CCExtractor · cfsmp3 · Dec 29, 2025 · Dec 29, 2025 · Dec 29, 2025 · Dec 29, 2025
diff --git a/.gitignore b/.gitignore
@@ -17,6 +17,7 @@ CVS
 mac/ccextractor
 linux/ccextractor
 linux/depend
+linux/build_scan/
 windows/x86_64-pc-windows-msvc/**
 windows/Debug/**
 windows/Debug-OCR/**
@@ -28,6 +29,7 @@ windows/Debug-Full/**
 windows/x64/**
 windows/ccextractor.VC.db
 build/
+build_*/
 
 ####
 # Python

diff --git a/src/lib_ccx/avc_functions.c b/src/lib_ccx/avc_functions.c
@@ -379,11 +379,10 @@ void sei_rbsp(struct avc_ctx *ctx, unsigned char *seibuf, unsigned char *seiend)
 	}
 	else
 	{
-		// TODO: This really really looks bad
-		mprint("WARNING: Unexpected SEI unit length...trying to continue.");
-		temp_debug = 1;
-		mprint("\n Failed block (at sei_rbsp) was:\n");
-		dump(CCX_DMT_GENERIC_NOTICES, (unsigned char *)seibuf, seiend - seibuf, 0, 0);
+		// Unexpected SEI length - common with malformed streams, don't spam output
+		dbg_print(CCX_DMT_VERBOSE, "WARNING: Unexpected SEI unit length (parsed to %p, expected %p)...trying to continue.\n",
+			  (void *)tbuf, (void *)(seiend - 1));
+		dump(CCX_DMT_VERBOSE, (unsigned char *)seibuf, seiend - seibuf, 0, 0);
 
 		ctx->num_unexpected_sei_length++;
 	}

diff --git a/src/lib_ccx/ccx_demuxer.c b/src/lib_ccx/ccx_demuxer.c
@@ -348,7 +348,6 @@ struct ccx_demuxer *init_demuxer(void *parent, struct demuxer_cfg *cfg)
 		{
 			ctx->pinfo[i].got_important_streams_min_pts[j] = UINT64_MAX;
 		}
-		ctx->pinfo[i].initialized_ocr = 0;
 		ctx->pinfo[i].version = 0xFF; // Not real in a real stream since it's 5 bits. FF => Not initialized
 	}
 

diff --git a/src/lib_ccx/ccx_demuxer.h b/src/lib_ccx/ccx_demuxer.h
@@ -35,7 +35,6 @@ struct program_info
 {
 	int pid;
 	int program_number;
-	int initialized_ocr; // Avoid initializing the OCR more than once
 	uint8_t analysed_PMT_once : 1;
 	uint8_t version;
 	uint8_t saved_section[1021];

diff --git a/src/lib_ccx/dvb_subtitle_decoder.c b/src/lib_ccx/dvb_subtitle_decoder.c
@@ -418,7 +418,7 @@ static void delete_regions(DVBSubContext *ctx)
  * @return DVB context kept as void* for abstraction
  *
  */
-void *dvbsub_init_decoder(struct dvb_config *cfg, int initialized_ocr)
+void *dvbsub_init_decoder(struct dvb_config *cfg)
 {
 	int i, r, g, b, a = 0;
 	DVBSubContext *ctx = (DVBSubContext *)malloc(sizeof(DVBSubContext));
@@ -442,8 +442,7 @@ void *dvbsub_init_decoder(struct dvb_config *cfg, int initialized_ocr)
 	}
 
 #ifdef ENABLE_OCR
-	if (!initialized_ocr)
-		ctx->ocr_ctx = init_ocr(ctx->lang_index);
+	ctx->ocr_ctx = init_ocr(ctx->lang_index);
 #endif
 	ctx->version = -1;
 

diff --git a/src/lib_ccx/dvb_subtitle_decoder.h b/src/lib_ccx/dvb_subtitle_decoder.h
@@ -42,7 +42,7 @@ extern "C"
 	 * @return DVB context kept as void* for abstraction
 	 *
 	 */
-	void *dvbsub_init_decoder(struct dvb_config *cfg, int initialized_ocr);
+	void *dvbsub_init_decoder(struct dvb_config *cfg);
 
 	int dvbsub_close_decoder(void **dvb_ctx);
 

diff --git a/src/lib_ccx/general_loop.c b/src/lib_ccx/general_loop.c
@@ -75,7 +75,7 @@ int ps_get_more_data(struct lib_ccx_ctx *ctx, struct demuxer_data **ppdata)
 				if (!ctx->demux_ctx->strangeheader)
 				{
 					mprint("\nNot a recognized header. Searching for next header.\n");
-					dump(CCX_DMT_GENERIC_NOTICES, nextheader, 6, 0, 0);
+					dump(CCX_DMT_PARSE, nextheader, 6, 0, 0);
 					// Only print the message once per loop / unrecognized header
 					ctx->demux_ctx->strangeheader = 1;
 				}
@@ -809,10 +809,6 @@ int process_data(struct encoder_ctx *enc_ctx, struct lib_cc_decode *dec_ctx, str
 			got = data_node->len;
 		}
 	}
-	else if (data_node->bufferdatatype == CCX_PRIVATE_MPEG2_CC)
-	{
-		got = data_node->len; // Do nothing. Still don't know how to process it
-	}
 	else if (data_node->bufferdatatype == CCX_RAW) // Raw two byte 608 data from DVR-MS/ASF
 	{
 		// The asf_get_more_data() loop sets current_pts when possible

diff --git a/src/lib_ccx/matroska.c b/src/lib_ccx/matroska.c
@@ -1173,7 +1173,7 @@ void parse_private_codec_data(struct matroska_ctx *mkv_ctx, char *codec_id_strin
 		memset((void *)&cnf, 0, sizeof(struct dvb_config));
 
 		parse_dvb_description(&cnf, codec_data, 8);
-		dec_ctx->private_data = dvbsub_init_decoder(&cnf, 0);
+		dec_ctx->private_data = dvbsub_init_decoder(&cnf);
 
 		free(codec_data);
 	}

diff --git a/src/lib_ccx/ts_functions.c b/src/lib_ccx/ts_functions.c
@@ -153,12 +153,11 @@ enum ccx_bufferdata_type get_buffer_type(struct cap_info *cinfo)
 	{
 		return CCX_TELETEXT;
 	}
-	else if (cinfo->stream == CCX_STREAM_TYPE_PRIVATE_MPEG2 && cinfo->codec == CCX_CODEC_ATSC_CC)
-	{
-		return CCX_PRIVATE_MPEG2_CC;
-	}
-	else if (cinfo->stream == CCX_STREAM_TYPE_PRIVATE_USER_MPEG2 && cinfo->codec == CCX_CODEC_ATSC_CC)
+	else if ((cinfo->stream == CCX_STREAM_TYPE_PRIVATE_MPEG2 ||
+		  cinfo->stream == CCX_STREAM_TYPE_PRIVATE_USER_MPEG2) &&
+		 cinfo->codec == CCX_CODEC_ATSC_CC)
 	{
+		// ATSC CC can be in either private stream type - process both as PES
 		return CCX_PES;
 	}
 	else
@@ -567,15 +566,6 @@ int copy_capbuf_demux_data(struct ccx_demuxer *ctx, struct demuxer_data **data,
 	if (!cinfo->capbuf || !cinfo->capbuflen)
 		return -1;
 
-	if (ptr->bufferdatatype == CCX_PRIVATE_MPEG2_CC)
-	{
-		dump(CCX_DMT_GENERIC_NOTICES, cinfo->capbuf, cinfo->capbuflen, 0, 1);
-		// Bogus data, so we return something
-		ptr->buffer[ptr->len++] = 0xFA;
-		ptr->buffer[ptr->len++] = 0x80;
-		ptr->buffer[ptr->len++] = 0x80;
-		return CCX_OK;
-	}
 	if (cinfo->codec == CCX_CODEC_TELETEXT)
 	{
 		memcpy(ptr->buffer + ptr->len, cinfo->capbuf, cinfo->capbuflen);

diff --git a/src/lib_ccx/ts_info.c b/src/lib_ccx/ts_info.c
@@ -173,7 +173,7 @@ static void *init_private_data(enum ccx_code_type codec)
 		case CCX_CODEC_TELETEXT:
 			return telxcc_init();
 		case CCX_CODEC_DVB:
-			return dvbsub_init_decoder(NULL, 0);
+			return dvbsub_init_decoder(NULL);
 		default:
 			return NULL;
 	}

diff --git a/src/lib_ccx/ts_tables.c b/src/lib_ccx/ts_tables.c
@@ -399,9 +399,7 @@ int parse_PMT(struct ccx_demuxer *ctx, unsigned char *buf, int len, struct progr
 					ret = parse_dvb_description(&cnf, es_info, desc_len);
 					if (ret < 0)
 						break;
-					ptr = dvbsub_init_decoder(&cnf, pinfo->initialized_ocr);
-					if (!pinfo->initialized_ocr)
-						pinfo->initialized_ocr = 1;
+					ptr = dvbsub_init_decoder(&cnf);
 					if (ptr == NULL)
 						break;
 					update_capinfo(ctx, elementary_PID, stream_type, CCX_CODEC_DVB, program_number, ptr);

diff --git a/src/rust/src/common.rs b/src/rust/src/common.rs
@@ -1070,7 +1070,6 @@ impl CType<program_info> for ProgramInfo {
         program_info {
             pid: self.pid,
             program_number: self.program_number,
-            initialized_ocr: self.initialized_ocr as c_int,
             _bitfield_align_1: [],
             _bitfield_1: bf1,
             version: self.version,

diff --git a/src/rust/src/ctorust.rs b/src/rust/src/ctorust.rs
@@ -540,7 +540,6 @@ impl FromCType<program_info> for ProgramInfo {
         Some(ProgramInfo {
             pid: info.pid,
             program_number: info.program_number,
-            initialized_ocr: info.initialized_ocr != 0,
             analysed_pmt_once: info._bitfield_1.get_bit(0) as u8,
             version: info.version,
             saved_section: info.saved_section,

diff --git a/src/rust/src/demuxer/common_types.rs b/src/rust/src/demuxer/common_types.rs
@@ -50,7 +50,6 @@ pub struct FileReport {
 pub struct ProgramInfo {
     pub pid: i32,
     pub program_number: i32,
-    pub initialized_ocr: bool, // Avoid initializing the OCR more than once
     pub analysed_pmt_once: u8, // 1-bit field
     pub version: u8,
     pub saved_section: [u8; SAVED_SECTIONS_PROGRAMINFO],
@@ -301,7 +300,6 @@ impl Default for ProgramInfo {
         ProgramInfo {
             pid: -1,
             program_number: 0,
-            initialized_ocr: false,
             analysed_pmt_once: 0,
             version: 0,
             saved_section: [0; SAVED_SECTIONS_PROGRAMINFO],
-Original file line number
+Diff line change
@@ Expand Up @@
     		{
     			ctx->pinfo[i].got_important_streams_min_pts[j] = UINT64_MAX;
     		}
-    		ctx->pinfo[i].initialized_ocr = 0;
     		ctx->pinfo[i].version = 0xFF; // Not real in a real stream since it's 5 bits. FF => Not initialized
     	}
@@ Expand Down @@