Add support for reading <font> tags in SubRip.

7 files changed, 99 insertions, 27 deletions

diff --git a/src/colour.cc b/src/colour.cc
index f163424..a47e633 100644
--- a/src/colour.cc
+++ b/src/colour.cc
@@ -25,16 +25,26 @@
 using std::string;
 using namespace sub;
 
-Colour::Colour (string argb_hex)
+Colour
+Colour::from_argb_hex (string argb_hex)
 {
 	int alpha, ir, ig, ib;
 	if (sscanf (argb_hex.c_str(), "%2x%2x%2x%2x", &alpha, &ir, &ig, &ib) < 4) {
 		throw XMLError ("could not parse colour string");
 	}
 
-	r = float (ir) / 255;
-	g = float (ig) / 255;
-	b = float (ib) / 255;
+	return Colour (float (ir) / 255, float (ig) / 255, float (ib) / 255);
+}
+
+Colour
+Colour::from_rgb_hex (string rgb_hex)
+{
+	int ir, ig, ib;
+	if (sscanf (rgb_hex.c_str(), "%2x%2x%2x", &ir, &ig, &ib) < 3) {
+		throw XMLError ("could not parse colour string");
+	}
+
+	return Colour (float (ir) / 255, float (ig) / 255, float (ib) / 255);
 }
 
 bool
diff --git a/src/colour.h b/src/colour.h
index 673bb9f..55d2239 100644
--- a/src/colour.h
+++ b/src/colour.h
@@ -35,14 +35,15 @@ public:
 		, g (0)
 		, b (0)
 	{}
-	
+
 	Colour (float r, float g, float b)
 		: r (r)
 		, g (g)
 		, b (b)
 	{}
 
-	Colour (std::string);
+	static Colour from_argb_hex (std::string);
+	static Colour from_rgb_hex (std::string);
 
 	/** red component (from 0 to 1) */
 	float r;
@@ -54,7 +55,7 @@ public:
 
 bool
 operator== (Colour const & a, Colour const & b);
-	
+
 }
 
 #endif
diff --git a/src/dcp/font.cc b/src/dcp/font.cc
index b2fd128..64787c7 100644
--- a/src/dcp/font.cc
+++ b/src/dcp/font.cc
@@ -35,7 +35,7 @@ dcp::Font::Font (cxml::ConstNodePtr node)
 	italic = node->optional_bool_attribute ("Italic");
 	optional<string> c = node->optional_string_attribute ("Color");
 	if (c) {
-		colour = Colour (c.get ());
+		colour = Colour::from_argb_hex (c.get ());
 	}
 	optional<string> const e = node->optional_string_attribute ("Effect");
 	if (e) {
@@ -43,15 +43,15 @@ dcp::Font::Font (cxml::ConstNodePtr node)
 	}
 	c = node->optional_string_attribute ("EffectColor");
 	if (c) {
-		effect_colour = Colour (c.get ());
+		effect_colour = Colour::from_argb_hex (c.get ());
 	}
 }
 
 dcp::Font::Font (std::list<boost::shared_ptr<Font> > const & font_nodes)
 	: size (0)
 	, italic (false)
-	, colour ("FFFFFFFF")
-	, effect_colour ("FFFFFFFF")
+	, colour (Colour::from_argb_hex ("FFFFFFFF"))
+	, effect_colour (Colour::from_argb_hex ("FFFFFFFF"))
 {
 	for (list<shared_ptr<Font> >::const_iterator i = font_nodes.begin(); i != font_nodes.end(); ++i) {
 		if ((*i)->id) {
diff --git a/src/subrip_reader.cc b/src/subrip_reader.cc
index 3f1c23d..134ca3e 100644
--- a/src/subrip_reader.cc
+++ b/src/subrip_reader.cc
@@ -21,12 +21,17 @@
 #include "exceptions.h"
 #include <boost/algorithm/string.hpp>
 #include <boost/lexical_cast.hpp>
+#include <boost/regex.hpp>
 #include <cstdio>
 #include <vector>
 
 using std::string;
 using std::vector;
+using std::list;
+using std::cout;
+using std::hex;
 using boost::lexical_cast;
+using boost::to_upper;
 using namespace sub;
 
 SubripReader::SubripReader (FILE* f)
@@ -60,7 +65,7 @@ SubripReader::SubripReader (FILE* f)
 			static_cast<unsigned char> (line[1]) == 0xbb &&
 			static_cast<unsigned char> (line[2]) == 0xbf
 			) {
-			
+
 			/* Skip Unicode byte order mark */
 			line = line.substr (3);
 		}
@@ -88,7 +93,7 @@ SubripReader::SubripReader (FILE* f)
 			to = convert_time (p[2]);
 
 			/* XXX: should not ignore coordinate specifications */
-			
+
 			state = CONTENT;
 			break;
 		}
@@ -132,7 +137,7 @@ SubripReader::convert_line (string t, int line_number, Time from, Time to)
 		TEXT,
 		TAG
 	} state = TEXT;
-	
+
 	string tag;
 
 	RawSubtitle p;
@@ -144,7 +149,10 @@ SubripReader::convert_line (string t, int line_number, Time from, Time to)
 	/* XXX: arbitrary */
 	p.vertical_position.lines = 32;
 	p.vertical_position.reference = TOP_OF_SUBTITLE;
-	
+
+	list<Colour> colours;
+	colours.push_back (Colour (1, 1, 1));
+
 	/* XXX: missing <font> support */
 	/* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 	   not work, I think.
@@ -179,6 +187,17 @@ SubripReader::convert_line (string t, int line_number, Time from, Time to)
 				} else if (tag == "/u") {
 					maybe_content (p);
 					p.underline = false;
+				} else if (boost::starts_with (tag, "font")) {
+					maybe_content (p);
+					boost::regex re (".*color=\"#([0123456789abcdef]+)\"");
+					boost::smatch match;
+					if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
+						p.colour = Colour::from_rgb_hex (match[1]);
+						colours.push_back (p.colour);
+					}
+				} else if (tag == "/font") {
+					colours.pop_back ();
+					p.colour = colours.back ();
 				}
 				tag.clear ();
 				state = TEXT;
diff --git a/src/wscript b/src/wscript
index 89c6e89..6afe1d9 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'
+    obj.uselib = 'CXML DCP BOOST_FILESYSTEM BOOST_LOCALE BOOST_REGEX'
     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/subrip_reader_test.cc b/test/subrip_reader_test.cc
index 91437f2..dcd506a 100644
--- a/test/subrip_reader_test.cc
+++ b/test/subrip_reader_test.cc
@@ -1,5 +1,5 @@
 /*
-    Copyright (C) 2014 Carl Hetherington <cth@carlh.net>
+    Copyright (C) 2014-2015 Carl Hetherington <cth@carlh.net>
 
     This program is free software; you can redistribute it and/or modify
     it under the terms of the GNU General Public License as published by
@@ -24,11 +24,13 @@
 #include <boost/test/unit_test.hpp>
 #include <boost/filesystem.hpp>
 #include <fstream>
+#include <cmath>
 
 using std::list;
 using std::cerr;
 using std::ifstream;
 using std::vector;
+using std::fabs;
 
 /* Test reading of a Subrip file */
 BOOST_AUTO_TEST_CASE (subrip_reader_test)
@@ -40,13 +42,13 @@ BOOST_AUTO_TEST_CASE (subrip_reader_test)
 
 	list<sub::Subtitle>::iterator i = subs.begin ();
 
-	
+
 	/* First subtitle */
 
 	BOOST_CHECK (i != subs.end ());
 	BOOST_CHECK_EQUAL (i->from, sub::Time::from_hms (0, 0, 41, 90));
 	BOOST_CHECK_EQUAL (i->to, sub::Time::from_hms (0, 0, 42, 210));
-	
+
 	list<sub::Line>::iterator j = i->lines.begin ();
 	BOOST_CHECK (j != i->lines.end ());
 	BOOST_CHECK_EQUAL (j->blocks.size(), 1);
@@ -74,13 +76,13 @@ BOOST_AUTO_TEST_CASE (subrip_reader_test)
 	BOOST_CHECK_EQUAL (j->vertical_position.reference.get(), sub::TOP_OF_SUBTITLE);
 	++i;
 
-	
+
 	/* Second subtitle */
-	
+
 	BOOST_CHECK (i != subs.end ());
 	BOOST_CHECK_EQUAL (i->from, sub::Time::from_hms (0, 1, 1, 10));
 	BOOST_CHECK_EQUAL (i->to, sub::Time::from_hms (0, 1, 2, 100));
-	
+
 	BOOST_CHECK_EQUAL (i->lines.size(), 1);
 	sub::Line l = i->lines.front ();
 	BOOST_CHECK_EQUAL (l.blocks.size(), 7);
@@ -88,7 +90,7 @@ BOOST_AUTO_TEST_CASE (subrip_reader_test)
 	BOOST_CHECK_EQUAL (l.vertical_position.reference.get(), sub::TOP_OF_SUBTITLE);
 
 	list<sub::Block>::iterator k = l.blocks.begin ();
-	
+
 	BOOST_CHECK (k != l.blocks.end ());
 	BOOST_CHECK_EQUAL (k->text, "This is some ");
 	BOOST_CHECK_EQUAL (k->font.get(), "Arial");
@@ -151,7 +153,7 @@ BOOST_AUTO_TEST_CASE (subrip_reader_test)
 	BOOST_CHECK_EQUAL (k->italic, false);
 	BOOST_CHECK_EQUAL (k->underline, false);
 	++k;
-	
+
 	BOOST_CHECK (k == l.blocks.end ());
 }
 
@@ -217,7 +219,7 @@ BOOST_AUTO_TEST_CASE (subrip_reader_test2)
 BOOST_AUTO_TEST_CASE (subrip_reader_convert_line_test)
 {
 	sub::SubripReader r;
-	
+
 	r.convert_line ("Hello world", 0, sub::Time (), sub::Time ());
 	BOOST_CHECK_EQUAL (r._subs.size(), 1);
 	BOOST_CHECK_EQUAL (r._subs.front().text, "Hello world");
@@ -261,7 +263,7 @@ BOOST_AUTO_TEST_CASE (subrip_reader_convert_line_test)
 
 	r.convert_line ("<b>This is <i>nesting</i> of subtitles</b>", 0, sub::Time (), sub::Time ());
 	BOOST_CHECK_EQUAL (r._subs.size(), 3);
-	list<sub::RawSubtitle>::iterator i = r._subs.begin ();	
+	list<sub::RawSubtitle>::iterator i = r._subs.begin ();
 	BOOST_CHECK_EQUAL (i->text, "This is ");
 	BOOST_CHECK_EQUAL (i->bold, true);
 	BOOST_CHECK_EQUAL (i->italic, false);
@@ -275,8 +277,39 @@ BOOST_AUTO_TEST_CASE (subrip_reader_convert_line_test)
 	BOOST_CHECK_EQUAL (i->italic, false);
 	++i;
 	r._subs.clear ();
-}
 
+	r.convert_line ("<font color=\"#ff0000\">some red text <b>in bold</b></font>", 0, sub::Time (), sub::Time ());
+	BOOST_CHECK_EQUAL (r._subs.size(), 2);
+	i = r._subs.begin ();
+	BOOST_CHECK_EQUAL (i->text, "some red text ");
+	BOOST_CHECK_EQUAL (i->bold, false);
+	BOOST_CHECK_CLOSE (i->colour.r, 1, 0.1);
+	BOOST_CHECK (fabs (i->colour.g) < 0.01);
+	BOOST_CHECK (fabs (i->colour.b) < 0.01);
+	++i;
+	BOOST_CHECK_EQUAL (i->text, "in bold");
+	BOOST_CHECK_EQUAL (i->bold, true);
+	BOOST_CHECK_CLOSE (i->colour.r, 1, 0.1);
+	BOOST_CHECK (fabs (i->colour.g) < 0.01);
+	BOOST_CHECK (fabs (i->colour.b) < 0.01);
+	r._subs.clear ();
+
+	r.convert_line ("<font color=\"#0000ff\">some blue text <b>in bold</b></font>", 0, sub::Time (), sub::Time ());
+	BOOST_CHECK_EQUAL (r._subs.size(), 2);
+	i = r._subs.begin ();
+	BOOST_CHECK_EQUAL (i->text, "some blue text ");
+	BOOST_CHECK_EQUAL (i->bold, false);
+	BOOST_CHECK (fabs (i->colour.r) < 0.01);
+	BOOST_CHECK (fabs (i->colour.g) < 0.01);
+	BOOST_CHECK_CLOSE (i->colour.b, 1, 0.1);
+	++i;
+	BOOST_CHECK_EQUAL (i->text, "in bold");
+	BOOST_CHECK_EQUAL (i->bold, true);
+	BOOST_CHECK (fabs (i->colour.r) < 0.01);
+	BOOST_CHECK (fabs (i->colour.g) < 0.01);
+	BOOST_CHECK_CLOSE (i->colour.b, 1, 0.1);
+	r._subs.clear ();
+}
 
 /** Test SubripReader::convert_time */
 BOOST_AUTO_TEST_CASE (subrip_reader_convert_time_test)
diff --git a/wscript b/wscript
index 2056713..12143d8 100644
--- a/wscript
+++ b/wscript
@@ -72,6 +72,15 @@ def configure(conf):
                    lib=['boost_locale%s' % boost_lib_suffix, 'boost_system%s' % boost_lib_suffix],
                    uselib_store='BOOST_LOCALE')
 
+    conf.check_cxx(fragment="""
+    			    #include <boost/regex.hpp>\n
+    			    int main() { boost::regex re ("foo"); }\n
+			    """,
+                   msg='Checking for boost regex library',
+                   libpath='/usr/local/lib',
+                   lib=['boost_regex%s' % boost_lib_suffix, 'boost_system%s' % boost_lib_suffix],
+                   uselib_store='BOOST_REGEX')
+
     if not conf.env.DISABLE_TESTS:
         conf.recurse('test')
     conf.recurse('asdcplib')