Merge branch 'master' into fix/issue-1478-cea708-writer-unwrap-panic

Author: x15sr71

.github/workflows/release.yml

@@ -34,10 +34,10 @@ jobs:
         LLVM_CONFIG_PATH: "C:\\Program Files\\LLVM\\bin\\llvm-config"
         CARGO_TARGET_DIR: "..\\..\\windows"
         BINDGEN_EXTRA_CLANG_ARGS: -fmsc-version=0
-      run: msbuild ccextractor.sln /p:Configuration=Release-Full /p:Platform=Win32
+      run: msbuild ccextractor.sln /p:Configuration=Release-Full /p:Platform=x64
       working-directory: ./windows
     - name: Copy files to directory for installer
-      run: mkdir installer; cp ./Release-Full/ccextractorwinfull.exe ./installer; cp ./Release-Full/*.dll ./installer
+      run: mkdir installer; cp ./x64/Release-Full/ccextractorwinfull.exe ./installer; cp ./x64/Release-Full/*.dll ./installer
       working-directory: ./windows
     - name: install WiX
       run: dotnet tool install --global wix --version 4.0.0-preview.0 &&  wix extension -g add WixToolset.UI.wixext

docs/CHANGES.TXT

@@ -1,107 +1,38 @@
-0.96 (2025-12-21)
+0.96 (2025-12-23)
 -----------------
 - Fix: Prevent CEA-708 decoder panic when using --stdout on Windows by safely handling writer output errors
+- New: Multi-page teletext extraction support (#665)
+  - Extract multiple teletext pages simultaneously with separate output files
+  - Use --tpage multiple times (e.g., --tpage 100 --tpage 200)
+  - Output files are named with page suffix (e.g., output_p100.srt, output_p200.srt)
+
 - New: Added --list-tracks (-L) option to list all tracks in media files without processing
-- Fix: Garbled captions from HDHomeRun and I/P-only H.264 streams (#1109)
-- Fix: Enable stdout output for CEA-708 captions on Windows (#1693)
-- Fix: McPoodle DVD raw format read/write - properly handle loop markers (#1524)
-- Fix: Variable shadowing in general_loop causing false "premature end of file" messages
-- Fix: Double-free crash in teletext cleanup when processing multiple files
-- Fix: Uninitialized memory and memory leaks found by Valgrind testing
-- Fix: Dangling pointers in Rust FFI copy_from_rust functions
-- New: Improve -out=report to show detected Teletext subtitle pages (#1034)
-- FIX: Include ATSC VCT virtual channel numbers and call signs in XMLTV output
-- FIX: Restore ATSC XMLTV generation with ETT parsing for extended descriptions, multi-segment handling, extended table ID's (EIT/VCT), corrected <programme> XMLTV formatting, buffer bounds fixes
-- Fix: DVB subtitle extraction improvements for Chinese broadcasts (#224):
-  - Fix crash in parse_PMT() due to missing bounds checks
-  - Fix negative timestamps in DVB subtitle output
-  - Fix crash in ignore_alpha_at_edge() OCR cropping
-  - Improve DVB subtitle OCR accuracy with image inversion
-  - Fix --ocrlang to accept Tesseract language names (chi_tra, chi_sim, etc.)
-  - Add case-insensitive matching for --dvblang parameter
-- FIX: Add HEVC/H.265 stream type recognition to prevent crashes on ATSC 3.0 streams
-- New: Add demuxer and file_functions module in lib_ccxr (#1662)
-- Fix: handle row_count decrease in CEA-708 C decoder
-- Fix: Bounds checks to prevent panic on malformed CEA-708 data
-- Fix: Multiprogram logic in is_decoder_processed_enough() causing false warnings
-- Fix: Write consistent 2-byte UTF-16BE encoding for CEA-708 captions (Japanese/Chinese)
-- New: Add --ttxtforcelatin option to force Latin G0 charset in Teletext
-- Fix: Add fallback for TS files without PAT/PMT tables
-- Fix: PTS jump handling to continue fts_now updates after jump
-- Fix: Null checks for unchecked memory allocations throughout codebase
-- Fix: Null checks and invalid UTF-8 handling in Rust FFI functions
-- Fix: Panics in timing code when processing multiple files
-- Fix: Caption start/end times to match FFmpeg timing in MP4/MPEG/TS
-- Fix: Correctly count and store multiple input files
-- Fix: Handle MP4 c608 tracks and improve garbage frame detection
-- Fix: Update fts_now for each frame in elementary streams
-- Fix: Preserve CR time during pop-on to roll-up transition
-- Fix: Defer min_pts until frame type is known
-- Fix: Skip leading non-I-frames when setting min_pts
-- Fix: Memory leaks in ts_tables_epg, ocr, and ccx_encoders_spupng
-- Fix: Buffer overruns in 708_output, mcc_encoder, utility, xds_decoder
-- Fix: Replace sprintf/strcpy with bounds-checked snprintf/strncpy in encoders
-- Fix: HHMMSSFFF format for ttxt output timestamps
-- Fix: Always emit position codes at start of SCC caption
-- Fix: Memory safety issues in ccx_decoders_common
-- Fix: Null checks after malloc calls in dvb_subtitle_decoder
-- Fix: Memory safety checks and memory leaks in Matroska parser
-
-0.95 (2025-09-15)
------------------
-- Fix: ARM64/aarch64 build failure due to c_char type mismatch in nal.rs
-- Fix: HardSubX OCR on Rust
-- Removed the Share Module
-- Fix: Regression failures on DVD files
-- Fix: Segmentation faults on MP4 files with CEA-708 captions
-- Refactor: Remove API structures from ccextractor
-- New: Add Encoder Module to Rust
-- Fix: Elementary stream regressions
-- Fix: Segmentation faults on XDS files
-- Fix: Clippy Errors Based on Rust 1.88
-- IMPROVEMENT: Refactor and optimize Dockerfile
-- Fix: Improved handling of IETF language tags in Matroska files (#1665)
-- New: Create unit test for rust code (#1615)
-- Breaking: Major argument flags revamp for CCExtractor (#1564 & #1619)
+  New: Chinese, Korean, Japanese support - proper encoding and OCR.
+  New: Correct McPoodle DVD raw format support
+  Fix: Timing is now frame perfect (using FFMpeg timing dump as reference) in all formats.
+  Fix: Solved garbling in all the pending issues we had on GitHub.
+  Fix: All causes of "premature end of file" messages due to bugs and not actual file cuts.
+  Fix: All memory leaks, double frees and usual C nastyness that valgrind could find.
+- Fix Include ATSC VCT virtual channel numbers and call signs in XMLTV output
+- Fix: Restore ATSC XMLTV generation with ETT parsing for extended descriptions, multi-segment handling, extended table ID's (EIT/VCT), corrected <programme> XMLTV formatting, buffer bounds fixes
+- Fix: Add HEVC/H.265 stream type recognition to prevent crashes on ATSC 3.0 streams.
+  Fix: Tolerance to damaged streams - recover where possible instead of terminating.
+  Issues closed: Over 40! Too many to list here, but each of them was either a bug squashed or a feature implemented.
+
+0.95 (2025-09-15 - never formally packaged)
+-----------------
 - New: Create a Docker image to simplify the CCExtractor usage without any environmental hustle (#1611)
-- New: Add time units module in lib_ccxr (#1623)
-- New: Add bits and levenshtein module in lib_ccxr (#1627)
-- New: Add constants module in lib_ccxr (#1624) 
-- New: Add log module in lib_ccxr (#1622)
-- New: Create `lib_ccxr` and `libccxr_exports` (#1621)
-- Fix: Unexpected behavior of get_write_interval (#1609)
-- Update: Bump rsmpeg to latest version for ffmpeg bindings (#1600)
 - New: Add SCC support for CEA-708 decoder (#1595)
-- Fix: respect `-stdout` even if multiple CC tracks are present in a Matroska input file (#1453)
-- Fix: crash in Rust decoder on ATSC1.0 TS Files (#1407)
-- Removed the --with-gui flag for linux/configure and mac/configure (use the Flutter GUI instead)
+  Refactor: Lots of code ported to Rust.
+- Fix: Improved handling of IETF language tags in Matroska files (#1665)
+- Breaking: Major argument flags revamp for CCExtractor (#1564 & #1619)
 - Fix: segmentation fault in using hardsubx
-- New: Add function (and command) that extracts closed caption subtitles as well as burnt-in subtitles from a file in a single pass. (As proposed in issue 726)
-- Refactored: the `general_loop` function has some code moved to a new function
 - Fix: WebVTT X-TIMESTAMP-MAP placement (#1463)
-- Disable X-TIMESTAMP-MAP by default (changed option --no-timestamp-map to --timestamp-map)
-- Fix: missing `#` in color attribute of font tag
 - Fix: ffmpeg 5.0, tesseract 5.0 compatibility and remove deprecated methods
 - Fix: tesseract 5.x traineddata location in ocr
-- Fix: fix autoconf tesseract detection problem (#1503)
-- Fix: add missing compile_info_real.h source to Autotools build
-- Fix: add missing `-lavfilter` for hardsubx linking
-- Fix: make webvtt-full work correctly with multi-byte utf-8 characters
-- Fix: encoding of solid block in latin-1 and unicode
-- Fix: McPoodle Broadcast Raw format for field 1
-- Fix: Incorrect skipping of packets
-- Fix: Repeated values for enums
-- Cleanup: Remove the (unmaintained) Nuklear GUI code
-- Cleanup: Reduce the amount of Windows build options in the project file
-- Fix: infinite loop in MP4 file type detector.
-- Improvement: Use Corrosion to build Rust code
 - Improvement: Ignore MXF Caption Essence Container version byte to enhance SRT subtitle extraction compatibility
 - New: Add tesseract page segmentation modes control with `--psm` flag
-- Fix: Resolve compile-time error about implicit declarations (#1646)
-- Fix: fatal out of memory error extracting from a VOB PS
-- Fix: Unit Test Rust failing due to changes in Rust Version 1.86.0
 - Fix: Support for MINGW-w64 cross compiling
-- Fix: Build with ENABLE_FFMPEG to support ffmpeg 5
 
 0.94 (2021-12-14)
 -----------------

src/lib_ccx/ccx_common_structs.h

@@ -84,6 +84,9 @@ struct cc_subtitle
 	/** Raw PTS value when this subtitle started (for DVB timing) */
 	LLONG start_pts;
 
+	/** Teletext page number (for multi-page extraction, issue #665) */
+	uint16_t teletext_page;
+
 	struct cc_subtitle *next;
 	struct cc_subtitle *prev;
 };

src/lib_ccx/ccx_encoders_common.c

@@ -719,6 +719,9 @@ void dinit_encoder(struct encoder_ctx **arg, LLONG current_fts)
 		write_subtitle_file_footer(ctx, ctx->out + i);
 	}
 
+	// Clean up teletext multi-page output files (issue #665)
+	dinit_teletext_outputs(ctx);
+
 	free_encoder_context(ctx->prev);
 	dinit_output_ctx(ctx);
 	freep(&ctx->subline);
@@ -838,6 +841,15 @@ struct encoder_ctx *init_encoder(struct encoder_cfg *opt)
 	ctx->segment_last_key_frame = 0;
 	ctx->nospupngocr = opt->nospupngocr;
 
+	// Initialize teletext multi-page output arrays (issue #665)
+	ctx->tlt_out_count = 0;
+	for (int i = 0; i < MAX_TLT_PAGES_EXTRACT; i++)
+	{
+		ctx->tlt_out[i] = NULL;
+		ctx->tlt_out_pages[i] = 0;
+		ctx->tlt_srt_counter[i] = 0;
+	}
+
 	ctx->prev = NULL;
 	return ctx;
 }
@@ -1298,3 +1310,168 @@ void switch_output_file(struct lib_ccx_ctx *ctx, struct encoder_ctx *enc_ctx, in
 	enc_ctx->cea_708_counter = 0;
 	enc_ctx->srt_counter = 0;
 }
+
+/**
+ * Get or create the output file for a specific teletext page (issue #665)
+ * Creates output files on-demand with suffix _pNNN (e.g., output_p891.srt)
+ * Returns NULL if we're in stdout mode or if too many pages are being extracted
+ */
+struct ccx_s_write *get_teletext_output(struct encoder_ctx *ctx, uint16_t teletext_page)
+{
+	// If teletext_page is 0, use the default output
+	if (teletext_page == 0 || ctx->out == NULL)
+		return ctx->out;
+
+	// Check if we're sending to stdout - can't do multi-page in that case
+	if (ctx->out[0].fh == STDOUT_FILENO)
+		return ctx->out;
+
+	// Check if we already have an output file for this page
+	for (int i = 0; i < ctx->tlt_out_count; i++)
+	{
+		if (ctx->tlt_out_pages[i] == teletext_page)
+			return ctx->tlt_out[i];
+	}
+
+	// If we only have one teletext page requested, use the default output
+	// (no suffix needed for backward compatibility)
+	extern struct ccx_s_teletext_config tlt_config;
+	if (tlt_config.num_user_pages <= 1 && !tlt_config.extract_all_pages)
+		return ctx->out;
+
+	// Need to create a new output file for this page
+	if (ctx->tlt_out_count >= MAX_TLT_PAGES_EXTRACT)
+	{
+		mprint("Warning: Too many teletext pages to extract (max %d), using default output for page %03d\n",
+		       MAX_TLT_PAGES_EXTRACT, teletext_page);
+		return ctx->out;
+	}
+
+	// Allocate the new write structure
+	struct ccx_s_write *new_out = (struct ccx_s_write *)malloc(sizeof(struct ccx_s_write));
+	if (!new_out)
+	{
+		mprint("Error: Memory allocation failed for teletext output\n");
+		return ctx->out;
+	}
+	memset(new_out, 0, sizeof(struct ccx_s_write));
+
+	// Create the filename with page suffix
+	const char *ext = get_file_extension(ctx->write_format);
+	char suffix[16];
+	snprintf(suffix, sizeof(suffix), "_p%03d", teletext_page);
+
+	char *basefilename = NULL;
+	if (ctx->out[0].filename != NULL)
+	{
+		basefilename = get_basename(ctx->out[0].filename);
+	}
+	else if (ctx->first_input_file != NULL)
+	{
+		basefilename = get_basename(ctx->first_input_file);
+	}
+	else
+	{
+		basefilename = strdup("untitled");
+	}
+
+	if (basefilename == NULL)
+	{
+		free(new_out);
+		return ctx->out;
+	}
+
+	char *filename = create_outfilename(basefilename, suffix, ext);
+	free(basefilename);
+
+	if (filename == NULL)
+	{
+		free(new_out);
+		return ctx->out;
+	}
+
+	// Open the file
+	new_out->filename = filename;
+	new_out->fh = open(filename, O_RDWR | O_CREAT | O_TRUNC | O_BINARY, S_IREAD | S_IWRITE);
+	if (new_out->fh == -1)
+	{
+		mprint("Error: Failed to open output file %s: %s\n", filename, strerror(errno));
+		free(filename);
+		free(new_out);
+		return ctx->out;
+	}
+
+	mprint("Creating teletext output file: %s\n", filename);
+
+	// Store in our array
+	int idx = ctx->tlt_out_count;
+	ctx->tlt_out[idx] = new_out;
+	ctx->tlt_out_pages[idx] = teletext_page;
+	ctx->tlt_srt_counter[idx] = 0;
+	ctx->tlt_out_count++;
+
+	// Write the subtitle file header
+	write_subtitle_file_header(ctx, new_out);
+
+	return new_out;
+}
+
+/**
+ * Get the SRT counter for a specific teletext page (issue #665)
+ * Returns pointer to the counter, or NULL if page not found
+ */
+unsigned int *get_teletext_srt_counter(struct encoder_ctx *ctx, uint16_t teletext_page)
+{
+	// If teletext_page is 0, use the default counter
+	if (teletext_page == 0)
+		return &ctx->srt_counter;
+
+	// Check if we're using multi-page mode
+	extern struct ccx_s_teletext_config tlt_config;
+	if (tlt_config.num_user_pages <= 1 && !tlt_config.extract_all_pages)
+		return &ctx->srt_counter;
+
+	// Find the counter for this page
+	for (int i = 0; i < ctx->tlt_out_count; i++)
+	{
+		if (ctx->tlt_out_pages[i] == teletext_page)
+			return &ctx->tlt_srt_counter[i];
+	}
+
+	// Not found, use default counter
+	return &ctx->srt_counter;
+}
+
+/**
+ * Clean up all teletext output files (issue #665)
+ */
+void dinit_teletext_outputs(struct encoder_ctx *ctx)
+{
+	if (!ctx)
+		return;
+
+	for (int i = 0; i < ctx->tlt_out_count; i++)
+	{
+		if (ctx->tlt_out[i] != NULL)
+		{
+			// Write footer
+			write_subtitle_file_footer(ctx, ctx->tlt_out[i]);
+
+			// Close file
+			if (ctx->tlt_out[i]->fh != -1)
+			{
+				close(ctx->tlt_out[i]->fh);
+			}
+
+			// Free filename
+			if (ctx->tlt_out[i]->filename != NULL)
+			{
+				free(ctx->tlt_out[i]->filename);
+			}
+
+			free(ctx->tlt_out[i]);
+			ctx->tlt_out[i] = NULL;
+		}
+	}
+	ctx->tlt_out_count = 0;
+}

src/lib_ccx/ccx_encoders_common.h

@@ -16,6 +16,11 @@
 #include "ccx_encoders_structs.h"
 #include "ccx_common_option.h"
 
+// Maximum number of teletext pages to extract simultaneously (issue #665)
+#ifndef MAX_TLT_PAGES_EXTRACT
+#define MAX_TLT_PAGES_EXTRACT 8
+#endif
+
 #define REQUEST_BUFFER_CAPACITY(ctx, length)                                                                       \
 	if (length > ctx->capacity)                                                                                \
 	{                                                                                                          \
@@ -169,6 +174,12 @@ struct encoder_ctx
 
 	// OCR in SPUPNG
 	int nospupngocr;
+
+	// Teletext multi-page output (issue #665)
+	struct ccx_s_write *tlt_out[MAX_TLT_PAGES_EXTRACT]; // Output files per teletext page
+	uint16_t tlt_out_pages[MAX_TLT_PAGES_EXTRACT];       // Page numbers for each output slot
+	unsigned int tlt_srt_counter[MAX_TLT_PAGES_EXTRACT]; // SRT counter per page
+	int tlt_out_count;                                    // Number of teletext output files
 };
 
 #define INITIAL_ENC_BUFFER_CAPACITY 2048
@@ -263,4 +274,9 @@ unsigned int get_font_encoded(struct encoder_ctx *ctx, unsigned char *buffer, in
 
 struct lib_ccx_ctx;
 void switch_output_file(struct lib_ccx_ctx *ctx, struct encoder_ctx *enc_ctx, int track_id);
+
+// Teletext multi-page output (issue #665)
+struct ccx_s_write *get_teletext_output(struct encoder_ctx *ctx, uint16_t teletext_page);
+unsigned int *get_teletext_srt_counter(struct encoder_ctx *ctx, uint16_t teletext_page);
+void dinit_teletext_outputs(struct encoder_ctx *ctx);
 #endif