Remove use of boost foreach.
[libsub.git] / src / ssa_reader.cc
index acaa46be4b0d5dc714474ffece2e9c09918f84b0..4a9be0e28b0d4300aaa01b191cdfbb141103bc93 100644 (file)
@@ -1,5 +1,5 @@
 /*
-    Copyright (C) 2016 Carl Hetherington <cth@carlh.net>
+    Copyright (C) 2016-2019 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
 #include "sub_assert.h"
 #include "raw_convert.h"
 #include "subtitle.h"
+#include "compose.hpp"
 #include <boost/algorithm/string.hpp>
 #include <boost/bind.hpp>
-#include <boost/foreach.hpp>
-#include <sstream>
 #include <iostream>
 #include <vector>
 
 using std::string;
-using std::stringstream;
 using std::vector;
 using std::map;
 using std::cout;
-using std::list;
 using boost::optional;
 using boost::function;
 using namespace boost::algorithm;
 using namespace sub;
 
 /** @param s Subtitle string encoded in UTF-8 */
-SSAReader::SSAReader (string const & s)
+SSAReader::SSAReader (string s)
 {
-       stringstream str (s);
-       this->read (boost::bind (&get_line_stringstream, &str));
+       this->read (boost::bind(&get_line_string, &s));
 }
 
 /** @param f Subtitle file encoded in UTF-8 */
@@ -53,26 +49,50 @@ SSAReader::SSAReader (FILE* f)
        this->read (boost::bind (&get_line_file, f));
 }
 
+Colour
+h_colour (string s)
+{
+       /* There are both BGR and ABGR versions of these colours */
+       if ((s.length() != 8 && s.length() != 10) || s[0] != '&' || s[1] != 'H') {
+               throw SSAError(String::compose("Badly formatted colour tag %1", s));
+       }
+       int ir, ig, ib;
+       /* XXX: ignoring alpha channel here; note that 00 is opaque and FF is transparent */
+       int const off = s.length() == 10 ? 4 : 2;
+       if (sscanf(s.c_str() + off, "%2x%2x%2x", &ib, &ig, &ir) < 3) {
+               throw SSAError(String::compose("Badly formatted colour tag %1", s));
+       }
+       return sub::Colour(ir / 255.0, ig / 255.0, ib / 255.0);
+}
+
 class Style
 {
 public:
        Style ()
-               : font_size (24)
+               : font_size (72)
                , primary_colour (255, 255, 255)
                , bold (false)
                , italic (false)
+               , underline (false)
+               , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
+               , vertical_reference (BOTTOM_OF_SCREEN)
+               , vertical_margin (0)
        {}
 
        Style (string format_line, string style_line)
-               : font_size (24)
+               : font_size (72)
                , primary_colour (255, 255, 255)
                , bold (false)
                , italic (false)
+               , underline (false)
+               , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
+               , vertical_reference (BOTTOM_OF_SCREEN)
+               , vertical_margin (0)
        {
                vector<string> keys;
-               split (keys, format_line, is_any_of (","));
+               split (keys, format_line, boost::is_any_of (","));
                vector<string> style;
-               split (style, style_line, is_any_of (","));
+               split (style, style_line, boost::is_any_of (","));
 
                SUB_ASSERT (!keys.empty());
                SUB_ASSERT (!style.empty());
@@ -88,17 +108,45 @@ public:
                        } 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]));
+                               primary_colour = colour (style[i]);
                        } else if (keys[i] == "BackColour") {
-                               back_colour = colour (raw_convert<int> (style[i]));
+                               back_colour = colour (style[i]);
                        } else if (keys[i] == "Bold") {
                                bold = style[i] == "-1";
                        } else if (keys[i] == "Italic") {
                                italic = style[i] == "-1";
+                       } else if (keys[i] == "Underline") {
+                               underline = style[i] == "-1";
                        } else if (keys[i] == "BorderStyle") {
                                if (style[i] == "1") {
                                        effect = SHADOW;
                                }
+                       } else if (keys[i] == "Alignment") {
+                               /* These values from libass' source code */
+                               switch ((raw_convert<int> (style[i]) - 1) % 3) {
+                               case 0:
+                                       horizontal_reference = LEFT_OF_SCREEN;
+                                       break;
+                               case 1:
+                                       horizontal_reference = HORIZONTAL_CENTRE_OF_SCREEN;
+                                       break;
+                               case 2:
+                                       horizontal_reference = RIGHT_OF_SCREEN;
+                                       break;
+                               }
+                               switch (raw_convert<int> (style[i]) & 12) {
+                               case 4:
+                                       vertical_reference = TOP_OF_SCREEN;
+                                       break;
+                               case 8:
+                                       vertical_reference = VERTICAL_CENTRE_OF_SCREEN;
+                                       break;
+                               case 0:
+                                       vertical_reference = BOTTOM_OF_SCREEN;
+                                       break;
+                               }
+                       } else if (keys[i] == "MarginV") {
+                               vertical_margin = raw_convert<int> (style[i]);
                        }
                }
        }
@@ -111,16 +159,27 @@ public:
        optional<Colour> back_colour;
        bool bold;
        bool italic;
+       bool underline;
        optional<Effect> effect;
+       HorizontalReference horizontal_reference;
+       VerticalReference vertical_reference;
+       int vertical_margin;
 
 private:
-       Colour colour (int c) const
+       Colour colour (string c) const
        {
-               return Colour (
-                       ((c & 0x0000ff) >>  0) / 255.0,
-                       ((c & 0x00ff00) >>  8) / 255.0,
-                       ((c & 0xff0000) >> 16) / 255.0
-                       );
+               if (c.length() > 0 && c[0] == '&') {
+                       /* &Hbbggrr or &Haabbggrr */
+                       return h_colour (c);
+               } else {
+                       /* integer */
+                       int i = raw_convert<int>(c);
+                       return Colour (
+                               ((i & 0x0000ff) >>  0) / 255.0,
+                               ((i & 0x00ff00) >>  8) / 255.0,
+                               ((i & 0xff0000) >> 16) / 255.0
+                               );
+               }
        }
 };
 
@@ -138,12 +197,74 @@ SSAReader::parse_time (string t) const
                );
 }
 
+void
+SSAReader::parse_style (RawSubtitle& sub, string style, int play_res_x, int play_res_y)
+{
+       if (style == "\\i1") {
+               sub.italic = true;
+       } else if (style == "\\i0" || style == "\\i") {
+               sub.italic = false;
+       } else if (style == "\\b1") {
+               sub.bold = true;
+       } else if (style == "\\b0") {
+               sub.bold = false;
+       } else if (style == "\\u1") {
+               sub.underline = true;
+       } else if (style == "\\u0") {
+               sub.underline = false;
+       } else if (style == "\\an1") {
+               sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
+               sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
+       } else if (style == "\\an2") {
+               sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
+               sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
+       } else if (style == "\\an3") {
+               sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
+               sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
+       } else if (style == "\\an4") {
+               sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
+               sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
+       } else if (style == "\\an5") {
+               sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
+               sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
+       } else if (style == "\\an6") {
+               sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
+               sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
+       } else if (style == "\\an7") {
+               sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
+               sub.vertical_position.reference = sub::TOP_OF_SCREEN;
+       } else if (style == "\\an8") {
+               sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
+               sub.vertical_position.reference = sub::TOP_OF_SCREEN;
+       } else if (style == "\\an9") {
+               sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
+               sub.vertical_position.reference = sub::TOP_OF_SCREEN;
+       } else if (boost::starts_with(style, "\\pos")) {
+               vector<string> bits;
+               boost::algorithm::split (bits, style, boost::is_any_of("(,"));
+               SUB_ASSERT (bits.size() == 3);
+               sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
+               sub.horizontal_position.proportional = raw_convert<float>(bits[1]) / play_res_x;
+               sub.vertical_position.reference = sub::TOP_OF_SCREEN;
+               sub.vertical_position.proportional = raw_convert<float>(bits[2]) / play_res_y;
+       } else if (boost::starts_with(style, "\\fs")) {
+               SUB_ASSERT (style.length() > 3);
+               sub.font_size.set_points (raw_convert<int>(style.substr(3)));
+       } else if (boost::starts_with(style, "\\c")) {
+               /* \c&Hbbggrr& */
+               if (style.length() <= 2) {
+                       throw SSAError(String::compose("Badly formatted colour tag %1", style));
+               }
+               sub.colour = h_colour (style.substr(2, style.length() - 3));
+       }
+}
+
 /** @param base RawSubtitle filled in with any required common values.
- *  @param line SSA line string.
+ *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
  */
-list<RawSubtitle>
-SSAReader::parse_line (RawSubtitle base, string line)
+vector<RawSubtitle>
+SSAReader::parse_line (RawSubtitle base, string line, int play_res_x, int play_res_y)
 {
        enum {
                TEXT,
@@ -151,12 +272,52 @@ SSAReader::parse_line (RawSubtitle base, string line)
                BACKSLASH
        } state = TEXT;
 
-       list<RawSubtitle> subs;
+       vector<RawSubtitle> subs;
        RawSubtitle current = base;
        string style;
 
-       current.vertical_position.line = 0;
-       current.vertical_position.reference = TOP_OF_SUBTITLE;
+       if (!current.vertical_position.reference) {
+               current.vertical_position.reference = BOTTOM_OF_SCREEN;
+       }
+
+       if (!current.vertical_position.proportional) {
+               current.vertical_position.proportional = 0;
+       }
+
+       /* We must have a font size, as there could be a margin specified
+          in pixels and in that case we must know how big the subtitle
+          lines are to work out the position on screen.
+       */
+       if (!current.font_size.points()) {
+               current.font_size.set_points (72);
+       }
+
+       /* Count the number of line breaks */
+       int line_breaks = 0;
+       if (line.length() > 0) {
+               for (size_t i = 0; i < line.length() - 1; ++i) {
+                       if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
+                               ++line_breaks;
+                       }
+               }
+       }
+
+       /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
+       double const line_size = current.font_size.proportional(792) * 1.2;
+
+       /* Tweak vertical_position accordingly */
+       switch (current.vertical_position.reference.get()) {
+       case TOP_OF_SCREEN:
+       case TOP_OF_SUBTITLE:
+               /* Nothing to do */
+               break;
+       case VERTICAL_CENTRE_OF_SCREEN:
+               current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
+               break;
+       case BOTTOM_OF_SCREEN:
+               current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
+               break;
+       }
 
        for (size_t i = 0; i < line.length(); ++i) {
                char const c = line[i];
@@ -166,32 +327,38 @@ SSAReader::parse_line (RawSubtitle base, string line)
                                state = STYLE;
                        } else if (c == '\\') {
                                state = BACKSLASH;
-                       } else {
+                       } else if (c != '\r' && c != '\n') {
                                current.text += c;
                        }
                        break;
                case STYLE:
-                       if (c == '}') {
+                       if (c == '}' || c == '\\') {
                                if (!current.text.empty ()) {
                                        subs.push_back (current);
                                        current.text = "";
                                }
-                               if (style == "i1") {
-                                       current.italic = true;
-                               } else if (style == "i0") {
-                                       current.italic = false;
-                               }
+                               parse_style (current, style, play_res_x, play_res_y);
                                style = "";
+                       }
+
+                       if (c == '}') {
                                state = TEXT;
                        } else {
                                style += c;
                        }
                        break;
                case BACKSLASH:
-                       if (c == 'n' && !current.text.empty ()) {
-                               subs.push_back (current);
-                               current.text = "";
-                               current.vertical_position.line = current.vertical_position.line.get() + 1;
+                       if (c == 'n' || c == 'N') {
+                               if (!current.text.empty ()) {
+                                       subs.push_back (current);
+                                       current.text = "";
+                               }
+                               /* Move down one line (1.2 times the font size) */
+                               if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
+                                       current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
+                               } else {
+                                       current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
+                               }
                        }
                        state = TEXT;
                        break;
@@ -214,6 +381,8 @@ SSAReader::read (function<optional<string> ()> get_line)
                EVENTS
        } part = INFO;
 
+       int play_res_x = 288;
+       int play_res_y = 288;
        map<string, Style> styles;
        string style_format_line;
        vector<string> event_format;
@@ -225,6 +394,7 @@ SSAReader::read (function<optional<string> ()> get_line)
                }
 
                trim (*line);
+               remove_unicode_bom (line);
 
                if (starts_with (*line, ";") || line->empty ()) {
                        continue;
@@ -234,7 +404,7 @@ SSAReader::read (function<optional<string> ()> get_line)
                        /* Section heading */
                        if (line.get() == "[Script Info]") {
                                part = INFO;
-                       } else if (line.get() == "[V4 Styles]") {
+                       } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
                                part = STYLES;
                        } else if (line.get() == "[Events]") {
                                part = EVENTS;
@@ -244,12 +414,17 @@ SSAReader::read (function<optional<string> ()> get_line)
 
                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);
+               string body = line->substr (colon + 1);
+               trim (body);
 
                switch (part) {
                case INFO:
+                       if (type == "PlayResX") {
+                               play_res_x = raw_convert<int> (body);
+                       } else if (type == "PlayResY") {
+                               play_res_y = raw_convert<int> (body);
+                       }
                        break;
                case STYLES:
                        if (type == "Format") {
@@ -263,7 +438,7 @@ SSAReader::read (function<optional<string> ()> get_line)
                case EVENTS:
                        if (type == "Format") {
                                split (event_format, body, is_any_of (","));
-                               BOOST_FOREACH (string& i, event_format) {
+                               for (auto& i: event_format) {
                                        trim (i);
                                }
                        } else if (type == "Dialogue") {
@@ -271,6 +446,15 @@ SSAReader::read (function<optional<string> ()> get_line)
                                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());
 
@@ -283,6 +467,10 @@ SSAReader::read (function<optional<string> ()> get_line)
                                        } else if (event_format[i] == "End") {
                                                sub.to = parse_time (event[i]);
                                        } else if (event_format[i] == "Style") {
+                                               /* libass trims leading '*'s from style names, commenting that
+                                                  "they seem to mean literally nothing".  Go figure...
+                                               */
+                                               trim_left_if (event[i], boost::is_any_of ("*"));
                                                SUB_ASSERT (styles.find(event[i]) != styles.end());
                                                Style style = styles[event[i]];
                                                sub.font = style.font_name;
@@ -291,19 +479,19 @@ SSAReader::read (function<optional<string> ()> get_line)
                                                sub.effect_colour = style.back_colour;
                                                sub.bold = style.bold;
                                                sub.italic = style.italic;
+                                               sub.underline = style.underline;
                                                sub.effect = style.effect;
-
-                                               /* XXX: arbitrary */
-                                               sub.vertical_position.lines = 32;
-                                               sub.vertical_position.reference = TOP_OF_SUBTITLE;
-                                               sub.vertical_position.line = 0;
-
+                                               sub.horizontal_position.reference = style.horizontal_reference;
+                                               sub.vertical_position.reference = style.vertical_reference;
+                                               sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
+                                       } else if (event_format[i] == "MarginV") {
+                                               sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
                                        } else if (event_format[i] == "Text") {
-                                               sub.text = event[i];
+                                               for (auto j: parse_line (sub, event[i], play_res_x, play_res_y)) {
+                                                       _subs.push_back (j);
+                                               }
                                        }
                                }
-
-                               _subs.push_back (sub);
                        }
                }