added ocr tesseract data (tessdata)

jozefbaranec · jozefbaranec · commit 7febf7034449 · 2018-06-20T23:20:54.000+02:00
resources folder update
diff --git a/.gitignore b/.gitignore
@@ -9,4 +9,3 @@ bin/PDFixSDKSample*
 bin/*.dll
 bin/*.dylib
 bin/output/
-bin/resources/tessdata
diff --git a/output/dummy.txt b/output/dummy.txt
@@ -0,0 +1 @@
+all saved files will appear here
diff --git a/resources/README.md b/resources/README.md
@@ -0,0 +1,6 @@
+resources required to run samples
+
+/test.pdf - sample PDF to process
+/watermark.png - image used for watermark
+/test.pfx - sample digital signature 
+/tessdata/ - folder for Tesseract OCR data (copy files from Tesserct official web site https://github.com/tesseract-ocr/tesseract/tree/master/tessdata)
diff --git a/resources/tessdata/Makefile.am b/resources/tessdata/Makefile.am
@@ -0,0 +1,34 @@
+datadir = @datadir@/tessdata
+
+data_DATA = pdf.ttf
+EXTRA_DIST = $(data_DATA)
+
+SUBDIRS = configs tessconfigs
+
+langdata = bul.traineddata mlt.traineddata chr.traineddata \
+	slk.traineddata dan-frak.traineddata eng.traineddata \
+	ces.traineddata afr.traineddata swa.traineddata \
+	kan.traineddata bel.traineddata ind.traineddata \
+	lit.traineddata nld.traineddata osd.traineddata \
+	mkd.traineddata est.traineddata fra.traineddata \
+	hin.traineddata lat_lid.traineddata nor.traineddata \
+	por.traineddata ron.traineddata swe.traineddata \
+	pol.traineddata ara.traineddata tel.traineddata \
+	ell.traineddata mal.traineddata vie.traineddata \
+	heb.traineddata deu.traineddata eus.traineddata \
+	ita_old.traineddata rus.traineddata sqi.traineddata \
+	spa.traineddata glg.traineddata slk-frak.traineddata \
+	equ.traineddata hrv.traineddata frk.traineddata \
+	cat.traineddata lav.traineddata ukr.traineddata \
+	enm.traineddata dan.traineddata fin.traineddata \
+	ben.traineddata srp.traineddata tha.traineddata \
+	hun.traineddata tgl.traineddata frm.traineddata \
+	slv.traineddata chi_sim.traineddata tam.traineddata \
+	tur.traineddata epo.traineddata msa.traineddata \
+	kor.traineddata isl.traineddata jpn.traineddata \
+	chi_tra.traineddata ita.traineddata spa_old.traineddata \
+	deu-frak.traineddata aze.traineddata
+
+uninstall-local:
+	cd $(DESTDIR)$(datadir); \
+	rm -f $(langdata)
diff --git a/resources/tessdata/configs/Makefile.am b/resources/tessdata/configs/Makefile.am
@@ -0,0 +1,3 @@
+datadir = @datadir@/tessdata/configs
+data_DATA = inter makebox box.train unlv ambigs.train lstm.train api_config kannada box.train.stderr quiet logfile digits hocr tsv linebox pdf rebox strokewidth bigram txt
+EXTRA_DIST = inter makebox box.train unlv ambigs.train lstm.train api_config kannada box.train.stderr quiet logfile digits hocr tsv linebox pdf rebox strokewidth bigram txt
diff --git a/resources/tessdata/configs/ambigs.train b/resources/tessdata/configs/ambigs.train
@@ -0,0 +1,7 @@
+tessedit_ambigs_training	1
+load_freq_dawg	0
+load_punc_dawg	0
+load_system_dawg	0
+load_number_dawg	0
+ambigs_debug_level	3
+load_fixed_length_dawgs	0
diff --git a/resources/tessdata/configs/api_config b/resources/tessdata/configs/api_config
@@ -0,0 +1 @@
+tessedit_zero_rejection T
diff --git a/resources/tessdata/configs/bazaar b/resources/tessdata/configs/bazaar
@@ -0,0 +1,4 @@
+load_system_dawg     F
+load_freq_dawg       F
+user_words_suffix    user-words
+user_patterns_suffix user-patterns
diff --git a/resources/tessdata/configs/bigram b/resources/tessdata/configs/bigram
@@ -0,0 +1,5 @@
+load_bigram_dawg	True
+tessedit_enable_bigram_correction	True
+tessedit_bigram_debug	3
+save_raw_choices	True
+save_alt_choices	True
diff --git a/resources/tessdata/configs/box.train b/resources/tessdata/configs/box.train
@@ -0,0 +1,14 @@
+disable_character_fragments T
+file_type                   .bl
+textord_fast_pitch_test	T
+tessedit_single_match	0
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+il1_adaption_test 1
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_resegment_from_boxes T
+tessedit_train_from_boxes T
+textord_no_rejects T
diff --git a/resources/tessdata/configs/box.train.stderr b/resources/tessdata/configs/box.train.stderr
@@ -0,0 +1,15 @@
+file_type .bl
+#tessedit_use_nn F
+textord_fast_pitch_test T
+tessedit_single_match 0
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+il1_adaption_test 1
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_resegment_from_boxes T
+tessedit_train_from_boxes T
+#textord_repeat_extraction F
+textord_no_rejects T
diff --git a/resources/tessdata/configs/digits b/resources/tessdata/configs/digits
@@ -0,0 +1 @@
+tessedit_char_whitelist 0123456789-.
diff --git a/resources/tessdata/configs/get.image b/resources/tessdata/configs/get.image
@@ -0,0 +1 @@
+tessedit_write_images T
diff --git a/resources/tessdata/configs/hocr b/resources/tessdata/configs/hocr
@@ -0,0 +1,3 @@
+tessedit_create_hocr 1
+tessedit_pageseg_mode 1
+hocr_font_info 0
diff --git a/resources/tessdata/configs/inter b/resources/tessdata/configs/inter
@@ -0,0 +1,2 @@
+interactive_display_mode				T
+tessedit_display_outwords		T
diff --git a/resources/tessdata/configs/kannada b/resources/tessdata/configs/kannada
@@ -0,0 +1,4 @@
+textord_skewsmooth_offset 8
+textord_skewsmooth_offset2 8
+textord_merge_desc 0.5
+textord_no_rejects 1
diff --git a/resources/tessdata/configs/linebox b/resources/tessdata/configs/linebox
@@ -0,0 +1,2 @@
+tessedit_resegment_from_line_boxes 1
+tessedit_make_boxes_from_boxes 1
diff --git a/resources/tessdata/configs/logfile b/resources/tessdata/configs/logfile
@@ -0,0 +1 @@
+debug_file tesseract.log
diff --git a/resources/tessdata/configs/lstm.train b/resources/tessdata/configs/lstm.train
@@ -0,0 +1,13 @@
+disable_character_fragments T
+file_type                   .bl
+textord_fast_pitch_test	T
+tessedit_single_match	0
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+il1_adaption_test 1
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_train_line_recognizer T
+textord_no_rejects T
diff --git a/resources/tessdata/configs/makebox b/resources/tessdata/configs/makebox
@@ -0,0 +1 @@
+tessedit_create_boxfile 1
diff --git a/resources/tessdata/configs/pdf b/resources/tessdata/configs/pdf
@@ -0,0 +1,2 @@
+tessedit_create_pdf 1
+tessedit_pageseg_mode 1
diff --git a/resources/tessdata/configs/quiet b/resources/tessdata/configs/quiet
@@ -0,0 +1 @@
+debug_file /dev/null
diff --git a/resources/tessdata/configs/rebox b/resources/tessdata/configs/rebox
@@ -0,0 +1,2 @@
+tessedit_resegment_from_boxes 1
+tessedit_make_boxes_from_boxes 1
diff --git a/resources/tessdata/configs/strokewidth b/resources/tessdata/configs/strokewidth
@@ -0,0 +1,12 @@
+textord_show_blobs 0
+textord_debug_tabfind 3
+textord_tabfind_show_partitions 1
+textord_tabfind_show_initial_partitions 1
+textord_tabfind_show_columns 1
+textord_tabfind_show_blocks 1
+textord_tabfind_show_initialtabs 1
+textord_tabfind_show_finaltabs 1
+textord_tabfind_show_strokewidths 1
+textord_tabfind_show_vlines 0
+textord_tabfind_show_images 1
+tessedit_dump_pageseg_images 0
diff --git a/resources/tessdata/configs/tsv b/resources/tessdata/configs/tsv
@@ -0,0 +1,2 @@
+tessedit_create_tsv 1
+tessedit_pageseg_mode 1
diff --git a/resources/tessdata/configs/txt b/resources/tessdata/configs/txt
@@ -0,0 +1,3 @@
+# This config file should be used with other cofig files which creates renderers.
+# usage example: tesseract eurotext.tif eurotext txt hocr pdf
+tessedit_create_txt 1
diff --git a/resources/tessdata/configs/unlv b/resources/tessdata/configs/unlv
@@ -0,0 +1,2 @@
+tessedit_write_unlv 1
+tessedit_pageseg_mode 6
diff --git a/resources/tessdata/eng.traineddata b/resources/tessdata/eng.traineddata
diff --git a/resources/tessdata/eng.user-patterns b/resources/tessdata/eng.user-patterns
@@ -0,0 +1,2 @@
+1-\d\d\d-GOOG-411
+www.\n\\\*.com
diff --git a/resources/tessdata/eng.user-words b/resources/tessdata/eng.user-words
@@ -0,0 +1,5 @@
+the
+quick
+brown
+fox
+jumped
diff --git a/resources/tessdata/osd.traineddata b/resources/tessdata/osd.traineddata
diff --git a/resources/tessdata/pdf.ttf b/resources/tessdata/pdf.ttf
diff --git a/resources/tessdata/tessconfigs/Makefile.am b/resources/tessdata/tessconfigs/Makefile.am
@@ -0,0 +1,3 @@
+datadir = @datadir@/tessdata/tessconfigs
+data_DATA = batch batch.nochop nobatch matdemo segdemo msdemo
+EXTRA_DIST = batch batch.nochop nobatch matdemo segdemo msdemo
diff --git a/resources/tessdata/tessconfigs/batch b/resources/tessdata/tessconfigs/batch
@@ -0,0 +1,2 @@
+# No content needed as all defaults are correct.
+
diff --git a/resources/tessdata/tessconfigs/batch.nochop b/resources/tessdata/tessconfigs/batch.nochop
@@ -0,0 +1,2 @@
+chop_enable 0
+wordrec_enable_assoc 0
diff --git a/resources/tessdata/tessconfigs/matdemo b/resources/tessdata/tessconfigs/matdemo
@@ -0,0 +1,7 @@
+#################################################
+# Adaptive Matcher Using PreAdapted Templates
+#################################################
+
+classify_enable_adaptive_debugger   1
+matcher_debug_flags         6
+matcher_debug_level       1
diff --git a/resources/tessdata/tessconfigs/msdemo b/resources/tessdata/tessconfigs/msdemo
@@ -0,0 +1,13 @@
+#################################################
+# Adaptive Matcher Using PreAdapted Templates
+#################################################
+
+classify_enable_adaptive_debugger	1
+matcher_debug_flags         6
+matcher_debug_level       1
+
+wordrec_display_splits          0
+wordrec_display_all_words       1
+wordrec_display_all_blobs       1
+wordrec_display_segmentations   2
+classify_debug_level		1
diff --git a/resources/tessdata/tessconfigs/nobatch b/resources/tessdata/tessconfigs/nobatch
@@ -0,0 +1 @@
+
diff --git a/resources/tessdata/tessconfigs/segdemo b/resources/tessdata/tessconfigs/segdemo
@@ -0,0 +1,10 @@
+#################################################
+# Adaptive Matcher Using PreAdapted Templates
+#################################################
+
+wordrec_display_splits          0
+wordrec_display_all_words       1
+wordrec_display_all_blobs       1
+wordrec_display_segmentations   2
+classify_debug_level		1
+stopper_debug_level 1
diff --git a/resources/test.pdf b/resources/test.pdf
diff --git a/resources/test.pfx b/resources/test.pfx
diff --git a/resources/watermark.png b/resources/watermark.png
diff --git a/src/Main.cpp b/src/Main.cpp
@@ -42,58 +42,62 @@ int main()
 {
   std::wstring email = L"YOUR_EMAIL";                                     // authorization email   
   std::wstring key = L"LICENSE_KEY";                                      // authorization license key
-  std::wstring open_path = GetAbsolutePath(L"resources/test.pdf");        // source PDF document
-  std::wstring config_path = GetAbsolutePath(L"resources/config.json");   // configuration file
+
+  std::wstring resources_dir = GetAbsolutePath(L"../resources");
+  std::wstring output_dir = GetAbsolutePath(L"../output");
+
+  std::wstring open_path = resources_dir + L"/test.pdf";        // source PDF document
+  std::wstring config_path = resources_dir + L"/config.json";   // configuration file
 
   try {
     std::cout << "Initialization" << std::endl;
     Initialization(email, key);
 
     std::cout << "AddComment" << std::endl;
-    AddComment(email, key, open_path, GetAbsolutePath(L"output/AddComment.pdf"));
+    AddComment(email, key, open_path, output_dir + L"/AddComment.pdf");
 
     std::cout << "AddTags" << std::endl;
-    AddTags(email, key, open_path, GetAbsolutePath(L"output/AddTags.pdf"), config_path);
+    AddTags(email, key, open_path, output_dir + L"/AddTags.pdf", config_path);
 
     std::cout << "AddWatermark" << std::endl;
     PdfWatermarkParams watermark_params;
-    AddWatermark(email, key, open_path, GetAbsolutePath(L"output/AddWatermark.pdf"), 
-      GetAbsolutePath(L"resources/watermark.png"), watermark_params);
+    AddWatermark(email, key, open_path, output_dir + L"/AddWatermark.pdf", 
+      resources_dir + L"/watermark.png", watermark_params);
 
     std::cout << "ConvertToHtml" << std::endl;
     PdfHtmlParams html_params;
-    ConvertToHtml(email, key, open_path, GetAbsolutePath(L"output/index.html"), config_path, 
+    ConvertToHtml(email, key, open_path, output_dir + L"/index.html", config_path, 
       html_params);
 
     std::cout << "DigitalSignature" << std::endl;
-    DigitalSignature(email, key, open_path, GetAbsolutePath(L"output/DigitalSignature.pdf"),
-      GetAbsolutePath(L"resources/test.pfx"), L"TEST_PASSWORD");
+    DigitalSignature(email, key, open_path, output_dir + L"/DigitalSignature.pdf",
+      resources_dir + L"/test.pfx", L"TEST_PASSWORD");
 
     std::cout << "DocumentMetadata" << std::endl;
-    DocumentMetadata(email, key, open_path, GetAbsolutePath(L"output/DocumentMetadata.pdf"),
-      GetAbsolutePath(L"output/metadata.xml"));
+    DocumentMetadata(email, key, open_path, output_dir + L"/DocumentMetadata.pdf",
+      output_dir + L"/metadata.xml");
 
     std::cout << "EmbedFonts" << std::endl;
-    EmbedFonts(email, key, open_path, GetAbsolutePath(L"output/EmbedFonts.pdf"));
+    EmbedFonts(email, key, open_path, output_dir + L"/EmbedFonts.pdf");
 
     std::cout << "ExportFormFieldValues" << std::endl;
     ExportFormFieldValues(email, key, open_path,
-      GetAbsolutePath(L"output/ExportFormFieldValues.txt"));
+      output_dir + L"/ExportFormFieldValues.txt");
 
     std::cout << "ExtractImages" << std::endl;
     PdfImageParams image_params;
-    ExtractImages(email, key, open_path, GetAbsolutePath(L"output/"), 800, 
+    ExtractImages(email, key, open_path, output_dir + L"/", 800, 
       image_params);
 
     std::cout << "ExtractTables" << std::endl;
-    ExtractTables(email, key, open_path, GetAbsolutePath(L"output/"));
+    ExtractTables(email, key, open_path, output_dir + L"/");
     
     std::cout << "ExtractText" << std::endl;
-    ExtractText(email, key, open_path, GetAbsolutePath(L"output/ExtractText.txt"), config_path);
+    ExtractText(email, key, open_path, output_dir + L"/ExtractText.txt", config_path);
 
     std::cout << "FlattenAnnots" << std::endl;
     PdfFlattenAnnotsParams flatten_annots_params;
-    FlattenAnnots(email, key, open_path, GetAbsolutePath(L"output/FlattenAnnots.pdf"), 
+    FlattenAnnots(email, key, open_path, output_dir + L"/FlattenAnnots.pdf", 
       flatten_annots_params);
 
     std::cout << "GetBookmarks" << std::endl;
@@ -103,13 +107,13 @@ int main()
     GetWhitespace(email, key, open_path);
     
     std::cout << "MakeAccessible" << std::endl;
-    MakeAccessible(email, key, open_path, GetAbsolutePath(L"output/MakeAccessible.pdf"),
+    MakeAccessible(email, key, open_path, output_dir + L"/MakeAccessible.pdf",
       L"en", L"Document Title", config_path);
 
     std::cout << "OcrWithTesseract" << std::endl;
     OcrTesseractParams ocr_params;
-    OcrWithTesseract(email, key, open_path, GetAbsolutePath(L"output/OcrTesseract.pdf"), 
-      GetAbsolutePath(L"resources/"), L"eng", ocr_params);
+    OcrWithTesseract(email, key, open_path, output_dir + L"/OcrTesseract.pdf", 
+      resources_dir, L"eng", ocr_params);
 
     std::cout << "PrintPage" << std::endl;
     PrintPage(email, key, open_path);
@@ -124,16 +128,16 @@ int main()
     RegisterEvent(email, key, open_path);
 
     std::cout << "RemoveComments" << std::endl;
-    RemoveComments(email, key, open_path, GetAbsolutePath(L"output/RemoveComments.pdf"));
+    RemoveComments(email, key, open_path, output_dir + L"/RemoveComments.pdf");
 
     std::cout << "RenderPage" << std::endl;
-    RenderPage(email, key, open_path, GetAbsolutePath(L"output/RenderPage.jpg"), 2.0, kRotate0);
+    RenderPage(email, key, open_path, output_dir + L"/RenderPage.jpg", 2.0, kRotate0);
     
     std::cout << "SetFieldFlags" << std::endl;
-    SetFieldFlags(email, key, open_path, GetAbsolutePath(L"output/SetFieldFlags.pdf"));
+    SetFieldFlags(email, key, open_path, output_dir + L"/SetFieldFlags.pdf");
     
     std::cout << "SetFormFieldValue" << std::endl;
-    SetFormFieldValue(email, key, open_path, GetAbsolutePath(L"output/SetFormFieldValue.pdf"));
+    SetFormFieldValue(email, key, open_path, output_dir + L"/SetFormFieldValue.pdf");
   }
   catch (std::exception& ex) {
     std::cout << ex.what();

Original file line number	Diff line number	Diff line change
`@@ -0,0 +1,3 @@`
	`1`	`+datadir = @datadir@/tessdata/configs`
	`2`	`+data_DATA = inter makebox box.train unlv ambigs.train lstm.train api_config kannada box.train.stderr quiet logfile digits hocr tsv linebox pdf rebox strokewidth bigram txt`
	`3`	`+EXTRA_DIST = inter makebox box.train unlv ambigs.train lstm.train api_config kannada box.train.stderr quiet logfile digits hocr tsv linebox pdf rebox strokewidth bigram txt`