src/ssa_reader.cc

   1 /*
   2     Copyright (C) 2016-2019 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 "compose.hpp"
  26 #include <boost/algorithm/string.hpp>
  27 #include <boost/bind/bind.hpp>
  28 #include <iostream>
  29 #include <vector>
  30
  31 using std::string;
  32 using std::vector;
  33 using std::map;
  34 using std::cout;
  35 using boost::optional;
  36 using boost::function;
  37 using namespace boost::algorithm;
  38 #if BOOST_VERSION >= 106100
  39 using namespace boost::placeholders;
  40 #endif
  41 using namespace sub;
  42
  43 /** @param s Subtitle string encoded in UTF-8 */
  44 SSAReader::SSAReader (string s)
  45 {
  46         this->read (boost::bind(&get_line_string, &s));
  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 Colour
  56 h_colour (string s)
  57 {
  58         /* There are both BGR and ABGR versions of these colours */
  59         if ((s.length() != 8 && s.length() != 10) || s[0] != '&' || s[1] != 'H') {
  60                 throw SSAError(String::compose("Badly formatted colour tag %1", s));
  61         }
  62         int ir, ig, ib;
  63         /* XXX: ignoring alpha channel here; note that 00 is opaque and FF is transparent */
  64         int const off = s.length() == 10 ? 4 : 2;
  65         if (sscanf(s.c_str() + off, "%2x%2x%2x", &ib, &ig, &ir) < 3) {
  66                 throw SSAError(String::compose("Badly formatted colour tag %1", s));
  67         }
  68         return sub::Colour(ir / 255.0, ig / 255.0, ib / 255.0);
  69 }
  70
  71 class Style
  72 {
  73 public:
  74         Style ()
  75                 : font_size (72)
  76                 , primary_colour (255, 255, 255)
  77                 , bold (false)
  78                 , italic (false)
  79                 , underline (false)
  80                 , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
  81                 , vertical_reference (BOTTOM_OF_SCREEN)
  82                 , vertical_margin (0)
  83         {}
  84
  85         Style (string format_line, string style_line)
  86                 : font_size (72)
  87                 , primary_colour (255, 255, 255)
  88                 , bold (false)
  89                 , italic (false)
  90                 , underline (false)
  91                 , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
  92                 , vertical_reference (BOTTOM_OF_SCREEN)
  93                 , vertical_margin (0)
  94         {
  95                 vector<string> keys;
  96                 split (keys, format_line, boost::is_any_of (","));
  97                 vector<string> style;
  98                 split (style, style_line, boost::is_any_of (","));
  99
 100                 SUB_ASSERT (!keys.empty());
 101                 SUB_ASSERT (!style.empty());
 102                 SUB_ASSERT (keys.size() == style.size());
 103
 104                 for (size_t i = 0; i < style.size(); ++i) {
 105                         trim (keys[i]);
 106                         trim (style[i]);
 107                         if (keys[i] == "Name") {
 108                                 name = style[i];
 109                         } else if (keys[i] == "Fontname") {
 110                                 font_name = style[i];
 111                         } else if (keys[i] == "Fontsize") {
 112                                 font_size = raw_convert<int> (style[i]);
 113                         } else if (keys[i] == "PrimaryColour") {
 114                                 primary_colour = colour (style[i]);
 115                         } else if (keys[i] == "BackColour") {
 116                                 back_colour = colour (style[i]);
 117                         } else if (keys[i] == "Bold") {
 118                                 bold = style[i] == "-1";
 119                         } else if (keys[i] == "Italic") {
 120                                 italic = style[i] == "-1";
 121                         } else if (keys[i] == "Underline") {
 122                                 underline = style[i] == "-1";
 123                         } else if (keys[i] == "BorderStyle") {
 124                                 if (style[i] == "1") {
 125                                         effect = SHADOW;
 126                                 }
 127                         } else if (keys[i] == "Alignment") {
 128                                 /* These values from libass' source code */
 129                                 switch ((raw_convert<int> (style[i]) - 1) % 3) {
 130                                 case 0:
 131                                         horizontal_reference = LEFT_OF_SCREEN;
 132                                         break;
 133                                 case 1:
 134                                         horizontal_reference = HORIZONTAL_CENTRE_OF_SCREEN;
 135                                         break;
 136                                 case 2:
 137                                         horizontal_reference = RIGHT_OF_SCREEN;
 138                                         break;
 139                                 }
 140                                 switch (raw_convert<int> (style[i]) & 12) {
 141                                 case 4:
 142                                         vertical_reference = TOP_OF_SCREEN;
 143                                         break;
 144                                 case 8:
 145                                         vertical_reference = VERTICAL_CENTRE_OF_SCREEN;
 146                                         break;
 147                                 case 0:
 148                                         vertical_reference = BOTTOM_OF_SCREEN;
 149                                         break;
 150                                 }
 151                         } else if (keys[i] == "MarginV") {
 152                                 vertical_margin = raw_convert<int> (style[i]);
 153                         }
 154                 }
 155         }
 156
 157         string name;
 158         optional<string> font_name;
 159         int font_size;
 160         Colour primary_colour;
 161         /** outline colour */
 162         optional<Colour> back_colour;
 163         bool bold;
 164         bool italic;
 165         bool underline;
 166         optional<Effect> effect;
 167         HorizontalReference horizontal_reference;
 168         VerticalReference vertical_reference;
 169         int vertical_margin;
 170
 171 private:
 172         Colour colour (string c) const
 173         {
 174                 if (c.length() > 0 && c[0] == '&') {
 175                         /* &Hbbggrr or &Haabbggrr */
 176                         return h_colour (c);
 177                 } else {
 178                         /* integer */
 179                         int i = raw_convert<int>(c);
 180                         return Colour (
 181                                 ((i & 0x0000ff) >>  0) / 255.0,
 182                                 ((i & 0x00ff00) >>  8) / 255.0,
 183                                 ((i & 0xff0000) >> 16) / 255.0
 184                                 );
 185                 }
 186         }
 187 };
 188
 189 Time
 190 SSAReader::parse_time (string t) const
 191 {
 192         vector<string> bits;
 193         split (bits, t, is_any_of (":."));
 194         SUB_ASSERT (bits.size() == 4);
 195         return Time::from_hms (
 196                 raw_convert<int> (bits[0]),
 197                 raw_convert<int> (bits[1]),
 198                 raw_convert<int> (bits[2]),
 199                 raw_convert<int> (bits[3]) * 10
 200                 );
 201 }
 202
 203 void
 204 SSAReader::parse_style (RawSubtitle& sub, string style, int play_res_x, int play_res_y)
 205 {
 206         if (style == "\\i1") {
 207                 sub.italic = true;
 208         } else if (style == "\\i0" || style == "\\i") {
 209                 sub.italic = false;
 210         } else if (style == "\\b1") {
 211                 sub.bold = true;
 212         } else if (style == "\\b0") {
 213                 sub.bold = false;
 214         } else if (style == "\\u1") {
 215                 sub.underline = true;
 216         } else if (style == "\\u0") {
 217                 sub.underline = false;
 218         } else if (style == "\\an1") {
 219                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 220                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 221         } else if (style == "\\an2") {
 222                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 223                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 224         } else if (style == "\\an3") {
 225                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 226                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 227         } else if (style == "\\an4") {
 228                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 229                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 230         } else if (style == "\\an5") {
 231                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 232                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 233         } else if (style == "\\an6") {
 234                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 235                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 236         } else if (style == "\\an7") {
 237                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 238                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 239         } else if (style == "\\an8") {
 240                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 241                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 242         } else if (style == "\\an9") {
 243                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 244                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 245         } else if (boost::starts_with(style, "\\pos")) {
 246                 vector<string> bits;
 247                 boost::algorithm::split (bits, style, boost::is_any_of("(,"));
 248                 SUB_ASSERT (bits.size() == 3);
 249                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 250                 sub.horizontal_position.proportional = raw_convert<float>(bits[1]) / play_res_x;
 251                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 252                 sub.vertical_position.proportional = raw_convert<float>(bits[2]) / play_res_y;
 253         } else if (boost::starts_with(style, "\\fs")) {
 254                 SUB_ASSERT (style.length() > 3);
 255                 sub.font_size.set_points (raw_convert<int>(style.substr(3)));
 256         } else if (boost::starts_with(style, "\\c")) {
 257                 /* \c&Hbbggrr& */
 258                 if (style.length() <= 2) {
 259                         throw SSAError(String::compose("Badly formatted colour tag %1", style));
 260                 }
 261                 sub.colour = h_colour (style.substr(2, style.length() - 3));
 262         }
 263 }
 264
 265 /** @param base RawSubtitle filled in with any required common values.
 266  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 267  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 268  */
 269 vector<RawSubtitle>
 270 SSAReader::parse_line (RawSubtitle base, string line, int play_res_x, int play_res_y)
 271 {
 272         enum {
 273                 TEXT,
 274                 STYLE,
 275                 BACKSLASH
 276         } state = TEXT;
 277
 278         vector<RawSubtitle> subs;
 279         RawSubtitle current = base;
 280         string style;
 281
 282         if (!current.vertical_position.reference) {
 283                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 284         }
 285
 286         if (!current.vertical_position.proportional) {
 287                 current.vertical_position.proportional = 0;
 288         }
 289
 290         /* We must have a font size, as there could be a margin specified
 291            in pixels and in that case we must know how big the subtitle
 292            lines are to work out the position on screen.
 293         */
 294         if (!current.font_size.points()) {
 295                 current.font_size.set_points (72);
 296         }
 297
 298         /* Count the number of line breaks */
 299         int line_breaks = 0;
 300         if (line.length() > 0) {
 301                 for (size_t i = 0; i < line.length() - 1; ++i) {
 302                         if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 303                                 ++line_breaks;
 304                         }
 305                 }
 306         }
 307
 308         /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
 309         double const line_size = current.font_size.proportional(792) * 1.2;
 310
 311         /* Tweak vertical_position accordingly */
 312         switch (current.vertical_position.reference.get()) {
 313         case TOP_OF_SCREEN:
 314         case TOP_OF_SUBTITLE:
 315                 /* Nothing to do */
 316                 break;
 317         case VERTICAL_CENTRE_OF_SCREEN:
 318                 current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 319                 break;
 320         case BOTTOM_OF_SCREEN:
 321                 current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
 322                 break;
 323         }
 324
 325         for (size_t i = 0; i < line.length(); ++i) {
 326                 char const c = line[i];
 327                 switch (state) {
 328                 case TEXT:
 329                         if (c == '{') {
 330                                 state = STYLE;
 331                         } else if (c == '\\') {
 332                                 state = BACKSLASH;
 333                         } else if (c != '\r' && c != '\n') {
 334                                 current.text += c;
 335                         }
 336                         break;
 337                 case STYLE:
 338                         if (c == '}' || c == '\\') {
 339                                 if (!current.text.empty ()) {
 340                                         subs.push_back (current);
 341                                         current.text = "";
 342                                 }
 343                                 parse_style (current, style, play_res_x, play_res_y);
 344                                 style = "";
 345                         }
 346
 347                         if (c == '}') {
 348                                 state = TEXT;
 349                         } else {
 350                                 style += c;
 351                         }
 352                         break;
 353                 case BACKSLASH:
 354                         if (c == 'n' || c == 'N') {
 355                                 if (!current.text.empty ()) {
 356                                         subs.push_back (current);
 357                                         current.text = "";
 358                                 }
 359                                 /* Move down one line (1.2 times the font size) */
 360                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 361                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 362                                 } else {
 363                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 364                                 }
 365                         }
 366                         state = TEXT;
 367                         break;
 368                 }
 369         }
 370
 371         if (!current.text.empty ()) {
 372                 subs.push_back (current);
 373         }
 374
 375         return subs;
 376 }
 377
 378 void
 379 SSAReader::read (function<optional<string> ()> get_line)
 380 {
 381         enum {
 382                 INFO,
 383                 STYLES,
 384                 EVENTS
 385         } part = INFO;
 386
 387         int play_res_x = 288;
 388         int play_res_y = 288;
 389         map<string, Style> styles;
 390         string style_format_line;
 391         vector<string> event_format;
 392
 393         while (true) {
 394                 optional<string> line = get_line ();
 395                 if (!line) {
 396                         break;
 397                 }
 398
 399                 trim (*line);
 400                 remove_unicode_bom (line);
 401
 402                 if (starts_with (*line, ";") || line->empty ()) {
 403                         continue;
 404                 }
 405
 406                 if (starts_with (*line, "[")) {
 407                         /* Section heading */
 408                         if (line.get() == "[Script Info]") {
 409                                 part = INFO;
 410                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 411                                 part = STYLES;
 412                         } else if (line.get() == "[Events]") {
 413                                 part = EVENTS;
 414                         }
 415                         continue;
 416                 }
 417
 418                 size_t const colon = line->find (":");
 419                 SUB_ASSERT (colon != string::npos);
 420                 string const type = line->substr (0, colon);
 421                 string body = line->substr (colon + 1);
 422                 trim (body);
 423
 424                 switch (part) {
 425                 case INFO:
 426                         if (type == "PlayResX") {
 427                                 play_res_x = raw_convert<int> (body);
 428                         } else if (type == "PlayResY") {
 429                                 play_res_y = raw_convert<int> (body);
 430                         }
 431                         break;
 432                 case STYLES:
 433                         if (type == "Format") {
 434                                 style_format_line = body;
 435                         } else if (type == "Style") {
 436                                 SUB_ASSERT (!style_format_line.empty ());
 437                                 Style s (style_format_line, body);
 438                                 styles[s.name] = s;
 439                         }
 440                         break;
 441                 case EVENTS:
 442                         if (type == "Format") {
 443                                 split (event_format, body, is_any_of (","));
 444                                 for (auto& i: event_format) {
 445                                         trim (i);
 446                                 }
 447                         } else if (type == "Dialogue") {
 448                                 SUB_ASSERT (!event_format.empty ());
 449                                 vector<string> event;
 450                                 split (event, body, is_any_of (","));
 451
 452                                 /* There may be commas in the subtitle part; reassemble any extra parts
 453                                    from when we just split it.
 454                                 */
 455                                 while (event.size() > event_format.size()) {
 456                                         string const ex = event.back ();
 457                                         event.pop_back ();
 458                                         event.back() += "," + ex;
 459                                 }
 460
 461                                 SUB_ASSERT (!event.empty());
 462                                 SUB_ASSERT (event_format.size() == event.size());
 463
 464                                 RawSubtitle sub;
 465
 466                                 for (size_t i = 0; i < event.size(); ++i) {
 467                                         trim (event[i]);
 468                                         if (event_format[i] == "Start") {
 469                                                 sub.from = parse_time (event[i]);
 470                                         } else if (event_format[i] == "End") {
 471                                                 sub.to = parse_time (event[i]);
 472                                         } else if (event_format[i] == "Style") {
 473                                                 /* libass trims leading '*'s from style names, commenting that
 474                                                    "they seem to mean literally nothing".  Go figure...
 475                                                 */
 476                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 477                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 478                                                 Style style = styles[event[i]];
 479                                                 sub.font = style.font_name;
 480                                                 sub.font_size = FontSize::from_points (style.font_size);
 481                                                 sub.colour = style.primary_colour;
 482                                                 sub.effect_colour = style.back_colour;
 483                                                 sub.bold = style.bold;
 484                                                 sub.italic = style.italic;
 485                                                 sub.underline = style.underline;
 486                                                 sub.effect = style.effect;
 487                                                 sub.horizontal_position.reference = style.horizontal_reference;
 488                                                 sub.vertical_position.reference = style.vertical_reference;
 489                                                 sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 490                                         } else if (event_format[i] == "MarginV") {
 491                                                 sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 492                                         } else if (event_format[i] == "Text") {
 493                                                 for (auto j: parse_line (sub, event[i], play_res_x, play_res_y)) {
 494                                                         _subs.push_back (j);
 495                                                 }
 496                                         }
 497                                 }
 498                         }
 499                 }
 500
 501         }
 502 }