Adding tesseract back

astutejoe · Jul 6, 2021 · 39536ab · 39536ab
1 parent 76a8849
commit 39536ab
Show file tree

Hide file tree

Showing 38 changed files with 146 additions and 0 deletions.
diff --git a/tesseract/tessdata/configs/Makefile.am b/tesseract/tessdata/configs/Makefile.am
@@ -0,0 +1,8 @@
+datadir = @datadir@/tessdata/configs
+data_DATA = inter makebox box.train unlv ambigs.train lstm.train lstmdebug
+data_DATA += api_config kannada box.train.stderr quiet logfile digits get.images
+data_DATA += lstmbox wordstrbox
+# Configurations for OCR output.
+data_DATA += alto hocr pdf tsv txt
+data_DATA += linebox rebox strokewidth bigram
+EXTRA_DIST = $(data_DATA)
diff --git a/tesseract/tessdata/configs/alto b/tesseract/tessdata/configs/alto
@@ -0,0 +1 @@
+tessedit_create_alto 1
diff --git a/tesseract/tessdata/configs/ambigs.train b/tesseract/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/tesseract/tessdata/configs/api_config b/tesseract/tessdata/configs/api_config
@@ -0,0 +1 @@
+tessedit_zero_rejection T
diff --git a/tesseract/tessdata/configs/bazaar b/tesseract/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/tesseract/tessdata/configs/bigram b/tesseract/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/tesseract/tessdata/configs/box.train b/tesseract/tessdata/configs/box.train
@@ -0,0 +1,12 @@
+disable_character_fragments T
+file_type                   .bl
+textord_fast_pitch_test	T
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+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/tesseract/tessdata/configs/box.train.stderr b/tesseract/tessdata/configs/box.train.stderr
@@ -0,0 +1,13 @@
+file_type .bl
+#tessedit_use_nn F
+textord_fast_pitch_test T
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+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/tesseract/tessdata/configs/digits b/tesseract/tessdata/configs/digits
@@ -0,0 +1 @@
+tessedit_char_whitelist 0123456789-.
diff --git a/tesseract/tessdata/configs/get.images b/tesseract/tessdata/configs/get.images
@@ -0,0 +1 @@
+tessedit_write_images T
diff --git a/tesseract/tessdata/configs/hocr b/tesseract/tessdata/configs/hocr
@@ -0,0 +1,2 @@
+tessedit_create_hocr 1
+hocr_font_info 0
diff --git a/tesseract/tessdata/configs/inter b/tesseract/tessdata/configs/inter
@@ -0,0 +1,2 @@
+interactive_display_mode				T
+tessedit_display_outwords		T
diff --git a/tesseract/tessdata/configs/kannada b/tesseract/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/tesseract/tessdata/configs/linebox b/tesseract/tessdata/configs/linebox
@@ -0,0 +1,2 @@
+tessedit_resegment_from_line_boxes 1
+tessedit_make_boxes_from_boxes 1
diff --git a/tesseract/tessdata/configs/logfile b/tesseract/tessdata/configs/logfile
@@ -0,0 +1 @@
+debug_file tesseract.log
diff --git a/tesseract/tessdata/configs/lstm.train b/tesseract/tessdata/configs/lstm.train
@@ -0,0 +1,11 @@
+file_type                   .bl
+textord_fast_pitch_test	T
+tessedit_zero_rejection T
+tessedit_minimal_rejection F
+tessedit_write_rep_codes F
+edges_children_fix F
+edges_childarea 0.65
+edges_boxarea 0.9
+tessedit_train_line_recognizer T
+textord_no_rejects T
+tessedit_init_config_only T
diff --git a/tesseract/tessdata/configs/lstmbox b/tesseract/tessdata/configs/lstmbox
@@ -0,0 +1 @@
+tessedit_create_lstmbox 1
diff --git a/tesseract/tessdata/configs/lstmdebug b/tesseract/tessdata/configs/lstmdebug
@@ -0,0 +1,4 @@
+stopper_debug_level 1
+classify_debug_level 1
+segsearch_debug_level 1
+language_model_debug_level 3
diff --git a/tesseract/tessdata/configs/makebox b/tesseract/tessdata/configs/makebox
@@ -0,0 +1 @@
+tessedit_create_boxfile 1
diff --git a/tesseract/tessdata/configs/pdf b/tesseract/tessdata/configs/pdf
@@ -0,0 +1 @@
+tessedit_create_pdf 1
diff --git a/tesseract/tessdata/configs/quiet b/tesseract/tessdata/configs/quiet
@@ -0,0 +1 @@
+debug_file /dev/null
diff --git a/tesseract/tessdata/configs/rebox b/tesseract/tessdata/configs/rebox
@@ -0,0 +1,2 @@
+tessedit_resegment_from_boxes 1
+tessedit_make_boxes_from_boxes 1
diff --git a/tesseract/tessdata/configs/strokewidth b/tesseract/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/tesseract/tessdata/configs/tsv b/tesseract/tessdata/configs/tsv
@@ -0,0 +1 @@
+tessedit_create_tsv 1
diff --git a/tesseract/tessdata/configs/txt b/tesseract/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/tesseract/tessdata/configs/unlv b/tesseract/tessdata/configs/unlv
@@ -0,0 +1,2 @@
+tessedit_write_unlv 1
+unlv_tilde_crunching T
diff --git a/tesseract/tessdata/configs/wordstrbox b/tesseract/tessdata/configs/wordstrbox
@@ -0,0 +1 @@
+tessedit_create_wordstrbox 1
diff --git a/tesseract/tessdata/eng.traineddata b/tesseract/tessdata/eng.traineddata
diff --git a/tesseract/tessdata/eng.user-patterns b/tesseract/tessdata/eng.user-patterns
@@ -0,0 +1,2 @@
+1-\d\d\d-GOOG-411
+www.\n\\\*.com
diff --git a/tesseract/tessdata/eng.user-words b/tesseract/tessdata/eng.user-words
@@ -0,0 +1,5 @@
+the
+quick
+brown
+fox
+jumped
diff --git a/tesseract/tessdata/pdf.ttf b/tesseract/tessdata/pdf.ttf
diff --git a/tesseract/tessdata/tessconfigs/Makefile.am b/tesseract/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/tesseract/tessdata/tessconfigs/batch b/tesseract/tessdata/tessconfigs/batch
@@ -0,0 +1 @@
+# No content needed as all defaults are correct.
diff --git a/tesseract/tessdata/tessconfigs/batch.nochop b/tesseract/tessdata/tessconfigs/batch.nochop
@@ -0,0 +1,2 @@
+chop_enable 0
+wordrec_enable_assoc 0
diff --git a/tesseract/tessdata/tessconfigs/matdemo b/tesseract/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/tesseract/tessdata/tessconfigs/msdemo b/tesseract/tessdata/tessconfigs/msdemo
@@ -0,0 +1,12 @@
+#################################################
+# 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_blobs       1
+wordrec_display_segmentations   2
+classify_debug_level		1
diff --git a/tesseract/tessdata/tessconfigs/nobatch b/tesseract/tessdata/tessconfigs/nobatch
@@ -0,0 +1 @@
+
diff --git a/tesseract/tessdata/tessconfigs/segdemo b/tesseract/tessdata/tessconfigs/segdemo
@@ -0,0 +1,9 @@
+#################################################
+# Adaptive Matcher Using PreAdapted Templates
+#################################################
+
+wordrec_display_splits          0
+wordrec_display_all_blobs       1
+wordrec_display_segmentations   2
+classify_debug_level		1
+stopper_debug_level 1
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		interactive_display_mode T
		tessedit_display_outwords T
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		tessedit_resegment_from_line_boxes 1
		tessedit_make_boxes_from_boxes 1
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1,2 @@
		tessedit_resegment_from_boxes 1
		tessedit_make_boxes_from_boxes 1
-Original file line number
+Diff line change
@@ -0,0 +1,5 @@
+    the
+    quick
+    brown
+    fox
+    jumped
Original file line number	Diff line number	Diff line change
		@@ -0,0 +1 @@
		# No content needed as all defaults are correct.