src/ssa_reader.cc

   1 /*
   2     Copyright (C) 2016 Carl Hetherington <cth@carlh.net>
   3
   4     This program is free software; you can redistribute it and/or modify
   5     it under the terms of the GNU General Public License as published by
   6     the Free Software Foundation; either version 2 of the License, or
   7     (at your option) any later version.
   8
   9     This program is distributed in the hope that it will be useful,
  10     but WITHOUT ANY WARRANTY; without even the implied warranty of
  11     MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12     GNU General Public License for more details.
  13
  14     You should have received a copy of the GNU General Public License
  15     along with this program; if not, write to the Free Software
  16     Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  17
  18 */
  19
  20 #include "ssa_reader.h"
  21 #include "util.h"
  22 #include "sub_assert.h"
  23 #include "raw_convert.h"
  24 #include "subtitle.h"
  25 #include <boost/algorithm/string.hpp>
  26 #include <boost/bind.hpp>
  27 #include <boost/foreach.hpp>
  28 #include <sstream>
  29 #include <iostream>
  30 #include <vector>
  31
  32 using std::string;
  33 using std::stringstream;
  34 using std::vector;
  35 using std::map;
  36 using std::cout;
  37 using std::list;
  38 using boost::optional;
  39 using boost::function;
  40 using namespace boost::algorithm;
  41 using namespace sub;
  42
  43 /** @param s Subtitle string encoded in UTF-8 */
  44 SSAReader::SSAReader (string const & s)
  45 {
  46         stringstream str (s);
  47         this->read (boost::bind (&get_line_stringstream, &str));
  48 }
  49
  50 /** @param f Subtitle file encoded in UTF-8 */
  51 SSAReader::SSAReader (FILE* f)
  52 {
  53         this->read (boost::bind (&get_line_file, f));
  54 }
  55
  56 class Style
  57 {
  58 public:
  59         Style ()
  60                 : font_size (24)
  61                 , primary_colour (255, 255, 255)
  62                 , bold (false)
  63                 , italic (false)
  64         {}
  65
  66         Style (string format_line, string style_line)
  67                 : font_size (24)
  68                 , primary_colour (255, 255, 255)
  69                 , bold (false)
  70                 , italic (false)
  71         {
  72                 vector<string> keys;
  73                 split (keys, format_line, is_any_of (","));
  74                 vector<string> style;
  75                 split (style, style_line, is_any_of (","));
  76
  77                 SUB_ASSERT (!keys.empty());
  78                 SUB_ASSERT (!style.empty());
  79                 SUB_ASSERT (keys.size() == style.size());
  80
  81                 for (size_t i = 0; i < style.size(); ++i) {
  82                         trim (keys[i]);
  83                         trim (style[i]);
  84                         if (keys[i] == "Name") {
  85                                 name = style[i];
  86                         } else if (keys[i] == "Fontname") {
  87                                 font_name = style[i];
  88                         } else if (keys[i] == "Fontsize") {
  89                                 font_size = raw_convert<int> (style[i]);
  90                         } else if (keys[i] == "PrimaryColour") {
  91                                 primary_colour = colour (raw_convert<int> (style[i]));
  92                         } else if (keys[i] == "BackColour") {
  93                                 back_colour = colour (raw_convert<int> (style[i]));
  94                         } else if (keys[i] == "Bold") {
  95                                 bold = style[i] == "-1";
  96                         } else if (keys[i] == "Italic") {
  97                                 italic = style[i] == "-1";
  98                         } else if (keys[i] == "BorderStyle") {
  99                                 if (style[i] == "1") {
 100                                         effect = SHADOW;
 101                                 }
 102                         }
 103                 }
 104         }
 105
 106         string name;
 107         optional<string> font_name;
 108         int font_size;
 109         Colour primary_colour;
 110         /** outline colour */
 111         optional<Colour> back_colour;
 112         bool bold;
 113         bool italic;
 114         optional<Effect> effect;
 115
 116 private:
 117         Colour colour (int c) const
 118         {
 119                 return Colour (
 120                         ((c & 0x0000ff) >>  0) / 255.0,
 121                         ((c & 0x00ff00) >>  8) / 255.0,
 122                         ((c & 0xff0000) >> 16) / 255.0
 123                         );
 124         }
 125 };
 126
 127 Time
 128 SSAReader::parse_time (string t) const
 129 {
 130         vector<string> bits;
 131         split (bits, t, is_any_of (":."));
 132         SUB_ASSERT (bits.size() == 4);
 133         return Time::from_hms (
 134                 raw_convert<int> (bits[0]),
 135                 raw_convert<int> (bits[1]),
 136                 raw_convert<int> (bits[2]),
 137                 raw_convert<int> (bits[3]) * 10
 138                 );
 139 }
 140
 141 /** @param base RawSubtitle filled in with any required common values.
 142  *  @param line SSA line string.
 143  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 144  */
 145 list<RawSubtitle>
 146 SSAReader::parse_line (RawSubtitle base, string line)
 147 {
 148         enum {
 149                 TEXT,
 150                 STYLE,
 151                 BACKSLASH
 152         } state = TEXT;
 153
 154         list<RawSubtitle> subs;
 155         RawSubtitle current = base;
 156         string style;
 157
 158         current.vertical_position.line = 0;
 159         /* XXX: arbitrary */
 160         current.vertical_position.lines = 32;
 161         current.vertical_position.reference = TOP_OF_SUBTITLE;
 162
 163         for (size_t i = 0; i < line.length(); ++i) {
 164                 char const c = line[i];
 165                 switch (state) {
 166                 case TEXT:
 167                         if (c == '{') {
 168                                 state = STYLE;
 169                         } else if (c == '\\') {
 170                                 state = BACKSLASH;
 171                         } else if (c != '\r' && c != '\n') {
 172                                 current.text += c;
 173                         }
 174                         break;
 175                 case STYLE:
 176                         if (c == '}') {
 177                                 if (!current.text.empty ()) {
 178                                         subs.push_back (current);
 179                                         current.text = "";
 180                                 }
 181                                 if (style == "i1") {
 182                                         current.italic = true;
 183                                 } else if (style == "i0") {
 184                                         current.italic = false;
 185                                 }
 186                                 style = "";
 187                                 state = TEXT;
 188                         } else {
 189                                 style += c;
 190                         }
 191                         break;
 192                 case BACKSLASH:
 193                         if ((c == 'n' || c == 'N') && !current.text.empty ()) {
 194                                 subs.push_back (current);
 195                                 current.text = "";
 196                                 current.vertical_position.line = current.vertical_position.line.get() + 1;
 197                         }
 198                         state = TEXT;
 199                         break;
 200                 }
 201         }
 202
 203         if (!current.text.empty ()) {
 204                 subs.push_back (current);
 205         }
 206
 207         return subs;
 208 }
 209
 210 void
 211 SSAReader::read (function<optional<string> ()> get_line)
 212 {
 213         enum {
 214                 INFO,
 215                 STYLES,
 216                 EVENTS
 217         } part = INFO;
 218
 219         map<string, Style> styles;
 220         string style_format_line;
 221         vector<string> event_format;
 222
 223         while (true) {
 224                 optional<string> line = get_line ();
 225                 if (!line) {
 226                         break;
 227                 }
 228
 229                 trim (*line);
 230
 231                 if (starts_with (*line, ";") || line->empty ()) {
 232                         continue;
 233                 }
 234
 235                 if (starts_with (*line, "[")) {
 236                         /* Section heading */
 237                         if (line.get() == "[Script Info]") {
 238                                 part = INFO;
 239                         } else if (line.get() == "[V4 Styles]") {
 240                                 part = STYLES;
 241                         } else if (line.get() == "[Events]") {
 242                                 part = EVENTS;
 243                         }
 244                         continue;
 245                 }
 246
 247                 size_t const colon = line->find (":");
 248                 SUB_ASSERT (colon != string::npos);
 249                 SUB_ASSERT (line->length() > colon + 1);
 250                 string const type = line->substr (0, colon);
 251                 string const body = line->substr (colon + 2);
 252
 253                 switch (part) {
 254                 case INFO:
 255                         break;
 256                 case STYLES:
 257                         if (type == "Format") {
 258                                 style_format_line = body;
 259                         } else if (type == "Style") {
 260                                 SUB_ASSERT (!style_format_line.empty ());
 261                                 Style s (style_format_line, body);
 262                                 styles[s.name] = s;
 263                         }
 264                         break;
 265                 case EVENTS:
 266                         if (type == "Format") {
 267                                 split (event_format, body, is_any_of (","));
 268                                 BOOST_FOREACH (string& i, event_format) {
 269                                         trim (i);
 270                                 }
 271                         } else if (type == "Dialogue") {
 272                                 SUB_ASSERT (!event_format.empty ());
 273                                 vector<string> event;
 274                                 split (event, body, is_any_of (","));
 275
 276                                 SUB_ASSERT (!event.empty());
 277                                 SUB_ASSERT (event_format.size() == event.size());
 278
 279                                 RawSubtitle sub;
 280
 281                                 for (size_t i = 0; i < event.size(); ++i) {
 282                                         trim (event[i]);
 283                                         if (event_format[i] == "Start") {
 284                                                 sub.from = parse_time (event[i]);
 285                                         } else if (event_format[i] == "End") {
 286                                                 sub.to = parse_time (event[i]);
 287                                         } else if (event_format[i] == "Style") {
 288                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 289                                                 Style style = styles[event[i]];
 290                                                 sub.font = style.font_name;
 291                                                 sub.font_size = FontSize::from_points (style.font_size);
 292                                                 sub.colour = style.primary_colour;
 293                                                 sub.effect_colour = style.back_colour;
 294                                                 sub.bold = style.bold;
 295                                                 sub.italic = style.italic;
 296                                                 sub.effect = style.effect;
 297
 298                                                 /* XXX: arbitrary */
 299                                                 sub.vertical_position.lines = 32;
 300                                                 sub.vertical_position.reference = TOP_OF_SUBTITLE;
 301                                                 sub.vertical_position.line = 0;
 302
 303                                         } else if (event_format[i] == "Text") {
 304                                                 BOOST_FOREACH (sub::RawSubtitle j, parse_line (sub, event[i])) {
 305                                                         _subs.push_back (j);
 306                                                 }
 307                                         }
 308                                 }
 309                         }
 310                 }
 311
 312         }
 313 }