[IMPROVEMENT] Implement subtitle modifications for each encoder (#1214)

* Implement subtitle modification for all 608 encoders

This is done by modifying the subtitles in `ccx_encoders_common.c`
rather than per encoder.

* Use `char *` instead of subtitle data to capitalize

* Implement subtitle modification for OCR encoders

* Remove signness warnings

* Remove two-word profanity

They do not work for the moment

* Deal with different encoding

* Mention in changelog

docs/CHANGES.TXT

@@ -6,6 +6,7 @@
 - New: Added more fatals on no arguments errors.
 - New: Updated ccextractor.cnf.sample.
 - New: Add support for censoring words ("Kid Friendly") (#1139)
+- New: Extend support of capitalization for all BITMAP and 608 subtitles (#1214)
 - Fix: ccx_demuxer_mxf.c: Parse framerate from MXF captions to fix caption timings.
 - Fix: hardsubx_decoder.c: Fix memory leaks using Leptonica API.
 - Fix: linux/Makefile.am: added some sources to enable rpms to be created.

src/lib_ccx/ccx_encoders_common.c

@@ -615,8 +615,6 @@ void write_cc_line_as_simplexml(struct eia608_screen *data, struct encoder_ctx *
 	char *cap = "<caption>";
 	char *cap1 = "</caption>";
 
-	correct_spelling_and_censor_words_608(context, line_number, data);
-
 	length = get_str_basic(context->subline, data->characters[line_number],
 			context->trim_subs, CCX_ENC_ASCII, context->encoding, CCX_DECODER_608_SCREEN_WIDTH);
 
@@ -1155,6 +1153,9 @@ int encode_sub(struct encoder_ctx *context, struct cc_subtitle *sub)
 					data->end_time += utc_refvalue * 1000;
 				}
 
+				for (int i = 0; i < CCX_DECODER_608_SCREEN_ROWS; ++i)
+					correct_spelling_and_censor_words(context, (char *) data->characters[i], CCX_DECODER_608_SCREEN_WIDTH);
+
 #ifdef PYTHON_API
 				pass_cc_buffer_to_python(data, context);
 #else
@@ -1228,7 +1229,26 @@ int encode_sub(struct encoder_ctx *context, struct cc_subtitle *sub)
 			}
 			freep(&sub->data);
 			break;
-		case CC_BITMAP:
+		case CC_BITMAP:;
+
+#ifdef ENABLE_OCR
+			struct cc_bitmap *rect;
+			int i;
+			for (i = 0, rect = sub->data; i < sub->nb_data; ++i, ++rect)
+			{
+				if (rect->ocr_text)
+				{
+					int len = strlen(rect->ocr_text);
+					correct_spelling_and_censor_words(context, rect->ocr_text, len);
+					for (int i = 0; i < len; ++i)
+					{
+						if ((unsigned char)rect->ocr_text[i] == 0x98) // asterisk in 608 encoding
+							rect->ocr_text[i] = '*';
+					}
+				}
+			}
+#endif
+
 			switch (context->write_format)
 			{
 				case CCX_OF_CCD:
@@ -1281,7 +1301,6 @@ int encode_sub(struct encoder_ctx *context, struct cc_subtitle *sub)
 				default:
 					break;
 			}
-
 			break;
 		case CC_RAW:
 			if (context->send_to_srv)

src/lib_ccx/ccx_encoders_helpers.c

@@ -61,7 +61,7 @@ const char *profane_builtin[] =
 	"goddamn",
 	"godsdamn",
 	"hell",
-	"holy shit",
+	"holy",
 	"horseshit",
 	"motherfucker",
 	"nigga",
@@ -84,17 +84,17 @@ int string_cmp(const void *p1, const void *p2)
 	return string_cmp_function(p1, p2, NULL);
 }
 
-void capitalize_word(size_t index, unsigned char *word)
+void capitalize_word(size_t index, char *word)
 {
 	memcpy(word, capitalization_list.words[index], strlen(capitalization_list.words[index]));
 }
 
-void censor_word(size_t index, unsigned char *word)
+void censor_word(size_t index, char *word)
 {
 	memset(word, 0x98, strlen(profane.words[index])); // 0x98 is the asterisk in EIA-608
 }
 
-void call_function_if_match(int line_num, struct eia608_screen *data, struct word_list *list, void (*modification)(size_t, unsigned char *))
+void call_function_if_match(char *line, struct word_list *list, void (*modification)(size_t, char *))
 {
 	char delim[64] = {
 		' ', '\n', '\r', 0x89, 0x99,
@@ -105,8 +105,8 @@ void call_function_if_match(int line_num, struct eia608_screen *data, struct wor
 		'.', '/', ':', '^', '_',
 		'{', '|', '}', '~', '\0' };
 
-	char *line = strdup(data->characters[line_num]);
-	char *c = strtok(line, delim);
+	char *line_token = strdup(line);
+	char *c = strtok(line_token, delim);
 
 	if (c != NULL)
 	{
@@ -116,21 +116,11 @@ void call_function_if_match(int line_num, struct eia608_screen *data, struct wor
 
 			if (index)
 			{
-				modification(index - list->words, data->characters[line_num] + (c - line));
+				modification(index - list->words, line + (c - line_token));
 			}
 		} while ((c = strtok(NULL, delim)) != NULL);
 	}
-	free(line);
-}
-
-void correct_case_with_dictionary(int line_num, struct eia608_screen *data)
-{
-	call_function_if_match(line_num, data, &capitalization_list, capitalize_word);
-}
-
-void censor_word_with_dictionary(int line_num, struct eia608_screen *data)
-{
-	call_function_if_match(line_num, data, &profane, censor_word);
+	free(line_token);
 }
 
 void telx_correct_case(char *sub_line)
@@ -180,17 +170,17 @@ int is_all_caps(struct encoder_ctx *context, int line_num, struct eia608_screen
 	return (saw_upper && !saw_lower); // 1 if we've seen upper and not lower, 0 otherwise
 }
 
-int clever_capitalize(struct encoder_ctx *context, int line_num, struct eia608_screen *data)
+int clever_capitalize(struct encoder_ctx *context, char *line, unsigned int length)
 {
 	// CFS: Tried doing to clever (see below) but some channels do all uppercase except for
 	// notes for deaf people (such as "(narrator)" which messes things up.
 	// First find out if we actually need to do it, don't mess with lines that come OK
 	//int doit = is_all_caps(context, line_num, data);
 	int doit = 1;
 
-	for (int i = 0; i < CCX_DECODER_608_SCREEN_WIDTH; i++)
+	for (int i = 0; i < length; i++)
 	{
-		switch (data->characters[line_num][i])
+		switch (line[i])
 		{
 		case ' ':
 		case 0x89: // This is a transparent space
@@ -206,9 +196,9 @@ int clever_capitalize(struct encoder_ctx *context, int line_num, struct eia608_s
 			if (doit)
 			{
 				if (context->new_sentence)
-					data->characters[line_num][i] = cctoupper(data->characters[line_num][i]);
+					line[i] = cctoupper(line[i]);
 				else
-					data->characters[line_num][i] = cctolower(data->characters[line_num][i]);
+					line[i] = cctolower(line[i]);
 			}
 			context->new_sentence = 0;
 			break;
@@ -455,17 +445,17 @@ int add_builtin_words(const char *builtin[], struct word_list *list)
 	return 0;
 }
 
-void correct_spelling_and_censor_words_608(struct encoder_ctx *context, int line_number, struct eia608_screen *data)
+void correct_spelling_and_censor_words(struct encoder_ctx *context, char *line, unsigned int length)
 {
 	if (context->sentence_cap)
 	{
-		if (clever_capitalize(context, line_number, data))
-			correct_case_with_dictionary(line_number, data);
+		if (clever_capitalize(context, line, length))
+			call_function_if_match(line, &capitalization_list, capitalize_word);
 	}
 
 	if (context->filter_profanity)
 	{
-		censor_word_with_dictionary(line_number, data);
+		call_function_if_match(line, &profane, censor_word);
 	}
 }
 

src/lib_ccx/ccx_encoders_helpers.h

@@ -23,9 +23,7 @@ struct ccx_encoders_helpers_settings_t {
 };
 
 // Helper functions
-void correct_case_with_dictionary(int line_num, struct eia608_screen *data);
 int is_all_caps(struct encoder_ctx *context, int line_num, struct eia608_screen *data);
-int clever_capitalize(struct encoder_ctx *context, int line_num, struct eia608_screen *data);
 void telx_correct_case(char *sub_line);
 unsigned get_decoder_line_encoded_for_gui(unsigned char *buffer, int line_num, struct eia608_screen *data);
 unsigned get_decoder_line_encoded(struct encoder_ctx *ctx, unsigned char *buffer, int line_num, struct eia608_screen *data);
@@ -36,7 +34,7 @@ int string_cmp_function(const void *p1, const void *p2, void *arg);
 int add_word(struct word_list *list, const char *word);
 
 int add_builtin_words(const char *builtin[], struct word_list *list);
-void correct_spelling_and_censor_words_608(struct encoder_ctx *context, int line_number, struct eia608_screen *data);
+void correct_spelling_and_censor_words(struct encoder_ctx *context, char *line, unsigned int length);
 
 unsigned encode_line (struct encoder_ctx *ctx, unsigned char *buffer, unsigned char *text);
 

src/lib_ccx/ccx_encoders_smptett.c

@@ -201,8 +201,6 @@ int write_cc_buffer_as_smptett(struct eia608_screen *data, struct encoder_ctx *c
 	{
 		if (data->row_used[row])
 		{
-			correct_spelling_and_censor_words_608(context, row, data);
-
 			float row1=0;
 			float col1=0;
 			int firstcol=-1;

src/lib_ccx/ccx_encoders_splitbysentence.c

@@ -762,7 +762,7 @@ struct cc_subtitle * reformat_cc_bitmap_through_sentence_buffer(struct cc_subtit
 	if (sub->flags & SUB_EOD_MARKER)
 		context->prev_start = sub->start_time;
 
-	str = paraof_ocrtext(sub, context->encoded_crlf, context->encoded_crlf_length);
+	str = paraof_ocrtext(sub, context);
 
 	if (str)
 	{

src/lib_ccx/ccx_encoders_spupng.c

@@ -449,7 +449,7 @@ int write_cc_bitmap_as_spupng(struct cc_subtitle *sub, struct encoder_ctx *conte
 	if (!context->nospupngocr)
 	{
 		char *str;
-		str = paraof_ocrtext(sub, context->encoded_crlf, context->encoded_crlf_length);
+		str = paraof_ocrtext(sub, context);
 		if (str)
 		{
 			write_spucomment(sp, str);

src/lib_ccx/ccx_encoders_srt.c

@@ -97,7 +97,7 @@ int write_cc_bitmap_as_srt(struct cc_subtitle *sub, struct encoder_ctx *context)
 	if(sub->flags & SUB_EOD_MARKER)
 		context->prev_start =  sub->start_time;
 
-	str = paraof_ocrtext(sub, context->encoded_crlf, context->encoded_crlf_length);
+	str = paraof_ocrtext(sub, context);
 	if (str)
 	{
         if(context->is_mkv == 1) {
@@ -210,8 +210,6 @@ int write_cc_buffer_as_srt(struct eia608_screen *data, struct encoder_ctx *conte
 	{
 		if (data->row_used[i])
 		{
-			correct_spelling_and_censor_words_608(context, i, data);
-
 			if (context->autodash && context->trim_subs)
 			{
 				int first=0, last=31, center1=-1, center2=-1;

src/lib_ccx/ccx_encoders_ssa.c

@@ -91,7 +91,7 @@ int write_cc_bitmap_as_ssa(struct cc_subtitle *sub, struct encoder_ctx *context)
 	if(sub->flags & SUB_EOD_MARKER)
 		context->prev_start =  sub->start_time;
 
-	str = paraof_ocrtext(sub, context->encoded_crlf, context->encoded_crlf_length);
+	str = paraof_ocrtext(sub, context);
 	if (str)
 	{
 		// SSA format - change "\r\n" to "\N"
@@ -194,8 +194,6 @@ int write_cc_buffer_as_ssa(struct eia608_screen *data, struct encoder_ctx *conte
 	{
 		if (data->row_used[i])
 		{
-			correct_spelling_and_censor_words_608(context, i, data);
-
 			if (context->autodash && context->trim_subs)
 			{
 				int first=0, last=31, center1=-1, center2=-1;

src/lib_ccx/ccx_encoders_transcript.c

@@ -32,7 +32,7 @@ int write_cc_bitmap_as_transcript(struct cc_subtitle *sub, struct encoder_ctx *c
 		if (context->prev_start != -1 || !(sub->flags & SUB_EOD_MARKER))
 		{
 			char *token = NULL;
-			token = paraof_ocrtext(sub, context->encoded_crlf, context->encoded_crlf_length);
+			token = paraof_ocrtext(sub, context);
 			if (context->transcript_settings->showStartTime)
 			{
 				char buf1[80];
@@ -239,9 +239,6 @@ int write_cc_subtitle_as_transcript(struct cc_subtitle *sub, struct encoder_ctx
 void write_cc_line_as_transcript2(struct eia608_screen *data, struct encoder_ctx *context, int line_number)
 {
 	int ret = 0;
-
-	correct_spelling_and_censor_words_608(context, line_number, data);
-
 	int length = get_str_basic(context->subline, data->characters[line_number],
 		context->trim_subs, CCX_ENC_ASCII, context->encoding, CCX_DECODER_608_SCREEN_WIDTH);
 

src/lib_ccx/ccx_encoders_webvtt.c

@@ -284,7 +284,7 @@ int write_cc_bitmap_as_webvtt(struct cc_subtitle *sub, struct encoder_ctx *conte
 	if (sub->flags & SUB_EOD_MARKER)
 		context->prev_start = sub->start_time;
 
-	str = paraof_ocrtext(sub, context->encoded_crlf, context->encoded_crlf_length);
+	str = paraof_ocrtext(sub, context);
 	if (str)
 	{
 		if (context->prev_start != -1 || !(sub->flags & SUB_EOD_MARKER))

src/lib_ccx/ocr.c

@@ -976,7 +976,7 @@ void add_ocrtext2str(char *dest, char *src, const unsigned char *crlf, unsigned
  * for all text detected from rectangles
  */
 
-char *paraof_ocrtext(struct cc_subtitle *sub, const unsigned char *crlf, unsigned crlf_length)
+char *paraof_ocrtext(struct cc_subtitle *sub, struct encoder_ctx *context)
 {
 	int i;
 	int len = 0;
@@ -1002,7 +1002,7 @@ char *paraof_ocrtext(struct cc_subtitle *sub, const unsigned char *crlf, unsigne
 	for(i = 0, rect = sub->data; i < sub->nb_data; i++, rect++)
 	{
 		if (!rect->ocr_text) continue;
-		add_ocrtext2str(str, rect->ocr_text, crlf, crlf_length);
+		add_ocrtext2str(str, rect->ocr_text, context->encoded_crlf, context->encoded_crlf_length);
 		free(rect->ocr_text);
 	}
 	return str;

src/lib_ccx/ocr.h

@@ -15,6 +15,6 @@ void delete_ocr (void** arg);
 void* init_ocr(int lang_index);
 char* ocr_bitmap(void* arg, png_color *palette,png_byte *alpha, unsigned char* indata,int w, int h, struct image_copy *copy);
 int ocr_rect(void* arg, struct cc_bitmap *rect, char **str, int bgcolor, int ocr_quantmode);
-char *paraof_ocrtext(struct cc_subtitle *sub, const unsigned char *crlf, unsigned crlf_length);
+char *paraof_ocrtext(struct cc_subtitle *sub, struct encoder_ctx *context);
 
 #endif