Removal of Sphinxbase dependency and update Pocketsphinx to v5.0.3

.gitignore

@@ -35,3 +35,4 @@ CppCheckResults.xml
 .eggs
 *.egg-info
 *.snap
+*result

doc/install.md

@@ -5,7 +5,7 @@ You could download release builds from [download page](https://sc0ty.github.io/s
 - C++11 compatible compiler (or better C++14);
 - pybind11;
 - ffmpeg libraries (4.0 or newer);
-- sphinxbase and pocketsphinx;
+- pocketsphinx;
 - Python interpreter (supporting Python 3.5 or newer);
 - Python modules listed in `requirements.txt` file;
 
@@ -30,10 +30,9 @@ Activate your virtual environment:
 source .env/bin/activate
 ```
 
-If you have ffmpeg, sphinxbase and pocketsphinx libraries installed and avaiable via `pkg-config`, it will be configured automatically. Otherwise, you must provide paths to these libraries manually, using evironment variables:
+If you have ffmpeg and pocketsphinx libraries installed and avaiable via `pkg-config`, it will be configured automatically. Otherwise, you must provide paths to these libraries manually, using evironment variables:
 ```
 export FFMPEG_DIR=PATH
-export SPHINXBASE_DIR=PATH
 export POCKETSPHINX_DIR=PATH
 export USE_PKG_CONFIG=no
 ```
@@ -75,13 +74,12 @@ python -m venv .env
 ```
 
 To build subsync, you need to provide dependencies first.
-Sphinxbase and pocketsphinx are published with Visual Studio solution file, which is [easy to use](https://github.com/cmusphinx/pocketsphinx#ms-windows-ms-visual-studio-2012-or-newer---we-test-with-vc-2012-express).
+pocketsphinx is published with a Visual Studio solution file, which is [easy to use](https://github.com/cmusphinx/pocketsphinx#ms-windows-ms-visual-studio-2012-or-newer---we-test-with-vc-2012-express).
 Building ffmpeg on the other hand is not that easy. You could use [official build](https://ffmpeg.zeranoe.com/builds/) instead.
 
 Configure dependencies paths:
 ```
 set FFMPEG_DIR=d:\projects\ffmpeg
-set SPHINXBASE_DIR=d:\projects\sphinxbase
 set POCKETSPHINX_DIR=d:\projects\pocketsphinx
 set USE_PKG_CONFIG=no
 ```

flake.nix

@@ -0,0 +1,64 @@
+{
+  inputs = {
+    nixpkgs = {
+      type = "github";
+      owner = "NixOS";
+      repo = "nixpkgs";
+      ref = "nixos-unstable";
+    };
+
+    # srcs
+    pocketsphinx-src = {
+      type = "github";
+      owner = "cmusphinx";
+      repo = "pocketsphinx";
+      ref = "v5.0.3";
+      flake = false;
+    };
+  };
+
+  outputs = {
+    self,
+    nixpkgs,
+    pocketsphinx-src,
+  }: let
+    supportedSystems = ["x86_64-linux"];
+
+    perSystem = attrs:
+      nixpkgs.lib.genAttrs supportedSystems (system: let
+        pkgs = nixpkgs.legacyPackages.${system};
+      in
+        attrs system pkgs);
+  in {
+    packages = perSystem (system: pkgs: let
+      mkDate = longDate: (pkgs.lib.concatStringsSep "-" [
+        (builtins.substring 0 4 longDate)
+        (builtins.substring 4 2 longDate)
+        (builtins.substring 6 2 longDate)
+      ]);
+      date = mkDate (self.lastModifiedDate or "19700101");
+    in {
+      pocketsphinx = pkgs.callPackage ./nix/pocketsphinx.nix {
+        inherit pocketsphinx-src;
+      };
+
+      subsync = pkgs.callPackage ./nix {
+        version = date;
+        inherit (self.packages.${system}) pocketsphinx;
+      };
+
+      default = self.packages.${system}.subsync;
+    });
+
+    formatter = perSystem (_: pkgs: pkgs.alejandra);
+
+    devShells = perSystem (_: pkgs: {
+      default = pkgs.mkShell {
+        packages = with pkgs; [
+          alejandra
+          # ... more dev packages
+        ];
+      };
+    });
+  };
+}

gizmo/general/logger.cpp

@@ -1,6 +1,7 @@
 #include "logger.h"
 #include "text/utf8.h"
-#include <sphinxbase/err.h>
+#include <pocketsphinx/err.h>
+#include <pocketsphinx.h>
 #include <string>
 #include <cstdarg>
 
@@ -60,7 +61,7 @@ static void sphinxLogCb(void *user_data, err_lvl_t level, const char *fmt, ...)
 {
 	(void) user_data;
 
-	if (g_loggerCallback && level != ERR_INFOCONT && level >= g_sphinxLogLevel)
+	if (g_loggerCallback && level >= g_sphinxLogLevel)
 	{
 		char line[MAX_LOG_SIZE];
 		va_list args;
@@ -73,7 +74,6 @@ static void sphinxLogCb(void *user_data, err_lvl_t level, const char *fmt, ...)
 		{
 			case ERR_DEBUG:    lvl = LOG_DEBUG;    break;
 			case ERR_INFO:     lvl = LOG_DEBUG;    break;
-			case ERR_INFOCONT: lvl = LOG_DEBUG;    break;
 			case ERR_WARN:     lvl = LOG_WARNING;  break;
 			case ERR_ERROR:    lvl = LOG_ERROR;    break;
 			case ERR_FATAL:    lvl = LOG_CRITICAL; break;

gizmo/media/speechrec.cpp

@@ -17,35 +17,39 @@ SpeechRecognition::SpeechRecognition() :
 	m_minProb(1.0f),
 	m_minLen(0)
 {
-	m_config = cmd_ln_parse_r(NULL, ps_args(), 0, NULL, TRUE);
+	m_config = ps_config_init(NULL);
 	if (m_config == NULL)
 		throw EXCEPTION("can't init Sphinx configuration")
-			.module("SpeechRecognition", "cmd_ln_parse_r");
+			.module("SpeechRecognition", "ps_config_init");
 }
 
 SpeechRecognition::~SpeechRecognition()
 {
-	cmd_ln_free_r(m_config);
+	ps_config_free(m_config);
 }
 
 void SpeechRecognition::setParam(const string &key, const string &val)
 {
-	arg_t const *args = ps_args();
+	ps_arg_t const *args = ps_args();
 
 	for (size_t i = 0; args[i].name != NULL; i++)
 	{
-		if (key == args[i].name)
+		if (key == string("-").append(args[i].name))
 		{
 			int type = args[i].type;
 			if (type & ARG_INTEGER)
-				cmd_ln_set_int_r(m_config, key.c_str(), atol(val.c_str()));
+				ps_config_set_int(m_config, args[i].name, atol(val.c_str()));
+
 			else if (type & ARG_FLOATING)
-				cmd_ln_set_float_r(m_config, key.c_str(), atof(val.c_str()));
+				ps_config_set_float(m_config, args[i].name, atof(val.c_str()));
+
 			else if (type & ARG_STRING)
-				cmd_ln_set_str_r(m_config, key.c_str(), val.c_str());
+				ps_config_set_str(m_config, args[i].name, val.c_str());
+
 			else if (type & ARG_BOOLEAN)
-				cmd_ln_set_boolean_r(m_config, key.c_str(),
+				ps_config_set_bool(m_config, args[i].name,
 						!(val.empty() || val == "0"));
+
 			else
 				throw EXCEPTION("invalid parameter type")
 					.module("SpeechRecognition", "setParameter")
@@ -89,14 +93,14 @@ void SpeechRecognition::start(const AVStream *stream)
 		throw EXCEPTION("can't init Sphinx engine")
 			.module("SpeechRecognition", "ps_init");
 
-	int32_t frate = cmd_ln_int32_r(m_config, "-frate");
+	int32_t frate = ps_config_int(m_config, "frate");
 	m_framePeriod = 1.0 / (double)frate;
 
 	if (frate == 0)
 		throw EXCEPTION("can't get frame rate value")
-			.module("SpeechRecognition", "cmd_ln_int32_r");
+			.module("SpeechRecognition", "ps_config_get");
 
-	if (ps_start_utt(m_ps))
+	if (ps_start_utt(m_ps) < 0)
 		throw EXCEPTION("can't start speech recognition")
 			.module("SpeechRecognition", "ps_start_utt");
 
@@ -108,7 +112,7 @@ void SpeechRecognition::stop()
 {
 	if (m_ps)
 	{
-		if (ps_end_utt(m_ps))
+		if (ps_end_utt(m_ps) < 0)
 			throw EXCEPTION("can't stop speech recognition")
 				.module("SpeechRecognition", "ps_end_utt");
 
@@ -143,13 +147,13 @@ void SpeechRecognition::feed(const AVFrame *frame)
 	}
 	if (!inSpeech && m_utteranceStarted)
 	{
-		if (ps_end_utt(m_ps))
+		if (ps_end_utt(m_ps) < 0)
 			throw EXCEPTION("can't end utterance")
 				.module("SpeechRecognition", "ps_end_utt");
 
 		parseUtterance();
 
-		if (ps_start_utt(m_ps))
+		if (ps_start_utt(m_ps) < 0)
 			throw EXCEPTION("can't start utterance")
 				.module("SpeechRecognition", "ps_start_utt");
 
@@ -163,21 +167,16 @@ void SpeechRecognition::flush()
 
 void SpeechRecognition::discontinuity()
 {
-	if (ps_end_utt(m_ps))
+	if (ps_end_utt(m_ps) < 0)
 		throw EXCEPTION("can't stop speech recognition")
 			.module("SpeechRecognition", "ps_end_utt");
 
 	if (m_utteranceStarted)
 		parseUtterance();
 
 	m_deltaTime = -1.0;
-	if (ps_start_stream(m_ps))
-	{
-		throw EXCEPTION("can't reset speech recognition engine")
-			.module("SpeechRecognition", "sphinx", "ps_start_stream");
-	}
 
-	if (ps_start_utt(m_ps))
+	if (ps_start_utt(m_ps) < 0)
 		throw EXCEPTION("can't start speech recognition")
 			.module("SpeechRecognition", "ps_start_utt");
 

gizmo/media/speechrec.h

@@ -38,7 +38,7 @@ class SpeechRecognition : public AVOutput
 
 	private:
 		ps_decoder_t *m_ps;
-		cmd_ln_t *m_config;
+		ps_config_t *m_config;
 
 		bool m_utteranceStarted;
 

gizmo/text/ssa.h

@@ -5,6 +5,7 @@
 #include <list>
 #include <set>
 #include <string>
+#include <cstdint>
 
 
 class SSAParser

gizmo/text/utf8.h

@@ -2,6 +2,7 @@
 #define __UTF8_H__
 
 #include <string>
+#include <cstdint>
 
 
 class Utf8

gizmo/util

@@ -0,0 +1 @@
+../pocketsphinx/src/util/

nix/default.nix

@@ -0,0 +1,55 @@
+{
+  ffmpeg,
+  pkg-config,
+  pocketsphinx,
+  python3Packages,
+  version,
+  ...
+}: let
+  inherit (builtins) concatStringsSep;
+in
+  python3Packages.buildPythonPackage {
+    pname = "subsync";
+    inherit version;
+    format = "other";
+
+    src = ../.;
+
+    buildInputs = [
+      ffmpeg
+      pkg-config
+      pocketsphinx
+    ];
+
+    nativeBuildInputs = with python3Packages; [
+      pip
+      setuptools
+      wheel
+    ];
+
+    propagatedBuildInputs = with python3Packages; [
+      certifi
+      cryptography
+      pybind11
+      pycryptodome
+      pysubs2
+      pyyaml
+      requests
+      utils
+    ];
+
+    # The tests are for the GUI
+    doCheck = false;
+
+    # 'pip install .' takes care of building the package
+    buildPhase = "";
+
+    installPhase = ''
+      python -m pip install . ${concatStringsSep " " [
+        "--no-index"
+        "--no-warn-script-location"
+        "--prefix=\"$out\""
+        "--no-cache"
+      ]}
+    '';
+  }

nix/pocketsphinx.nix

@@ -0,0 +1,14 @@
+{
+  cmake,
+  pocketsphinx-src,
+  stdenv,
+  ...
+}:
+stdenv.mkDerivation {
+  pname = "pocketsphinx";
+  version = pocketsphinx-src.shortRev;
+
+  src = pocketsphinx-src;
+
+  buildInputs = [cmake];
+}