March 2011 - Olena-patches

last-svn-commit-785-g6b548c5 Regen mk files.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

* scribo/headers.mk, * scribo/tests/unit_test/unit-tests.mk: Regen. * scribo/tests/unit_test/cond_tests_qt: Add scribo/io/xml/load.hh. --- scribo/ChangeLog | 9 +++++++++ scribo/headers.mk | 13 ++++++++++++- scribo/tests/unit_test/cond_tests_qt | 1 + scribo/tests/unit_test/unit-tests.mk | 30 +++++++++++++++++++++++++++--- 4 files changed, 49 insertions(+), 4 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index bbd0cf2..4eae5f8 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,3 +1,12 @@ +2011-03-02 Guillaume Lazzara <z(a)lrde.epita.fr> + + Regen mk files. + + * scribo/headers.mk, + * scribo/tests/unit_test/unit-tests.mk: Regen. + + * scribo/tests/unit_test/cond_tests_qt: Add scribo/io/xml/load.hh. + 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> * scribo/primitive/extract/non_text.hh: Remove debug. diff --git a/scribo/headers.mk b/scribo/headers.mk index 7603f8a..0b21a3f 100644 --- a/scribo/headers.mk +++ b/scribo/headers.mk @@ -14,17 +14,22 @@ scribo/binarization/sauvola_ms.hh \ scribo/binarization/sauvola_ms_split.hh \ scribo/binarization/sauvola_threshold_image.hh \ scribo/canvas/integral_browsing.hh \ +scribo/convert/from_base64.hh \ scribo/convert/from_qimage.hh \ +scribo/convert/to_base64.hh \ scribo/core/all.hh \ scribo/core/central_sites.hh \ scribo/core/component_info.hh \ scribo/core/component_set.hh \ scribo/core/concept/dmax_functor.hh \ scribo/core/concept/link_functor.hh \ +scribo/core/concept/serializable.hh \ +scribo/core/concept/serialize_visitor.hh \ scribo/core/def/lbl_type.hh \ scribo/core/document.hh \ scribo/core/erase_objects.hh \ scribo/core/init_integral_image.hh \ +scribo/core/internal/doc_xml_serializer.hh \ scribo/core/line_info.hh \ scribo/core/line_links.hh \ scribo/core/line_set.hh \ @@ -93,6 +98,12 @@ scribo/fun/v2b/label_to_bool.hh \ scribo/fun/v2b/objects_large_filter.hh \ scribo/fun/v2b/objects_small_filter.hh \ scribo/io/text_boxes/save.hh \ +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 \ +scribo/io/xml/internal/print_box_coords.hh \ +scribo/io/xml/internal/print_page_preambule.hh \ +scribo/io/xml/load.hh \ scribo/io/xml/save.hh \ scribo/make/all.hh \ scribo/make/debug_filename.hh \ @@ -115,7 +126,6 @@ scribo/primitive/extract/all.hh \ scribo/primitive/extract/canvas.hh \ scribo/primitive/extract/cells.hh \ scribo/primitive/extract/components.hh \ -scribo/primitive/extract/elements.hh \ scribo/primitive/extract/horizontal_separators.hh \ scribo/primitive/extract/lines_discontinued.hh \ scribo/primitive/extract/lines_h_discontinued.hh \ @@ -130,6 +140,7 @@ scribo/primitive/extract/lines_v_pattern.hh \ scribo/primitive/extract/lines_v_single.hh \ scribo/primitive/extract/lines_v_thick.hh \ scribo/primitive/extract/lines_v_thick_and_single.hh \ +scribo/primitive/extract/non_text.hh \ scribo/primitive/extract/separators.hh \ scribo/primitive/extract/separators_nonvisible.hh \ scribo/primitive/extract/vertical_separators.hh \ diff --git a/scribo/tests/unit_test/cond_tests_qt b/scribo/tests/unit_test/cond_tests_qt index 5428215..577f9e6 100644 --- a/scribo/tests/unit_test/cond_tests_qt +++ b/scribo/tests/unit_test/cond_tests_qt @@ -1 +1,2 @@ scribo/convert/from_qimage.hh +scribo/io/xml/load.hh diff --git a/scribo/tests/unit_test/unit-tests.mk b/scribo/tests/unit_test/unit-tests.mk index 6a8f29d..71b0b5b 100644 --- a/scribo/tests/unit_test/unit-tests.mk +++ b/scribo/tests/unit_test/unit-tests.mk @@ -19,11 +19,15 @@ endif HAVE_MAGICKXX # Starting a conditional unit test list. if HAVE_QT check_PROGRAMS += \ -scribo_convert_from_qimage +scribo_convert_from_qimage \ +scribo_io_xml_load scribo_convert_from_qimage_CPPFLAGS= ${QT_CPPFLAGS} ${AM_CPPFLAGS} scribo_convert_from_qimage_LDFLAGS= ${QT_LDFLAGS} ${AM_LDFLAGS} scribo_convert_from_qimage_SOURCES = scribo_convert_from_qimage.cc +scribo_io_xml_load_CPPFLAGS= ${QT_CPPFLAGS} ${AM_CPPFLAGS} +scribo_io_xml_load_LDFLAGS= ${QT_LDFLAGS} ${AM_LDFLAGS} +scribo_io_xml_load_SOURCES = scribo_io_xml_load.cc endif HAVE_QT # Starting a conditional unit test list. @@ -79,16 +83,21 @@ scribo_binarization_sauvola_ms \ scribo_binarization_sauvola_ms_split \ scribo_binarization_sauvola_threshold_image \ scribo_canvas_integral_browsing \ +scribo_convert_from_base64 \ +scribo_convert_to_base64 \ scribo_core_all \ scribo_core_central_sites \ scribo_core_component_info \ scribo_core_component_set \ scribo_core_concept_dmax_functor \ scribo_core_concept_link_functor \ +scribo_core_concept_serializable \ +scribo_core_concept_serialize_visitor \ scribo_core_def_lbl_type \ scribo_core_document \ scribo_core_erase_objects \ scribo_core_init_integral_image \ +scribo_core_internal_doc_xml_serializer \ scribo_core_line_info \ scribo_core_line_links \ scribo_core_line_set \ @@ -157,6 +166,11 @@ scribo_fun_v2b_label_to_bool \ scribo_fun_v2b_objects_large_filter \ scribo_fun_v2b_objects_small_filter \ scribo_io_text_boxes_save \ +scribo_io_xml_internal_extended_page_xml_visitor \ +scribo_io_xml_internal_full_xml_visitor \ +scribo_io_xml_internal_page_xml_visitor \ +scribo_io_xml_internal_print_box_coords \ +scribo_io_xml_internal_print_page_preambule \ scribo_io_xml_save \ scribo_make_all \ scribo_make_debug_filename \ @@ -179,7 +193,6 @@ scribo_primitive_extract_all \ scribo_primitive_extract_canvas \ scribo_primitive_extract_cells \ scribo_primitive_extract_components \ -scribo_primitive_extract_elements \ scribo_primitive_extract_horizontal_separators \ scribo_primitive_extract_lines_discontinued \ scribo_primitive_extract_lines_h_discontinued \ @@ -194,6 +207,7 @@ scribo_primitive_extract_lines_v_pattern \ scribo_primitive_extract_lines_v_single \ scribo_primitive_extract_lines_v_thick \ scribo_primitive_extract_lines_v_thick_and_single \ +scribo_primitive_extract_non_text \ scribo_primitive_extract_separators \ scribo_primitive_extract_separators_nonvisible \ scribo_primitive_extract_vertical_separators \ @@ -286,16 +300,21 @@ scribo_binarization_sauvola_ms_SOURCES = scribo_binarization_sauvola_ms.cc scribo_binarization_sauvola_ms_split_SOURCES = scribo_binarization_sauvola_ms_split.cc scribo_binarization_sauvola_threshold_image_SOURCES = scribo_binarization_sauvola_threshold_image.cc scribo_canvas_integral_browsing_SOURCES = scribo_canvas_integral_browsing.cc +scribo_convert_from_base64_SOURCES = scribo_convert_from_base64.cc +scribo_convert_to_base64_SOURCES = scribo_convert_to_base64.cc scribo_core_all_SOURCES = scribo_core_all.cc scribo_core_central_sites_SOURCES = scribo_core_central_sites.cc scribo_core_component_info_SOURCES = scribo_core_component_info.cc scribo_core_component_set_SOURCES = scribo_core_component_set.cc scribo_core_concept_dmax_functor_SOURCES = scribo_core_concept_dmax_functor.cc scribo_core_concept_link_functor_SOURCES = scribo_core_concept_link_functor.cc +scribo_core_concept_serializable_SOURCES = scribo_core_concept_serializable.cc +scribo_core_concept_serialize_visitor_SOURCES = scribo_core_concept_serialize_visitor.cc scribo_core_def_lbl_type_SOURCES = scribo_core_def_lbl_type.cc scribo_core_document_SOURCES = scribo_core_document.cc scribo_core_erase_objects_SOURCES = scribo_core_erase_objects.cc scribo_core_init_integral_image_SOURCES = scribo_core_init_integral_image.cc +scribo_core_internal_doc_xml_serializer_SOURCES = scribo_core_internal_doc_xml_serializer.cc scribo_core_line_info_SOURCES = scribo_core_line_info.cc scribo_core_line_links_SOURCES = scribo_core_line_links.cc scribo_core_line_set_SOURCES = scribo_core_line_set.cc @@ -364,6 +383,11 @@ scribo_fun_v2b_label_to_bool_SOURCES = scribo_fun_v2b_label_to_bool.cc scribo_fun_v2b_objects_large_filter_SOURCES = scribo_fun_v2b_objects_large_filter.cc scribo_fun_v2b_objects_small_filter_SOURCES = scribo_fun_v2b_objects_small_filter.cc scribo_io_text_boxes_save_SOURCES = scribo_io_text_boxes_save.cc +scribo_io_xml_internal_extended_page_xml_visitor_SOURCES = scribo_io_xml_internal_extended_page_xml_visitor.cc +scribo_io_xml_internal_full_xml_visitor_SOURCES = scribo_io_xml_internal_full_xml_visitor.cc +scribo_io_xml_internal_page_xml_visitor_SOURCES = scribo_io_xml_internal_page_xml_visitor.cc +scribo_io_xml_internal_print_box_coords_SOURCES = scribo_io_xml_internal_print_box_coords.cc +scribo_io_xml_internal_print_page_preambule_SOURCES = scribo_io_xml_internal_print_page_preambule.cc scribo_io_xml_save_SOURCES = scribo_io_xml_save.cc scribo_make_all_SOURCES = scribo_make_all.cc scribo_make_debug_filename_SOURCES = scribo_make_debug_filename.cc @@ -386,7 +410,6 @@ scribo_primitive_extract_all_SOURCES = scribo_primitive_extract_all.cc scribo_primitive_extract_canvas_SOURCES = scribo_primitive_extract_canvas.cc scribo_primitive_extract_cells_SOURCES = scribo_primitive_extract_cells.cc scribo_primitive_extract_components_SOURCES = scribo_primitive_extract_components.cc -scribo_primitive_extract_elements_SOURCES = scribo_primitive_extract_elements.cc scribo_primitive_extract_horizontal_separators_SOURCES = scribo_primitive_extract_horizontal_separators.cc scribo_primitive_extract_lines_discontinued_SOURCES = scribo_primitive_extract_lines_discontinued.cc scribo_primitive_extract_lines_h_discontinued_SOURCES = scribo_primitive_extract_lines_h_discontinued.cc @@ -401,6 +424,7 @@ scribo_primitive_extract_lines_v_pattern_SOURCES = scribo_primitive_extract_line scribo_primitive_extract_lines_v_single_SOURCES = scribo_primitive_extract_lines_v_single.cc scribo_primitive_extract_lines_v_thick_SOURCES = scribo_primitive_extract_lines_v_thick.cc scribo_primitive_extract_lines_v_thick_and_single_SOURCES = scribo_primitive_extract_lines_v_thick_and_single.cc +scribo_primitive_extract_non_text_SOURCES = scribo_primitive_extract_non_text.cc scribo_primitive_extract_separators_SOURCES = scribo_primitive_extract_separators.cc scribo_primitive_extract_separators_nonvisible_SOURCES = scribo_primitive_extract_separators_nonvisible.cc scribo_primitive_extract_vertical_separators_SOURCES = scribo_primitive_extract_vertical_separators.cc -- 1.5.6.5

1 0

last-svn-commit-784-gd12d660 scribo/primitive/extract/non_text.hh: Remove debug.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

--- scribo/ChangeLog | 4 ++++ scribo/scribo/primitive/extract/non_text.hh | 21 --------------------- 2 files changed, 4 insertions(+), 21 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 4a9dcbf..bbd0cf2 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/primitive/extract/non_text.hh: Remove debug. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Add new debug tools. * scribo/src/debug/Makefile.am: New targets. diff --git a/scribo/scribo/primitive/extract/non_text.hh b/scribo/scribo/primitive/extract/non_text.hh index a017f1f..8528782 100644 --- a/scribo/scribo/primitive/extract/non_text.hh +++ b/scribo/scribo/primitive/extract/non_text.hh @@ -58,10 +58,6 @@ # 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 { @@ -139,16 +135,9 @@ 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>(), @@ -178,19 +167,9 @@ namespace scribo 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); - // DEBUG - io::pbm::save(elts, "elements.pbm"); - // END OF DEBUG - output = primitive::extract::components(elts, c8(), nlabels); } -- 1.5.6.5

1 0

last-svn-commit-783-gca041c0 Add new debug tools.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

* scribo/src/debug/Makefile.am: New targets. * scribo/src/debug/show_separators.cc, * scribo/src/debug/show_stoppers.cc, * scribo/src/debug/show_text_lines.cc: New. --- scribo/ChangeLog | 10 ++ scribo/src/debug/Makefile.am | 12 ++- .../lines_pattern.cc => debug/show_separators.cc} | 74 +++++++++----- scribo/src/debug/show_stoppers.cc | 104 ++++++++++++++++++++ .../show_text_lines.cc} | 100 +++++++++++-------- 5 files changed, 225 insertions(+), 75 deletions(-) copy scribo/src/{primitive/extract/lines_pattern.cc => debug/show_separators.cc} (50%) create mode 100644 scribo/src/debug/show_stoppers.cc copy scribo/src/{text/pbm_lines_recognition.cc => debug/show_text_lines.cc} (52%) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 74c7a03..4a9dcbf 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,15 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Add new debug tools. + + * scribo/src/debug/Makefile.am: New targets. + + * scribo/src/debug/show_separators.cc, + * scribo/src/debug/show_stoppers.cc, + * scribo/src/debug/show_text_lines.cc: New. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/text/extract_lines.hh: Update code. 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> diff --git a/scribo/src/debug/Makefile.am b/scribo/src/debug/Makefile.am index d203f5c..cdb1f30 100644 --- a/scribo/src/debug/Makefile.am +++ b/scribo/src/debug/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. # @@ -20,6 +20,7 @@ include $(top_srcdir)/scribo/scribo.mk noinst_PROGRAMS = \ show_info_x_height \ show_info_median_inter_characters \ + show_separators \ show_links_left_right_links_validation \ show_links_bbox_h_ratio \ show_links_bbox_overlap \ @@ -43,11 +44,14 @@ noinst_PROGRAMS = \ show_objects_large_small \ show_objects_small \ show_objects_thick \ - show_objects_thin + show_objects_thin \ + show_stoppers \ + show_text_lines show_info_x_height_SOURCES = show_info_x_height.cc show_info_median_inter_characters_SOURCES = show_info_median_inter_characters.cc +show_separators_SOURCES = show_separators.cc show_links_left_right_links_validation_SOURCES = show_links_left_right_links_validation.cc show_links_bbox_h_ratio_SOURCES = show_links_bbox_h_ratio.cc show_links_bbox_overlap_SOURCES = show_links_bbox_overlap.cc @@ -72,6 +76,8 @@ show_objects_large_small_SOURCES = show_objects_large_small.cc show_objects_small_SOURCES = show_objects_small.cc show_objects_thick_SOURCES = show_objects_thick.cc show_objects_thin_SOURCES = show_objects_thin.cc +show_stoppers_SOURCES = show_stoppers.cc +show_text_lines_SOURCES = show_text_lines.cc if HAVE_MAGICKXX diff --git a/scribo/src/primitive/extract/lines_pattern.cc b/scribo/src/debug/show_separators.cc similarity index 50% copy from scribo/src/primitive/extract/lines_pattern.cc copy to scribo/src/debug/show_separators.cc index cb56b36..ef80da6 100644 --- a/scribo/src/primitive/extract/lines_pattern.cc +++ b/scribo/src/debug/show_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. // @@ -14,33 +13,25 @@ // // 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. #include <mln/core/image/image2d.hh> +#include <mln/core/alias/box2d.hh> +#include <mln/core/alias/neighb2d.hh> #include <mln/io/pbm/all.hh> -#include <mln/pw/all.hh> -#include <mln/data/fill.hh> -#include <mln/core/image/dmorph/image_if.hh> +#include <mln/arith/plus.hh> +#include <mln/labeling/foreground.hh> #include <scribo/primitive/extract/lines_h_pattern.hh> #include <scribo/primitive/extract/lines_v_pattern.hh> #include <scribo/debug/usage.hh> +#include <scribo/debug/save_bboxes_image.hh> const char *args_desc[][2] = { - { "input.pbm", "A binary image." }, - { "length", " Minimum line length. (Common value : 51)" }, + { "length", " Minimum line length. (default: 101)" }, { "delta", " Distance between the object pixel and the background pixel" - "(Common value : 4)" }, + "(default: 4)" }, {0, 0} }; @@ -49,10 +40,10 @@ int main(int argc, char *argv[]) { using namespace mln; - if (argc != 5) + if (argc != 7 && argc != 5) return scribo::debug::usage(argv, - "Extract horizontal lines patterns", - "input.pbm length delta output.pbm", + "Extract horizontal and vertical lines patterns", + "input.pbm output.pbm input_wo_seps.pbm output.ppm length delta", args_desc); trace::entering("main"); @@ -60,19 +51,46 @@ int main(int argc, char *argv[]) image2d<bool> input; io::pbm::load(input, argv[1]); + // Set default option value + unsigned + length = 101, + delta = 4; + if (argc > 5) + { + length = atoi(argv[5]); + delta = atoi(argv[6]); + } + + util::timer t; + t.start(); + image2d<bool> - h_lines = scribo::primitive::extract::lines_h_pattern(input, - atoi(argv[2]), - atoi(argv[3])); + h_lines = scribo::primitive::extract::lines_h_pattern(input, length, delta); image2d<bool> - v_lines = scribo::primitive::extract::lines_v_pattern(input, - atoi(argv[2]), - atoi(argv[3])); + v_lines = scribo::primitive::extract::lines_v_pattern(input, length, delta); + + v_lines += h_lines; + + t.stop(); + std::cout << t << "s" << std::endl; + + // Save binary image. + io::pbm::save(v_lines, argv[2]); - data::fill((v_lines | pw::value(h_lines)).rw(), true); + // Save input without separators + { + image2d<bool> in_wo_seps = duplicate(input); + data::fill((in_wo_seps | pw::value(v_lines)).rw(), false); + io::pbm::save(in_wo_seps, argv[3]); + } - io::pbm::save(v_lines, argv[4]); + // Save bbox image + value::int_u8 nlabels; + image2d<value::int_u8> lbl = labeling::foreground(v_lines, c4(), nlabels); + mln::util::array<box2d> + bbox = labeling::compute(accu::shape::bbox<point2d>(), lbl, nlabels); + scribo::debug::save_bboxes_image(input, bbox, argv[4], literal::red); trace::exiting("main"); } diff --git a/scribo/src/debug/show_stoppers.cc b/scribo/src/debug/show_stoppers.cc new file mode 100644 index 0000000..a108da1 --- /dev/null +++ b/scribo/src/debug/show_stoppers.cc @@ -0,0 +1,104 @@ +// 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/>. + +#include <mln/core/image/image2d.hh> +#include <mln/core/alias/box2d.hh> +#include <mln/core/alias/neighb2d.hh> +#include <mln/io/pbm/all.hh> +#include <mln/arith/plus.hh> +#include <mln/labeling/foreground.hh> + +#include <scribo/primitive/extract/lines_h_pattern.hh> +#include <scribo/primitive/extract/lines_v_pattern.hh> +#include <scribo/primitive/extract/separators_nonvisible.hh> +#include <scribo/debug/usage.hh> +#include <scribo/debug/save_bboxes_image.hh> + + +const char *args_desc[][2] = +{ + { "length", " Minimum line length. (default: 101)" }, + { "delta", " Distance between the object pixel and the background pixel" + "(default: 4)" }, + {0, 0} +}; + + +int main(int argc, char *argv[]) +{ + using namespace mln; + + if (argc != 7 && argc != 5) + return scribo::debug::usage(argv, + "Extract horizontal, vertical lines and stoppers", + "input.pbm output.pbm input_wo_seps.pbm output.ppm length delta", + args_desc); + + trace::entering("main"); + + image2d<bool> input; + io::pbm::load(input, argv[1]); + + // Set default option value + unsigned + length = 101, + delta = 4; + if (argc > 5) + { + length = atoi(argv[5]); + delta = atoi(argv[6]); + } + + util::timer t; + + t.start(); + image2d<bool> + h_lines = scribo::primitive::extract::lines_h_pattern(input, length, delta); + + image2d<bool> + v_lines = scribo::primitive::extract::lines_v_pattern(input, length, delta); + + v_lines += h_lines; + + t.stop(); + + // Save input without separators + { + image2d<bool> in_wo_seps = duplicate(input); + data::fill((in_wo_seps | pw::value(v_lines)).rw(), false); + io::pbm::save(in_wo_seps, argv[3]); + } + + // Save bbox image + value::int_u8 nlabels; + image2d<value::int_u8> lbl = labeling::foreground(v_lines, c4(), nlabels); + mln::util::array<box2d> + bbox = labeling::compute(accu::shape::bbox<point2d>(), lbl, nlabels); + scribo::debug::save_bboxes_image(input, bbox, argv[4], literal::red); + + t.resume(); + + // Non visible separators + v_lines += scribo::primitive::extract::separators_nonvisible(input); + + t.stop(); + std::cout << t << "s" << std::endl; + + // Save binary image. + io::pbm::save(v_lines, argv[2]); + + trace::exiting("main"); +} diff --git a/scribo/src/text/pbm_lines_recognition.cc b/scribo/src/debug/show_text_lines.cc similarity index 52% copy from scribo/src/text/pbm_lines_recognition.cc copy to scribo/src/debug/show_text_lines.cc index 3451449..a031072 100644 --- a/scribo/src/text/pbm_lines_recognition.cc +++ b/scribo/src/debug/show_text_lines.cc @@ -1,4 +1,4 @@ -// Copyright (C) 2010 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -23,15 +23,15 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. -#ifdef HAVE_CONFIG_H -# include <config.h> -#endif - #include <iostream> #include <mln/core/image/image2d.hh> #include <mln/io/pbm/load.hh> #include <mln/value/int_u16.hh> +#include <mln/draw/box_plain.hh> +#include <mln/pw/all.hh> +#include <mln/core/image/dmorph/image_if.hh> +#include <mln/data/convert.hh> #include <scribo/text/recognition.hh> @@ -41,19 +41,15 @@ #include <scribo/core/object_links.hh> #include <scribo/core/object_groups.hh> -#include <scribo/text/merging.hh> -#include <scribo/primitive/extract/components.hh> -#include <scribo/primitive/link/with_single_left_link_dmax_ratio.hh> -#include <scribo/primitive/group/from_single_link.hh> +#include <scribo/text/extract_lines.hh> #include <scribo/io/text_boxes/save.hh> +#include <scribo/debug/save_bboxes_image.hh> + const char *args_desc[][2] = { - { "input.pbm", "A binary image. 'True' for objects, 'False'\ -for the background." }, - { "lang", "Must be set to \"eng\", \"fra\", \"deu\", \"spa\", \"ita\" (Default \"fra\")" }, {0, 0} }; @@ -64,10 +60,10 @@ int main(int argc, char* argv[]) using namespace scribo; using namespace mln; - if (argc != 2 && argc != 3 && argc != 4) + if (argc != 6) return scribo::debug::usage(argv, - "Text recognition", - "input.pbm [lang] [file]", + "Show text lines", + "input.pbm input_seps.pbm out.ppm out.pbm comps.pbm", args_desc); trace::entering("main"); @@ -75,37 +71,53 @@ int main(int argc, char* argv[]) image2d<bool> input; mln::io::pbm::load(input, argv[1]); - typedef image2d<value::int_u16> L; - value::int_u16 ncomps; - component_set<L> - comps = scribo::primitive::extract::components(input, c8(), ncomps); - std::cout << "ncomps = " << ncomps << std::endl; - - object_links<L> - links = scribo::primitive::link::with_single_left_link_dmax_ratio(comps, 2); - - object_groups<L> - groups = scribo::primitive::group::from_single_link(links); + image2d<bool> input_seps; + mln::io::pbm::load(input_seps, argv[2]); - line_set<L> lines(groups); - lines = text::merging(lines); + util::timer t; + t.start(); - std::string str = argv[2]; - const char *lang; - if (argc < 3 || (str != "eng" && str != "fra" - && str != "deu" && str != "spa" - && str != "ita")) - lang = "fra"; - else - lang = argv[2]; - - char *output = 0; - if (argc >= 4) - output = argv[3]; - - scribo::text::recognition(lines, lang); - - scribo::io::text_boxes::save(lines, output); + typedef image2d<value::int_u16> L; + line_set<L> lines = text::extract_lines(input, c8(), input_seps); + + t.stop(); + std::cout << t << "s" << std::endl; + + // Color image with line bboxes. + { + image2d<value::rgb8> output = data::convert(value::rgb8(), input); + for_all_lines(l, lines) + if (lines(l).is_textline()) + mln::draw::box(output, lines(l).bbox(), literal::red); + mln::io::ppm::save(output, argv[3]); + } + + // Binary image with line bboxes. + { + image2d<bool> output; + initialize(output, input); + data::fill(output, false); + for_all_lines(l, lines) + if (lines(l).is_textline()) + mln::draw::box_plain(output, lines(l).bbox(), true); + mln::io::pbm::save(output, argv[4]); + } + + { + image2d<bool> output; + initialize(output, input); + data::fill(output, false); + for_all_lines(l, lines) + if (lines(l).is_textline()) + { + const util::array<component_id_t>& comps = lines(l).components(); + const L& lbl = lines.components().labeled_image(); + for_all_elements(c, comps) + data::fill((output | lines.components()(comps(c)).bbox()).rw(), + ((input | lines.components()(comps(c)).bbox()) | (pw::value(lbl) == comps(c)))); + } + mln::io::pbm::save(output, argv[5]); + } trace::exiting("main"); } -- 1.5.6.5

1 0

last-svn-commit-782-gba41452 scribo/text/extract_lines.hh: Update code.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

--- scribo/ChangeLog | 4 ++ scribo/scribo/text/extract_lines.hh | 68 ++++++++++++++++++++++------------ 2 files changed, 48 insertions(+), 24 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index f740230..74c7a03 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/text/extract_lines.hh: Update code. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make use of mln::fun::v2v::rgb_to_luma. * scribo/toolchain/internal/text_in_doc_preprocess_functor.hh, diff --git a/scribo/scribo/text/extract_lines.hh b/scribo/scribo/text/extract_lines.hh index 1a25cc8..b81cb79 100644 --- a/scribo/scribo/text/extract_lines.hh +++ b/scribo/scribo/text/extract_lines.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. // @@ -33,13 +33,14 @@ # include <mln/core/concept/image.hh> -# include <mln/value/label_16.hh> +# include <mln/value/int_u16.hh> # include <scribo/core/line_set.hh> # include <scribo/primitive/extract/components.hh> # include <scribo/primitive/link/merge_double_link.hh> +# include <scribo/primitive/link/internal/dmax_width_and_height.hh> # include <scribo/primitive/link/with_single_left_link_dmax_ratio.hh> # include <scribo/primitive/link/with_single_right_link_dmax_ratio.hh> @@ -65,11 +66,18 @@ namespace scribo /*! ** \param[in] input A binary image. ** \param[in] nbh A neighborhood used for labeling. + ** \param[in] seps A binary image with separator information. ** ** \return A set of lines. */ template <typename I, typename N> - line_set<mln_ch_value(I,value::label_16)> + line_set<mln_ch_value(I,value::int_u16)> + extract_lines(const Image& input_, const Neighborhood<N>& nbh_, + const mln_ch_value(I,bool)& separators); + + /// \overload + template <typename I, typename N> + line_set<mln_ch_value(I,value::int_u16)> extract_lines(const Image& input, const Neighborhood<N>& nbh); @@ -77,8 +85,18 @@ namespace scribo template <typename I, typename N> - line_set<mln_ch_value(I,value::label_16)> - extract_lines(const Image& input_, const Neighborhood<N>& nbh_) + line_set<mln_ch_value(I,value::int_u16)> + extract_lines(const Image& input, const Neighborhood<N>& nbh) + { + mln_ch_value(I,bool) seps; + return extract_lines(input, nbh, seps); + } + + + template <typename I, typename N> + line_set<mln_ch_value(I,value::int_u16)> + extract_lines(const Image& input_, const Neighborhood<N>& nbh_, + const mln_ch_value(I,bool)& separators) { trace::entering("scribo::text::extract_lines"); @@ -88,44 +106,46 @@ namespace scribo mln_precondition(input.is_valid()); mln_precondition(nbh.is_valid()); - typedef mln_ch_value(I,value::label_16) L; - /// Finding comps. - value::label_16 ncomps; + typedef mln_ch_value(I,value::int_u16) L; + value::int_u16 ncomps; component_set<L> - comps = scribo::primitive::extract::components(input, nbh, ncomps); + comps = scribo::primitive::extract::components(input, c8(), ncomps); /// First filtering. - component_set<L> filtered_comps - = scribo::filter::components_small(comps, 6); + comps = scribo::filter::components_small(comps, 3); + + /// Use separators. + if (exact(separators).is_valid()) + comps.add_separators(separators); /// Linking potential comps object_links<L> left_link - = primitive::link::with_single_left_link_dmax_ratio(filtered_comps); + = primitive::link::with_single_left_link_dmax_ratio(comps, + primitive::link::internal::dmax_width_and_height(1), + anchor::MassCenter); object_links<L> right_link - = primitive::link::with_single_right_link_dmax_ratio(filtered_comps); - + = primitive::link::with_single_right_link_dmax_ratio(comps, + primitive::link::internal::dmax_width_and_height(1), + anchor::MassCenter); // Validating left and right links. object_links<L> merged_links = primitive::link::merge_double_link(left_link, - right_link); + right_link); - // Remove links if bboxes have too different sizes. - object_links<L> hratio_filtered_links - = filter::object_links_bbox_h_ratio(merged_links, 2.0f); + object_links<L> hratio_filtered_links + = filter::object_links_bbox_h_ratio(merged_links, 2.5f); object_groups<L> groups = primitive::group::from_single_link(hratio_filtered_links); - - line_set<L> line = scribo::make::line_set(groups); - line = text::merging(line); - + line_set<L> lines(groups); + lines = text::merging(lines); trace::exiting("scribo::text::extract_lines"); - return line; + return lines; } # endif // ! MLN_INCLUDE_ONLY -- 1.5.6.5

1 0

last-svn-commit-781-g6fdc4cc Make use of mln::fun::v2v::rgb_to_luma.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

* scribo/toolchain/internal/text_in_doc_preprocess_functor.hh, * src/binarization/ppm_sauvola.cc, * src/binarization/ppm_sauvola_ms.cc, * src/binarization/ppm_sauvola_ms_fg.cc, * src/binarization/ppm_sauvola_ms_split.cc, * src/binarization/sauvola.cc, * src/binarization/sauvola_debug.cc, * src/binarization/sauvola_ms.cc, * src/binarization/sauvola_ms_debug.cc, * src/binarization/sauvola_ms_fg.cc, * src/text_in_picture.cc, * src/text_in_picture_neg.cc, * src/text_recognition_in_picture.cc: Here. --- scribo/ChangeLog | 18 ++++++++++++++++++ .../internal/text_in_doc_preprocess_functor.hh | 7 ++++--- scribo/src/binarization/ppm_sauvola.cc | 9 +++++---- scribo/src/binarization/ppm_sauvola_ms.cc | 9 +++++---- scribo/src/binarization/ppm_sauvola_ms_fg.cc | 8 ++++---- scribo/src/binarization/ppm_sauvola_ms_split.cc | 5 ++--- scribo/src/binarization/sauvola.cc | 4 ++-- scribo/src/binarization/sauvola_debug.cc | 8 ++++---- scribo/src/binarization/sauvola_ms.cc | 9 +++++---- scribo/src/binarization/sauvola_ms_debug.cc | 9 +++++---- scribo/src/binarization/sauvola_ms_fg.cc | 8 ++++---- scribo/src/text_in_picture.cc | 7 ++++--- scribo/src/text_in_picture_neg.cc | 5 ++--- scribo/src/text_recognition_in_picture.cc | 9 +++++---- 14 files changed, 69 insertions(+), 46 deletions(-) diff --git a/scribo/ChangeLog b/scribo/ChangeLog index cf51ca8..f740230 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,23 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make use of mln::fun::v2v::rgb_to_luma. + + * scribo/toolchain/internal/text_in_doc_preprocess_functor.hh, + * src/binarization/ppm_sauvola.cc, + * src/binarization/ppm_sauvola_ms.cc, + * src/binarization/ppm_sauvola_ms_fg.cc, + * src/binarization/ppm_sauvola_ms_split.cc, + * src/binarization/sauvola.cc, + * src/binarization/sauvola_debug.cc, + * src/binarization/sauvola_ms.cc, + * src/binarization/sauvola_ms_debug.cc, + * src/binarization/sauvola_ms_fg.cc, + * src/text_in_picture.cc, + * src/text_in_picture_neg.cc, + * src/text_recognition_in_picture.cc: Here. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/io/xml/load.hh: New XML loader. 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> diff --git a/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh b/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh index 6a9506b..6c0dd5a 100644 --- a/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh +++ b/scribo/scribo/toolchain/internal/text_in_doc_preprocess_functor.hh @@ -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. // @@ -29,7 +30,7 @@ #include <mln/core/concept/image.hh> #include <mln/data/transform.hh> #include <mln/data/convert.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <mln/subsampling/antialiased.hh> #include <scribo/binarization/sauvola.hh> @@ -208,7 +209,7 @@ namespace scribo on_new_progress_label("Convert to gray-scale image"); image2d<value::int_u8> intensity_ima = mln::data::transform(input_rgb, - mln::fun::v2v::rgb_to_int_u<8>()); + mln::fun::v2v::rgb_to_luma<value::int_u8>()); on_progress(); diff --git a/scribo/src/binarization/ppm_sauvola.cc b/scribo/src/binarization/ppm_sauvola.cc index f0cd355..4732a02 100644 --- a/scribo/src/binarization/ppm_sauvola.cc +++ b/scribo/src/binarization/ppm_sauvola.cc @@ -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. // @@ -27,7 +27,7 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola.hh> #include <scribo/debug/usage.hh> @@ -74,7 +74,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_gl = data::transform(input, mln::fun::v2v::rgb_to_int_u<8>()); + input_gl = data::transform(input, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> out = scribo::binarization::sauvola(input_gl, w, k); diff --git a/scribo/src/binarization/ppm_sauvola_ms.cc b/scribo/src/binarization/ppm_sauvola_ms.cc index eb694c2..cff28f9 100644 --- a/scribo/src/binarization/ppm_sauvola_ms.cc +++ b/scribo/src/binarization/ppm_sauvola_ms.cc @@ -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. // @@ -29,7 +29,7 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/debug/usage.hh> @@ -110,7 +110,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input_1, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/binarization/ppm_sauvola_ms_fg.cc b/scribo/src/binarization/ppm_sauvola_ms_fg.cc index 20f237f..a17ce9f 100644 --- a/scribo/src/binarization/ppm_sauvola_ms_fg.cc +++ b/scribo/src/binarization/ppm_sauvola_ms_fg.cc @@ -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. // @@ -31,7 +31,7 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/preprocessing/split_bg_fg.hh> @@ -122,7 +122,7 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/binarization/ppm_sauvola_ms_split.cc b/scribo/src/binarization/ppm_sauvola_ms_split.cc index bb98b38..92a0817 100644 --- a/scribo/src/binarization/ppm_sauvola_ms_split.cc +++ b/scribo/src/binarization/ppm_sauvola_ms_split.cc @@ -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. // @@ -29,7 +29,6 @@ #include <mln/io/ppm/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> #include <scribo/binarization/sauvola_ms_split.hh> #include <scribo/debug/usage.hh> diff --git a/scribo/src/binarization/sauvola.cc b/scribo/src/binarization/sauvola.cc index 0273071..e8047ab 100644 --- a/scribo/src/binarization/sauvola.cc +++ b/scribo/src/binarization/sauvola.cc @@ -29,7 +29,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola.hh> #include <scribo/debug/usage.hh> @@ -77,7 +77,7 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input, mln::fun::v2v::rgb_to_luma<value::int_u8>()); image2d<bool> out = scribo::binarization::sauvola(input_1_gl, w, k); diff --git a/scribo/src/binarization/sauvola_debug.cc b/scribo/src/binarization/sauvola_debug.cc index f723851..0c335a8 100644 --- a/scribo/src/binarization/sauvola_debug.cc +++ b/scribo/src/binarization/sauvola_debug.cc @@ -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. // @@ -32,7 +32,7 @@ #include <mln/data/convert.hh> #include <mln/data/saturate.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/local_threshold.hh> #include <scribo/binarization/sauvola.hh> @@ -116,7 +116,7 @@ int main(int argc, char *argv[]) image2d<value::int_u8> gima = data::transform(input, - mln::fun::v2v::rgb_to_int_u<8>()); + mln::fun::v2v::rgb_to_luma<value::int_u8>()); image2d<bool> diff --git a/scribo/src/binarization/sauvola_ms.cc b/scribo/src/binarization/sauvola_ms.cc index 541b9e5..6d60ab4 100644 --- a/scribo/src/binarization/sauvola_ms.cc +++ b/scribo/src/binarization/sauvola_ms.cc @@ -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. // @@ -31,7 +31,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/debug/usage.hh> @@ -115,7 +115,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input_1, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/binarization/sauvola_ms_debug.cc b/scribo/src/binarization/sauvola_ms_debug.cc index 6bf9837..70c1a9a 100644 --- a/scribo/src/binarization/sauvola_ms_debug.cc +++ b/scribo/src/binarization/sauvola_ms_debug.cc @@ -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. // @@ -29,7 +29,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/debug/usage.hh> @@ -124,7 +124,8 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - input_1_gl = data::transform(input_1, mln::fun::v2v::rgb_to_int_u<8>()); + input_1_gl = data::transform(input_1, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize. diff --git a/scribo/src/binarization/sauvola_ms_fg.cc b/scribo/src/binarization/sauvola_ms_fg.cc index 4227db4..7ff9321 100644 --- a/scribo/src/binarization/sauvola_ms_fg.cc +++ b/scribo/src/binarization/sauvola_ms_fg.cc @@ -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. // @@ -31,7 +31,7 @@ #include <mln/io/magick/load.hh> #include <mln/io/pbm/save.hh> #include <mln/data/transform.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <scribo/binarization/sauvola_ms.hh> #include <scribo/preprocessing/split_bg_fg.hh> @@ -102,7 +102,7 @@ int main(int argc, char *argv[]) // Convert to Gray level image. image2d<value::int_u8> - fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + fg_gl = data::transform(fg, mln::fun::v2v::rgb_to_luma<value::int_u8>()); // Binarize image2d<bool> diff --git a/scribo/src/text_in_picture.cc b/scribo/src/text_in_picture.cc index e2f30de..77cf7b6 100644 --- a/scribo/src/text_in_picture.cc +++ b/scribo/src/text_in_picture.cc @@ -40,7 +40,7 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <mln/subsampling/antialiased.hh> @@ -236,7 +236,8 @@ int main(int argc, char* argv[]) std::cout << "** Using split_bg_fg" << std::endl; image2d<value::rgb8> fg = preprocessing::split_bg_fg(input_rgb, lambda, 32).second(); - intensity_ima = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + intensity_ima = data::transform(fg, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); t_ = timer_; std::cout << "Foreground extracted. " << t_ << std::endl; @@ -253,7 +254,7 @@ int main(int argc, char* argv[]) timer_.start(); std::cout << "** Using data::transform(intensity)" << std::endl; intensity_ima = data::transform(input_rgb, - mln::fun::v2v::rgb_to_int_u<8>()); + mln::fun::v2v::rgb_to_luma<value::int_u8>()); t_ = timer_; std::cout << "Intensity image " << t_ << std::endl; } diff --git a/scribo/src/text_in_picture_neg.cc b/scribo/src/text_in_picture_neg.cc index c1a4317..8d818e0 100644 --- a/scribo/src/text_in_picture_neg.cc +++ b/scribo/src/text_in_picture_neg.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. // @@ -48,8 +49,6 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> - #include <mln/data/wrap.hh> #include <mln/draw/box.hh> diff --git a/scribo/src/text_recognition_in_picture.cc b/scribo/src/text_recognition_in_picture.cc index a5f93a7..8b76f48 100644 --- a/scribo/src/text_recognition_in_picture.cc +++ b/scribo/src/text_recognition_in_picture.cc @@ -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. // @@ -51,7 +51,7 @@ #include <mln/value/rgb8.hh> #include <mln/value/label_16.hh> -#include <mln/fun/v2v/rgb_to_int_u.hh> +#include <mln/fun/v2v/rgb_to_luma.hh> #include <mln/data/wrap.hh> @@ -253,7 +253,8 @@ int main(int argc, char* argv[]) // Extract foreground image2d<value::rgb8> fg = preprocessing::split_bg_fg(input_rgb, lambda, 32).second(); - intensity_ima = data::transform(fg, mln::fun::v2v::rgb_to_int_u<8>()); + intensity_ima = data::transform(fg, + mln::fun::v2v::rgb_to_luma<value::int_u8>()); // // Perform an initial rotation if needed. // // input_rgb = geom::rotate(input_rgb, -45, literal::black); -- 1.5.6.5

1 0

last-svn-commit-780-g7f4a367 mln/labeling/fill_holes.hh: Improve speed.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

--- milena/ChangeLog | 4 ++++ milena/mln/labeling/fill_holes.hh | 17 +++++++++-------- 2 files changed, 13 insertions(+), 8 deletions(-) diff --git a/milena/ChangeLog b/milena/ChangeLog index edefc8e..2eed916 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * mln/labeling/fill_holes.hh: Improve speed. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * mln/fun/v2v/rgb_to_luma.hh: New function for grayscale conversion. diff --git a/milena/mln/labeling/fill_holes.hh b/milena/mln/labeling/fill_holes.hh index e76c489..5e487d9 100644 --- a/milena/mln/labeling/fill_holes.hh +++ b/milena/mln/labeling/fill_holes.hh @@ -1,4 +1,5 @@ -// Copyright (C) 2007, 2008, 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2007, 2008, 2009, 2011 EPITA Research and Development +// Laboratory (LRDE) // // This file is part of Olena. // @@ -33,6 +34,8 @@ # include <mln/labeling/background.hh> # include <mln/labeling/compute.hh> +# include <mln/data/transform.hh> + # include <mln/core/image/dmorph/image_if.hh> # include <mln/accu/math/count.hh> @@ -57,7 +60,7 @@ namespace mln /// \see mln::labeling::background /// template <typename I, typename N, typename L> - I + mln_concrete(I) fill_holes(const Image& input, const Neighborhood<N>& nbh, L& nlabels); @@ -66,7 +69,7 @@ namespace mln template <typename I, typename N, typename L> inline - I + mln_concrete(I) fill_holes(const Image& input, const Neighborhood<N>& nbh, L& nlabels) { @@ -77,10 +80,6 @@ namespace mln mln_precondition(exact(input).is_valid()); mln_precondition(exact(nbh).is_valid()); - mln_ch_value(I, bool) output; - initialize(output, input); - data::fill(output, false); - mln_ch_value(I, L) lbls = labeling::background(input, nbh, nlabels); accu::math::count<mln_value(I)> a_; @@ -99,7 +98,9 @@ namespace mln } } - data::fill((output | (pw::value(lbls) != bg_lbl)).rw(), true); + util::array<bool> bg_relbl(arr.nelements(), true); + bg_relbl(bg_lbl) = false; + mln_ch_value(I, bool) output = data::transform(lbls, bg_relbl); trace::exiting("labeling::fill_holes"); return output; -- 1.5.6.5

1 0

last-svn-commit-779-ge270275 mln/fun/v2v/rgb_to_luma.hh: New function for grayscale conversion.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

--- milena/ChangeLog | 5 +++ .../achromatism.hh => mln/fun/v2v/rgb_to_luma.hh} | 32 ++++++++++++------- 2 files changed, 25 insertions(+), 12 deletions(-) copy milena/{sandbox/green/mln/fun/v2v/achromatism.hh => mln/fun/v2v/rgb_to_luma.hh} (67%) diff --git a/milena/ChangeLog b/milena/ChangeLog index c4bed2c..edefc8e 100644 --- a/milena/ChangeLog +++ b/milena/ChangeLog @@ -1,3 +1,8 @@ +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + + * mln/fun/v2v/rgb_to_luma.hh: New function for grayscale + conversion. + 2011-02-17 Guillaume Lazzara <z(a)lrde.epita.fr> * mln/util/array.hh: Add last() method. diff --git a/milena/sandbox/green/mln/fun/v2v/achromatism.hh b/milena/mln/fun/v2v/rgb_to_luma.hh similarity index 67% copy from milena/sandbox/green/mln/fun/v2v/achromatism.hh copy to milena/mln/fun/v2v/rgb_to_luma.hh index 72b545c..304a798 100644 --- a/milena/sandbox/green/mln/fun/v2v/achromatism.hh +++ b/milena/mln/fun/v2v/rgb_to_luma.hh @@ -1,4 +1,4 @@ -// Copyright (C) 2008, 2009 EPITA Research and Development Laboratory (LRDE) +// Copyright (C) 2011 EPITA Research and Development Laboratory (LRDE) // // This file is part of Olena. // @@ -23,10 +23,10 @@ // exception does not however invalidate any other reasons why the // executable file might be covered by the GNU General Public License. -#ifndef MLN_FUN_V2V_ACHROMATISM_HH -# define MLN_FUN_V2V_ACHROMATISM_HH +#ifndef MLN_FUN_V2V_RGB_TO_LUMA_HH +# define MLN_FUN_V2V_RGB_TO_LUMA_HH -# include <mln/value/rgb8.hh> +# include <mln/value/rgb.hh> namespace mln { @@ -37,22 +37,30 @@ namespace mln namespace v2v { - struct achromatism : public Function_v2v< achromatism > + template <typename T_luma> + struct rgb_to_luma : public Function_v2v< rgb_to_luma<T_luma> > { - typedef float result; + typedef T_luma result; + + template <typename T_rgb> + T_luma operator()(const T_rgb& rgb) const; - float operator()(const value::rgb8 rgb) const; }; + # ifndef MLN_INCLUDE_ONLY - float achromatism::operator()(const value::rgb8 rgb) const + template <typename T_luma> + template <typename T_rgb> + inline + T_luma + rgb_to_luma<T_luma>::operator()(const T_rgb& rgb) const { - return (math::abs(rgb.red() - rgb.green()) - + math::abs(rgb.red() - rgb.blue()) - + math::abs(rgb.green() - rgb.blue()))/3.0; + float luma = 0.299 * rgb.red() + 0.587 * rgb.green() + 0.114 * rgb.blue(); + return unsigned(luma + 0.49999); } + # endif // !MLN_INCLUDE_ONLY } // end of namespace fun::v2v @@ -61,4 +69,4 @@ namespace mln } // end of namespace mln -#endif // ! MLN_FUN_V2V_ACHROMATISM_HH +#endif // ! MLN_FUN_V2V_RGB_TO_LUMA_HH -- 1.5.6.5

1 0

last-svn-commit-778-g134e242 scribo/io/xml/load.hh: New XML loader.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

--- scribo/ChangeLog | 4 + scribo/scribo/io/xml/load.hh | 525 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 529 insertions(+), 0 deletions(-) create mode 100644 scribo/scribo/io/xml/load.hh diff --git a/scribo/ChangeLog b/scribo/ChangeLog index 6571137..cf51ca8 100644 --- a/scribo/ChangeLog +++ b/scribo/ChangeLog @@ -1,5 +1,9 @@ 2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + * scribo/io/xml/load.hh: New XML loader. + +2011-03-01 Guillaume Lazzara <z(a)lrde.epita.fr> + Make XML output more flexible. * scribo/core/component_info.hh, diff --git a/scribo/scribo/io/xml/load.hh b/scribo/scribo/io/xml/load.hh new file mode 100644 index 0000000..e0f4548 --- /dev/null +++ b/scribo/scribo/io/xml/load.hh @@ -0,0 +1,525 @@ +// 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_LOAD_HH +# define SCRIBO_IO_XML_LOAD_HH + +/// \file +/// +/// \brief Load document information from XML. + +# include <QtXml> + +# include <libgen.h> +# include <fstream> +# include <sstream> + +# include <map> + +# include <mln/core/image/image2d.hh> + +# include <mln/data/wrap.hh> +# include <mln/value/int_u8.hh> +# include <mln/io/pgm/save.hh> +# include <mln/io/pbm/save.hh> + +# include <scribo/core/document.hh> +# include <scribo/core/component_set.hh> +# include <scribo/core/line_set.hh> +# include <scribo/core/line_info.hh> + +# include <scribo/convert/from_base64.hh> + +namespace scribo +{ + + namespace io + { + + namespace xml + { + using namespace mln; + + /*! \brief Load document information from 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… + + */ + template <typename L> + void + load(document<L>& doc, const std::string& input_name); + + +# ifndef MLN_INCLUDE_ONLY + + namespace internal + { + + + enum Mode + { + None, + ComponentSet, + ComponentInfo, + LabeledImage, + SeparatorsImage, + ObjectLinks, + ObjectGroups, + Point, + Link, + Group, + Line, + LineLinks, + LineLink, + TextData, + TextRegion, + CompIdList, + CompId, + Page + }; + + + struct ModeData + { + const char *name; + Mode mode; + }; + + + static const ModeData mode_data[] = { + { "component_set", ComponentSet }, + { "component_info", ComponentInfo }, + { "labeled_image", LabeledImage }, + { "separators_image", SeparatorsImage }, + { "object_links", ObjectLinks }, + { "object_groups", ObjectGroups }, + { "point", Point }, + { "link", Link }, + { "group", Group }, + { "line", Line }, + { "line_links", LineLinks }, + { "line_link", LineLink }, + { "text_data", TextData }, + { "text_region", TextRegion }, + { "compid_list", CompIdList }, + { "compid", CompId }, + { "page", Page }, + { 0, None } + }; + + + template <typename L> + class xml_handler : public QXmlDefaultHandler + { + + typedef mln_ch_value(L,bool) B; + + public: + xml_handler() : current_paragraph_id(1) { lines_data.append(line_info<L>()); } // line info id starts from 1. + + virtual + bool + startElement(const QString& /*namespaceURI*/, const QString& /*localName*/, + const QString& qName, const QXmlAttributes& atts ) + { + mode.push(find_mode(qName)); + + switch (mode.top()) + { + + // Component Set + case ComponentSet: + { + comp_set_data = new scribo::internal::component_set_data<L>(); + // qDebug() << qName << " - atts.value(\"nelements\").toInt() = " << atts.value("nelements").toInt();; + comp_set_data->soft_init(atts.value("nelements").toInt()); + } + break; + + + // Component Info + case ComponentInfo: + { + component_info info(atts.value("id").toInt(), + mln::make::box2d(atts.value("pmin_y").toInt(), + atts.value("pmin_x").toInt(), + atts.value("pmax_y").toInt(), + atts.value("pmax_x").toInt()), + mln::point2d(atts.value("mass_center_y").toInt(), + atts.value("mass_center_x").toInt()), + atts.value("card").toInt()); + + + info.update_tag(component::str2tag(atts.value("tag").toUtf8().constData())); + info.update_type(component::str2type(atts.value("type").toUtf8().constData())); + + comp_set_data->infos_.append(info); + } + break; + + + // Object links + case ObjectLinks: + { + // qDebug() << "object_links created"; + links = object_links<L>(components); + } + break; + + + // Object groups + case ObjectGroups: + { + // qDebug() << "object_groups created"; + groups = object_groups<L>(links); + } + break; + + + // Text data + case TextData: + { + // qDebug() << "TextData"; + + // Reserve space for line data. + lines_data.resize(atts.value("nlines").toInt() + 1); + // qDebug() << "line_set created"; + lines = line_set<L>(groups, lines_data); + llinks = line_links<L>(lines); + par_data = new scribo::internal::paragraph_set_data<L>(llinks, atts.value("nparagraphs").toInt()); + // std::cout << par_data->pars_.nelements() << " - " << llinks.nelements() << " - " << lines.nelements() << std::endl; + } + break; + + + // Text Region + case TextRegion: + { + // qDebug() << "TextRegion"; + + current_paragraph = paragraph_info<L>(llinks); + } + break; + + + // Line link + case LineLink: + { + llinks(atts.value("from").toInt()) = atts.value("to").toInt(); + } + break; + + + // Line + case Line: + { + current_line_id = atts.value("id").toInt(); + + line_data = new scribo::internal::line_info_data<L>(lines, mln::util::array<component_id_t>()); + + line_data->holder_ = lines; + line_data->text_ = atts.value("text").toUtf8().constData(); + + line_data->hidden_ = false; + line_data->tag_ = static_cast<line::Tag>(atts.value("tag").toInt()); + + line_data->baseline_ = atts.value("baseline").toInt(); + line_data->meanline_ = atts.value("meanline").toInt(); + line_data->x_height_ = atts.value("x_height").toInt(); + line_data->d_height_ = atts.value("d_height").toInt(); + line_data->a_height_ = atts.value("a_height").toInt(); + line_data->char_space_ = atts.value("kerning").toInt(); + line_data->char_width_ = atts.value("char_width").toInt(); + line_data->word_space_ = 0; + + line_data->reading_direction_ = line::LeftToRight; + line_data->type_ = line::str2type(atts.value("txt_text_type").toAscii().constData()); + line_data->reverse_video_ = (atts.value("txt_reverse_video") == "false" ? false : true); + line_data->orientation_ = 0; + line_data->reading_orientation_ = atts.value("txt_reading_orientation").toInt(); + line_data->indented_ = (atts.value("txt_indented") == "false" ? false : true); + + bbox.init(); + } + break; + + + // CompIdList + case CompIdList: + { + + } + break; + + + // CompId + case CompId: + { + line_data->components_.append(atts.value("value").toInt()); + } + break; + + + // Point + case Point: + { + point2d p(atts.value("y").toInt(), atts.value("x").toInt()); + bbox.take(p); + } + break; + + + // Labeled Image + case LabeledImage: + { + width = atts.value("width").toInt(); + height = atts.value("height").toInt(); + comp_set_data->ima_ = L(mln::make::box2d(height, width), 0); // No border + } + break; + + + // Separator Image + case SeparatorsImage: + { + width = atts.value("width").toInt(); + height = atts.value("height").toInt(); + comp_set_data->separators_ = B(mln::make::box2d(height, width), 0); // No border + } + break; + + + // Link + case Link: + { + links(atts.value("from").toInt()) = atts.value("to").toInt(); + } + break; + + + // Group + case Group: + { + groups(atts.value("object_id").toInt()) = atts.value("group_id").toInt(); + } + break; + + + // DEFAULT + default: + ; + } + + return true; + } + + + virtual + bool + endElement(const QString& /*namespaceURI*/, const QString& /*localName*/, const QString& /*qName*/) + { + switch(mode.top()) + { + // Component set + case ComponentSet: + { + // qDebug() << "Component set done"; + components = component_set<L>(comp_set_data); + } + break; + + // Line + case Line: + { + // qDebug() << "Line done"; + line_data->bbox_ = bbox; + lines_data(current_line_id) = line_info<L>(current_line_id, line_data); + lines_data(current_line_id).update_ebbox(); + + // Add this line to the current paragraph. + current_paragraph.add_line(lines_data(current_line_id)); + } + break; + + // TextRegion + case TextRegion: + { + // qDebug() << TextRegion; + par_data->pars_(current_paragraph_id++) = current_paragraph; + } + break; + + // Page + case Page: + { + // qDebug() << "Page done"; + lines.update_line_data_(lines_data); + parset = paragraph_set<L>(par_data); + } + break; + + // DEFAULT + default: + ; + + } + + mode.pop(); + return true; + } + + + + bool characters(const QString & ch) + { + switch (mode.top()) + { + case LabeledImage: + { + scribo::convert::from_base64(ch, comp_set_data->ima_); + } + break; + + case SeparatorsImage: + { + scribo::convert::from_base64(ch, comp_set_data->separators_); + } + break; + + default: + ; + } + + return true; + } + + +// private: // Methods + + Mode find_mode(const QString& qName) + { + for (int i = 0; mode_data[i].name; ++i) + if (mode_data[i].name == qName) + return mode_data[i].mode; + return None; + } + + +// private: // Attributes + + QStack<Mode> mode; + + // Shape + accu::shape::bbox<point2d> bbox; + + unsigned width; + unsigned height; + + // Components + mln::util::tracked_ptr<scribo::internal::component_set_data<L> > comp_set_data; + component_set<L> components; + + object_links<L> links; + object_groups<L> groups; + + // Lines + unsigned current_line_id; + scribo::internal::line_info_data<L> *line_data; + + line_links<L> llinks; + + unsigned current_paragraph_id; + paragraph_info<L> current_paragraph; + scribo::internal::paragraph_set_data<L> *par_data; + paragraph_set<L> parset; + + mln::util::array<line_info<L> > lines_data; + line_set<L> lines; + }; + + + + + + + + + + + template <typename L> + void + load_extended(document<L>& doc, + const std::string& output_name) + { + xml_handler<L> handler; + QXmlSimpleReader reader; + reader.setContentHandler(&handler); + + QFile file(output_name.c_str()); + if (!file.open(QFile::ReadOnly | QFile::Text)) + { + qDebug() << "Cannot read file"; + return; + } + + QXmlInputSource xmlInputSource(&file); + if (reader.parse(xmlInputSource)) + qDebug() << "Loaded successfuly"; + + doc.set_paragraphs(handler.parset); + } + + } // end of namespace scribo::io::xml::internal + + + // FACADE + + template <typename L> + void + load(document<L>& doc, + const std::string& output_name) + { + internal::load_extended(doc, output_name); + } + + +# endif // ! MLN_INCLUDE_ONLY + + } // end of namespace scribo::io::xml + + } // end of namespace scribo::io + +} // end of namespace scribo + + +#endif // ! SCRIBO_IO_XML_LOAD_HH + -- 1.5.6.5

1 0

last-svn-commit-777-g6985484 Make XML output more flexible.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

* 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

1 0

last-svn-commit-776-g8c287df Set component type during component extraction.
by Guillaume Lazzara 28 Mar '11

28 Mar '11

* 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

1 0