Simplify decode c api

src/exe/cimbar_recv2/recv2.cpp

@@ -150,17 +150,28 @@ int main(int argc, char** argv)
 			// attempt save
 			uint32_t fileId = res;
 
-			unsigned size = cimbard_get_filesize(fileId);
-			std::string filename = fmt::format("0.{}", size);
-			std::cerr << "Saving file " << filename << " of size " << size << std::endl;
+			int size = cimbard_get_filesize(fileId);
+
+			std::string filename;
+			filename.resize(255, '\0');
+			int fnsize = cimbard_get_filename(fileId, filename.data(), filename.size());
+			if (fnsize > 0)
+				filename.resize(fnsize);
+			else // fallback
+				filename = fmt::format("0.{}", size);
+			std::cerr << "Saving file " << filename << " of (compressed) size " << size << std::endl;
+
+			std::string file_path = fmt::format("{}/{}", outpath, filename);
+			std::ofstream outs(file_path, std::ios::binary);
 
 			std::vector<unsigned char> data;
-			data.resize(size);
-			int res = cimbard_finish_copy(fileId, data.data(), size);
-			if (res != 0)
-				std::cerr << "failed fountain_finish_copy " << res << std::endl;
+			data.resize(cimbard_get_decompress_bufsize());
 
-			decompress_on_store<std::ofstream>(outpath, true)(filename, data);
+			int res = 1;
+			while ((res = cimbard_decompress_read(fileId, data.data(), data.size())) > 0)
+				outs.write(reinterpret_cast<const char*>(data.data()), res);
+			if (res < 0)
+				std::cerr << "failed cimbard_decompress_read " << res << std::endl;
 		}
 	}
 

src/lib/cimbar_js/CMakeLists.txt

@@ -14,8 +14,6 @@ set(SOURCES
 
 	cimbar_recv_js.h
 	cimbar_recv_js.cpp
-	cimbar_zstd_js.h
-	cimbar_zstd_js.cpp
 )
 endif()
 
@@ -66,16 +64,14 @@ set(LINK_WASM_EXPORTED_FUNCTIONS
 	${LINK_WASM_EXPORTED_FUNCTIONS}
 
 	, "_cimbard_get_report"
+	, "_cimbard_get_bufsize"
 	, "_cimbard_scan_extract_decode"
 	, "_cimbard_fountain_decode"
-	, "_cimbard_get_bufsize"
 	, "_cimbard_get_filesize"
 	, "_cimbard_get_filename"
-	, "_cimbard_finish_copy"
+	, "_cimbard_get_decompress_bufsize"
+	, "_cimbard_decompress_read"
 	, "_cimbard_configure_decode"
-	, "_cimbarz_init_decompress"
-	, "_cimbarz_get_bufsize"
-	, "_cimbarz_decompress_read"
 )
 endif()
 

src/lib/cimbar_js/cimbar_recv_js.cpp

@@ -19,10 +19,16 @@
 
 
 namespace {
-
+	// for decode
 	std::shared_ptr<fountain_decoder_sink> _sink;
-	std::string _reporting;
 
+	// for decompress
+	// we support only one decompress at a time!
+	uint32_t _decId = 0;
+	std::vector<uchar> _reassembled;
+	std::unique_ptr<cimbar::zstd_decompressor<std::stringstream>> _dec;
+
+	std::string _reporting;
 	cv::Mat _debugFrame;
 
 	TimeAccumulator _tScanExtract;
@@ -31,6 +37,48 @@ namespace {
 	// settings
 	int _modeVal = 68;
 
+	// set up stateful decompressor
+	// for api simplicity, this is coupled to recover_contents()
+	// ... but we *could* split them up
+	int init_decompress(uint32_t id)
+	{
+		if (id != _decId)
+			return -11;
+		if (_dec)
+			_dec.reset();
+		_dec = std::make_unique<cimbar::zstd_decompressor<std::stringstream>>();
+		if (!_dec)
+			return -12;
+		_dec->init_decompress(reinterpret_cast<char*>(_reassembled.data()), _reassembled.size());
+		return 0;
+	}
+
+	// recovers file contents into _reassembled, and sets _decId = id if so.
+	// if the contents cannot be recovered, amounts to a no-op with error
+	int recover_contents(uint32_t id)
+	{
+		if (id != _decId)
+		{
+			if (!_sink)
+				return -1;
+			if (_sink->is_done(id))
+				return -2; // it's gone man
+
+			_reassembled.resize(cimbard_get_filesize(id));
+			if (!_sink->recover(id, _reassembled.data(), _reassembled.size()))
+				return -3;
+			_decId = id;
+
+			int res = init_decompress(id);
+			if (res < 0)
+				return res;
+		}
+		if (_reassembled.empty())
+			return -5;
+
+		return 0;
+	}
+
 	unsigned fountain_chunks_per_frame()
 	{
 		return cimbar::Config::fountain_chunks_per_frame(
@@ -166,14 +214,24 @@ int64_t cimbard_fountain_decode(const unsigned char* buffer, unsigned size)
 	return res;
 }
 
-int cimbard_get_filesize(uint32_t id)
+// mostly for internal use, but also helpful for debugging
+unsigned cimbard_get_filesize(uint32_t id)
 {
 	FountainMetadata md(id);
 	return md.file_size();
 }
 
-int cimbard_get_filename(const uchar* finbuffer, unsigned size, char* filename, unsigned fnsize)
+// do recover() if does not exist (fail with appropriate neg values), if it does use the bytes.
+// stateful against a map (same as cimbard_decompress_read()
+int cimbard_get_filename(uint32_t id, char* filename, unsigned fnsize)
 {
+	int	res = recover_contents(id);
+	if (res < 0)
+		return res;
+
+	const uchar* finbuffer = _reassembled.data();
+	unsigned size = _reassembled.size();
+
 	std::string fn = cimbar::zstd_header_check::get_filename(finbuffer, size);
 	if (!fn.empty())
 		fn = File::basename(fn);
@@ -186,15 +244,29 @@ int cimbard_get_filename(const uchar* finbuffer, unsigned size, char* filename,
 	return fn.size();
 }
 
-// if fountain_decode returned a >0 value, call this to retrieve the reassembled file
-// bouth fountain_*() calls should be from the same js webworker/thread
-int cimbard_finish_copy(uint32_t id, uchar* buffer, unsigned size)
+int cimbard_decompress_read(uint32_t id, unsigned char* buffer, unsigned size)
 {
-	if (!_sink)
-		return -1;
-	if (!_sink->recover(id, buffer, size))
-		return -2;
-	return 0;
+	int	res = recover_contents(id);
+	if (res < 0)
+		return res;
+
+	if (!_dec)
+		return -13;
+	if (!_dec->good())
+		return -14;
+
+	_dec->str(std::string());
+	_dec->write_once();
+	std::string temp = _dec->str();
+	if (size > temp.size())
+		size = temp.size();
+	std::copy(temp.data(), temp.data()+size, buffer);
+	return size;
+}
+
+int cimbard_get_decompress_bufsize()
+{
+	return ZSTD_DStreamOutSize();
 }
 
 int cimbard_configure_decode(int mode_val)
@@ -215,4 +287,12 @@ int cimbard_configure_decode(int mode_val)
 	return 0;
 }
 
+// testing
+unsigned char* cimbard_get_reassembled_file_buff()
+{
+	if (_reassembled.empty())
+		return nullptr;
+	return _reassembled.data();
+}
+
 }

src/lib/cimbar_js/cimbar_recv_js.h

@@ -20,20 +20,24 @@ int cimbard_scan_extract_decode(const unsigned char* imgdata, unsigned imgw, uns
 // persists state, the return value (if >0) corresponds to a uint32_t id
 int64_t cimbard_fountain_decode(const unsigned char* buffer, unsigned size);
 
-// get filesize from id
-int cimbard_get_filesize(uint32_t id);
-
-// if fountain_decode returned a >0 value, call this to retrieve the reassembled file
-// wherever a uint32_t id is passed, it should be on the same js thread
-// ... or at least in the same js shared memory...
-// as the fountain_decode() call
-int cimbard_finish_copy(uint32_t id, unsigned char* finbuffer, unsigned size);
-
-// get filename from reassembled file
-int cimbard_get_filename(const unsigned char* finbuffer, unsigned size, char* filename, unsigned fnsize);
+// get compressed filesize from id
+// you probably don't need to use this.
+unsigned cimbard_get_filesize(uint32_t id);
+
+// if fountain_decode returned a >0 value,
+//  get filename and (partial) contents from reassembled file
+// wherever a uint32_t id is passed, it should be in the
+//  same js shared memory as the fountain_decode() call
+// cimbard_decompress_read() will return 0 when all file contents have been read
+int cimbard_get_filename(uint32_t id, char* filename, unsigned fnsize);
+int cimbard_get_decompress_bufsize();
+int cimbard_decompress_read(uint32_t id, unsigned char* buffer, unsigned size);
 
 int cimbard_configure_decode(int mode_val);
 
+// testing usage only!
+unsigned char* cimbard_get_reassembled_file_buff();
+
 #ifdef __cplusplus
 }
 #endif

src/lib/cimbar_js/test/cimbar_jsTest.cpp

@@ -5,7 +5,6 @@
 #include "cimb_translator/Config.h"
 #include "cimbar_js/cimbar_js.h"
 #include "cimbar_js/cimbar_recv_js.h"
-#include "cimbar_js/cimbar_zstd_js.h"
 #include "serialize/format.h"
 
 #include <iostream>
@@ -55,27 +54,17 @@ TEST_CASE( "cimbar_jsTest/testRoundtrip", "[unit]" )
 	{
 		uint32_t fileId = res;
 
-		unsigned size = cimbard_get_filesize(fileId);
-		assertEquals( 5282, size );
-
-		std::vector<unsigned char> data;
-		data.resize(size);
-		int res = cimbard_finish_copy(fileId, data.data(), size);
-		assertEquals( 0, res );
-
 		std::string actualFilename;
 		actualFilename.resize(255);
-		int fnsz = cimbard_get_filename(data.data(), size, actualFilename.data(), actualFilename.size());
+		int fnsz = cimbard_get_filename(fileId, actualFilename.data(), actualFilename.size());
 		assertEquals( 21, fnsz );
 		actualFilename.resize(fnsz);
 		assertEquals( "foobar-c语言版.txt", actualFilename );
 
-		assertEquals(0, cimbarz_init_decompress(data.data(), data.size()));
-
 		std::vector<unsigned char> zstdbuff;
-		zstdbuff.resize(cimbarz_get_bufsize());
+		zstdbuff.resize(cimbard_get_decompress_bufsize());
 
-		int outsize = cimbarz_decompress_read(zstdbuff.data(), zstdbuff.size());
+		int outsize = cimbard_decompress_read(fileId, zstdbuff.data(), zstdbuff.size());
 		assertEquals(contents.size(), outsize);
 		std::string_view finalOutput{reinterpret_cast<char*>(zstdbuff.data()), contents.size()};