Hacks.libssa

8 files changed, 133 insertions, 302 deletions

diff --git a/src/ssa_reader.cc b/src/ssa_reader.cc
index c76a78d..a4daf24 100644
--- a/src/ssa_reader.cc
+++ b/src/ssa_reader.cc
@@ -22,6 +22,7 @@
 #include "sub_assert.h"
 #include "raw_convert.h"
 #include "subtitle.h"
+#include <ass/ass.h>
 #include <boost/algorithm/string.hpp>
 #include <boost/bind.hpp>
 #include <boost/foreach.hpp>
@@ -43,281 +44,77 @@ using namespace sub;
 /** @param s Subtitle string encoded in UTF-8 */
 SSAReader::SSAReader (string const & s)
 {
-	stringstream str (s);
-	this->read (boost::bind (&get_line_stringstream, &str));
+	char* buffer = new char[s.length() + 1];
+	strcpy (buffer, s.c_str());
+	read (buffer, s.length() + 1);
 }
 
 /** @param f Subtitle file encoded in UTF-8 */
 SSAReader::SSAReader (FILE* f)
 {
-	this->read (boost::bind (&get_line_file, f));
-}
-
-class Style
-{
-public:
-	Style ()
-		: font_size (24)
-		, primary_colour (255, 255, 255)
-		, bold (false)
-		, italic (false)
-	{}
-
-	Style (string format_line, string style_line)
-		: font_size (24)
-		, primary_colour (255, 255, 255)
-		, bold (false)
-		, italic (false)
-	{
-		vector<string> keys;
-		split (keys, format_line, is_any_of (","));
-		vector<string> style;
-		split (style, style_line, is_any_of (","));
-
-		SUB_ASSERT (!keys.empty());
-		SUB_ASSERT (!style.empty());
-		SUB_ASSERT (keys.size() == style.size());
-
-		for (size_t i = 0; i < style.size(); ++i) {
-			trim (keys[i]);
-			trim (style[i]);
-			if (keys[i] == "Name") {
-				name = style[i];
-			} else if (keys[i] == "Fontname") {
-				font_name = style[i];
-			} else if (keys[i] == "Fontsize") {
-				font_size = raw_convert<int> (style[i]);
-			} else if (keys[i] == "PrimaryColour") {
-				primary_colour = colour (raw_convert<int> (style[i]));
-			} else if (keys[i] == "BackColour") {
-				back_colour = colour (raw_convert<int> (style[i]));
-			} else if (keys[i] == "Bold") {
-				bold = style[i] == "-1";
-			} else if (keys[i] == "Italic") {
-				italic = style[i] == "-1";
-			} else if (keys[i] == "BorderStyle") {
-				if (style[i] == "1") {
-					effect = SHADOW;
-				}
-			}
-		}
-	}
-
-	string name;
-	optional<string> font_name;
-	int font_size;
-	Colour primary_colour;
-	/** outline colour */
-	optional<Colour> back_colour;
-	bool bold;
-	bool italic;
-	optional<Effect> effect;
+	fseek (f, 0, SEEK_END);
+	long const size = ftell (f);
+	fseek (f, 0, SEEK_SET);
+	char* buffer = new char[size];
+	fread (buffer, size, 1, f);
 
-private:
-	Colour colour (int c) const
-	{
-		return Colour (
-			((c & 0x0000ff) >>  0) / 255.0,
-			((c & 0x00ff00) >>  8) / 255.0,
-			((c & 0xff0000) >> 16) / 255.0
-			);
-	}
-};
-
-Time
-SSAReader::parse_time (string t) const
-{
-	vector<string> bits;
-	split (bits, t, is_any_of (":."));
-	SUB_ASSERT (bits.size() == 4);
-	return Time::from_hms (
-		raw_convert<int> (bits[0]),
-		raw_convert<int> (bits[1]),
-		raw_convert<int> (bits[2]),
-		raw_convert<int> (bits[3]) * 10
-		);
+	read (buffer, size);
 }
 
-/** @param base RawSubtitle filled in with any required common values.
- *  @param line SSA line string.
- *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
- */
-list<RawSubtitle>
-SSAReader::parse_line (RawSubtitle base, string line)
+void
+SSAReader::read (char* buffer, int size)
 {
-	enum {
-		TEXT,
-		STYLE,
-		BACKSLASH
-	} state = TEXT;
-
-	list<RawSubtitle> subs;
-	RawSubtitle current = base;
-	string style;
-
-	current.vertical_position.line = 0;
-	/* XXX: arbitrary */
-	current.vertical_position.lines = 32;
-	current.vertical_position.reference = TOP_OF_SUBTITLE;
-
-	for (size_t i = 0; i < line.length(); ++i) {
-		char const c = line[i];
-		switch (state) {
-		case TEXT:
-			if (c == '{') {
-				state = STYLE;
-			} else if (c == '\\') {
-				state = BACKSLASH;
-			} else if (c != '\r' && c != '\n') {
-				current.text += c;
-			}
-			break;
-		case STYLE:
-			if (c == '}') {
-				if (!current.text.empty ()) {
-					subs.push_back (current);
-					current.text = "";
-				}
-				if (style == "i1") {
-					current.italic = true;
-				} else if (style == "i0") {
-					current.italic = false;
-				}
-				style = "";
-				state = TEXT;
-			} else {
-				style += c;
-			}
-			break;
-		case BACKSLASH:
-			if ((c == 'n' || c == 'N') && !current.text.empty ()) {
-				subs.push_back (current);
-				current.text = "";
-				current.vertical_position.line = current.vertical_position.line.get() + 1;
-			}
-			state = TEXT;
-			break;
-		}
+	ASS_Library* library = ass_library_init ();
+	if (!library) {
+		delete[] buffer;
+		throw std::runtime_error ("Could not initialise libass");
 	}
 
-	if (!current.text.empty ()) {
-		subs.push_back (current);
-	}
+	char encoding[] = "UTF-8";
 
-	return subs;
-}
+	ASS_Track* track = ass_read_memory (library, buffer, size, encoding);
+	delete[] buffer;
 
-void
-SSAReader::read (function<optional<string> ()> get_line)
-{
-	enum {
-		INFO,
-		STYLES,
-		EVENTS
-	} part = INFO;
+	for (int i = 0; i < track->n_events; ++i) {
+		sub::RawSubtitle sub;
 
-	map<string, Style> styles;
-	string style_format_line;
-	vector<string> event_format;
+		ASS_Event const & event = track->events[i];
+		ASS_Style const & style = track->styles[event.Style];
 
-	while (true) {
-		optional<string> line = get_line ();
-		if (!line) {
-			break;
-		}
+		sub.text = event.Text;
+		sub.font = style.FontName;
+		sub.font_size.set_proportional (style.FontSize / track->PlayResY);
 
-		trim (*line);
-		remove_unicode_bom (line);
+		/* XXX: effect, effect_colour */
 
-		if (starts_with (*line, ";") || line->empty ()) {
-			continue;
-		}
+		sub.bold = style.Bold;
+		sub.italic = style.Italic;
+		sub.underline = style.Underline;
 
-		if (starts_with (*line, "[")) {
-			/* Section heading */
-			if (line.get() == "[Script Info]") {
-				part = INFO;
-			} else if (line.get() == "[V4 Styles]") {
-				part = STYLES;
-			} else if (line.get() == "[Events]") {
-				part = EVENTS;
-			}
-			continue;
-		}
+		cout << "align " << style.Alignment << "\n";
 
-		size_t const colon = line->find (":");
-		SUB_ASSERT (colon != string::npos);
-		SUB_ASSERT (line->length() > colon + 1);
-		string const type = line->substr (0, colon);
-		string const body = line->substr (colon + 2);
+		int valign = style.Alignment & 12;
 
-		switch (part) {
-		case INFO:
+		switch (valign) {
+		case VALIGN_SUB:
+			sub.vertical_position.reference = BOTTOM_OF_SCREEN;
 			break;
-		case STYLES:
-			if (type == "Format") {
-				style_format_line = body;
-			} else if (type == "Style") {
-				SUB_ASSERT (!style_format_line.empty ());
-				Style s (style_format_line, body);
-				styles[s.name] = s;
-			}
+		case VALIGN_CENTER:
+			sub.vertical_position.reference = CENTRE_OF_SCREEN;
 			break;
-		case EVENTS:
-			if (type == "Format") {
-				split (event_format, body, is_any_of (","));
-				BOOST_FOREACH (string& i, event_format) {
-					trim (i);
-				}
-			} else if (type == "Dialogue") {
-				SUB_ASSERT (!event_format.empty ());
-				vector<string> event;
-				split (event, body, is_any_of (","));
-
-				/* There may be commas in the subtitle part; reassemble any extra parts
-				   from when we just split it.
-				*/
-				while (event.size() > event_format.size()) {
-					string const ex = event.back ();
-					event.pop_back ();
-					event.back() += "," + ex;
-				}
-
-				SUB_ASSERT (!event.empty());
-				SUB_ASSERT (event_format.size() == event.size());
-
-				RawSubtitle sub;
-
-				for (size_t i = 0; i < event.size(); ++i) {
-					trim (event[i]);
-					if (event_format[i] == "Start") {
-						sub.from = parse_time (event[i]);
-					} else if (event_format[i] == "End") {
-						sub.to = parse_time (event[i]);
-					} else if (event_format[i] == "Style") {
-						SUB_ASSERT (styles.find(event[i]) != styles.end());
-						Style style = styles[event[i]];
-						sub.font = style.font_name;
-						sub.font_size = FontSize::from_points (style.font_size);
-						sub.colour = style.primary_colour;
-						sub.effect_colour = style.back_colour;
-						sub.bold = style.bold;
-						sub.italic = style.italic;
-						sub.effect = style.effect;
+		case VALIGN_TOP:
+			sub.vertical_position.reference = TOP_OF_SCREEN;
+			break;
+		}
 
-						/* XXX: arbitrary */
-						sub.vertical_position.lines = 32;
-						sub.vertical_position.reference = TOP_OF_SUBTITLE;
-						sub.vertical_position.line = 0;
+		sub.vertical_position.proportional = style.MarginV / track->PlayResY;
 
-					} else if (event_format[i] == "Text") {
-						BOOST_FOREACH (sub::RawSubtitle j, parse_line (sub, event[i])) {
-							_subs.push_back (j);
-						}
-					}
-				}
-			}
-		}
+		sub.from = Time::from_milliseconds (event.Start);
+		sub.to = sub.from + Time::from_milliseconds (event.Duration);
 
+		_subs.push_back (sub);
 	}
+
+	ass_free_track (track);
+	ass_library_done (library);
 }
diff --git a/src/ssa_reader.h b/src/ssa_reader.h
index 1fe64f3..b80aec4 100644
--- a/src/ssa_reader.h
+++ b/src/ssa_reader.h
@@ -35,11 +35,8 @@ public:
 	SSAReader (FILE* f);
 	SSAReader (std::string const & subs);
 
-	static std::list<RawSubtitle> parse_line (RawSubtitle base, std::string line);
-
 private:
-	void read (boost::function<boost::optional<std::string> ()> get_line);
-	Time parse_time (std::string t) const;
+	void read (char* buffer, int size);
 };
 
 }
diff --git a/src/sub_time.cc b/src/sub_time.cc
index c4e00ba..1c7d294 100644
--- a/src/sub_time.cc
+++ b/src/sub_time.cc
@@ -155,6 +155,12 @@ Time::from_hms (int h, int m, int s, int ms)
 	return Time (h * 3600 + m * 60 + s, ms, Rational (1000, 1));
 }
 
+Time
+Time::from_milliseconds (int64_t m)
+{
+	return Time (m / 1000, m % 1000, Rational (1000, 1));
+}
+
 /** Create a Time from a number of frames.
  *  rate must be integer.
  */
@@ -172,11 +178,9 @@ Time::all_as_seconds () const
 	return _seconds + double(milliseconds ()) / 1000;
 }
 
-/** Add a time to this one.  This time must have an integer _rate
- *  and t must have the same rate.
- */
-void
-Time::add (Time t)
+/** This time must have an integer _rate and t must have the same rate */
+Time &
+Time::operator+= (Time const & t)
 {
 	SUB_ASSERT (_rate);
 
@@ -190,6 +194,17 @@ Time::add (Time t)
 		_frames -= _rate.get().integer_fraction();
 		++_seconds;
 	}
+
+	return *this;
+}
+
+/** This time must have an integer _rate and t must have the same rate */
+Time
+Time::operator+ (Time const & t)
+{
+	Time r = *this;
+	r += t;
+	return r;
 }
 
 void
diff --git a/src/sub_time.h b/src/sub_time.h
index 93088cb..f5d15d7 100644
--- a/src/sub_time.h
+++ b/src/sub_time.h
@@ -42,13 +42,16 @@ public:
 
 	double all_as_seconds () const;
 
-	void add (Time t);
 	void scale (float f);
 
 	static Time from_hmsf (int h, int m, int s, int f, boost::optional<Rational> rate = boost::optional<Rational> ());
 	static Time from_hms (int h, int m, int s, int ms);
+	static Time from_milliseconds (int64_t ms);
 	static Time from_frames (int frames, Rational rate);
 
+	Time operator+ (Time const & other);
+	Time& operator+= (Time const & other);
+
 private:
 	friend bool operator< (Time const & a, Time const & b);
 	friend bool operator> (Time const & a, Time const & b);
diff --git a/src/wscript b/src/wscript
index 9136ea1..e493a07 100644
--- a/src/wscript
+++ b/src/wscript
@@ -8,7 +8,7 @@ def build(bld):
 
     obj.name = 'libsub%s' % bld.env.API_VERSION
     obj.target = 'sub%s' % bld.env.API_VERSION
-    obj.uselib = 'CXML DCP BOOST_FILESYSTEM BOOST_LOCALE BOOST_REGEX ASDCPLIB_CTH'
+    obj.uselib = 'CXML DCP BOOST_FILESYSTEM BOOST_LOCALE BOOST_REGEX ASDCPLIB_CTH LIBASS'
     obj.use = 'libkumu-libsub%s libasdcp-libsub%s' % (bld.env.API_VERSION, bld.env.API_VERSION)
     obj.export_includes = ['.']
     obj.source = """
diff --git a/test/data/test.ssa b/test/data/test.ssa
new file mode 100644
index 0000000..578c3d4
--- /dev/null
+++ b/test/data/test.ssa
@@ -0,0 +1,20 @@
+[Script Info]
+Title: libsub test
+ScriptType: v4.00
+WrapStyle: 0
+ScaledBorderAndShadow: yes
+PlayResX: 1920
+PlayResY: 1080
+
+[V4 Styles]
+Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, TertiaryColour, BackColour, Bold, Italic, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, AlphaLevel, Encoding
+Style: Default,Arial,20,16777215,255,0,0,0,0,1,2,2,2,10,10,10,0,1
+
+[Fonts]
+
+[Graphics]
+
+[Events]
+Format: Marked, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
+Dialogue: Marked=0,0:00:01.23,0:00:04.55,Default,,0,0,0,,Hello world
+Dialogue: Marked=0,0:00:05.74,0:00:11.00,Default,,0,0,900,,This is vertically moved
diff --git a/test/ssa_reader_test.cc b/test/ssa_reader_test.cc
index 76043a7..9a6dd56 100644
--- a/test/ssa_reader_test.cc
+++ b/test/ssa_reader_test.cc
@@ -42,11 +42,13 @@ BOOST_AUTO_TEST_CASE (ssa_reader_test)
 	BOOST_CHECK_EQUAL (i->to, sub::Time::from_hms (0, 2, 41, 790));
 	list<sub::Line>::iterator j = i->lines.begin();
 	BOOST_REQUIRE (j != i->lines.end ());
+	BOOST_CHECK_EQUAL (j->vertical_position.proportional.get(), 0);
+	BOOST_CHECK_EQUAL (j->vertical_position.reference.get(), sub::BOTTOM_OF_SCREEN);
 	BOOST_REQUIRE_EQUAL (j->blocks.size(), 1);
 	sub::Block b = j->blocks.front ();
 	BOOST_CHECK_EQUAL (b.text, "Et les enregistrements de ses ondes delta ?");
 	BOOST_CHECK_EQUAL (b.font.get(), "Wolf_Rain");
-	BOOST_CHECK_EQUAL (b.font_size.points().get(), 56);
+	BOOST_CHECK_EQUAL (b.font_size.proportional().get(), 56.0 / 1024);
 	BOOST_CHECK_EQUAL (b.bold, false);
 	BOOST_CHECK_EQUAL (b.italic, false);
 	BOOST_CHECK_EQUAL (b.underline, false);
@@ -61,7 +63,7 @@ BOOST_AUTO_TEST_CASE (ssa_reader_test)
 	b = j->blocks.front ();
 	BOOST_CHECK_EQUAL (b.text, "Toujours rien.");
 	BOOST_CHECK_EQUAL (b.font.get(), "Wolf_Rain");
-	BOOST_CHECK_EQUAL (b.font_size.points().get(), 56);
+	BOOST_CHECK_EQUAL (b.font_size.proportional().get(), 56.0 / 1024);
 	BOOST_CHECK_EQUAL (b.bold, false);
 	BOOST_CHECK_EQUAL (b.italic, false);
 	BOOST_CHECK_EQUAL (b.underline, false);
@@ -70,44 +72,6 @@ BOOST_AUTO_TEST_CASE (ssa_reader_test)
 	BOOST_CHECK (i == subs.end());
 }
 
-BOOST_AUTO_TEST_CASE (ssa_reader_line_test1)
-{
-	sub::RawSubtitle base;
-	list<sub::RawSubtitle> r = sub::SSAReader::parse_line (base, "This is a line with some {i1}italics{i0} and then\\nthere is a new line.");
-
-	list<sub::RawSubtitle>::const_iterator i = r.begin ();
-	BOOST_CHECK_EQUAL (i->text, "This is a line with some ");
-	BOOST_CHECK_EQUAL (i->italic, false);
-	++i;
-	BOOST_REQUIRE (i != r.end ());
-
-	BOOST_CHECK_EQUAL (i->text, "italics");
-	BOOST_CHECK_EQUAL (i->italic, true);
-	++i;
-	BOOST_REQUIRE (i != r.end ());
-
-	BOOST_CHECK_EQUAL (i->text, " and then");
-	BOOST_CHECK_EQUAL (i->italic, false);
-	++i;
-	BOOST_REQUIRE (i != r.end ());
-
-	BOOST_CHECK_EQUAL (i->text, "there is a new line.");
-	++i;
-	BOOST_REQUIRE (i == r.end ());
-}
-
-BOOST_AUTO_TEST_CASE (ssa_reader_line_test2)
-{
-	sub::RawSubtitle base;
-	list<sub::RawSubtitle> r = sub::SSAReader::parse_line (base, "{i1}It's all just italics{i0}");
-
-	list<sub::RawSubtitle>::const_iterator i = r.begin ();
-	BOOST_CHECK_EQUAL (i->text, "It's all just italics");
-	BOOST_CHECK_EQUAL (i->italic, true);
-	++i;
-	BOOST_REQUIRE (i == r.end ());
-}
-
 static void
 test (boost::filesystem::path p)
 {
@@ -123,3 +87,34 @@ BOOST_AUTO_TEST_CASE (ssa_reader_test2)
 {
 	test ("DKH_UT_EN20160601def.ssa");
 }
+
+/** Test a file with some bits to exercise the parser */
+BOOST_AUTO_TEST_CASE (ssa_reader_test3)
+{
+	boost::filesystem::path p = "test/data/test.ssa";
+	FILE* f = fopen (p.string().c_str(), "r");
+	sub::SSAReader reader (f);
+	fclose (f);
+	list<sub::Subtitle> subs = sub::collect<std::list<sub::Subtitle> > (reader.subtitles ());
+
+	BOOST_REQUIRE_EQUAL (subs.size(), 2);
+
+	list<sub::Subtitle>::const_iterator i = subs.begin ();
+	BOOST_CHECK_EQUAL (i->from, sub::Time::from_hms (0, 0, 1, 230));
+	BOOST_CHECK_EQUAL (i->to, sub::Time::from_hms (0, 0, 4, 550));
+	BOOST_REQUIRE_EQUAL (i->lines.size(), 1);
+	list<sub::Line>::const_iterator j = i->lines.begin();
+	BOOST_REQUIRE_EQUAL (j->blocks.size(), 1);
+	sub::Block b = j->blocks.front ();
+	BOOST_CHECK_EQUAL (b.text, "Hello world");
+	++i;
+
+	BOOST_CHECK_EQUAL (i->from, sub::Time::from_hms (0, 0, 5, 740));
+	BOOST_CHECK_EQUAL (i->to, sub::Time::from_hms (0, 0, 11, 0));
+	BOOST_REQUIRE_EQUAL (i->lines.size(), 1);
+	j = i->lines.begin();
+	BOOST_REQUIRE_EQUAL (j->blocks.size(), 1);
+	b = j->blocks.front ();
+	BOOST_CHECK_EQUAL (b.text, "This is vertically moved");
+	++i;
+}
diff --git a/wscript b/wscript
index 96d5da7..4fc9e37 100644
--- a/wscript
+++ b/wscript
@@ -18,6 +18,9 @@ def configure(conf):
     conf.env.append_value('CXXFLAGS', ['-Wall', '-Wextra', '-D_FILE_OFFSET_BITS=64', '-D__STDC_FORMAT_MACROS'])
     conf.env.append_value('CXXFLAGS', ['-DLIBSUB_VERSION="%s"' % VERSION])
 
+    # Disable libxml++ deprecation warnings for now
+    conf.env.append_value('CXXFLAGS', ['-Wno-deprecated-declarations'])
+
     conf.env.ENABLE_DEBUG = conf.options.enable_debug
     conf.env.STATIC = conf.options.static
     conf.env.TARGET_WINDOWS = conf.options.target_windows
@@ -30,6 +33,7 @@ def configure(conf):
         conf.env.append_value('CXXFLAGS', '-O3')
 
     conf.check_cfg(package='openssl', args='--cflags --libs', uselib_store='OPENSSL', mandatory=True)
+    conf.check_cfg(package='libass', args='--cflags --libs', uselib_store='LIBASS', mandatory=True)
 
     if conf.options.static:
         conf.env.HAVE_CXML = 1