() + 1; } @@ -467,7 +480,7 @@ namespace scribo int line_info<L>::descent() const { - return baseline_ - d_height(); + return data_->baseline_ - d_height(); } @@ -475,7 +488,7 @@ namespace scribo unsigned line_info<L>::x_height() const { - return x_height_; + return data_->x_height_; } @@ -483,7 +496,7 @@ namespace scribo int line_info<L>::d_height() const { - return d_height_; + return data_->d_height_; } @@ -491,7 +504,7 @@ namespace scribo int line_info<L>::a_height() const { - return a_height_; + return data_->a_height_; } @@ -499,7 +512,7 @@ namespace scribo unsigned line_info<L>::char_space() const { - return char_space_; + return data_->char_space_; } @@ -507,7 +520,7 @@ namespace scribo unsigned line_info<L>::char_width() const { - return char_width_; + return data_->char_width_; } @@ -515,7 +528,7 @@ namespace scribo unsigned line_info<L>::word_space() const { - return word_space_; + return data_->word_space_; } @@ -523,14 +536,14 @@ namespace scribo line::ReadingDirection line_info<L>::reading_direction() const { - return reading_direction_; + return data_->reading_direction_; } template <typename L> line::Type line_info<L>::type() const { - return type_; + return data_->type_; } @@ -538,10 +551,10 @@ namespace scribo void line_info<L>::update_components_type(component::Type type) { - for_all_elements(i, components_) + for_all_elements(i, data_->components_) { - unsigned c = components_[i]; - holder_.components_()(c).update_type(type); + unsigned c = data_->components_[i]; + data_->holder_.components_()(c).update_type(type); } } @@ -550,7 +563,7 @@ namespace scribo void line_info<L>::update_type(line::Type type) { - type_ = type; + data_->type_ = type; // Some line types may involve updating components type as well. if (type == line::Punctuation) @@ -564,7 +577,7 @@ namespace scribo bool line_info<L>::reverse_video() const { - return reverse_video_; + return data_->reverse_video_; } @@ -572,7 +585,7 @@ namespace scribo float line_info<L>::orientation() const { - return orientation_; + return data_->orientation_; } @@ -580,7 +593,7 @@ namespace scribo float line_info<L>::reading_orientation() const { - return reading_orientation_; + return data_->reading_orientation_; } @@ -588,21 +601,21 @@ namespace scribo bool line_info<L>::indented() const { - return indented_; + return data_->indented_; } template <typename L> bool line_info<L>::has_text() const { - return !text_.empty(); + return !data_->text_.empty(); } template <typename L> const std::string& line_info<L>::text() const { - return text_; + return data_->text_; } @@ -610,7 +623,7 @@ namespace scribo void line_info<L>::update_text(const std::string& str) { - text_ = str; + data_->text_ = str; } @@ -624,9 +637,19 @@ namespace scribo template <typename L> bool + line_info<L>::is_textline() const + { + return is_valid() + && !is_hidden() + && type() == line::Text; + } + + + template <typename L> + bool line_info<L>::is_hidden() const { - return hidden_; + return data_->hidden_; } @@ -634,7 +657,7 @@ namespace scribo void line_info<L>::set_hidden(bool b) { - hidden_ = b; + data_->hidden_ = b; } @@ -665,8 +688,8 @@ namespace scribo void line_info<L>::update_ebbox() { - int A = a_height_ - x_height_; - int D = - d_height_; + int A = data_->a_height_ - data_->x_height_; + int D = - data_->d_height_; if (A <= 2 && D > 2) A = D; if (D <= 2 && A > 2) @@ -674,10 +697,12 @@ namespace scribo int delta = delta_of_line(); - ebbox_ = mln::make::box2d(meanline_ - A, bbox().pmin().col() - delta, - baseline_ + D, bbox().pmax().col() + delta); + data_->ebbox_ = mln::make::box2d(data_->meanline_ - A, + bbox().pmin().col() - delta, + data_->baseline_ + D, + bbox().pmax().col() + delta); - ebbox_.crop_wrt(holder_.components().labeled_image().domain()); + data_->ebbox_.crop_wrt(data_->holder_.components().labeled_image().domain()); } @@ -729,20 +754,20 @@ namespace scribo // Adjusting ebboxes with the highest delta and merging ebboxes. int d_delta = other.delta_of_line() - this->delta_of_line(); if (d_delta < 0) // other.delta_of_line() < this->delta_of_line() - ebbox_.merge(enlarge(other.ebbox(), - d_delta)); + data_->ebbox_.merge(enlarge(other.ebbox(), - d_delta)); else { - mln::box2d b = ebbox_; - ebbox_ = other.bbox(); - ebbox_.merge(enlarge(b, d_delta)); + mln::box2d b = data_->ebbox_; + data_->ebbox_ = other.bbox(); + data_->ebbox_.merge(enlarge(b, d_delta)); } - ebbox_.crop_wrt(holder_.components().labeled_image().domain()); + data_->ebbox_.crop_wrt(data_->holder_.components().labeled_image().domain()); } else // /other/ IS NOT a text line. { - ebbox_.merge(other.ebbox()); - ebbox_.merge(merged_ebbox(*this, other)); + data_->ebbox_.merge(other.ebbox()); + data_->ebbox_.merge(merged_ebbox(*this, other)); } } else // /this/ is NOT a text line @@ -755,15 +780,15 @@ namespace scribo } update_type(line::Text); - ebbox_.merge(other.ebbox()); - ebbox_.merge(merged_ebbox(other, *this)); + data_->ebbox_.merge(other.ebbox()); + data_->ebbox_.merge(merged_ebbox(other, *this)); } // Merging bboxes. - bbox_.merge(other.bbox()); + data_->bbox_.merge(other.bbox()); // Make sure the ebbox is included in the image domain. - ebbox_.crop_wrt(holder_.components().labeled_image().domain()); + data_->ebbox_.crop_wrt(data_->holder_.components().labeled_image().domain()); } @@ -771,14 +796,14 @@ namespace scribo void line_info<L>::fast_merge(line_info<L>& other, bool hide) { - tag_ = line::Needs_Precise_Stats_Update; + data_->tag_ = line::Needs_Precise_Stats_Update; other.update_tag(line::Merged); other.set_hidden(hide); // Update bbox and ebbox update_bbox_and_ebox(other); - components_.append(other.components()); + data_->components_.append(other.components()); } @@ -795,7 +820,7 @@ namespace scribo line_info<L>::force_stats_update() { typedef mln_site(L) P; - const component_set<L>& comp_set = holder_.components(); + const component_set<L>& comp_set = data_->holder_.components(); // Init. typedef mln::value::int_u<12> median_data_t; @@ -814,21 +839,21 @@ namespace scribo // Workaround to avoid overflow with int_u<12> in median accumulators. // // FIXME: not optimal... - for_all_elements(i, components_) + for_all_elements(i, data_->components_) { - unsigned c = components_(i); + unsigned c = data_->components_(i); // Ignore punctuation for stats computation but not for bbox // computation. - if (holder_.components()(c).type() == component::Punctuation) + if (data_->holder_.components()(c).type() == component::Punctuation) continue; ref_line = mln::math::min(comp_set(c).bbox().pmin().row(), ref_line); } - for_all_elements(i, components_) + for_all_elements(i, data_->components_) { - unsigned c = components_(i); + unsigned c = data_->components_(i); const mln::box2d& bb = comp_set(c).bbox(); @@ -837,7 +862,7 @@ namespace scribo // Ignore punctuation for stats computation but not for bbox // computation. - if (holder_.components()(c).type() == component::Punctuation) + if (data_->holder_.components()(c).type() == component::Punctuation) continue; @@ -853,11 +878,11 @@ namespace scribo // (right link) (left link) // Space between characters. - if (holder_.links()(c) != c) + if (data_->holder_.links()(c) != c) { int space = bb.pmin().col() - - comp_set(holder_.links()(c)).bbox().pmax().col() - 1; + - comp_set(data_->holder_.links()(c)).bbox().pmax().col() - 1; // -- Ignore overlapped characters. if (space > 0) @@ -884,31 +909,31 @@ namespace scribo // Finalization { - tag_ = line::None; - bbox_ = bbox.to_result(); + data_->tag_ = line::None; + data_->bbox_ = bbox.to_result(); // Char space if (char_space.card() < 2) - char_space_ = 0; + data_->char_space_ = 0; else - char_space_ = char_space.to_result(); + data_->char_space_ = char_space.to_result(); // Char width if (card() == 2) - char_width_ = (comp_set(components_[0]).bbox().width() - + comp_set(components_[1]).bbox().width()) / 2; + data_->char_width_ = (comp_set(data_->components_[0]).bbox().width() + + comp_set(data_->components_[1]).bbox().width()) / 2; else - char_width_ = char_width.to_result(); + data_->char_width_ = char_width.to_result(); mln::def::coord absolute_baseline_r = baseline.to_result() + ref_line, absolute_meanline_r = meanline.to_result() + ref_line; - baseline_ = absolute_baseline_r; - meanline_ = absolute_meanline_r; - x_height_ = baseline_ - meanline_ + 1; - d_height_ = baseline_ - bbox.to_result().pmax().row(); - a_height_ = baseline_ - bbox.to_result().pmin().row() + 1; + data_->baseline_ = absolute_baseline_r; + data_->meanline_ = absolute_meanline_r; + data_->x_height_ = data_->baseline_ - data_->meanline_ + 1; + data_->d_height_ = data_->baseline_ - bbox.to_result().pmax().row(); + data_->a_height_ = data_->baseline_ - bbox.to_result().pmin().row() + 1; //FIXME // @@ -929,7 +954,7 @@ namespace scribo const line_set<L>& line_info<L>::holder() const { - return holder_; + return data_->holder_; } diff --git a/scribo/scribo/core/line_links.hh b/scribo/scribo/core/line_links.hh index 85c45e8..de62158 100644 --- a/scribo/scribo/core/line_links.hh +++ b/scribo/scribo/core/line_links.hh @@ -139,8 +139,8 @@ namespace scribo template <typename L> line_links<L>::line_links() + : data_(0) { - data_ = new data_t(); } diff --git a/scribo/scribo/core/line_set.hh b/scribo/scribo/core/line_set.hh index 29795b2..bfa9240 100644 --- a/scribo/scribo/core/line_set.hh +++ b/scribo/scribo/core/line_set.hh @@ -99,6 +99,10 @@ namespace scribo /// Constructor from object groups. line_set(const object_groups<L>& groups); + + /// Constructor useful for delayed construction (loading from file). + line_set(const object_groups<L>& groups, + const mln::util::array<line_info<L> >& line_data); /// @} /// Compute line stats and fill the underlying information. @@ -155,8 +159,14 @@ namespace scribo /// @} + /// Return false if it is not initialized (built with the default + /// constructor). bool is_valid() const; + + void update_line_data_(const mln::util::array<line_info<L> >& line_data); + + private: /// Duplicate the underlying image and create a new line_set. void init_(const line_set<L>& model); @@ -233,6 +243,15 @@ namespace scribo template <typename L> + inline + line_set<L>::line_set(const object_groups<L>& groups, + const mln::util::array<line_info<L> >& line_data) + { + data_ = new internal::line_set_data<L>(line_data, groups); + } + + + template <typename L> void line_set<L>::compute_lines(const object_groups<L>& groups) { @@ -396,7 +415,15 @@ namespace scribo bool line_set<L>::is_valid() const { - return data_->links_.is_valid() && data_->groups_.is_valid(); + return data_ && data_->groups_.is_valid(); + } + + template <typename L> + inline + void + line_set<L>::update_line_data_(const mln::util::array<line_info<L> >& line_data) + { + data_->infos_ = line_data; } template <typename L> diff --git a/scribo/scribo/core/paragraph_info.hh b/scribo/scribo/core/paragraph_info.hh index a8c623a..17f847f 100644 --- a/scribo/scribo/core/paragraph_info.hh +++ b/scribo/scribo/core/paragraph_info.hh @@ -113,6 +113,16 @@ namespace scribo } + template <typename L> + std::ostream& + operator<<(std::ostream& ostr, const paragraph_info<L>& info) + { + return ostr << "paragraph_info(" + << "line_ids=" << info.line_ids() + << ", bbox=" << info.bbox() + << ")" << std::endl; + } + # endif // ! MLN_INCLUDE_ONLY } // end of namespace scribo diff --git a/scribo/scribo/core/paragraph_set.hh b/scribo/scribo/core/paragraph_set.hh index 355eaa9..6597189 100644 --- a/scribo/scribo/core/paragraph_set.hh +++ b/scribo/scribo/core/paragraph_set.hh @@ -36,6 +36,25 @@ namespace scribo { + namespace internal + { + + /// Data structure for \c scribo::paragraph_set. + template <typename L> + struct paragraph_set_data + { + paragraph_set_data(); + paragraph_set_data(const line_links<L>& llines, unsigned npars); + + mln::util::array<paragraph_info<L> > pars_; + line_set<L> lines_; + line_links<L> links_; + }; + + } // end of namespace scribo::internal + + + /*! \brief Paragraph container. Paragraph ids start from 1. @@ -46,6 +65,7 @@ namespace scribo { public: paragraph_set(); + paragraph_set(internal::paragraph_set_data<L>* data); paragraph_set(const line_links<L>& llinks, unsigned npars); unsigned nelements() const; @@ -57,9 +77,10 @@ namespace scribo const line_set<L>& lines() const; + const line_links<L>& links() const; + private: - mln::util::array<paragraph_info<L> > pars_; - line_set<L> lines_; + mln::util::tracked_ptr< internal::paragraph_set_data<L> > data_; }; @@ -77,37 +98,72 @@ namespace scribo # ifndef MLN_INCLUDE_ONLY + // paragraph_set_data<L> > + + namespace internal + { + + // data< paragraph_set<L> > + + + template <typename L> + inline + paragraph_set_data<L>::paragraph_set_data() + { + } + + + template <typename L> + inline + paragraph_set_data<L>::paragraph_set_data(const line_links<L>& llinks, unsigned npars) + : pars_(npars + 1, paragraph_info<L>(llinks)), links_(llinks) + { + lines_ = llinks.lines(); + } + + } // end of namespace mln::internal + + template <typename L> paragraph_set<L>::paragraph_set() + : data_(0) { } template <typename L> + paragraph_set<L>::paragraph_set(internal::paragraph_set_data<L>* data) + { + data_ = data; + } + + template <typename L> paragraph_set<L>::paragraph_set(const line_links<L>& llinks, unsigned npars) - : pars_(npars + 1, paragraph_info<L>(llinks)) { - lines_ = llinks.lines(); + data_ = new internal::paragraph_set_data<L>(llinks, npars); } template <typename L> unsigned paragraph_set<L>::nelements() const { - return pars_.nelements() - 1; + mln_precondition(data_ != 0); + return data_->pars_.nelements() - 1; } template <typename L> paragraph_info<L>& paragraph_set<L>::operator()(unsigned i) { - return pars_[i]; + mln_precondition(data_ != 0); + return data_->pars_[i]; } template <typename L> const paragraph_info<L>& paragraph_set<L>::operator()(unsigned i) const { - return pars_[i]; + mln_precondition(data_ != 0); + return data_->pars_[i]; } @@ -115,7 +171,7 @@ namespace scribo bool paragraph_set<L>::is_valid() const { - return !pars_.is_empty(); + return data_ && !data_->pars_.is_empty(); } @@ -123,7 +179,17 @@ namespace scribo const line_set<L>& paragraph_set<L>::lines() const { - return lines_; + mln_precondition(data_ != 0); + return data_->lines_; + } + + + template <typename L> + const line_links<L>& + paragraph_set<L>::links() const + { + mln_precondition(data_ != 0); + return data_->links_; } -- 1.5.6.5

13 years, 10 months

last-svn-commit-771-gf67cec4 Add new conversion routines from string to tag.

by Guillaume Lazzara

* scribo/core/tag/component.hh, * scribo/core/tag/line.hh: Here. --- scribo/ChangeLog | 7 +++++++ scribo/scribo/core/tag/component.hh | 35 +++++++++++++++++++++++++++++++++++ scribo/scribo/core/tag/line.hh | 34 +++++++++++++++++++++++++++++++++- 3 files changed, 75 insertions(+), 1 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index ab028c3..ab498cc 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,12 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Add new conversion routines from string to tag. + + * scribo/core/tag/component.hh, + * scribo/core/tag/line.hh: Here. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Various changes in scribo core structures. * scribo/scribo/core/component_info.hh: Remove useless line_id. diff --git a/scribo/scribo/core/tag/component.hh b/scribo/scribo/core/tag/component.hh index f773932..10b86a6 100644 --- a/scribo/scribo/core/tag/component.hh +++ b/scribo/scribo/core/tag/component.hh @@ -61,6 +61,14 @@ namespace scribo Image }; + + std::ostream& operator<<(std::ostream& ostr, const Tag& tag); + Tag str2tag(const std::string& str); + + std::ostream& operator<<(std::ostream& ostr, const Type& type); + Type str2type(const std::string& str); + + # ifndef MLN_INCLUDE_ONLY @@ -85,6 +93,16 @@ namespace scribo inline + Tag str2tag(const std::string& str) + { + if (str == "Ignored") + return Ignored; + + return None; + } + + + inline std::ostream& operator<<(std::ostream& ostr, const Type& type) { @@ -116,6 +134,23 @@ namespace scribo } + inline + Type str2type(const std::string& str) + { + if (str == "Character") + return Character; + else if (str == "Separator") + return Separator; + else if (str == "Noise") + return Noise; + else if (str == "Punctuation") + return Punctuation; + else if (str == "Image") + return Image; + + return Undefined; + } + # endif // ! MLN_INCLUDE_ONLY diff --git a/scribo/scribo/core/tag/line.hh b/scribo/scribo/core/tag/line.hh index 0444a11..3fb1fdf 100644 --- a/scribo/scribo/core/tag/line.hh +++ b/scribo/scribo/core/tag/line.hh @@ -96,7 +96,7 @@ namespace scribo std::ostream& operator<<(std::ostream& ostr, const Type& type); - + Type str2type(const std::string& str); # ifndef MLN_INCLUDE_ONLY @@ -208,6 +208,38 @@ namespace scribo return ostr << str; } + inline + Type str2type(const std::string& str) + { + if (str == "caption") + return Caption; + else if (str == "credit") + return Credit; + else if (str == "drop-capital") + return DropCapital; + else if (str == "floating") + return Floating; + else if (str == "footer") + return Footer; + else if (str == "header") + return Header; + else if (str == "heading") + return Heading; + else if (str == "page-number") + return PageNumber; + else if (str == "paragraph") + return Paragraph; + + // Values unsupported by the XSD + else if(str == "punctuation") + return Punctuation; + else if (str == "text") + return Text; + + return Undefined; + } + + # endif // ! MLN_INCLUDE_ONLY -- 1.5.6.5

13 years, 10 months

last-svn-commit-772-g4756e17 Improve and cleanup whitespace separator detection.

by Guillaume Lazzara

* scribo/core/tag/anchor.hh: Add new anchors. * scribo/filter/internal/alignment_angle.hh, * scribo/primitive/link/internal/compute_anchor.hh: Support new anchors. * scribo/primitive/extract/separators_nonvisible.hh: Remove dead code and comment debug code. * scribo/toolchain/internal/content_in_doc_functor.hh: Detect horizontal whitespace. --- scribo/ChangeLog | 16 + scribo/scribo/core/tag/anchor.hh | 5 +- scribo/scribo/filter/internal/alignment_angle.hh | 144 +++-- .../primitive/extract/separators_nonvisible.hh | 673 +++++++++----------- .../primitive/link/internal/compute_anchor.hh | 26 +- .../toolchain/internal/content_in_doc_functor.hh | 13 +- 6 files changed, 447 insertions(+), 430 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index ab498cc..d783c5c 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,21 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Improve and cleanup whitespace separator detection. + + * scribo/core/tag/anchor.hh: Add new anchors. + + * scribo/filter/internal/alignment_angle.hh, + * scribo/primitive/link/internal/compute_anchor.hh: Support new + anchors. + + * scribo/primitive/extract/separators_nonvisible.hh: Remove dead + code and comment debug code. + + * scribo/toolchain/internal/content_in_doc_functor.hh: Detect + horizontal whitespace. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Add new conversion routines from string to tag. * scribo/core/tag/component.hh, diff --git a/scribo/scribo/core/tag/anchor.hh b/scribo/scribo/core/tag/anchor.hh index 262a32d..bb6780c 100644 --- a/scribo/scribo/core/tag/anchor.hh +++ b/scribo/scribo/core/tag/anchor.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2009, 2011 EPITA Research and Development Laboratory +// (LRDE) // // This file is part of Olena. // @@ -43,9 +44,11 @@ namespace scribo ActualRight, Right, TopLeft, + TopStrictLeft, TopRight, BottomLeft, BottomRight, + BottomStrictRight, Invalid }; diff --git a/scribo/scribo/filter/internal/alignment_angle.hh b/scribo/scribo/filter/internal/alignment_angle.hh index 5b88012..33855a8 100644 --- a/scribo/scribo/filter/internal/alignment_angle.hh +++ b/scribo/scribo/filter/internal/alignment_angle.hh @@ -1,4 +1,4 @@ -// Copyright (C) 2010 EPITA Research and Development Laboratory +// Copyright (C) 2010, 2011 EPITA Research and Development Laboratory // (LRDE) // // This file is part of Olena. @@ -73,76 +73,98 @@ namespace scribo unsigned current_object, unsigned nbh_object, anchor::Type anchor) { - trace::entering("scribo::filter::internal::alignment_angle_rad"); + trace::entering("scribo::filter::internal::alignment_angle"); mln_precondition(comps.is_valid()); - float dr, dc; + float dr, dc, result = 0; if (nbh_object == current_object) return 0; - // Center - if (anchor == anchor::Center) + switch(anchor) { - dr = math::abs(comps(current_object).bbox().pcenter().row() - - comps(nbh_object).bbox().pcenter().row()); - dc = math::abs(comps(current_object).bbox().pcenter().col() - - comps(nbh_object).bbox().pcenter().col()); - - return std::atan(dr / dc); - } - - // Top - else if (anchor == anchor::Top) - { - dr = math::abs(comps(current_object).bbox().pmin().row() - - comps(nbh_object).bbox().pmin().row()); - dc = math::abs(comps(current_object).bbox().pcenter().col() - - comps(nbh_object).bbox().pcenter().col()); - - return std::atan(dr / dc); + // Center + case anchor::Center: + { + dr = math::abs(comps(current_object).bbox().pcenter().row() + - comps(nbh_object).bbox().pcenter().row()); + dc = math::abs(comps(current_object).bbox().pcenter().col() + - comps(nbh_object).bbox().pcenter().col()); + + result = std::atan(dr / dc); + } + break; + + // Mass Center + case anchor::MassCenter: + { + dr = math::abs(comps(current_object).mass_center().row() + - comps(nbh_object).mass_center().row()); + dc = math::abs(comps(current_object).mass_center().col() + - comps(nbh_object).mass_center().col()); + + result = std::atan(dr / dc); + } + break; + + // Top + case anchor::TopStrictLeft: + case anchor::Top: + { + dr = math::abs(comps(current_object).bbox().pmin().row() + - comps(nbh_object).bbox().pmin().row()); + dc = math::abs(comps(current_object).bbox().pcenter().col() + - comps(nbh_object).bbox().pcenter().col()); + + result = std::atan(dr / dc); + } + break; + + // Bottom + case anchor::BottomStrictRight: + case anchor::Bottom: + { + dr = math::abs(comps(current_object).bbox().pmax().row() + - comps(nbh_object).bbox().pmax().row()); + dc = math::abs(comps(current_object).bbox().pcenter().col() + - comps(nbh_object).bbox().pcenter().col()); + + result = std::atan(dr / dc); + } + break; + + // Left + case anchor::Left: + { + dr = math::abs(comps(current_object).bbox().pcenter().row() + - comps(nbh_object).bbox().pcenter().row()); + dc = math::abs(comps(current_object).bbox().pmin().col() + - comps(nbh_object).bbox().pmin().col()); + + result = std::atan(dc / dr); + } + break; + + // Right + case anchor::Right: + { + dr = math::abs(comps(current_object).bbox().pcenter().row() + - comps(nbh_object).bbox().pcenter().row()); + dc = math::abs(comps(current_object).bbox().pmax().col() + - comps(nbh_object).bbox().pmax().col()); + + result = std::atan(dc / dr); + } + break; + + default: + trace::warning("scribo::filter::internal::alignment_angle," + " Invalid anchor value... Aborting computation."); } - // Bottom - else if (anchor == anchor::Bottom) - { - dr = math::abs(comps(current_object).bbox().pmax().row() - - comps(nbh_object).bbox().pmax().row()); - dc = math::abs(comps(current_object).bbox().pcenter().col() - - comps(nbh_object).bbox().pcenter().col()); - - return std::atan(dr / dc); - } - - // Left - else if (anchor == anchor::Left) - { - dr = math::abs(comps(current_object).bbox().pcenter().row() - - comps(nbh_object).bbox().pcenter().row()); - dc = math::abs(comps(current_object).bbox().pmin().col() - - comps(nbh_object).bbox().pmin().col()); - - return std::atan(dc / dr); - } - - // Right - else if (anchor == anchor::Right) - { - dr = math::abs(comps(current_object).bbox().pcenter().row() - - comps(nbh_object).bbox().pcenter().row()); - dc = math::abs(comps(current_object).bbox().pmax().col() - - comps(nbh_object).bbox().pmax().col()); - - return std::atan(dc / dr); - } - - else - trace::warning("Invalid anchor value... Aborting computation."); - - trace::exiting("scribo::filter::internal::alignment_angle_rad"); - return 0; - + trace::exiting("scribo::filter::internal::alignment_angle"); + return result; } # endif // ! MLN_INCLUDE_ONLY diff --git a/scribo/scribo/primitive/extract/separators_nonvisible.hh b/scribo/scribo/primitive/extract/separators_nonvisible.hh index 81ebd64..4e31650 100644 --- a/scribo/scribo/primitive/extract/separators_nonvisible.hh +++ b/scribo/scribo/primitive/extract/separators_nonvisible.hh @@ -86,6 +86,8 @@ #include <scribo/preprocessing/denoise_fg.hh> #include <scribo/preprocessing/rotate_90.hh> +#include <scribo/primitive/link/internal/compute_anchor.hh> + #include <scribo/primitive/link/internal/dmax_default.hh> #include <scribo/primitive/link/with_single_right_link_dmax_ratio.hh> @@ -128,80 +130,6 @@ namespace scribo namespace internal { - template <typename L> - void filter_bad_groups(object_groups<L>& top_groups, - object_groups<L>& bot_groups) - { - const component_set<L>& comps = top_groups.components(); - const L& lbl = comps.labeled_image(); - - for_all_groups(c, top_groups) - { - box2d b = comps(c).bbox(); - b.enlarge(0, comps(c).bbox().height()); - b.crop_wrt(lbl.domain()); - - typedef mln_value(L) V; - - const V* top_ptr = & lbl(b.pmin()); - const V* bot_ptr = & lbl(point2d(b.pmax().row(), b.pmin().col())); - - unsigned ntop = 0, nbot = 0; - for (unsigned n = 0; n < b.width(); ++n) - { - if (*top_ptr) - ++ntop; - if (*bot_ptr) - ++nbot; - } - - if (ntop / b.width() > 0.50f) - top_groups(c) = c; - - if (nbot / b.width() > 0.50f) - bot_groups(c) = c; - } - - } - - - template <typename L> - mln_site(L) - my_anchors(const component_set<L>& comps, - unsigned current_object, - anchor::Type anchor) - { - mln_site(L) sp;// = comps(current_object).bbox().pcenter(); - - unsigned h = comps(current_object).bbox().height(); - - switch (anchor) - { - default: - return sp; - - - // Bounding box top center - case anchor::Top: // FIXME: rename as TopLeft - sp.col() = comps(current_object).bbox().pmin().col(); - sp.row() = comps(current_object).bbox().pmin().row() - + math::min(2u, (h + 1) / 2 - 1); - break; - - - // Bounding box bottom center - case anchor::Bottom: // FIXME: rename as BottomLeft - sp.col() = comps(current_object).bbox().pmax().col(); - sp.row() = comps(current_object).bbox().pmax().row() - - math::min(2u, (h + 1) / 2 - 1); - break; - } - - return sp; - } - - - using namespace primitive::link::internal; template <typename L, typename E> @@ -225,12 +153,17 @@ namespace scribo anchor::Horizontal, dmax_default(dmax)), anchor(anchor_), - debug_(data::convert(value::rgb8(), input)), - debug_angle_(data::convert(value::rgb8(), input)), _debug_(debug) { + (void) input; // FIXME : remove this argument min_alpha_rad = (min_angle / 180.0f) * math::pi; max_alpha_rad = (max_angle / 180.0f) * math::pi; + + // if (_debug_) + // { + // debug_ = data::convert(value::rgb8(), input); + // debug_angle_ = data::convert(value::rgb8(), input); + // } } void compute_next_site_(P& p) @@ -247,7 +180,8 @@ namespace scribo mln_site(L) start_point_(unsigned current_object, anchor::Type anchor) { - return my_anchors(this->components_, current_object, anchor); + return link::internal::compute_anchor(this->components_, + current_object, anchor); } @@ -303,27 +237,29 @@ namespace scribo { super_::validate_link_(current_object, start_point, p, anchor); - if (_debug_) - { - mln_site(L) - p1 = my_anchors(this->components_, current_object, anchor), - p2 = my_anchors(this->components_, this->labeled_image_(p), - anchor); - mln::draw::line(debug_, p1, p2, literal::green); - - - float - angle = filter::internal::alignment_angle(this->components_, - current_object, - this->labeled_image_(p), - anchor); - angle = (angle * 180.0f) / math::pi; - angle = angle * 20.0f + 1.0f; - mln::draw::line(debug_angle_, p1, p2, - value::rgb8(unsigned(angle), - unsigned(angle), - unsigned(angle))); - } + // if (_debug_) + // { + // mln_site(L) + // p1 = link::internal::compute_anchor(this->components_, + // current_object, anchor), + // p2 = link::internal::compute_anchor(this->components_, + // this->labeled_image_(p), + // anchor); + // mln::draw::line(debug_, p1, p2, literal::green); + + + // float + // angle = filter::internal::alignment_angle(this->components_, + // current_object, + // this->labeled_image_(p), + // anchor); + // angle = (angle * 180.0f) / math::pi; + // angle = angle * 20.0f + 1.0f; + // mln::draw::line(debug_angle_, p1, p2, + // value::rgb8(unsigned(angle), + // unsigned(angle), + // unsigned(angle))); + // } } void invalidate_link_(unsigned current_object, @@ -333,33 +269,36 @@ namespace scribo { super_::invalidate_link_(current_object, start_point, p, anchor); - if (_debug_) - { - if (this->labeled_image_.domain().has(p) && this->labeled_image_(p) != 0) - { - mln_site(L) - p1 = my_anchors(this->components_, current_object, anchor), - p2 = my_anchors(this->components_, this->labeled_image_(p), - anchor); - if (this->labeled_image_.domain().has(p2) && norm::l1_distance(p1.to_vec(), p2.to_vec()) < 300) - { - mln::draw::line(debug_, p1, p2, literal::red); - } - - - float - angle = filter::internal::alignment_angle(this->components_, - current_object, - this->labeled_image_(p), - anchor); - angle = (angle * 180.0f) / math::pi; - angle = angle * 20.0f + 1.0f; - mln::draw::line(debug_angle_, p1, p2, - value::rgb8(unsigned(angle), - unsigned(angle), - unsigned(angle))); - } - } + // if (_debug_) + // { + // if (this->labeled_image_.domain().has(p) && this->labeled_image_(p) != 0) + // { + // mln_site(L) + // p1 = link::internal::compute_anchor(this->components_, + // current_object, anchor), + // p2 = link::internal::compute_anchor(this->components_, + // this->labeled_image_(p), + // anchor); + // if (this->labeled_image_.domain().has(p2) + // && norm::l1_distance(p1.to_vec(), p2.to_vec()) < 300) + // { + // mln::draw::line(debug_, p1, p2, literal::red); + // } + + + // float + // angle = filter::internal::alignment_angle(this->components_, + // current_object, + // this->labeled_image_(p), + // anchor); + // angle = (angle * 180.0f) / math::pi; + // angle = angle * 20.0f + 1.0f; + // mln::draw::line(debug_angle_, p1, p2, + // value::rgb8(unsigned(angle), + // unsigned(angle), + // unsigned(angle))); + // } + // } } @@ -368,8 +307,8 @@ namespace scribo anchor::Type anchor; - mln_ch_value(L, value::rgb8) debug_; - mln_ch_value(L, value::rgb8) debug_angle_; + // mln_ch_value(L, value::rgb8) debug_; + // mln_ch_value(L, value::rgb8) debug_angle_; bool _debug_; }; @@ -392,7 +331,8 @@ namespace scribo float max_angle, anchor::Type anchor, bool debug) - : super_(input, components, dmax, min_angle, max_angle, anchor, debug) + : super_(input, components, dmax, min_angle, + max_angle, anchor, debug) { } @@ -427,7 +367,8 @@ namespace scribo float max_angle, anchor::Type anchor, bool debug) - : super_(input, components, dmax, min_angle, max_angle, anchor, debug) + : super_(input, components, dmax, min_angle, + max_angle, anchor, debug) { } @@ -455,6 +396,8 @@ namespace scribo mln_concrete(I) separators_nonvisible(const Image& in_) { + trace::entering("scribo::primitive::extract::separators_nonvisible"); + const I& in = exact(in_); mln_precondition(in.is_valid()); typedef mln_value(I) Vi; @@ -469,42 +412,35 @@ namespace scribo util::timer t; util::timer gt; - // Load (OK) - t.start(); - float t_ = t; - std::cout << "Image loaded - " << t_ << std::endl; - gt.start(); - // Remove horizontal lines. - t.restart(); + // // Remove horizontal lines. + // t.restart(); - mln_concrete(I) hlines = primitive::extract::lines_h_pattern(in, 50, 3); - mln_concrete(I) input = primitive::remove::separators(in, hlines); + // mln_concrete(I) hlines = primitive::extract::lines_h_pattern(in, 50, 3); + // mln_concrete(I) input = primitive::remove::separators(in, hlines); - t_ = t; - std::cout << "Horizontal lines removed - " << t_ << std::endl; + // t_ = t; + // std::cout << "Horizontal lines removed - " << t_ << std::endl; // Closing structural - Connect characters. - t.restart(); + t.start(); win::hline2d vl(17); - mln_concrete(I) input_clo = morpho::closing::structural(input, vl); + mln_concrete(I) input_clo = morpho::closing::structural(in, vl); -// input_clo = scribo::preprocessing::rotate_90(input_clo, true); - - t_ = t; + float t_ = t; std::cout << "closing_structural - " << t_ << std::endl; - if (_debug_) - { - // Restore input orientation. - input = scribo::preprocessing::rotate_90(input, false); + // if (_debug_) + // { + // // Restore input orientation. + // input = scribo::preprocessing::rotate_90(input, false); - io::pbm::save(input_clo, "input_clo.pbm"); - } + // io::pbm::save(input_clo, "input_clo.pbm"); + // } // Rotate (OK) t.restart(); @@ -526,9 +462,9 @@ namespace scribo t_ = t; std::cout << "extract::components - " << t_ << std::endl; - if (_debug_) - io::pgm::save(data::convert(value::int_u8(), components.labeled_image()), - "lbl.pgm"); + // if (_debug_) + // io::pgm::save(data::convert(value::int_u8(), components.labeled_image()), + // "lbl.pgm"); unsigned dmax = 5; @@ -543,39 +479,42 @@ namespace scribo { // Right internal::single_right_dmax_ratio_aligned_functor<L> - functor(input_clo, components, dmax, min_angle, max_angle, anchor::Top, _debug_); -// top_right = primitive::link::impl::compute_fastest(functor, anchor::Top); - top_right = primitive::link::compute(functor, anchor::Top); + functor(input_clo, components, dmax, min_angle, max_angle, + anchor::TopStrictLeft, _debug_); +// top_right = primitive::link::impl::compute_fastest(functor, anchor::TopStrictLeft); + top_right = primitive::link::compute(functor, anchor::TopStrictLeft); t.stop(); - if (_debug_) - { - io::ppm::save(functor.debug_, "right_top.ppm"); - io::ppm::save(functor.debug_angle_, "right_top_angle.ppm"); - } + // if (_debug_) + // { + // io::ppm::save(functor.debug_, "right_top.ppm"); + // io::ppm::save(functor.debug_angle_, "right_top_angle.ppm"); + // } t.resume(); // Left internal::single_left_dmax_ratio_aligned_functor<L> - lfunctor(input_clo, components, dmax, min_angle, max_angle, anchor::Top, _debug_); - top_left = primitive::link::compute(lfunctor, anchor::Top); + lfunctor(input_clo, components, dmax, min_angle, max_angle, + anchor::TopStrictLeft, _debug_); + top_left = primitive::link::compute(lfunctor, anchor::TopStrictLeft); t.stop(); - if (_debug_) - { - io::ppm::save(lfunctor.debug_, "left_top.ppm"); - io::ppm::save(lfunctor.debug_angle_, "left_top_angle.ppm"); + // if (_debug_) + // { + // io::ppm::save(lfunctor.debug_, "left_top.ppm"); + // io::ppm::save(lfunctor.debug_angle_, "left_top_angle.ppm"); - mln_ch_value(I, value::rgb8) output = duplicate(functor.debug_); - data::paste((lfunctor.debug_ | (pw::value(lfunctor.debug_) != pw::cst(literal::black))) | (pw::value(lfunctor.debug_) != pw::cst(literal::white)), output); + // mln_ch_value(I, value::rgb8) output = duplicate(functor.debug_); + // data::paste((lfunctor.debug_ | (pw::value(lfunctor.debug_) != pw::cst(literal::black))) + // | (pw::value(lfunctor.debug_) != pw::cst(literal::white)), output); - io::ppm::save(output, "left_right_top.ppm"); - } + // io::ppm::save(output, "left_right_top.ppm"); + // } t.resume(); } @@ -585,105 +524,53 @@ namespace scribo { // Right internal::single_right_dmax_ratio_aligned_functor<L> - functor(input_clo, components, dmax, min_angle, max_angle, anchor::Bottom, _debug_); - bot_right = primitive::link::compute(functor, anchor::Bottom); + functor(input_clo, components, dmax, min_angle, max_angle, + anchor::BottomStrictRight, _debug_); + bot_right = primitive::link::compute(functor, anchor::BottomStrictRight); t.stop(); - if (_debug_) - { - io::ppm::save(functor.debug_, "right_bot.ppm"); - io::ppm::save(functor.debug_angle_, "right_bot_angle.ppm"); - } + // if (_debug_) + // { + // io::ppm::save(functor.debug_, "right_bot.ppm"); + // io::ppm::save(functor.debug_angle_, "right_bot_angle.ppm"); + // } t.resume(); // Left internal::single_left_dmax_ratio_aligned_functor<L> - lfunctor(input_clo, components, dmax, min_angle, max_angle, anchor::Bottom, _debug_); - bot_left = primitive::link::compute(lfunctor, anchor::Bottom); + lfunctor(input_clo, components, dmax, min_angle, max_angle, + anchor::BottomStrictRight, _debug_); + bot_left = primitive::link::compute(lfunctor, anchor::BottomStrictRight); t.stop(); - if (_debug_) - { - io::ppm::save(lfunctor.debug_, "left_bot.ppm"); - io::ppm::save(lfunctor.debug_angle_, "left_bot_angle.ppm"); - } + // if (_debug_) + // { + // io::ppm::save(lfunctor.debug_, "left_bot.ppm"); + // io::ppm::save(lfunctor.debug_angle_, "left_bot_angle.ppm"); - if (_debug_) - { - mln_ch_value(I, value::rgb8) output = duplicate(functor.debug_); - data::paste((lfunctor.debug_ | (pw::value(lfunctor.debug_) != pw::cst(literal::black))) | (pw::value(lfunctor.debug_) != pw::cst(literal::white)), output); + // mln_ch_value(I, value::rgb8) output = duplicate(functor.debug_); + // data::paste((lfunctor.debug_ | (pw::value(lfunctor.debug_) != pw::cst(literal::black))) + // | (pw::value(lfunctor.debug_) != pw::cst(literal::white)), output); - io::ppm::save(output, "left_right_bot.ppm"); - } + // io::ppm::save(output, "left_right_bot.ppm"); + // } } t_ = t; std::cout << "links - " << t_ << std::endl; - + // Merge links and build CC groups t.restart(); - std::cout << "group - top" << std::endl; object_groups<L> top_groups = primitive::group::from_double_link_any(top_left, top_right); - std::cout << "group - bot" << std::endl; object_groups<L> bot_groups = primitive::group::from_double_link_any(bot_left, bot_right); t_ = t; std::cout << "group - " << t_ << std::endl; - t.restart(); - util::array<accu::shape::bbox<point2d> > - btop_accu(top_groups.nelements()), - bbot_accu(bot_groups.nelements()); - - - for_all_groups(c, top_groups) - { - btop_accu(top_groups(c)).take(components(c).bbox()); - bbot_accu(bot_groups(c)).take(components(c).bbox()); - } - t_ = t; - std::cout << "groups to group bboxes - " << t_ << std::endl; - - - - if (_debug_) - { - - mln_ch_value(I, value::rgb8) - wo_filtering = data::convert(value::rgb8(), input); - - for_all_comp_data(d, btop_accu) - { - if (btop_accu(d).is_valid()) - { - mln::draw::line(wo_filtering, - btop_accu(d).to_result().pmin(), - point2d(btop_accu(d).to_result().pmin().row(), - btop_accu(d).to_result().pmax().col()), - literal::green); - - } - } - - for_all_comp_data(d, bbot_accu) - { - if (bbot_accu(d).is_valid()) - { - mln::draw::line(wo_filtering, - point2d(bbot_accu(d).to_result().pmax().row(), - bbot_accu(d).to_result().pmin().col()), - bbot_accu(d).to_result().pmax(), - literal::green); - } - - } - io::ppm::save(wo_filtering, "wo_filtering.ppm"); - } - - + // Filter CC groups t.restart(); top_groups = filter::object_groups_small(top_groups, min_card); bot_groups = filter::object_groups_small(bot_groups, min_card); @@ -691,7 +578,7 @@ namespace scribo std::cout << "small groups - " << t_ << std::endl; - + // Compute group bboxes t.restart(); util::array<accu::shape::bbox<point2d> > top_accu(top_groups.nelements()), @@ -708,78 +595,35 @@ namespace scribo - - - - - t.restart(); mln_concrete(I) separators; initialize(separators, input_clo); + + // FIXME: any way to fill border AND data at the same time? data::fill(separators, false); + extension::fill(separators, false); + t_ = t; std::cout << "Initialize separators image - " << t_ << std::endl; - mln_ch_value(I, value::rgb8) both; - - if (_debug_) - both = data::convert(value::rgb8(), input); - - - t.restart(); for_all_comp_data(d, top_accu) { - if (top_accu(d).is_valid() || btop_accu(d).is_valid()) + if (top_accu(d).is_valid()) { - if (top_accu(d).is_valid()) - { - if (_debug_) - mln::draw::line(both, - top_accu(d).to_result().pmin(), - point2d(top_accu(d).to_result().pmin().row(), - top_accu(d).to_result().pmax().col()), - literal::green); - - mln::draw::line(separators, - top_accu(d).to_result().pmin(), - point2d(top_accu(d).to_result().pmin().row(), - top_accu(d).to_result().pmax().col()), - true); - } - else - if (_debug_ && btop_accu(d).is_valid()) - mln::draw::line(both, - btop_accu(d).to_result().pmin(), - point2d(btop_accu(d).to_result().pmin().row(), - btop_accu(d).to_result().pmax().col()), - literal::yellow); - + mln::draw::line(separators, + top_accu(d).to_result().pmin(), + point2d(top_accu(d).to_result().pmin().row(), + top_accu(d).to_result().pmax().col()), + true); } - if (bot_accu(d).is_valid() || bbot_accu(d).is_valid()) + if (bot_accu(d).is_valid()) { - if (bot_accu(d).is_valid()) - { - if (_debug_) - mln::draw::line(both, - point2d(bot_accu(d).to_result().pmax().row(), - bot_accu(d).to_result().pmin().col()), - bot_accu(d).to_result().pmax(), - literal::green); - - mln::draw::line(separators, - point2d(bot_accu(d).to_result().pmax().row(), - bot_accu(d).to_result().pmin().col()), - bot_accu(d).to_result().pmax(), - true); - } - else - if (_debug_ && bbot_accu(d).is_valid()) - mln::draw::line(both, - point2d(bbot_accu(d).to_result().pmax().row(), - bbot_accu(d).to_result().pmin().col()), - bbot_accu(d).to_result().pmax(), - literal::yellow); + mln::draw::line(separators, + point2d(bot_accu(d).to_result().pmax().row(), + bot_accu(d).to_result().pmin().col()), + bot_accu(d).to_result().pmax(), + true); } } @@ -787,22 +631,112 @@ namespace scribo std::cout << "Drawing output image - " << t_ << std::endl; - if (_debug_) - { - io::ppm::save(both, "both.ppm"); - io::pbm::save(separators, "separators.pbm"); - } + // if (_debug_) + // { + // // Restore input orientation. + // mln_concrete(I) input = scribo::preprocessing::rotate_90(in, false); + + + // // Debug group bboxes (includes all bboxes before filtering) + // util::array<accu::shape::bbox<point2d> > + // btop_accu(top_groups.nelements()), + // bbot_accu(bot_groups.nelements()); + + + // for_all_groups(c, top_groups) + // { + // btop_accu(top_groups(c)).take(components(c).bbox()); + // bbot_accu(bot_groups(c)).take(components(c).bbox()); + // } + + // mln_ch_value(I, value::rgb8) + // wo_filtering = data::convert(value::rgb8(), input); + + // for_all_comp_data(d, btop_accu) + // { + // if (btop_accu(d).is_valid()) + // { + // mln::draw::line(wo_filtering, + // btop_accu(d).to_result().pmin(), + // point2d(btop_accu(d).to_result().pmin().row(), + // btop_accu(d).to_result().pmax().col()), + // literal::green); + + // } + // } + + // for_all_comp_data(d, bbot_accu) + // { + // if (bbot_accu(d).is_valid()) + // { + // mln::draw::line(wo_filtering, + // point2d(bbot_accu(d).to_result().pmax().row(), + // bbot_accu(d).to_result().pmin().col()), + // bbot_accu(d).to_result().pmax(), + // literal::green); + // } + + // } + // io::ppm::save(wo_filtering, "wo_filtering.ppm"); + + // mln_ch_value(I, value::rgb8) both = data::convert(value::rgb8(), input); + + // for_all_comp_data(d, top_accu) + // { + // if (top_accu(d).is_valid() || btop_accu(d).is_valid()) + // { + // if (top_accu(d).is_valid()) + // { + // mln::draw::line(both, + // top_accu(d).to_result().pmin(), + // point2d(top_accu(d).to_result().pmin().row(), + // top_accu(d).to_result().pmax().col()), + // literal::green); + // } + // else + // if (btop_accu(d).is_valid()) + // mln::draw::line(both, + // btop_accu(d).to_result().pmin(), + // point2d(btop_accu(d).to_result().pmin().row(), + // btop_accu(d).to_result().pmax().col()), + // literal::yellow); + + // } + // if (bot_accu(d).is_valid() || bbot_accu(d).is_valid()) + // { + // if (bot_accu(d).is_valid()) + // { + // mln::draw::line(both, + // point2d(bot_accu(d).to_result().pmax().row(), + // bot_accu(d).to_result().pmin().col()), + // bot_accu(d).to_result().pmax(), + // literal::green); + // } + // else + // if (bbot_accu(d).is_valid()) + // mln::draw::line(both, + // point2d(bbot_accu(d).to_result().pmax().row(), + // bbot_accu(d).to_result().pmin().col()), + // bbot_accu(d).to_result().pmax(), + // literal::yellow); + // } + + // } + + // io::ppm::save(both, "both.ppm"); + // io::pbm::save(separators, "separators.pbm"); + // } // Hit or miss { - if (_debug_) - { - mln_concrete(I) input_with_seps = duplicate(input_clo); - data::paste(separators | pw::value(separators), input_with_seps); + // if (_debug_) + // { + // mln_concrete(I) input_with_seps = duplicate(input_clo); + // data::paste(separators | pw::value(separators), input_with_seps); - io::pbm::save(input_with_seps, "input_with_seps.pbm"); - } + // io::pbm::save(input_with_seps, "input_with_seps.pbm"); + // } t.restart(); unsigned length = 25; @@ -819,11 +753,12 @@ namespace scribo typedef mln_ch_value(I,unsigned) J; J tmp = accu::transform_line(accu, input_clo, length, 1); + t_ = t; std::cout << "* accu::transform_line - " << t_ << std::endl; - if (_debug_) - io::pgm::save(data::convert(value::int_u8(), tmp), "tmp.pgm"); + // if (_debug_) + // io::pgm::save(data::convert(value::int_u8(), tmp), "tmp.pgm"); t.restart(); @@ -841,25 +776,53 @@ namespace scribo unsigned invalid_ratio = unsigned(length * 0.30f); - mln_piter(I) p(separators.domain()); - for_all(p) - if (separators(p)) - { - unsigned lbl = sep_lbl(p); + extension::adjust_fill(tmp, 21, 0); + + value::int_u8 *sep_lbl_ptr = sep_lbl.buffer() + sep_lbl.index_of_point(sep_lbl.domain().pmin()); + bool *separators_ptr = separators.buffer() + separators.index_of_point(separators.domain().pmin()); + unsigned *tmp_ptr = tmp.buffer() + tmp.index_of_point(tmp.domain().pmin());; + int idx1 = tmp.delta_index(dp1); + int idx2 = tmp.delta_index(dp2); + + unsigned nrows = separators.nrows(); + unsigned ncols = separators.ncols(); - unsigned - top_count = tmp(p + dp1), - bot_count = tmp(p + dp2); + unsigned row_idx_sep_lbl = sep_lbl.delta_index(dpoint2d(+1, - ncols)); + unsigned row_idx_separators = separators.delta_index(dpoint2d(+1, - ncols)); + unsigned row_idx_tmp = tmp.delta_index(dpoint2d(+1, - ncols)); - // This site is wrapped between two lines of text so we don't - // want it. - if (top_count >= invalid_ratio + 1 - && bot_count >= invalid_ratio + 1) + for (unsigned row = 0; row < nrows; ++row) + { + for (unsigned col = 0; col < ncols; ++col) + { + if (*separators_ptr) { - relbl(lbl) = false; + unsigned lbl = *sep_lbl_ptr; + + unsigned + top_count = *(tmp_ptr + idx1), + bot_count = *(tmp_ptr + idx2); + + // This site is wrapped between two lines of text so we don't + // want it. + if (top_count >= invalid_ratio + 1 + && bot_count >= invalid_ratio + 1) + { + relbl(lbl) = false; + } } + + ++tmp_ptr; + ++sep_lbl_ptr; + ++separators_ptr; } + tmp_ptr += row_idx_tmp; + sep_lbl_ptr += row_idx_sep_lbl; + separators_ptr += row_idx_separators; + } + + t_ = t; std::cout << "* reading data - " << t_ << std::endl; @@ -870,37 +833,35 @@ namespace scribo mln_concrete(I) output = data::convert(bool(), sep_lbl); - if (_debug_) - { - io::pbm::save(output, "separators_hom.pbm"); - io::pbm::save(separators, "separators_filtered.pbm"); - } + // if (_debug_) + // { + // io::pbm::save(output, "separators_hom.pbm"); + // io::pbm::save(separators, "separators_filtered.pbm"); - t.restart(); - value::int_u16 ncomps; - component_set<L> comps = primitive::extract::components(output, c8(), ncomps); - mln_ch_value(I, value::rgb8) both; + // // value::int_u16 ncomps; + // // component_set<L> comps = primitive::extract::components(output, c8(), ncomps); + // // mln_ch_value(I, value::rgb8) both; - both = data::convert(value::rgb8(), input); + // // both = data::convert(value::rgb8(), input); - // Needed since the rotated image origin is (0,0). Rotation does - // not preserve rotated coordinates. - dpoint2d dp(input.domain().pcenter() - input_clo.domain().pcenter()); + // // // Needed since the rotated image origin is (0,0). + // // dpoint2d dp(input.domain().pcenter() - input_clo.domain().pcenter()); + + // // for_all_comps(c, comps) + // // { + // // box2d b = geom::rotate(comps(c).bbox(), -90, input_clo.domain().pcenter()); + // // mln::draw::line(both, + // // b.pmin() + dp, + // // b.pmax() + dp, + // // literal::green); + // // } + // } - for_all_comps(c, comps) - { - box2d b = geom::rotate(comps(c).bbox(), -90, input_clo.domain().pcenter()); - mln::draw::line(both, - b.pmin() + dp, - b.pmax() + dp, - literal::green); - } - t_ = t; - std::cout << "Output image - " << t_ << std::endl; gt.stop(); t_ = gt; - std::cout << "Total time: " << t_ << std::endl; + std::cout << "Non visible separators: " << t_ << std::endl; + trace::exiting("scribo::primitive::extract::separators_nonvisible"); return scribo::preprocessing::rotate_90(output, true); } } diff --git a/scribo/scribo/primitive/link/internal/compute_anchor.hh b/scribo/scribo/primitive/link/internal/compute_anchor.hh index 1c52b91..c1a9106 100644 --- a/scribo/scribo/primitive/link/internal/compute_anchor.hh +++ b/scribo/scribo/primitive/link/internal/compute_anchor.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -88,10 +88,8 @@ namespace scribo { typedef mln_site(L) P; - unsigned h = components(current_object).bbox().pmax().row() - - components(current_object).bbox().pmin().row(); - unsigned w = components(current_object).bbox().pmax().col() - - components(current_object).bbox().pmin().col(); + unsigned h = components(current_object).bbox().height(); + unsigned w = components(current_object).bbox().width(); mln_site(L) sp = components(current_object).bbox().pcenter(); @@ -113,6 +111,22 @@ namespace scribo break; + // Bounding box top left + case anchor::TopStrictLeft: + sp.col() = components(current_object).bbox().pmin().col(); + sp.row() = components(current_object).bbox().pmin().row() + + math::min(2u, (h + 1) / 2 - 1); + break; + + + // Bounding box bottom right + case anchor::BottomStrictRight: + sp.col() = components(current_object).bbox().pmax().col(); + sp.row() = components(current_object).bbox().pmax().row() + - math::min(2u, (h + 1) / 2 - 1); + break; + + // Bounding box bottom center case anchor::Bottom: if (h < 30) diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index ed691e8..48098ba 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -34,11 +34,11 @@ # include <scribo/core/line_set.hh> # include <scribo/core/paragraph_set.hh> -# include <scribo/primitive/extract/elements.hh> +# include <scribo/primitive/extract/non_text.hh> # include <scribo/primitive/extract/components.hh> -# include <scribo/primitive/extract/vertical_separators.hh> +//# include <scribo/primitive/extract/vertical_separators.hh> +# include <scribo/primitive/extract/separators.hh> # include <scribo/primitive/extract/separators_nonvisible.hh> -# include <scribo/primitive/extract/elements.hh> # include <scribo/primitive/identify.hh> @@ -168,16 +168,17 @@ namespace scribo input_cleaned = exact(processed_image); if (enable_line_seps) { - on_new_progress_label("Find vertical separators..."); + on_new_progress_label("Find vertical and horizontal separators..."); - // Vertical separators - separators = primitive::extract::vertical_separators(processed_image, 81); + // Vertical and horizontal separators + separators = primitive::extract::separators(processed_image, 81); on_progress(); on_new_progress_label("Remove separators..."); input_cleaned = primitive::remove::separators(processed_image, separators); + doc.set_line_separators(separators); on_progress(); } -- 1.5.6.5

13 years, 10 months

last-svn-commit-773-gc65020d Small fixes in Scribo.

by Guillaume Lazzara

* scribo/core/macros.hh: Update comments. * scribo/text/merging.hh: Add comments and fix line data swap. * scribo/text/recognition.hh: Make use of is_textline. * src/text_in_picture.cc: Initialize ImageMagick. --- scribo/ChangeLog | 12 ++++++++++++ scribo/scribo/core/macros.hh | 7 +++---- scribo/scribo/text/merging.hh | 6 +++--- scribo/scribo/text/recognition.hh | 2 +- 4 files changed, 19 insertions(+), 8 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index d783c5c..0aa4ba2 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,17 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Small fixes in Scribo. + + * scribo/core/macros.hh: Update comments. + + * scribo/text/merging.hh: Add comments and fix line data swap. + + * scribo/text/recognition.hh: Make use of is_textline. + + * src/text_in_picture.cc: Initialize ImageMagick. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Improve and cleanup whitespace separator detection. * scribo/core/tag/anchor.hh: Add new anchors. diff --git a/scribo/scribo/core/macros.hh b/scribo/scribo/core/macros.hh index 887539f..c6de1ff 100644 --- a/scribo/scribo/core/macros.hh +++ b/scribo/scribo/core/macros.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -33,8 +33,6 @@ # define for_all_elements(E, S) \ for (unsigned E = 0; E < S.nelements(); ++E) - -// FIXME: we want to replace previous macros by these ones. # define for_all_comps(C, S) \ for (unsigned C = 1; C <= S.nelements(); ++C) @@ -56,6 +54,7 @@ # define for_all_line_comps(E, S) \ for_all_elements(E, S) +// Internal use only. # define for_all_lines_info(E, S) \ for_all_comp_data(E, S) diff --git a/scribo/scribo/text/merging.hh b/scribo/scribo/text/merging.hh index 3087465..f433e51 100644 --- a/scribo/scribo/text/merging.hh +++ b/scribo/scribo/text/merging.hh @@ -192,7 +192,7 @@ namespace scribo { // we transfer data from the largest item to the root one. scribo::line_info<L> tmp = lines(l1); - lines(l1) = lines(l2); + std::swap(lines(l1), lines(l2)); lines(l1).fast_merge(tmp); // We must set manually the tag for lines(l2) since it is @@ -504,8 +504,8 @@ namespace scribo void one_merge_pass(unsigned ith_pass, const box2d& domain, - std::vector<scribo::line_id_t>& v, - scribo::line_set<L>& lines, + std::vector<scribo::line_id_t>& v, // Ids sorted by bbox size. + scribo::line_set<L>& lines, // Tagged Lines (looks_like_a_text_line?) mln::util::array<unsigned>& parent) { image2d<unsigned> billboard(domain); diff --git a/scribo/scribo/text/recognition.hh b/scribo/scribo/text/recognition.hh index 59f269e..3a9742b 100644 --- a/scribo/scribo/text/recognition.hh +++ b/scribo/scribo/text/recognition.hh @@ -127,7 +127,7 @@ namespace scribo /// Use text bboxes with Tesseract for_all_lines(i, lines) { - if (! lines(i).is_valid() || lines(i).is_hidden() || lines(i).type() != line::Text) + if (! lines(i).is_textline()) continue; mln_domain(I) box = lines(i).bbox(); -- 1.5.6.5

13 years, 10 months

last-svn-commit-774-gcc8c9ed Add new tools in Scribo.

by Guillaume Lazzara

* src/primitive/extract/Makefile.am, * src/primitive/remove/Makefile.am: Add new targets. * src/primitive/extract/separators_nonvisible.cc, * src/primitive/remove/separators.cc: New. --- scribo/ChangeLog | 10 ++++++++++ scribo/src/primitive/extract/Makefile.am | 2 ++ .../extract/separators_nonvisible.cc} | 19 ++++++++++--------- scribo/src/primitive/{ => remove}/Makefile.am | 11 ++++++----- .../negate.cc => primitive/remove/separators.cc} | 20 ++++++++++++-------- 5 files changed, 40 insertions(+), 22 deletions(-) copy scribo/src/{misc/negate.cc => primitive/extract/separators_nonvisible.cc} (79%) copy scribo/src/primitive/{ => remove}/Makefile.am (82%) copy scribo/src/{misc/negate.cc => primitive/remove/separators.cc} (76%) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 0aa4ba2..8b7ad7f 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,15 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Add new tools in Scribo. + + * src/primitive/extract/Makefile.am, + * src/primitive/remove/Makefile.am: Add new targets. + + * src/primitive/extract/separators_nonvisible.cc, + * src/primitive/remove/separators.cc: New. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + Small fixes in Scribo. * scribo/core/macros.hh: Update comments. diff --git a/scribo/src/primitive/extract/Makefile.am b/scribo/src/primitive/extract/Makefile.am index a46cd68..22d6bfd 100644 --- a/scribo/src/primitive/extract/Makefile.am +++ b/scribo/src/primitive/extract/Makefile.am @@ -22,6 +22,7 @@ noinst_PROGRAMS = \ discontinued_lines \ discontinued_vlines \ discontinued_hlines \ + separators_nonvisible \ thick_vlines \ thick_hlines \ lines_pattern @@ -29,6 +30,7 @@ noinst_PROGRAMS = \ discontinued_lines_SOURCES = discontinued_lines.cc discontinued_vlines_SOURCES = discontinued_vlines.cc discontinued_hlines_SOURCES = discontinued_hlines.cc +separators_nonvisible_SOURCES = separators_nonvisible.cc thick_vlines_SOURCES = thick_vlines.cc thick_hlines_SOURCES = thick_hlines.cc lines_pattern_SOURCES = lines_pattern.cc diff --git a/scribo/src/misc/negate.cc b/scribo/src/primitive/extract/separators_nonvisible.cc similarity index 79% copy from scribo/src/misc/negate.cc copy to scribo/src/primitive/extract/separators_nonvisible.cc index da6fad6..82d4787 100644 --- a/scribo/src/misc/negate.cc +++ b/scribo/src/primitive/extract/separators_nonvisible.cc @@ -1,5 +1,4 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -25,35 +24,37 @@ // executable file might be covered by the GNU General Public License. #include <mln/core/image/image2d.hh> -#include <mln/logical/not.hh> #include <mln/io/pbm/all.hh> +#include <mln/data/convert.hh> +#include <scribo/primitive/extract/separators_nonvisible2.hh> #include <scribo/debug/usage.hh> - const char *args_desc[][2] = { { "input.pbm", "A binary image." }, + { "output.pbm", "Output image." }, {0, 0} }; int main(int argc, char *argv[]) { - mln::trace::entering("main"); using namespace mln; + using namespace scribo; if (argc != 3) return scribo::debug::usage(argv, - "Negate a binary image", + "Extract non visible separators (whitespaces)", "input.pbm output.pbm", args_desc); + trace::entering("main"); + image2d<bool> input; io::pbm::load(input, argv[1]); - io::pbm::save(logical::not_(input), argv[2]); - - mln::trace::exiting("main"); + io::pbm::save(primitive::extract::separators_nonvisible(input), argv[2]); + trace::exiting("main"); } diff --git a/scribo/src/primitive/Makefile.am b/scribo/src/primitive/remove/Makefile.am similarity index 82% copy from scribo/src/primitive/Makefile.am copy to scribo/src/primitive/remove/Makefile.am index 7e46a66..a673886 100644 --- a/scribo/src/primitive/Makefile.am +++ b/scribo/src/primitive/remove/Makefile.am @@ -1,4 +1,4 @@ -# Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE). +# Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE). # # This file is part of Olena. # @@ -16,7 +16,8 @@ include $(top_srcdir)/scribo/scribo.mk -SUBDIRS = \ - extract \ - find \ - group + +noinst_PROGRAMS = \ + separators + +separators_SOURCES = separators.cc diff --git a/scribo/src/misc/negate.cc b/scribo/src/primitive/remove/separators.cc similarity index 76% copy from scribo/src/misc/negate.cc copy to scribo/src/primitive/remove/separators.cc index da6fad6..46e977f 100644 --- a/scribo/src/misc/negate.cc +++ b/scribo/src/primitive/remove/separators.cc @@ -1,5 +1,4 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -25,35 +24,40 @@ // executable file might be covered by the GNU General Public License. #include <mln/core/image/image2d.hh> -#include <mln/logical/not.hh> #include <mln/io/pbm/all.hh> +#include <mln/data/convert.hh> +#include <scribo/primitive/extract/separators.hh> +#include <scribo/primitive/remove/separators.hh> #include <scribo/debug/usage.hh> - const char *args_desc[][2] = { { "input.pbm", "A binary image." }, + { "output.pbm", "Output image." }, {0, 0} }; int main(int argc, char *argv[]) { - mln::trace::entering("main"); using namespace mln; + using namespace scribo; if (argc != 3) return scribo::debug::usage(argv, - "Negate a binary image", + "Remove visible separators", "input.pbm output.pbm", args_desc); + trace::entering("main"); + image2d<bool> input; io::pbm::load(input, argv[1]); - io::pbm::save(logical::not_(input), argv[2]); + image2d<bool> seps = primitive::extract::separators(input, 81); - mln::trace::exiting("main"); + io::pbm::save(primitive::remove::separators(input, seps), argv[2]); + trace::exiting("main"); } -- 1.5.6.5

13 years, 10 months

last-svn-commit-775-gfd62480 configure.ac: scribo/src/primitive/remove.

by Guillaume Lazzara

--- ChangeLog | 4 ++++ configure.ac | 1 + 2 files changed, 5 insertions(+), 0 deletions(-) diff --git a/ChangeLog b/ChangeLog index 9fedefa..30f8b74 100644 --- a/ChangeLog +++ b/ChangeLog @@ -1,5 +1,9 @@ 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + * configure.ac: scribo/src/primitive/remove. + +2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> + * configure.ac: configure scribo/tests/convert. 2011-03-14 Thierry GERAUD <thierry.geraud(a)lrde.epita.fr> diff --git a/configure.ac b/configure.ac index e30f010..44f359b 100644 --- a/configure.ac +++ b/configure.ac @@ -332,6 +332,7 @@ AC_CONFIG_FILES([ scribo/src/primitive/extract/Makefile scribo/src/primitive/find/Makefile scribo/src/primitive/group/Makefile + scribo/src/primitive/remove/Makefile scribo/src/table/Makefile scribo/src/text/Makefile scribo/src/toolchain/Makefile -- 1.5.6.5

13 years, 10 months

last-svn-commit-776-g8c287df Set component type during component extraction.

by Guillaume Lazzara

* scribo/core/component_info.hh, * scribo/core/component_set.hh, * scribo/core/document.hh, * scribo/core/tag/component.hh, * scribo/primitive/extract/components.hh, * scribo/primitive/identify.hh: Explicitly set component type to Separator when extracting separator components. --- scribo/ChangeLog | 12 ++++++ scribo/scribo/core/component_info.hh | 8 ++- scribo/scribo/core/component_set.hh | 50 ++++++++++++++++--------- scribo/scribo/core/document.hh | 6 ++- scribo/scribo/core/tag/component.hh | 20 ++++++--- scribo/scribo/primitive/extract/components.hh | 18 ++++++--- scribo/scribo/primitive/identify.hh | 2 +- 7 files changed, 79 insertions(+), 37 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 8b7ad7f..330338a 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,15 @@ +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + + Set component type during component extraction. + + * scribo/core/component_info.hh, + * scribo/core/component_set.hh, + * scribo/core/document.hh, + * scribo/core/tag/component.hh, + * scribo/primitive/extract/components.hh, + * scribo/primitive/identify.hh: Explicitly set component type to + Separator when extracting separator components. + 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> Add new tools in Scribo. diff --git a/scribo/scribo/core/component_info.hh b/scribo/scribo/core/component_info.hh index 1b03318..6fc73f8 100644 --- a/scribo/scribo/core/component_info.hh +++ b/scribo/scribo/core/component_info.hh @@ -53,7 +53,8 @@ namespace scribo component_info(const component_id_t& id, const mln::box2d& bbox, const mln::point2d& mass_center, - unsigned card); + unsigned card, + component::Type type = component::Undefined); component_id_t id() const; const mln::box2d& bbox() const; @@ -101,9 +102,10 @@ namespace scribo component_info::component_info(const component_id_t& id, const mln::box2d& bbox, const mln::point2d& mass_center, - unsigned card) + unsigned card, + component::Type type) : id_(id), bbox_(bbox), mass_center_(mass_center), card_(card), - tag_(component::None), type_(component::Undefined) + tag_(component::None), type_(type) { } diff --git a/scribo/scribo/core/component_set.hh b/scribo/scribo/core/component_set.hh index 7ddcf16..442e8d6 100644 --- a/scribo/scribo/core/component_set.hh +++ b/scribo/scribo/core/component_set.hh @@ -86,16 +86,20 @@ namespace scribo component_set_data(); component_set_data(const L& ima, const mln_value(L)& ncomps); component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs); + const mln::util::array<pair_accu_t>& attribs, + component::Type type = component::Undefined); component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs); + const mln::util::array<pair_data_t>& attribs, + component::Type type = component::Undefined); component_set_data(const L& ima, const mln_value(L)& ncomps, const mln::util::array<scribo::component_info>& infos); - void fill_infos(const mln::util::array<pair_accu_t>& attribs); + void fill_infos(const mln::util::array<pair_accu_t>& attribs, + component::Type type = component::Undefined); - void fill_infos(const mln::util::array<pair_data_t>& attribs); + void fill_infos(const mln::util::array<pair_data_t>& attribs, + component::Type type = component::Undefined); // Useful while constructing incrementaly (XML loading). void soft_init(const mln_value(L) ncomps); @@ -141,10 +145,12 @@ namespace scribo /// Constructor from an image \p ima, the number of labels \p ncomps and /// attributes values (bounding box and mass center). component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs); + const mln::util::array<pair_accu_t>& attribs, + component::Type type = component::Undefined); component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs); + const mln::util::array<pair_data_t>& attribs, + component::Type type = component::Undefined); /// @} /// Return the component count. @@ -284,26 +290,28 @@ namespace scribo inline component_set_data<L>::component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs) + const mln::util::array<pair_accu_t>& attribs, + component::Type type) : ima_(ima), ncomps_(ncomps) { initialize(separators_, ima); // FIXME: do we really want that? mln::data::fill(separators_, false); - fill_infos(attribs); + fill_infos(attribs, type); } template <typename L> inline component_set_data<L>::component_set_data(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs) + const mln::util::array<pair_data_t>& attribs, + component::Type type) : ima_(ima), ncomps_(ncomps) { initialize(separators_, ima); // FIXME: do we really want that? mln::data::fill(separators_, false); - fill_infos(attribs); + fill_infos(attribs, type); } template <typename L> @@ -321,7 +329,8 @@ namespace scribo template <typename L> inline void - component_set_data<L>::fill_infos(const mln::util::array<pair_accu_t>& attribs) + component_set_data<L>::fill_infos(const mln::util::array<pair_accu_t>& attribs, + component::Type type) { typedef mln_site(L) P; @@ -331,7 +340,8 @@ namespace scribo for_all_comp_data(i, attribs) { component_info info(i, attribs[i].first(), - attribs[i].second(), attribs[i].second_accu().nsites()); + attribs[i].second(), attribs[i].second_accu().nsites(), + type); infos_.append(info); } } @@ -339,7 +349,8 @@ namespace scribo template <typename L> inline void - component_set_data<L>::fill_infos(const mln::util::array<pair_data_t>& attribs) + component_set_data<L>::fill_infos(const mln::util::array<pair_data_t>& attribs, + component::Type type) { typedef mln_site(L) P; @@ -349,7 +360,8 @@ namespace scribo for_all_comp_data(i, attribs) { component_info info(i, attribs[i].first, - attribs[i].second.first, attribs[i].second.second); + attribs[i].second.first, attribs[i].second.second, + type); infos_.append(info); } } @@ -397,9 +409,10 @@ namespace scribo template <typename L> inline component_set<L>::component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_accu_t>& attribs) + const mln::util::array<pair_accu_t>& attribs, + component::Type type) { - data_ = new internal::component_set_data<L>(ima, ncomps, attribs); + data_ = new internal::component_set_data<L>(ima, ncomps, attribs, type); } @@ -407,9 +420,10 @@ namespace scribo inline component_set<L>::component_set(const L& ima, const mln_value(L)& ncomps, - const mln::util::array<pair_data_t>& attribs) + const mln::util::array<pair_data_t>& attribs, + component::Type type) { - data_ = new internal::component_set_data<L>(ima, ncomps, attribs); + data_ = new internal::component_set_data<L>(ima, ncomps, attribs, type); } diff --git a/scribo/scribo/core/document.hh b/scribo/scribo/core/document.hh index e5ac825..ef0869e 100644 --- a/scribo/scribo/core/document.hh +++ b/scribo/scribo/core/document.hh @@ -297,7 +297,8 @@ namespace scribo mln_value(L) ncomps; whitespace_seps_comps_ = primitive::extract::components(whitespace_seps, - mln::c8(), ncomps); + mln::c8(), ncomps, + component::WhitespaceSeparator); } @@ -333,7 +334,8 @@ namespace scribo mln_value(L) ncomps; line_seps_comps_ = primitive::extract::components(line_seps, - mln::c8(), ncomps); + mln::c8(), ncomps, + component::LineSeparator); } diff --git a/scribo/scribo/core/tag/component.hh b/scribo/scribo/core/tag/component.hh index 10b86a6..7cd2ede 100644 --- a/scribo/scribo/core/tag/component.hh +++ b/scribo/scribo/core/tag/component.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -55,7 +55,8 @@ namespace scribo { Undefined = 0, Character, - Separator, + LineSeparator, + WhitespaceSeparator, Noise, Punctuation, Image @@ -116,8 +117,11 @@ namespace scribo case Character: str = "Character"; break; - case Separator: - str = "Separator"; + case LineSeparator: + str = "LineSeparator"; + break; + case WhitespaceSeparator: + str = "WhitespaceSeparator"; break; case Noise: str = "Noise"; @@ -139,8 +143,10 @@ namespace scribo { if (str == "Character") return Character; - else if (str == "Separator") - return Separator; + else if (str == "LineSeparator") + return LineSeparator; + else if (str == "WhitespaceSeparator") + return WhitespaceSeparator; else if (str == "Noise") return Noise; else if (str == "Punctuation") diff --git a/scribo/scribo/primitive/extract/components.hh b/scribo/scribo/primitive/extract/components.hh index 4994d4b..849dd7b 100644 --- a/scribo/scribo/primitive/extract/components.hh +++ b/scribo/scribo/primitive/extract/components.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2009, 2011 EPITA Research and Development Laboratory +// (LRDE) // // This file is part of Olena. // @@ -68,6 +69,7 @@ namespace scribo /// and background to 'false'. /// \param[in] nbh A neighborhood to be used for labeling. /// \param[in,out] ncomponents Will store the numbers of components found. + /// \param[in] type The default component type set to components. /// /// \return An image of labeled components. // @@ -75,7 +77,8 @@ namespace scribo inline component_set<mln_ch_value(I,V)> components(const Image& input, - const Neighborhood<N>& nbh, V& ncomponents); + const Neighborhood<N>& nbh, V& ncomponents, + component::Type type = component::Undefined); # ifndef MLN_INCLUDE_ONLY @@ -88,7 +91,8 @@ namespace scribo inline void components_tests(const Image& input, - const Neighborhood<N>& nbh, V& ncomponents) + const Neighborhood<N>& nbh, V& ncomponents, + component::Type type) { mlc_equal(mln_value(I),bool)::check(); // mlc_is_a(V, mln::value::Symbolic)::check(); @@ -97,6 +101,7 @@ namespace scribo (void) input; (void) nbh; (void) ncomponents; + (void) type; } @@ -107,11 +112,12 @@ namespace scribo inline component_set<mln_ch_value(I,V)> components(const Image& input, - const Neighborhood<N>& nbh, V& ncomponents) + const Neighborhood<N>& nbh, V& ncomponents, + component::Type type = component::Undefined) { trace::entering("scribo::components"); - internal::components_tests(input, nbh, ncomponents); + internal::components_tests(input, nbh, ncomponents, type); typedef mln_ch_value(I,V) L; typedef mln::accu::shape::bbox<mln_site(L)> bbox_accu_t; @@ -129,7 +135,7 @@ namespace scribo pair_accu_t()); component_set<L> - output(results.first(), ncomponents, results.second().second()); + output(results.first(), ncomponents, results.second().second(), type); trace::exiting("scribo::components"); return output; diff --git a/scribo/scribo/primitive/identify.hh b/scribo/scribo/primitive/identify.hh index 81a7d16..1bed712 100644 --- a/scribo/scribo/primitive/identify.hh +++ b/scribo/scribo/primitive/identify.hh @@ -61,7 +61,7 @@ namespace scribo std::swap(min, max); if (max/min > 10) - output(c).update_type(component::Separator); + output(c).update_type(component::LineSeparator); } mln::trace::exiting("scribo::primitive::identify"); -- 1.5.6.5

13 years, 10 months

last-svn-commit-777-g6985484 Make XML output more flexible.

by Guillaume Lazzara

* scribo/core/component_info.hh, * scribo/core/component_set.hh, * scribo/core/document.hh, * scribo/core/line_info.hh, * scribo/core/line_links.hh, * scribo/core/object_groups.hh, * scribo/core/object_links.hh, * scribo/core/paragraph_set.hh: Make these classes serializable. * scribo/core/concept/serializable.hh, * scribo/core/concept/serialize_visitor.hh: New concepts. * scribo/core/internal/doc_xml_serializer.hh: New. Base implementation. * scribo/io/xml/internal/extended_page_xml_visitor.hh, * scribo/io/xml/internal/full_xml_visitor.hh, * scribo/io/xml/internal/page_xml_visitor.hh: New. Visitors producing different XML outputs. * scribo/io/xml/internal/html_markups_replace.hh, * scribo/io/xml/internal/print_box_coords.hh, * scribo/io/xml/internal/print_page_preambule.hh: New. Tools for XML output. * scribo/io/xml/save.hh: Make use of visitors. * scribo/toolchain/internal/content_in_doc_functor.hh: Set default XML output type. * src/content_in_doc.cc: Produce several XML output. --- scribo/ChangeLog | 38 ++ scribo/demo/viewer/runner.cc | 5 +- scribo/scribo/core/component_info.hh | 3 +- scribo/scribo/core/component_set.hh | 7 +- scribo/scribo/core/concept/serializable.hh | 64 +++ scribo/scribo/core/concept/serialize_visitor.hh | 49 +++ scribo/scribo/core/document.hh | 8 +- scribo/scribo/core/internal/doc_xml_serializer.hh | 140 ++++++ scribo/scribo/core/line_info.hh | 21 +- scribo/scribo/core/line_links.hh | 3 +- scribo/scribo/core/object_groups.hh | 4 +- scribo/scribo/core/object_links.hh | 8 +- scribo/scribo/core/paragraph_set.hh | 4 +- .../io/xml/internal/extended_page_xml_visitor.hh | 283 ++++++++++++ scribo/scribo/io/xml/internal/full_xml_visitor.hh | 456 ++++++++++++++++++++ .../scribo/io/xml/internal/html_markups_replace.hh | 97 +++++ scribo/scribo/io/xml/internal/page_xml_visitor.hh | 222 ++++++++++ scribo/scribo/io/xml/internal/print_box_coords.hh | 92 ++++ .../scribo/io/xml/internal/print_page_preambule.hh | 95 ++++ scribo/scribo/io/xml/save.hh | 388 +++-------------- .../toolchain/internal/content_in_doc_functor.hh | 9 +- scribo/src/content_in_doc.cc | 4 +- 22 files changed, 1660 insertions(+), 340 deletions(-) create mode 100644 scribo/scribo/core/concept/serializable.hh create mode 100644 scribo/scribo/core/concept/serialize_visitor.hh create mode 100644 scribo/scribo/core/internal/doc_xml_serializer.hh create mode 100644 scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh create mode 100644 scribo/scribo/io/xml/internal/full_xml_visitor.hh create mode 100644 scribo/scribo/io/xml/internal/html_markups_replace.hh create mode 100644 scribo/scribo/io/xml/internal/page_xml_visitor.hh create mode 100644 scribo/scribo/io/xml/internal/print_box_coords.hh create mode 100644 scribo/scribo/io/xml/internal/print_page_preambule.hh diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 330338a..6571137 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,43 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make XML output more flexible. + + * scribo/core/component_info.hh, + * scribo/core/component_set.hh, + * scribo/core/document.hh, + * scribo/core/line_info.hh, + * scribo/core/line_links.hh, + * scribo/core/object_groups.hh, + * scribo/core/object_links.hh, + * scribo/core/paragraph_set.hh: Make these classes serializable. + + * scribo/core/concept/serializable.hh, + * scribo/core/concept/serialize_visitor.hh: New concepts. + + * scribo/core/internal/doc_xml_serializer.hh: New. Base + implementation. + + * scribo/io/xml/internal/extended_page_xml_visitor.hh, + * scribo/io/xml/internal/full_xml_visitor.hh, + * scribo/io/xml/internal/page_xml_visitor.hh: New. Visitors + producing different XML outputs. + + * scribo/io/xml/internal/html_markups_replace.hh, + * scribo/io/xml/internal/print_box_coords.hh, + * scribo/io/xml/internal/print_page_preambule.hh: New. Tools for + XML output. + + * scribo/io/xml/save.hh: Make use of visitors. + + * scribo/toolchain/internal/content_in_doc_functor.hh: Set default + XML output type. + + * src/content_in_doc.cc: Produce several XML output. + + * demo/viewer/runner.cc: Update call to io::xml::save. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Set component type during component extraction. * scribo/core/component_info.hh, diff --git a/scribo/demo/viewer/runner.cc b/scribo/demo/viewer/runner.cc index 86ff5dc..a3cc883 100644 --- a/scribo/demo/viewer/runner.cc +++ b/scribo/demo/viewer/runner.cc @@ -1,4 +1,5 @@ -// Copyright (C) 2010 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2010, 2011 EPITA Research and Development Laboratory +// (LRDE) // // This file is part of Olena. // @@ -156,7 +157,7 @@ void runner::process(const image2d<value::rgb8>& original_ima, f.enable_whitespace_seps = (find_seps == defs::Whitespaces || find_seps == defs::LinesAndWhitespaces); - f.allow_xml_extensions = true; + f.xml_format = scribo::io::xml::PageExtended; f.save_doc_as_xml = true; diff --git a/scribo/scribo/core/component_info.hh b/scribo/scribo/core/component_info.hh index 6fc73f8..f825aee 100644 --- a/scribo/scribo/core/component_info.hh +++ b/scribo/scribo/core/component_info.hh @@ -36,6 +36,7 @@ # include <mln/core/alias/point2d.hh> # include <mln/util/object_id.hh> +# include <scribo/core/concept/serializable.hh> # include <scribo/core/tag/component.hh> # include <scribo/core/tag/line.hh> @@ -44,7 +45,7 @@ namespace scribo typedef mln::util::object_id<scribo::ComponentId, unsigned> component_id_t; - class component_info + class component_info : public Serializable<component_info> { typedef mln::util::object_id<scribo::ComponentId, unsigned> component_id_t; diff --git a/scribo/scribo/core/component_set.hh b/scribo/scribo/core/component_set.hh index 442e8d6..a63ed6c 100644 --- a/scribo/scribo/core/component_set.hh +++ b/scribo/scribo/core/component_set.hh @@ -30,6 +30,10 @@ /// \file /// /// \brief Definition of a component set. +/// +/// \fixme component_set should always set a component type in order +/// to be fully supported by visitors. + # include <mln/core/concept/site_set.hh> # include <mln/core/concept/function.hh> @@ -59,6 +63,7 @@ # include <scribo/core/macros.hh> # include <scribo/core/component_info.hh> +# include <scribo/core/concept/serializable.hh> namespace scribo @@ -115,7 +120,7 @@ namespace scribo template <typename L> - class component_set + class component_set : public Serializable<component_set<L> > { typedef mln::accu::shape::bbox<mln_site(L)> bbox_accu_t; typedef mln::accu::center<mln_site(L)> center_accu_t; diff --git a/scribo/scribo/core/concept/serializable.hh b/scribo/scribo/core/concept/serializable.hh new file mode 100644 index 0000000..6e661a6 --- /dev/null +++ b/scribo/scribo/core/concept/serializable.hh @@ -0,0 +1,64 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_CORE_CONCEPT_SERIALIZABLE_HH +# define SCRIBO_CORE_CONCEPT_SERIALIZABLE_HH + +/// \file +/// +/// Concept for serializer visitors. + +# include <mln/core/concept/object.hh> +# include <scribo/core/concept/serialize_visitor.hh> + +namespace scribo +{ + + /// \brief Link functor concept. + template <typename E> + class Serializable : public mln::Object<E> + { + public: + template <typename E2> + void accept(const SerializeVisitor<E2>& visitor) const; + }; + + +# ifndef MLN_INCLUDE_ONLY + + template <typename E> + template <typename E2> + void + Serializable<E>::accept(const SerializeVisitor<E2>& visitor) const + { + exact(visitor).visit(exact(*this)); + } + +# endif // ! MLN_INCLUDE_ONLY + + +} // end of namespace scribo + +#endif // SCRIBO_CORE_CONCEPT_SERIALIZABLE_HH diff --git a/scribo/scribo/core/concept/serialize_visitor.hh b/scribo/scribo/core/concept/serialize_visitor.hh new file mode 100644 index 0000000..e5e598f --- /dev/null +++ b/scribo/scribo/core/concept/serialize_visitor.hh @@ -0,0 +1,49 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_CORE_CONCEPT_SERIALIZE_VISITOR_HH +# define SCRIBO_CORE_CONCEPT_SERIALIZE_VISITOR_HH + +/// \file +/// +/// Concept for serializer visitors. + +# include <mln/core/concept/object.hh> + +namespace scribo +{ + + /// \brief Link functor concept. + template <typename E> + class SerializeVisitor : public mln::Object<E> + { + public: + // void visit(..); + }; + + +} // end of namespace scribo + +#endif // SCRIBO_CORE_CONCEPT_SERIALIZE_VISITOR_HH diff --git a/scribo/scribo/core/document.hh b/scribo/scribo/core/document.hh index ef0869e..372f0a4 100644 --- a/scribo/scribo/core/document.hh +++ b/scribo/scribo/core/document.hh @@ -40,13 +40,15 @@ # include <scribo/core/line_set.hh> # include <scribo/core/paragraph_set.hh> +# include <scribo/core/concept/serializable.hh> + # include <scribo/primitive/extract/components.hh> namespace scribo { template <typename L> - struct document + struct document : public Serializable<document<L> > { public: @@ -98,7 +100,7 @@ namespace scribo private: - const char *filename_; + std::string filename_; mln::image2d<mln::value::rgb8> image_; paragraph_set<L> parset_; @@ -142,7 +144,7 @@ namespace scribo const char * document<L>::filename() const { - return filename_; + return filename_.c_str(); } diff --git a/scribo/scribo/core/internal/doc_xml_serializer.hh b/scribo/scribo/core/internal/doc_xml_serializer.hh new file mode 100644 index 0000000..b64c9d4 --- /dev/null +++ b/scribo/scribo/core/internal/doc_xml_serializer.hh @@ -0,0 +1,140 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_CORE_INTERNAL_DOC_XML_SERIALIZER_HH +# define SCRIBO_CORE_INTERNAL_DOC_XML_SERIALIZER_HH + +/// \file +/// +/// Concept for serializer visitors. + +# include <scribo/core/concept/serialize_visitor.hh> + +# include <scribo/core/document.hh> +# include <scribo/core/component_set.hh> +# include <scribo/core/component_info.hh> +# include <scribo/core/paragraph_set.hh> +# include <scribo/core/object_groups.hh> +# include <scribo/core/object_links.hh> +# include <scribo/core/line_links.hh> +# include <scribo/core/line_info.hh> + +namespace scribo +{ + + /// \brief Link functor concept. + template <typename E> + class doc_xml_serializer : public SerializeVisitor<E> + { + public: + // Visit overloads + template <typename L> + void visit(const document<L>& doc) const; + + template <typename L> + void visit(const line_links<L>& llinks) const; + + template <typename L> + void visit(const object_groups<L>& groups) const; + + template <typename L> + void visit(const object_links<L>& links) const; + + template <typename L> + void visit(const component_set<L>& comp_set) const; + + void visit(const component_info& info) const; + + template <typename L> + void visit(const paragraph_set<L>& parset) const; + + template <typename L> + void visit(const line_info<L>& line) const; + }; + + +# ifndef MLN_INCLUDE_ONLY + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const document<L>& doc) const + { + } + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const line_links<L>& llinks) const + { + } + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const object_groups<L>& groups) const + { + } + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const object_links<L>& links) const + { + } + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const component_set<L>& comp_set) const + { + } + + template <typename E> + void + doc_xml_serializer<E>::visit(const component_info& info) const + { + } + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const paragraph_set<L>& parset) const + { + } + + template <typename E> + template <typename L> + void + doc_xml_serializer<E>::visit(const line_info<L>& line) const + { + } + +# endif // ! MLN_INCLUDE_ONLY + + +} // end of namespace scribo + +#endif // SCRIBO_CORE_INTERNAL_DOC_XML_SERIALIZER_HH diff --git a/scribo/scribo/core/line_info.hh b/scribo/scribo/core/line_info.hh index c82160a..33a1529 100644 --- a/scribo/scribo/core/line_info.hh +++ b/scribo/scribo/core/line_info.hh @@ -53,6 +53,11 @@ # include <scribo/core/line_set.hh> # include <scribo/core/component_set.hh> +# include <scribo/io/xml/internal/html_markups_replace.hh> + +# include <scribo/core/concept/serializable.hh> + + namespace scribo { @@ -114,6 +119,7 @@ namespace scribo bool indented_; std::string text_; + std::string html_text_; // Line set holding this element. line_set<L> holder_; @@ -125,7 +131,7 @@ namespace scribo template <typename L> - class line_info + class line_info : public Serializable<line_info<L> > { typedef internal::line_info_data<L> data_t; typedef mln::util::object_id<scribo::ComponentId, unsigned> component_id_t; @@ -198,6 +204,7 @@ namespace scribo bool has_text() const; const std::string& text() const; + const std::string& html_text() const; void update_text(const std::string& str); bool is_valid() const; @@ -604,6 +611,7 @@ namespace scribo return data_->indented_; } + template <typename L> bool line_info<L>::has_text() const @@ -611,6 +619,7 @@ namespace scribo return !data_->text_.empty(); } + template <typename L> const std::string& line_info<L>::text() const @@ -620,10 +629,19 @@ namespace scribo template <typename L> + const std::string& + line_info<L>::html_text() const + { + return data_->html_text_; + } + + + template <typename L> void line_info<L>::update_text(const std::string& str) { data_->text_ = str; + data_->html_text_ = scribo::io::xml::internal::html_markups_replace(str); } @@ -987,6 +1005,7 @@ namespace scribo << ", indented=" << info.indented() << ", hidden=" << info.is_hidden() << ", text=" << info.text() + << ", html_text=" << info.html_text() << ")" << std::endl; } diff --git a/scribo/scribo/core/line_links.hh b/scribo/scribo/core/line_links.hh index de62158..fdd09a5 100644 --- a/scribo/scribo/core/line_links.hh +++ b/scribo/scribo/core/line_links.hh @@ -34,6 +34,7 @@ # include <mln/util/array.hh> # include <mln/util/tracked_ptr.hh> +# include <scribo/core/concept/serializable.hh> # include <scribo/core/line_set.hh> @@ -69,7 +70,7 @@ namespace scribo /// \brief Line group representation. // template <typename L> - class line_links + class line_links : public Serializable<line_links<L> > { typedef internal::line_links_data<L> data_t; diff --git a/scribo/scribo/core/object_groups.hh b/scribo/scribo/core/object_groups.hh index 9d9fb25..bbfaf6e 100644 --- a/scribo/scribo/core/object_groups.hh +++ b/scribo/scribo/core/object_groups.hh @@ -36,6 +36,8 @@ # include <scribo/core/object_links.hh> # include <scribo/core/component_set.hh> +# include <scribo/core/concept/serializable.hh> + namespace scribo { @@ -69,7 +71,7 @@ namespace scribo /// \brief Object group representation. // template <typename L> - class object_groups + class object_groups : public Serializable<object_groups<L> > { typedef internal::object_groups_data<L> data_t; diff --git a/scribo/scribo/core/object_links.hh b/scribo/scribo/core/object_links.hh index af7dc38..2c2eea1 100644 --- a/scribo/scribo/core/object_links.hh +++ b/scribo/scribo/core/object_links.hh @@ -1,5 +1,5 @@ -// Copyright (C) 2009, 2010 EPITA Research and Development Laboratory -// (LRDE) +// Copyright (C) 2009, 2010, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -37,6 +37,8 @@ # include <scribo/core/component_set.hh> +# include <scribo/core/concept/serializable.hh> + namespace scribo { @@ -70,7 +72,7 @@ namespace scribo /// \brief Object group representation. // template <typename L> - class object_links + class object_links : public Serializable<object_links<L> > { typedef internal::object_links_data<L> data_t; diff --git a/scribo/scribo/core/paragraph_set.hh b/scribo/scribo/core/paragraph_set.hh index 6597189..5451069 100644 --- a/scribo/scribo/core/paragraph_set.hh +++ b/scribo/scribo/core/paragraph_set.hh @@ -33,6 +33,8 @@ # include <scribo/core/line_set.hh> # include <scribo/core/paragraph_info.hh> +# include <scribo/core/concept/serializable.hh> + namespace scribo { @@ -61,7 +63,7 @@ namespace scribo */ template <typename L> - class paragraph_set + class paragraph_set : public Serializable<paragraph_set<L> > { public: paragraph_set(); diff --git a/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh new file mode 100644 index 0000000..5d8a672 --- /dev/null +++ b/scribo/scribo/io/xml/internal/extended_page_xml_visitor.hh @@ -0,0 +1,283 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_INTERNAL_EXTENDED_PAGE_XML_VISITOR_HH +# define SCRIBO_IO_XML_INTERNAL_EXTENDED_PAGE_XML_VISITOR_HH + +/// \file +/// +/// Extended XML PAGE format serializer Visitor. + +# include <fstream> +# include <scribo/core/internal/doc_xml_serializer.hh> +# include <scribo/core/document.hh> +# include <scribo/core/component_set.hh> +# include <scribo/core/paragraph_set.hh> +# include <scribo/core/object_groups.hh> +# include <scribo/core/object_links.hh> +# include <scribo/core/line_links.hh> +# include <scribo/core/line_info.hh> + +# include <scribo/convert/to_base64.hh> + +# include <scribo/io/xml/internal/print_box_coords.hh> +# include <scribo/io/xml/internal/print_page_preambule.hh> + + +namespace scribo +{ + + namespace io + { + + namespace xml + { + + namespace internal + { + + + class extended_page_xml_visitor : public doc_xml_serializer<extended_page_xml_visitor> + { + public: + // Constructor + extended_page_xml_visitor(std::ofstream& out); + + // Visit overloads + template <typename L> + void visit(const document<L>& doc) const; + + template <typename L> + void visit(const component_set<L>& comp_set) const; + + void visit(const component_info& info) const; + + template <typename L> + void visit(const paragraph_set<L>& parset) const; + + template <typename L> + void visit(const line_info<L>& line) const; + + private: // Attributes + std::ofstream& output; + }; + + + +# ifndef MLN_INCLUDE_ONLY + + + inline + extended_page_xml_visitor::extended_page_xml_visitor(std::ofstream& out) + : output(out) + { + } + + + + /// Document + // + template <typename L> + void + extended_page_xml_visitor::visit(const document<L>& doc) const + { + // Preambule + print_PAGE_preambule(output, doc, false); + + // Text + if (doc.has_text()) + doc.paragraphs().accept(*this); + + + // Page elements (Pictures, ...) + if (doc.has_elements()) + doc.elements().accept(*this); + + // Whitespace seraparators + if (doc.has_whitespace_seps()) + doc.whitespace_seps_comps().accept(*this); + + output << " </page>" << std::endl; + output << "</pcGts>" << std::endl; + + } + + /// Component Set + // + template <typename L> + void + extended_page_xml_visitor::visit(const component_set<L>& comp_set) const + { + for_all_comps(c, comp_set) + if (comp_set(c).is_valid()) + comp_set(c).accept(*this); + } + + + /// Component_info + // + inline + void + extended_page_xml_visitor::visit(const component_info& info) const + { + switch (info.type()) + { + case component::WhitespaceSeparator: + { + output << " <whitespace_separator_region id=\"wss" + << info.id() + << "\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </whitespace_separator_region>" << std::endl; + break; + } + + case component::LineSeparator: + { + output << " <separator_region id=\"sr" << info.id() + << "\" sep_orientation=\"0.000000\" " + << " sep_colour=\"Black\" " + << " sep_bgcolour=\"White\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </separator_region>" << std::endl; + break; + } + + + default: + case component::Image: + { + output << " <image_region id=\"ir" << info.id() + << "\" img_colour_type=\"24_Bit_Colour\"" + << " img_orientation=\"0.000000\" " + << " img_emb_text=\"No\" " + << " img_bgcolour=\"White\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </image_region>" << std::endl; + break; + } + } + } + + + /// Paragraph Set + // + template <typename L> + void + extended_page_xml_visitor::visit(const paragraph_set<L>& parset) const + { + const line_set<L>& lines = parset.lines(); + + for_all_paragraphs(p, parset) + { + const mln::util::array<line_id_t>& line_ids = parset(p).line_ids(); + + // FIXME: compute that information on the whole paragraph + // and use them here. + line_id_t fid = line_ids(0); + output << " <text_region id=\"" << p + << "\" txt_orientation=\"" << lines(fid).orientation() + << "\" txt_reading_orientation=\"" << lines(fid).reading_orientation() + << "\" txt_reading_direction=\"" << lines(fid).reading_direction() + << "\" txt_text_type=\"" << lines(fid).type() + << "\" txt_reverse_video=\"" << (lines(fid).reverse_video() ? "true" : "false") + << "\" txt_indented=\"" << (lines(fid).indented() ? "true" : "false") + << "\" kerning=\"" << lines(fid).char_space(); + + // EXTENSIONS - Not officially supported + output << "\" baseline=\"" << lines(fid).baseline() + << "\" meanline=\"" << lines(fid).meanline() + << "\" x_height=\"" << lines(fid).x_height() + << "\" d_height=\"" << lines(fid).d_height() + << "\" a_height=\"" << lines(fid).a_height() + << "\" char_width=\"" << lines(fid).char_width(); + // End of EXTENSIONS + output << "\">" + << std::endl; + + internal::print_box_coords(output, parset(p).bbox(), " "); + + // EXTENSIONS - Not officially supported + for_all_paragraph_lines(lid, line_ids) + { + line_id_t l = line_ids(lid); + lines(l).accept(*this); + } + // End of EXTENSIONS + + output << " </text_region>" << std::endl; + } + } + + + template <typename L> + void + extended_page_xml_visitor::visit(const line_info<L>& line) const + { + if (line.has_text()) + { + output << " <line text=\"" << line.html_text() << "\" "; + } + else + output << " <line " << std::endl; + + output << "id=\"" << line.id() + << "\" txt_orientation=\"" << line.orientation() + << "\" txt_reading_orientation=\"" << line.reading_orientation() + << "\" txt_reading_direction=\"" << line.reading_direction() + << "\" txt_text_type=\"" << line.type() + << "\" txt_reverse_video=\"" << (line.reverse_video() ? "true" : "false") + << "\" txt_indented=\"" << (line.indented() ? "true" : "false") + << "\" kerning=\"" << line.char_space() + << "\" baseline=\"" << line.baseline() + << "\" meanline=\"" << line.meanline() + << "\" x_height=\"" << line.x_height() + << "\" d_height=\"" << line.d_height() + << "\" a_height=\"" << line.a_height() + << "\" char_width=\"" << line.char_width() + << "\">" << std::endl; + + internal::print_box_coords(output, line.bbox(), " "); + + output << " </line>" << std::endl; + } + +#endif // MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml::internal + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + +#endif // SCRIBO_IO_XML_INTERNAL_EXTENDED_PAGE_XML_VISITOR_HH diff --git a/scribo/scribo/io/xml/internal/full_xml_visitor.hh b/scribo/scribo/io/xml/internal/full_xml_visitor.hh new file mode 100644 index 0000000..9c5bd1d --- /dev/null +++ b/scribo/scribo/io/xml/internal/full_xml_visitor.hh @@ -0,0 +1,456 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_INTERNAL_FULL_XML_VISITOR_HH +# define SCRIBO_IO_XML_INTERNAL_FULL_XML_VISITOR_HH + +/// \file +/// +/// XML serializer Visitor. + +# include <fstream> +# include <scribo/core/internal/doc_xml_serializer.hh> +# include <scribo/core/document.hh> +# include <scribo/core/component_set.hh> +# include <scribo/core/paragraph_set.hh> +# include <scribo/core/object_groups.hh> +# include <scribo/core/object_links.hh> +# include <scribo/core/line_links.hh> +# include <scribo/core/line_info.hh> + +# include <scribo/convert/to_base64.hh> + +# include <scribo/io/xml/internal/print_box_coords.hh> +# include <scribo/io/xml/internal/print_page_preambule.hh> + + +namespace scribo +{ + + namespace io + { + + namespace xml + { + + namespace internal + { + + + class full_xml_visitor : public doc_xml_serializer<full_xml_visitor> + { + public: + // Constructor + full_xml_visitor(std::ofstream& out); + + // Visit overloads + template <typename L> + void visit(const document<L>& doc) const; + + template <typename L> + void visit(const line_links<L>& llinks) const; + + template <typename L> + void visit(const object_groups<L>& groups) const; + + template <typename L> + void visit(const object_links<L>& links) const; + + template <typename L> + void visit(const component_set<L>& comp_set) const; + + void visit(const component_info& info) const; + + template <typename L> + void visit(const paragraph_set<L>& parset) const; + + template <typename L> + void visit(const line_info<L>& line) const; + + private: // Attributes + std::ofstream& output; + }; + + + +# ifndef MLN_INCLUDE_ONLY + + + inline + full_xml_visitor::full_xml_visitor(std::ofstream& out) + : output(out) + { + } + + + + /// Document + // + template <typename L> + void + full_xml_visitor::visit(const document<L>& doc) const + { + print_PAGE_preambule(output, doc, false); + + // Text + if (doc.has_text()) + { + const line_set<L>& lines = doc.lines(); + + // Save component/link/group information (Extension) + { + // Component set + lines.components().accept(*this); + + // Object link + lines.links().accept(*this); + + // Object group + lines.groups().accept(*this); + } + // End of EXTENSIONS + + const paragraph_set<L>& parset = doc.paragraphs(); + + // Save paragraphs related information (Extension) + { + // General text information + output << " <text_data nlines=\"" << lines.nelements() << "\" " + << " nparagraphs=\"" << parset.nelements() << "\" />" << std::endl; + + // line_links + parset.links().accept(*this); + } + + // Paragraph and lines + parset.accept(*this); + } + + + // Page elements (Pictures, ...) + if (doc.has_elements()) + { + const component_set<L>& elts = doc.elements(); + for_all_comps(e, elts) + if (elts(e).is_valid()) + elts(e).accept(*this); + } + + + // line seraparators + if (doc.has_line_seps()) + { + const component_set<L>& + line_seps_comps = doc.line_seps_comps(); + + for_all_comps(c, line_seps_comps) + line_seps_comps(c).accept(*this); + } + + + // Whitespace seraparators + if (doc.has_whitespace_seps()) + { + const component_set<L>& + whitespace_seps_comps = doc.whitespace_seps_comps(); + + for_all_comps(c, whitespace_seps_comps) + whitespace_seps_comps(c).accept(*this); + } + + output << " </page>" << std::endl; + output << "</pcGts>" << std::endl; + + } + + + /// Line Links + // + template <typename L> + void + full_xml_visitor::visit(const line_links<L>& llinks) const + { + output << " <line_links>" << std::endl; + for_all_links(l, llinks) + { + output << " <line_link" + << " from=\"" << l + << "\" to=\"" << llinks(l) + << "\"/>" << std::endl; + } + output << " </line_links>" << std::endl; + } + + + /// Object Groups + // + template <typename L> + void + full_xml_visitor::visit(const object_groups<L>& groups) const + { + output << " <object_groups>" << std::endl; + for_all_groups(g, groups) + { + output << " <group " + << " object_id=\"" << g + << "\" group_id=\"" << groups(g) + << "\"/>" << std::endl; + } + output << " </object_groups>" << std::endl; + } + + + /// Object Links + // + template <typename L> + void + full_xml_visitor::visit(const object_links<L>& links) const + { + output << " <object_links>" << std::endl; + for_all_links(l, links) + { + output << " <link" + << " from=\"" << l + << "\" to=\"" << links(l) + << "\"/>" << std::endl; + } + output << " </object_links>" << std::endl; + } + + + /// Component Set + // + template <typename L> + void + full_xml_visitor::visit(const component_set<L>& comp_set) const + { + output << " <component_set nelements=\"" << comp_set.nelements() + << "\">" << std::endl; + for_all_comps(c, comp_set) + { + output << " <component_info" + << " id=\"" << comp_set(c).id() + << "\" mass_center_x=\"" << comp_set(c).mass_center().col() + << "\" mass_center_y=\"" << comp_set(c).mass_center().row() + << "\" card=\"" << comp_set(c).card() + << "\" tag=\"" << comp_set(c).tag() + << "\" type=\"" << comp_set(c).type() + << "\" pmin_x=\"" << comp_set(c).bbox().pmin().col() + << "\" pmin_y=\"" << comp_set(c).bbox().pmin().row() + << "\" pmax_x=\"" << comp_set(c).bbox().pmax().col() + << "\" pmax_y=\"" << comp_set(c).bbox().pmax().row() + << "\"/>" << std::endl; + } + + + // Save labeled image + { + const L& lbl = comp_set.labeled_image(); + output << "<labeled_image " + << " height=\"" << lbl.domain().height() + << "\" width=\"" << lbl.domain().width() << "\">" + << "<![CDATA["; + + util::array<unsigned char> lbl64; + convert::to_base64(lbl, lbl64); + output.write((const char *)lbl64.std_vector().data(), + lbl64.nelements()); + + output << "]]></labeled_image>" << std::endl; + } + + // Save separators image + { + const mln_ch_value(L,bool)& seps = comp_set.separators(); + output << "<separators_image " + << " height=\"" << seps.domain().height() + << "\" width=\"" << seps.domain().width() << "\">" + << "<![CDATA["; + + util::array<unsigned char> seps64; + convert::to_base64(seps, seps64); + output.write((const char *)seps64.std_vector().data(), + seps64.nelements()); + + output << "]]></separators_image>" << std::endl; + } + + output << "</component_set>" << std::endl; + } + + + /// Component_info + // + inline + void + full_xml_visitor::visit(const component_info& info) const + { + switch (info.type()) + { + case component::WhitespaceSeparator: + { + output << " <whitespace_separator_region id=\"wss" + << info.id() + << "\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </whitespace_separator_region>" << std::endl; + break; + } + + case component::LineSeparator: + { + output << " <separator_region id=\"sr" << info.id() + << "\" sep_orientation=\"0.000000\" " + << " sep_colour=\"Black\" " + << " sep_bgcolour=\"White\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </separator_region>" << std::endl; + break; + } + + + default: + case component::Image: + { + output << " <image_region id=\"ir" << info.id() + << "\" img_colour_type=\"24_Bit_Colour\"" + << " img_orientation=\"0.000000\" " + << " img_emb_text=\"No\" " + << " img_bgcolour=\"White\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </image_region>" << std::endl; + break; + } + } + } + + /// Paragraph Set + // + template <typename L> + void + full_xml_visitor::visit(const paragraph_set<L>& parset) const + { + const line_set<L>& lines = parset.lines(); + + for_all_paragraphs(p, parset) + { + const mln::util::array<line_id_t>& line_ids = parset(p).line_ids(); + + // FIXME: compute that information on the whole paragraph + // and use them here. + line_id_t fid = line_ids(0); + output << " <text_region id=\"" << p + << "\" txt_orientation=\"" << lines(fid).orientation() + << "\" txt_reading_orientation=\"" << lines(fid).reading_orientation() + << "\" txt_reading_direction=\"" << lines(fid).reading_direction() + << "\" txt_text_type=\"" << lines(fid).type() + << "\" txt_reverse_video=\"" << (lines(fid).reverse_video() ? "true" : "false") + << "\" txt_indented=\"" << (lines(fid).indented() ? "true" : "false") + << "\" kerning=\"" << lines(fid).char_space(); + + // EXTENSIONS - Not officially supported + output << "\" baseline=\"" << lines(fid).baseline() + << "\" meanline=\"" << lines(fid).meanline() + << "\" x_height=\"" << lines(fid).x_height() + << "\" d_height=\"" << lines(fid).d_height() + << "\" a_height=\"" << lines(fid).a_height() + << "\" char_width=\"" << lines(fid).char_width(); + // End of EXTENSIONS + output << "\">" + << std::endl; + + internal::print_box_coords(output, parset(p).bbox(), " "); + + + // EXTENSIONS - Not officially supported + for_all_paragraph_lines(lid, line_ids) + { + line_id_t l = line_ids(lid); + + lines(l).accept(*this); + } + + output << " </text_region>" << std::endl; + } + } + + + template <typename L> + void + full_xml_visitor::visit(const line_info<L>& line) const + { + if (line.has_text()) + { + output << " <line text=\"" << line.html_text() << "\" "; + } + else + output << " <line " << std::endl; + + output << "id=\"" << line.id() + << "\" txt_orientation=\"" << line.orientation() + << "\" txt_reading_orientation=\"" << line.reading_orientation() + << "\" txt_reading_direction=\"" << line.reading_direction() + << "\" txt_text_type=\"" << line.type() + << "\" txt_reverse_video=\"" << (line.reverse_video() ? "true" : "false") + << "\" txt_indented=\"" << (line.indented() ? "true" : "false") + << "\" kerning=\"" << line.char_space() + << "\" baseline=\"" << line.baseline() + << "\" meanline=\"" << line.meanline() + << "\" x_height=\"" << line.x_height() + << "\" d_height=\"" << line.d_height() + << "\" a_height=\"" << line.a_height() + << "\" char_width=\"" << line.char_width() + << "\">" << std::endl; + + internal::print_box_coords(output, line.bbox(), " "); + + output << " <compid_list>" << std::endl; + + for_all_line_comps(c, line.components()) + output << " <compid value=\"" + << line.components()(c) << "\" />" << std::endl; + + output << " </compid_list>" << std::endl; + + output << " </line>" << std::endl; + } + +#endif // MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml::internal + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + +#endif // SCRIBO_IO_XML_INTERNAL_FULL_XML_VISITOR_HH diff --git a/scribo/scribo/io/xml/internal/html_markups_replace.hh b/scribo/scribo/io/xml/internal/html_markups_replace.hh new file mode 100644 index 0000000..76f8107 --- /dev/null +++ b/scribo/scribo/io/xml/internal/html_markups_replace.hh @@ -0,0 +1,97 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_INTERNAL_HTML_MARKUPS_REPLACE_HH +# define SCRIBO_IO_XML_INTERNAL_HTML_MARKUPS_REPLACE_HH + +/// \file +/// +/// \brief Replace HTML markups characters by their corresponding +/// markups. + + +namespace scribo +{ + + namespace io + { + + namespace xml + { + + namespace internal + { + + /*! \brief Replace HTML markups characters by their corresponding + markups. + */ + inline + std::string + html_markups_replace(std::string& input); + + +# ifndef MLN_INCLUDE_ONLY + + static inline std::map<char, std::string> init_map() + { + std::map<char, std::string> html_map; + html_map['\"'] = """; + html_map['<'] = "<"; + html_map['>'] = ">"; + html_map['&'] = "&"; + return html_map; + } + + + inline + std::string + html_markups_replace(const std::string& input) + { + static std::map<char, std::string> map = init_map(); + + std::string output = input; + for (unsigned i = 0; i < input.size(); ++i) + { + std::map<char, std::string>::iterator it = map.find(output.at(i)); + if (it != map.end()) + { + output.replace(i, 1, it->second); + i += it->second.size() - 1; + } + } + return output; + } + +# endif // ! MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml::internal + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + +#endif // ! SCRIBO_IO_XML_INTERNAL_HTML_MARKUPS_REPLACE_HH diff --git a/scribo/scribo/io/xml/internal/page_xml_visitor.hh b/scribo/scribo/io/xml/internal/page_xml_visitor.hh new file mode 100644 index 0000000..52d8f12 --- /dev/null +++ b/scribo/scribo/io/xml/internal/page_xml_visitor.hh @@ -0,0 +1,222 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_INTERNAL_PAGE_XML_VISITOR_HH +# define SCRIBO_IO_XML_INTERNAL_PAGE_XML_VISITOR_HH + +/// \file +/// +/// PAGE format XML serializer Visitor. + +# include <fstream> + +# include <scribo/core/internal/doc_xml_serializer.hh> +# include <scribo/convert/to_base64.hh> + +# include <scribo/io/xml/internal/print_box_coords.hh> +# include <scribo/io/xml/internal/print_page_preambule.hh> + + +namespace scribo +{ + + namespace io + { + + namespace xml + { + + namespace internal + { + + /*! \brief Save document information as XML. + + We use a XML Schema part of the PAGE (Page Analysis and Ground + truth Elements) image representation framework. + + This schema was used in the Page Segmentation COMPetition + (PSCOMP) for ICDAR 2009. + + Its XSD file is located here: + http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16/pagecontent… + + */ + class page_xml_visitor : public doc_xml_serializer<page_xml_visitor> + { + public: + // Constructor + page_xml_visitor(std::ofstream& out); + + // Visit overloads + template <typename L> + void visit(const document<L>& doc) const; + + template <typename L> + void visit(const component_set<L>& comp_set) const; + + void visit(const component_info& info) const; + + template <typename L> + void visit(const paragraph_set<L>& parset) const; + + private: // Attributes + std::ofstream& output; + }; + + + +# ifndef MLN_INCLUDE_ONLY + + + inline + page_xml_visitor::page_xml_visitor(std::ofstream& out) + : output(out) + { + } + + + + /// Document + // + template <typename L> + void + page_xml_visitor::visit(const document<L>& doc) const + { + // Preambule + print_PAGE_preambule(output, doc, true); + + // Text + if (doc.has_text()) + doc.paragraphs().accept(*this); + + // Page elements (Pictures, ...) + if (doc.has_elements()) + doc.elements().accept(*this); + + // line seraparators + if (doc.has_line_seps()) + doc.line_seps_comps().accept(*this); + + output << " </page>" << std::endl; + output << "</pcGts>" << std::endl; + } + + + /// Component Set + // + template <typename L> + void + page_xml_visitor::visit(const component_set<L>& comp_set) const + { + for_all_comps(c, comp_set) + if (comp_set(c).is_valid()) + comp_set(c).accept(*this); + } + + + /// Component_info + // + inline + void + page_xml_visitor::visit(const component_info& info) const + { + switch (info.type()) + { + case component::LineSeparator: + { + output << " <separator_region id=\"sr" << info.id() + << "\" sep_orientation=\"0.000000\" " + << " sep_colour=\"Black\" " + << " sep_bgcolour=\"White\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </separator_region>" << std::endl; + break; + } + + + default: + case component::Image: + { + output << " <image_region id=\"ir" << info.id() + << "\" img_colour_type=\"24_Bit_Colour\"" + << " img_orientation=\"0.000000\" " + << " img_emb_text=\"No\" " + << " img_bgcolour=\"White\">" << std::endl; + + internal::print_box_coords(output, info.bbox(), " "); + + output << " </image_region>" << std::endl; + break; + } + } + } + + + /// Paragraph Set + // + template <typename L> + void + page_xml_visitor::visit(const paragraph_set<L>& parset) const + { + const line_set<L>& lines = parset.lines(); + + for_all_paragraphs(p, parset) + { + const mln::util::array<line_id_t>& line_ids = parset(p).line_ids(); + + // FIXME: compute that information on the whole paragraph + // and use them here. + line_id_t fid = line_ids(0); + output << " <text_region id=\"" << p + << "\" txt_orientation=\"" << lines(fid).orientation() + << "\" txt_reading_orientation=\"" << lines(fid).reading_orientation() + << "\" txt_reading_direction=\"" << lines(fid).reading_direction() + << "\" txt_text_type=\"" << lines(fid).type() + << "\" txt_reverse_video=\"" << (lines(fid).reverse_video() ? "true" : "false") + << "\" txt_indented=\"" << (lines(fid).indented() ? "true" : "false") + << "\" kerning=\"" << lines(fid).char_space() + << "\">" + << std::endl; + + internal::print_box_coords(output, parset(p).bbox(), " "); + + output << " </text_region>" << std::endl; + } + } + + +#endif // MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml::internal + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + +#endif // SCRIBO_IO_XML_INTERNAL_PAGE_XML_VISITOR_HH diff --git a/scribo/scribo/io/xml/internal/print_box_coords.hh b/scribo/scribo/io/xml/internal/print_box_coords.hh new file mode 100644 index 0000000..d3aeedf --- /dev/null +++ b/scribo/scribo/io/xml/internal/print_box_coords.hh @@ -0,0 +1,92 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_INTERNAL_PRINT_BOX_COORDS_HH +# define SCRIBO_IO_XML_INTERNAL_PRINT_BOX_COORDS_HH + +/// \file +/// +/// \brief Prints box2d coordinates to XML data. + +# include <mln/core/alias/box2d.hh> + +namespace scribo +{ + + namespace io + { + + namespace xml + { + + namespace internal + { + + /*! \brief Prints box2d coordinates to XML data. + */ + void + print_box_coords(std::ofstream& ostr, const box2d& b, + const char *space); + + +# ifndef MLN_INCLUDE_ONLY + + + inline + void + print_box_coords(std::ofstream& ostr, const box2d& b, + const char *space) + { + std::string sc = space; + std::string sp = sc + " "; + ostr << sc << "<coords>" << std::endl + << sp << "<point x=\"" << b.pmin().col() + << "\" y=\"" << b.pmin().row() << "\"/>" + << std::endl + << sp << "<point x=\"" << b.pmax().col() + << "\" y=\"" << b.pmin().row() << "\"/>" + << std::endl + << sp << "<point x=\"" << b.pmax().col() + << "\" y=\"" << b.pmax().row() << "\"/>" + << std::endl + << sp << "<point x=\"" << b.pmin().col() + << "\" y=\"" << b.pmax().row() << "\"/>" + << std::endl + << sc << "</coords>" << std::endl; + + } + + +# endif // ! MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml::internal + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + +#endif // ! SCRIBO_IO_XML_INTERNAL_PRINT_BOX_COORDS_HH diff --git a/scribo/scribo/io/xml/internal/print_page_preambule.hh b/scribo/scribo/io/xml/internal/print_page_preambule.hh new file mode 100644 index 0000000..b5ae891 --- /dev/null +++ b/scribo/scribo/io/xml/internal/print_page_preambule.hh @@ -0,0 +1,95 @@ +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) +// +// This file is part of Olena. +// +// Olena is free software: you can redistribute it and/or modify it under +// the terms of the GNU General Public License as published by the Free +// Software Foundation, version 2 of the License. +// +// Olena is distributed in the hope that it will be useful, +// but WITHOUT ANY WARRANTY; without even the implied warranty of +// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +// General Public License for more details. +// +// You should have received a copy of the GNU General Public License +// along with Olena. If not, see <http://www.gnu.org/licenses/>. +// +// As a special exception, you may use this file as part of a free +// software project without restriction. Specifically, if other files +// instantiate templates or use macros or inline functions from this +// file, or you compile this file and link it with other files to produce +// an executable, this file does not by itself cause the resulting +// executable to be covered by the GNU General Public License. This +// exception does not however invalidate any other reasons why the +// executable file might be covered by the GNU General Public License. + +#ifndef SCRIBO_IO_XML_INTERNAL_PRINT_PAGE_PREAMBULE_HH +# define SCRIBO_IO_XML_INTERNAL_PRINT_PAGE_PREAMBULE_HH + +/// \file +/// +/// \brief Print PAGE XML format preambule. + +# include <mln/core/alias/box2d.hh> + +namespace scribo +{ + + namespace io + { + + namespace xml + { + + namespace internal + { + + /// \brief Print PAGE XML format preambule. + template <typename L> + void print_PAGE_preambule(std::ofstream& output, + const document<L>& doc, + bool with_validation); + + +# ifndef MLN_INCLUDE_ONLY + + template <typename L> + void print_PAGE_preambule(std::ofstream& output, + const document<L>& doc, + bool with_validation) + { + output << "<?xml version=\"1.0\"?>" << std::endl; + + if (with_validation) + output << "<pcGts xmlns=\"http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16\" " + << "xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" " + << "xsi:schemaLocation=\"http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16 " + << "http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16/pagecontent…" " + << "pcGtsId=\"" << doc.filename() << "\">" << std::endl; + else + output << "<pcGts>" << std::endl; + + output << " <pcMetadata>" << std::endl; + output << " <pcCreator>LRDE</pcCreator>" << std::endl; + output << " <pcCreated/>" << std::endl; + output << " <pcLastChange/>" << std::endl; + output << " <pcComments>Generated by Scribo from Olena.</pcComments>" << std::endl; + output << " </pcMetadata>" << std::endl; + + output << " <page image_filename=\"" << doc.filename() + << "\" image_width=\"" << doc.width() + << "\" image_height=\"" << doc.height() + << "\">" << std::endl; + } + +# endif // ! MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml::internal + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + +#endif // ! SCRIBO_IO_XML_INTERNAL_PRINT_PAGE_PREAMBULE_HH diff --git a/scribo/scribo/io/xml/save.hh b/scribo/scribo/io/xml/save.hh index 7011e87..30579d0 100644 --- a/scribo/scribo/io/xml/save.hh +++ b/scribo/scribo/io/xml/save.hh @@ -38,7 +38,11 @@ # include <map> # include <scribo/core/document.hh> -# include <scribo/core/line_set.hh> + +# include <scribo/io/xml/internal/full_xml_visitor.hh> +# include <scribo/io/xml/internal/extended_page_xml_visitor.hh> +# include <scribo/io/xml/internal/page_xml_visitor.hh> + namespace scribo { @@ -49,360 +53,104 @@ namespace scribo namespace xml { - /*! \brief Save document information as XML. + /*! \brief Supported XML formats + + Page : PRima PAGE format. Used in ICDAR 2009. - We use a XML Schema part of the PAGE (Page Analysis and Ground - truth Elements) image representation framework. + PageExtended : Enriched PRima PAGE format with scribo data. + + Full : Enriched PRima PAGE format with scribo data. This + format can be reloaded in Scribo. + */ + enum Format + { + Page, + PageExtended, + Full + //Hocr + }; - This schema was used in the Page Segmentation COMPetition - (PSCOMP) for ICDAR 2009. - Its XSD file is located here: - http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16/pagecontent… + /*! \brief Save document information as XML. */ template <typename L> void - save(const document<L>& doc, - const std::string& output_name, - bool allow_extensions); + save(const document<L>& doc, const std::string& output_name, + Format format); # ifndef MLN_INCLUDE_ONLY + namespace internal { - inline - std::string& - html_markups_replace(std::string& input, - std::map<char, std::string>& map) - { - for (unsigned i = 0; i < input.size(); ++i) - { - std::map<char, std::string>::iterator it = map.find(input.at(i)); - if (it != map.end()) - { - input.replace(i, 1, it->second); - i += it->second.size() - 1; - } - } - return input; - } - - - inline - void print_box_coords(std::ofstream& ostr, const box2d& b, - const char *space) + template <typename L> + void save_page(const document<L>& doc, std::ofstream& output) { - std::string sc = space; - std::string sp = sc + " "; - ostr << sc << "<coords>" << std::endl - << sp << "<point x=\"" << b.pmin().col() - << "\" y=\"" << b.pmin().row() << "\"/>" - << std::endl - << sp << "<point x=\"" << b.pmax().col() - << "\" y=\"" << b.pmin().row() << "\"/>" - << std::endl - << sp << "<point x=\"" << b.pmax().col() - << "\" y=\"" << b.pmax().row() << "\"/>" - << std::endl - << sp << "<point x=\"" << b.pmin().col() - << "\" y=\"" << b.pmax().row() << "\"/>" - << std::endl - << sc << "</coords>" << std::endl; - + scribo::io::xml::internal::page_xml_visitor f(output); + doc.accept(f); } - - template <typename L> - void - save(const document<L>& doc, - const std::string& output_name) + void save_page_extended(const document<L>& doc, std::ofstream& output) { - trace::entering("scribo::io::xml:save_text_lines"); - - std::ofstream file(output_name.c_str()); - if (! file) - { - std::cerr << "error: cannot open file '" << doc.filename() << "'!"; - abort(); - } - - std::map<char, std::string> html_map; - html_map['\"'] = """; - html_map['<'] = "<"; - html_map['>'] = ">"; - html_map['&'] = "&"; - - file << "<?xml version=\"1.0\"?>" << std::endl; - file << "<pcGts xmlns=\"http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16\" xmlns:xsi=\"http://www.w3.org/2001/XMLSchema-instance\" xsi:schemaLocation=\"http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16 http://schema.primaresearch.org/PAGE/gts/pagecontent/2009-03-16/pagecontent…" pcGtsId=\"" << doc.filename() << "\">" << std::endl; - - file << " <pcMetadata>" << std::endl; - file << " <pcCreator>LRDE</pcCreator>" << std::endl; - file << " <pcCreated/>" << std::endl; - file << " <pcLastChange/>" << std::endl; - file << " <pcComments>Generated by Scribo from Olena.</pcComments>" << std::endl; - file << " </pcMetadata>" << std::endl; - - file << " <page image_filename=\"" << doc.filename() - << "\" image_width=\"" << doc.width() - << "\" image_height=\"" << doc.height() - << "\">" << std::endl; - - // Text - if (doc.has_text()) - { - const line_set<L>& lines = doc.lines(); - const paragraph_set<L>& parset = doc.paragraphs(); - - for_all_paragraphs(p, parset) - { - const mln::util::array<line_id_t>& line_ids = parset(p).line_ids(); - - // FIXME: compute that information on the whole paragraph - // and use them here. - line_id_t fid = line_ids(0); - file << " <text_region id=\"" << p - << "\" txt_orientation=\"" << lines(fid).orientation() - << "\" txt_reading_orientation=\"" << lines(fid).reading_orientation() - << "\" txt_reading_direction=\"" << lines(fid).reading_direction() - << "\" txt_text_type=\"" << lines(fid).type() - << "\" txt_reverse_video=\"" << (lines(fid).reverse_video() ? "true" : "false") - << "\" txt_indented=\"" << (lines(fid).indented() ? "true" : "false") - << "\" kerning=\"" << lines(fid).char_space() - << "\">" - << std::endl; - - internal::print_box_coords(file, parset(p).bbox(), " "); - - file << " </text_region>" << std::endl; - } - } - - // Page elements (Pictures, ...) - if (doc.has_elements()) - { - const component_set<L>& elts = doc.elements(); - for_all_comps(e, elts) - if (elts(e).is_valid()) - { - file << " <image_region id=\"ir" << elts(e).id() - << "\" img_colour_type=\"24_Bit_Colour\"" - << " img_orientation=\"0.000000\" " - << " img_emb_text=\"No\" " - << " img_bgcolour=\"White\">" << std::endl; - - internal::print_box_coords(file, elts(e).bbox(), " "); - - file << " </image_region>" << std::endl; - } - } - - - file << " </page>" << std::endl; - file << "</pcGts>" << std::endl; - - trace::exiting("scribo::io::xml::save_text_lines"); + scribo::io::xml::internal::extended_page_xml_visitor f(output); + doc.accept(f); } - - - template <typename L> - void - save_extended(const document<L>& doc, - const std::string& output_name) + void save_full(const document<L>& doc, std::ofstream& output) { - trace::entering("scribo::io::xml:save_text_lines"); - - std::ofstream file(output_name.c_str()); - if (! file) - { - std::cerr << "error: cannot open file '" << doc.filename() << "'!"; - abort(); - } - - std::map<char, std::string> html_map; - html_map['\"'] = """; - html_map['<'] = "<"; - html_map['>'] = ">"; - html_map['&'] = "&"; - - file << "<?xml version=\"1.0\"?>" << std::endl; - file << "<pcGts>" << std::endl; - - file << " <pcMetadata>" << std::endl; - file << " <pcCreator>LRDE</pcCreator>" << std::endl; - file << " <pcCreated/>" << std::endl; - file << " <pcLastChange/>" << std::endl; - file << " <pcComments>Generated by Scribo from Olena.</pcComments>" << std::endl; - file << " </pcMetadata>" << std::endl; - - file << " <page image_filename=\"" << doc.filename() - << "\" image_width=\"" << doc.width() - << "\" image_height=\"" << doc.height() - << "\">" << std::endl; - - // Text - if (doc.has_text()) - { - const line_set<L>& lines = doc.lines(); - const paragraph_set<L>& parset = doc.paragraphs(); - - for_all_paragraphs(p, parset) - { - const mln::util::array<line_id_t>& line_ids = parset(p).line_ids(); - - // FIXME: compute that information on the whole paragraph - // and use them here. - line_id_t fid = line_ids(0); - file << " <text_region id=\"" << p - << "\" txt_orientation=\"" << lines(fid).orientation() - << "\" txt_reading_orientation=\"" << lines(fid).reading_orientation() - << "\" txt_reading_direction=\"" << lines(fid).reading_direction() - << "\" txt_text_type=\"" << lines(fid).type() - << "\" txt_reverse_video=\"" << (lines(fid).reverse_video() ? "true" : "false") - << "\" txt_indented=\"" << (lines(fid).indented() ? "true" : "false") - << "\" kerning=\"" << lines(fid).char_space(); - - // EXTENSIONS - Not officially supported - file << "\" baseline=\"" << lines(fid).baseline() - << "\" meanline=\"" << lines(fid).meanline() - << "\" x_height=\"" << lines(fid).x_height() - << "\" d_height=\"" << lines(fid).d_height() - << "\" a_height=\"" << lines(fid).a_height() - << "\" char_width=\"" << lines(fid).char_width(); - // End of EXTENSIONS - file << "\">" - << std::endl; - - internal::print_box_coords(file, parset(p).bbox(), " "); - - - // EXTENSIONS - Not officially supported - for_all_paragraph_lines(lid, line_ids) - { - line_id_t l = line_ids(lid); - - if (lines(l).has_text()) - { - std::string tmp = lines(l).text(); - tmp = internal::html_markups_replace(tmp, html_map); - - file << " <line text=\"" << tmp << "\" "; - } - else - file << " <line " << std::endl; - - file << "id=\"" << lines(l).id() - << "\" txt_orientation=\"" << lines(l).orientation() - << "\" txt_reading_orientation=\"" << lines(l).reading_orientation() - << "\" txt_reading_direction=\"" << lines(l).reading_direction() - << "\" txt_text_type=\"" << lines(l).type() - << "\" txt_reverse_video=\"" << (lines(l).reverse_video() ? "true" : "false") - << "\" txt_indented=\"" << (lines(l).indented() ? "true" : "false") - << "\" kerning=\"" << lines(l).char_space() - << "\" baseline=\"" << lines(l).baseline() - << "\" meanline=\"" << lines(l).meanline() - << "\" x_height=\"" << lines(l).x_height() - << "\" d_height=\"" << lines(l).d_height() - << "\" a_height=\"" << lines(l).a_height() - << "\" char_width=\"" << lines(l).char_width() - << "\">" << std::endl; - - internal::print_box_coords(file, lines(l).bbox(), " "); - - file << " </line>" << std::endl; - } - - file << " </text_region>" << std::endl; - } - } - // End of EXTENSIONS - - // Page elements (Pictures, ...) - if (doc.has_elements()) - { - const component_set<L>& elts = doc.elements(); - for_all_comps(e, elts) - if (elts(e).is_valid()) - { - switch (elts(e).type()) - { - case component::Separator: - { - file << " <separator_region id=\"sr" << elts(e).id() - << "\" sep_orientation=\"0.000000\" " - << " sep_colour=\"Black\" " - << " sep_bgcolour=\"White\">" << std::endl; - - internal::print_box_coords(file, elts(e).bbox(), " "); - - file << " </separator_region>" << std::endl; - break; - break; - } - - default: - case component::Image: - { - file << " <image_region id=\"ir" << elts(e).id() - << "\" img_colour_type=\"24_Bit_Colour\"" - << " img_orientation=\"0.000000\" " - << " img_emb_text=\"No\" " - << " img_bgcolour=\"White\">" << std::endl; - - internal::print_box_coords(file, elts(e).bbox(), " "); - - file << " </image_region>" << std::endl; - break; - } - } - } - } - - - // Whitespace seraparators - if (doc.has_whitespace_seps()) - { - const component_set<L>& - whitespace_seps_comps = doc.whitespace_seps_comps(); - - for_all_comps(c, whitespace_seps_comps) - { - file << " <whitespace_separator_region id=\"wss" - << whitespace_seps_comps(c).id() - << "\">" << std::endl; - - internal::print_box_coords(file, whitespace_seps_comps(c).bbox(), " "); - - file << " </whitespace_separator_region>" << std::endl; - } - } - - file << " </page>" << std::endl; - file << "</pcGts>" << std::endl; - - trace::exiting("scribo::io::xml::save_text_lines"); + scribo::io::xml::internal::full_xml_visitor f(output); + doc.accept(f); } } // end of namespace scribo::io::xml::internal + // FACADE template <typename L> void save(const document<L>& doc, const std::string& output_name, - bool allow_extensions) + Format format) { - if (allow_extensions) - internal::save_extended(doc, output_name); - else - internal::save(doc, output_name); + trace::entering("scribo::io::xml::save"); + + // Open file + std::ofstream output(output_name.c_str()); + if (! output) + { + std::cerr << "scribo::io::xml::save - ERROR: cannot open file '" + << doc.filename() << "'!"; + return; + } + + // Choose saving method. + switch (format) + { + case Page: + internal::save_page(doc, output); + break; + + case PageExtended: + internal::save_page_extended(doc, output); + break; + + case Full: + internal::save_full(doc, output); + break; + + default: + trace::warning("scribo::io::xml::save - Invalid XML format! Skip saving..."); + } + + output.close(); + trace::exiting("scribo::io::xml::save"); } diff --git a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh index 48098ba..dcbb4f7 100644 --- a/scribo/scribo/toolchain/internal/content_in_doc_functor.hh +++ b/scribo/scribo/toolchain/internal/content_in_doc_functor.hh @@ -36,7 +36,6 @@ # include <scribo/primitive/extract/non_text.hh> # include <scribo/primitive/extract/components.hh> -//# include <scribo/primitive/extract/vertical_separators.hh> # include <scribo/primitive/extract/separators.hh> # include <scribo/primitive/extract/separators_nonvisible.hh> @@ -114,7 +113,7 @@ namespace scribo bool enable_whitespace_seps; bool enable_debug; bool save_doc_as_xml; - bool allow_xml_extensions; + scribo::io::xml::Format xml_format; //============ // Parameters @@ -139,7 +138,7 @@ namespace scribo enable_whitespace_seps(true), enable_debug(false), save_doc_as_xml(false), - allow_xml_extensions(true), + xml_format(scribo::io::xml::PageExtended), ocr_language("eng"), output_file("/tmp/foo.xml"), doc(doc_filename) @@ -189,7 +188,7 @@ namespace scribo // Whitespace separators on_new_progress_label("Find whitespace separators..."); - whitespaces = primitive::extract::separators_nonvisible(processed_image); + whitespaces = primitive::extract::separators_nonvisible(input_cleaned); on_progress(); } @@ -483,7 +482,7 @@ namespace scribo { on_new_progress_label("Saving results"); - scribo::io::xml::save(doc, output_file, allow_xml_extensions); + scribo::io::xml::save(doc, output_file, xml_format); on_xml_saved(); on_progress(); diff --git a/scribo/src/content_in_doc.cc b/scribo/src/content_in_doc.cc index 9748b28..d8d4e52 100644 --- a/scribo/src/content_in_doc.cc +++ b/scribo/src/content_in_doc.cc @@ -172,7 +172,9 @@ int main(int argc, char* argv[]) debug); // Saving results - scribo::io::xml::save(doc, argv[2], true); + scribo::io::xml::save(doc, argv[2], scribo::io::xml::PageExtended); + scribo::io::xml::save(doc, "page.xml", scribo::io::xml::Page); + scribo::io::xml::save(doc, "full.xml", scribo::io::xml::Full); trace::exiting("main"); } -- 1.5.6.5

13 years, 10 months

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Olena-patches