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 <locked_sstream.h>
  26 #include <boost/algorithm/string.hpp>
  27 #include <boost/bind.hpp>
  28 #include <boost/foreach.hpp>
  29 #include <iostream>
  30 #include <vector>
  31
  32 using std::string;
  33 using std::vector;
  34 using std::map;
  35 using std::cout;
  36 using std::list;
  37 using boost::optional;
  38 using boost::function;
  39 using namespace boost::algorithm;
  40 using namespace sub;
  41
  42 /** @param s Subtitle string encoded in UTF-8 */
  43 SSAReader::SSAReader (string const & s)
  44 {
  45         locked_stringstream str (s);
  46         this->read (boost::bind (&get_line_stringstream, &str));
  47 }
  48
  49 /** @param f Subtitle file encoded in UTF-8 */
  50 SSAReader::SSAReader (FILE* f)
  51 {
  52         this->read (boost::bind (&get_line_file, f));
  53 }
  54
  55 class Style
  56 {
  57 public:
  58         Style ()
  59                 : font_size (72)
  60                 , primary_colour (255, 255, 255)
  61                 , bold (false)
  62                 , italic (false)
  63                 , underline (false)
  64                 , vertical_reference (BOTTOM_OF_SCREEN)
  65                 , vertical_margin (0)
  66         {}
  67
  68         Style (string format_line, string style_line)
  69                 : font_size (72)
  70                 , primary_colour (255, 255, 255)
  71                 , bold (false)
  72                 , italic (false)
  73                 , underline (false)
  74                 , vertical_reference (BOTTOM_OF_SCREEN)
  75                 , vertical_margin (0)
  76         {
  77                 vector<string> keys;
  78                 split (keys, format_line, is_any_of (","));
  79                 vector<string> style;
  80                 split (style, style_line, is_any_of (","));
  81
  82                 SUB_ASSERT (!keys.empty());
  83                 SUB_ASSERT (!style.empty());
  84                 SUB_ASSERT (keys.size() == style.size());
  85
  86                 for (size_t i = 0; i < style.size(); ++i) {
  87                         trim (keys[i]);
  88                         trim (style[i]);
  89                         if (keys[i] == "Name") {
  90                                 name = style[i];
  91                         } else if (keys[i] == "Fontname") {
  92                                 font_name = style[i];
  93                         } else if (keys[i] == "Fontsize") {
  94                                 font_size = raw_convert<int> (style[i]);
  95                         } else if (keys[i] == "PrimaryColour") {
  96                                 primary_colour = colour (raw_convert<int> (style[i]));
  97                         } else if (keys[i] == "BackColour") {
  98                                 back_colour = colour (raw_convert<int> (style[i]));
  99                         } else if (keys[i] == "Bold") {
 100                                 bold = style[i] == "-1";
 101                         } else if (keys[i] == "Italic") {
 102                                 italic = style[i] == "-1";
 103                         } else if (keys[i] == "Underline") {
 104                                 underline = style[i] == "-1";
 105                         } else if (keys[i] == "BorderStyle") {
 106                                 if (style[i] == "1") {
 107                                         effect = SHADOW;
 108                                 }
 109                         } else if (keys[i] == "Alignment") {
 110                                 /* These values from libass' source code */
 111                                 switch (raw_convert<int> (style[i]) & 12) {
 112                                 case 4:
 113                                         vertical_reference = TOP_OF_SCREEN;
 114                                         break;
 115                                 case 8:
 116                                         vertical_reference = CENTRE_OF_SCREEN;
 117                                         break;
 118                                 case 0:
 119                                         vertical_reference = BOTTOM_OF_SCREEN;
 120                                         break;
 121                                 }
 122                         } else if (keys[i] == "MarginV") {
 123                                 vertical_margin = raw_convert<int> (style[i]);
 124                         }
 125                 }
 126         }
 127
 128         string name;
 129         optional<string> font_name;
 130         int font_size;
 131         Colour primary_colour;
 132         /** outline colour */
 133         optional<Colour> back_colour;
 134         bool bold;
 135         bool italic;
 136         bool underline;
 137         optional<Effect> effect;
 138         VerticalReference vertical_reference;
 139         int vertical_margin;
 140
 141 private:
 142         Colour colour (int c) const
 143         {
 144                 return Colour (
 145                         ((c & 0x0000ff) >>  0) / 255.0,
 146                         ((c & 0x00ff00) >>  8) / 255.0,
 147                         ((c & 0xff0000) >> 16) / 255.0
 148                         );
 149         }
 150 };
 151
 152 Time
 153 SSAReader::parse_time (string t) const
 154 {
 155         vector<string> bits;
 156         split (bits, t, is_any_of (":."));
 157         SUB_ASSERT (bits.size() == 4);
 158         return Time::from_hms (
 159                 raw_convert<int> (bits[0]),
 160                 raw_convert<int> (bits[1]),
 161                 raw_convert<int> (bits[2]),
 162                 raw_convert<int> (bits[3]) * 10
 163                 );
 164 }
 165
 166 /** @param base RawSubtitle filled in with any required common values.
 167  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 168  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 169  */
 170 list<RawSubtitle>
 171 SSAReader::parse_line (RawSubtitle base, string line)
 172 {
 173         enum {
 174                 TEXT,
 175                 STYLE,
 176                 BACKSLASH
 177         } state = TEXT;
 178
 179         list<RawSubtitle> subs;
 180         RawSubtitle current = base;
 181         string style;
 182
 183         if (!current.vertical_position.reference) {
 184                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 185         }
 186
 187         if (!current.vertical_position.proportional) {
 188                 current.vertical_position.proportional = 0;
 189         }
 190
 191         /* We must have a font size, as there could be a margin specified
 192            in pixels and in that case we must know how big the subtitle
 193            lines are to work out the position on screen.
 194         */
 195         if (!current.font_size.points()) {
 196                 current.font_size.set_points (72);
 197         }
 198
 199         /* Count the number of line breaks */
 200         int line_breaks = 0;
 201         for (size_t i = 0; i < line.length() - 1; ++i) {
 202                 if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 203                         ++line_breaks;
 204                 }
 205         }
 206
 207         /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
 208         double const line_size = current.font_size.proportional(792) * 1.2;
 209
 210         /* Tweak vertical_position accordingly */
 211         switch (current.vertical_position.reference.get()) {
 212         case TOP_OF_SCREEN:
 213         case TOP_OF_SUBTITLE:
 214                 /* Nothing to do */
 215                 break;
 216         case CENTRE_OF_SCREEN:
 217                 current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 218                 break;
 219         case BOTTOM_OF_SCREEN:
 220                 current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
 221                 break;
 222         }
 223
 224         for (size_t i = 0; i < line.length(); ++i) {
 225                 char const c = line[i];
 226                 switch (state) {
 227                 case TEXT:
 228                         if (c == '{') {
 229                                 state = STYLE;
 230                         } else if (c == '\\') {
 231                                 state = BACKSLASH;
 232                         } else if (c != '\r' && c != '\n') {
 233                                 current.text += c;
 234                         }
 235                         break;
 236                 case STYLE:
 237                         if (c == '}' || c == '\\') {
 238                                 if (!current.text.empty ()) {
 239                                         subs.push_back (current);
 240                                         current.text = "";
 241                                 }
 242                                 if (style == "\\i1") {
 243                                         current.italic = true;
 244                                 } else if (style == "\\i0" || style == "\\i") {
 245                                         current.italic = false;
 246                                 } else if (style == "\\b1") {
 247                                         current.bold = true;
 248                                 } else if (style == "\\b0") {
 249                                         current.bold = false;
 250                                 } else if (style == "\\u1") {
 251                                         current.underline = true;
 252                                 } else if (style == "\\u0") {
 253                                         current.underline = false;
 254                                 } else if (style == "\\an1" || style == "\\an2" || style == "\\an3") {
 255                                         current.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 256                                 } else if (style == "\\an4" || style == "\\an5" || style == "\\an6") {
 257                                         current.vertical_position.reference = sub::CENTRE_OF_SCREEN;
 258                                 } else if (style == "\\an7" || style == "\\an8" || style == "\\an9") {
 259                                         current.vertical_position.reference = sub::TOP_OF_SCREEN;
 260                                 }
 261
 262                                 style = "";
 263                         }
 264
 265                         if (c == '}') {
 266                                 state = TEXT;
 267                         } else {
 268                                 style += c;
 269                         }
 270                         break;
 271                 case BACKSLASH:
 272                         if (c == 'n' || c == 'N') {
 273                                 if (!current.text.empty ()) {
 274                                         subs.push_back (current);
 275                                         current.text = "";
 276                                 }
 277                                 /* Move down one line (1.2 times the font size) */
 278                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 279                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 280                                 } else {
 281                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 282                                 }
 283                         }
 284                         state = TEXT;
 285                         break;
 286                 }
 287         }
 288
 289         if (!current.text.empty ()) {
 290                 subs.push_back (current);
 291         }
 292
 293         return subs;
 294 }
 295
 296 void
 297 SSAReader::read (function<optional<string> ()> get_line)
 298 {
 299         enum {
 300                 INFO,
 301                 STYLES,
 302                 EVENTS
 303         } part = INFO;
 304
 305         int play_res_y = 288;
 306         map<string, Style> styles;
 307         string style_format_line;
 308         vector<string> event_format;
 309
 310         while (true) {
 311                 optional<string> line = get_line ();
 312                 if (!line) {
 313                         break;
 314                 }
 315
 316                 trim (*line);
 317                 remove_unicode_bom (line);
 318
 319                 if (starts_with (*line, ";") || line->empty ()) {
 320                         continue;
 321                 }
 322
 323                 if (starts_with (*line, "[")) {
 324                         /* Section heading */
 325                         if (line.get() == "[Script Info]") {
 326                                 part = INFO;
 327                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 328                                 part = STYLES;
 329                         } else if (line.get() == "[Events]") {
 330                                 part = EVENTS;
 331                         }
 332                         continue;
 333                 }
 334
 335                 size_t const colon = line->find (":");
 336                 SUB_ASSERT (colon != string::npos);
 337                 string const type = line->substr (0, colon);
 338                 string body = line->substr (colon + 1);
 339                 trim (body);
 340
 341                 switch (part) {
 342                 case INFO:
 343                         if (type == "PlayResY") {
 344                                 play_res_y = raw_convert<int> (body);
 345                         }
 346                         break;
 347                 case STYLES:
 348                         if (type == "Format") {
 349                                 style_format_line = body;
 350                         } else if (type == "Style") {
 351                                 SUB_ASSERT (!style_format_line.empty ());
 352                                 Style s (style_format_line, body);
 353                                 styles[s.name] = s;
 354                         }
 355                         break;
 356                 case EVENTS:
 357                         if (type == "Format") {
 358                                 split (event_format, body, is_any_of (","));
 359                                 BOOST_FOREACH (string& i, event_format) {
 360                                         trim (i);
 361                                 }
 362                         } else if (type == "Dialogue") {
 363                                 SUB_ASSERT (!event_format.empty ());
 364                                 vector<string> event;
 365                                 split (event, body, is_any_of (","));
 366
 367                                 /* There may be commas in the subtitle part; reassemble any extra parts
 368                                    from when we just split it.
 369                                 */
 370                                 while (event.size() > event_format.size()) {
 371                                         string const ex = event.back ();
 372                                         event.pop_back ();
 373                                         event.back() += "," + ex;
 374                                 }
 375
 376                                 SUB_ASSERT (!event.empty());
 377                                 SUB_ASSERT (event_format.size() == event.size());
 378
 379                                 RawSubtitle sub;
 380
 381                                 for (size_t i = 0; i < event.size(); ++i) {
 382                                         trim (event[i]);
 383                                         if (event_format[i] == "Start") {
 384                                                 sub.from = parse_time (event[i]);
 385                                         } else if (event_format[i] == "End") {
 386                                                 sub.to = parse_time (event[i]);
 387                                         } else if (event_format[i] == "Style") {
 388                                                 /* libass trims leading '*'s from style names, commenting that
 389                                                    "they seem to mean literally nothing".  Go figure...
 390                                                 */
 391                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 392                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 393                                                 Style style = styles[event[i]];
 394                                                 sub.font = style.font_name;
 395                                                 sub.font_size = FontSize::from_points (style.font_size);
 396                                                 sub.colour = style.primary_colour;
 397                                                 sub.effect_colour = style.back_colour;
 398                                                 sub.bold = style.bold;
 399                                                 sub.italic = style.italic;
 400                                                 sub.underline = style.underline;
 401                                                 sub.effect = style.effect;
 402                                                 sub.vertical_position.reference = style.vertical_reference;
 403                                                 sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 404                                         } else if (event_format[i] == "MarginV") {
 405                                                 sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 406                                         } else if (event_format[i] == "Text") {
 407                                                 BOOST_FOREACH (sub::RawSubtitle j, parse_line (sub, event[i])) {
 408                                                         _subs.push_back (j);
 409                                                 }
 410                                         }
 411                                 }
 412                         }
 413                 }
 414
 415         }
 416 }