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.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 using namespace sub;
  39
  40 /** @param s Subtitle string encoded in UTF-8 */
  41 SSAReader::SSAReader (string s)
  42 {
  43         this->read (boost::bind(&get_line_string, &s));
  44 }
  45
  46 /** @param f Subtitle file encoded in UTF-8 */
  47 SSAReader::SSAReader (FILE* f)
  48 {
  49         this->read (boost::bind (&get_line_file, f));
  50 }
  51
  52 Colour
  53 h_colour (string s)
  54 {
  55         /* There are both BGR and ABGR versions of these colours */
  56         if ((s.length() != 8 && s.length() != 10) || s[0] != '&' || s[1] != 'H') {
  57                 throw SSAError(String::compose("Badly formatted colour tag %1", s));
  58         }
  59         int ir, ig, ib;
  60         /* XXX: ignoring alpha channel here; note that 00 is opaque and FF is transparent */
  61         int const off = s.length() == 10 ? 4 : 2;
  62         if (sscanf(s.c_str() + off, "%2x%2x%2x", &ib, &ig, &ir) < 3) {
  63                 throw SSAError(String::compose("Badly formatted colour tag %1", s));
  64         }
  65         return sub::Colour(ir / 255.0, ig / 255.0, ib / 255.0);
  66 }
  67
  68 class Style
  69 {
  70 public:
  71         Style ()
  72                 : font_size (72)
  73                 , primary_colour (255, 255, 255)
  74                 , bold (false)
  75                 , italic (false)
  76                 , underline (false)
  77                 , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
  78                 , vertical_reference (BOTTOM_OF_SCREEN)
  79                 , vertical_margin (0)
  80         {}
  81
  82         Style (string format_line, string style_line)
  83                 : font_size (72)
  84                 , primary_colour (255, 255, 255)
  85                 , bold (false)
  86                 , italic (false)
  87                 , underline (false)
  88                 , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
  89                 , vertical_reference (BOTTOM_OF_SCREEN)
  90                 , vertical_margin (0)
  91         {
  92                 vector<string> keys;
  93                 split (keys, format_line, boost::is_any_of (","));
  94                 vector<string> style;
  95                 split (style, style_line, boost::is_any_of (","));
  96
  97                 SUB_ASSERT (!keys.empty());
  98                 SUB_ASSERT (!style.empty());
  99                 SUB_ASSERT (keys.size() == style.size());
 100
 101                 for (size_t i = 0; i < style.size(); ++i) {
 102                         trim (keys[i]);
 103                         trim (style[i]);
 104                         if (keys[i] == "Name") {
 105                                 name = style[i];
 106                         } else if (keys[i] == "Fontname") {
 107                                 font_name = style[i];
 108                         } else if (keys[i] == "Fontsize") {
 109                                 font_size = raw_convert<int> (style[i]);
 110                         } else if (keys[i] == "PrimaryColour") {
 111                                 primary_colour = colour (style[i]);
 112                         } else if (keys[i] == "BackColour") {
 113                                 back_colour = colour (style[i]);
 114                         } else if (keys[i] == "Bold") {
 115                                 bold = style[i] == "-1";
 116                         } else if (keys[i] == "Italic") {
 117                                 italic = style[i] == "-1";
 118                         } else if (keys[i] == "Underline") {
 119                                 underline = style[i] == "-1";
 120                         } else if (keys[i] == "BorderStyle") {
 121                                 if (style[i] == "1") {
 122                                         effect = SHADOW;
 123                                 }
 124                         } else if (keys[i] == "Alignment") {
 125                                 /* These values from libass' source code */
 126                                 switch ((raw_convert<int> (style[i]) - 1) % 3) {
 127                                 case 0:
 128                                         horizontal_reference = LEFT_OF_SCREEN;
 129                                         break;
 130                                 case 1:
 131                                         horizontal_reference = HORIZONTAL_CENTRE_OF_SCREEN;
 132                                         break;
 133                                 case 2:
 134                                         horizontal_reference = RIGHT_OF_SCREEN;
 135                                         break;
 136                                 }
 137                                 switch (raw_convert<int> (style[i]) & 12) {
 138                                 case 4:
 139                                         vertical_reference = TOP_OF_SCREEN;
 140                                         break;
 141                                 case 8:
 142                                         vertical_reference = VERTICAL_CENTRE_OF_SCREEN;
 143                                         break;
 144                                 case 0:
 145                                         vertical_reference = BOTTOM_OF_SCREEN;
 146                                         break;
 147                                 }
 148                         } else if (keys[i] == "MarginV") {
 149                                 vertical_margin = raw_convert<int> (style[i]);
 150                         }
 151                 }
 152         }
 153
 154         string name;
 155         optional<string> font_name;
 156         int font_size;
 157         Colour primary_colour;
 158         /** outline colour */
 159         optional<Colour> back_colour;
 160         bool bold;
 161         bool italic;
 162         bool underline;
 163         optional<Effect> effect;
 164         HorizontalReference horizontal_reference;
 165         VerticalReference vertical_reference;
 166         int vertical_margin;
 167
 168 private:
 169         Colour colour (string c) const
 170         {
 171                 if (c.length() > 0 && c[0] == '&') {
 172                         /* &Hbbggrr or &Haabbggrr */
 173                         return h_colour (c);
 174                 } else {
 175                         /* integer */
 176                         int i = raw_convert<int>(c);
 177                         return Colour (
 178                                 ((i & 0x0000ff) >>  0) / 255.0,
 179                                 ((i & 0x00ff00) >>  8) / 255.0,
 180                                 ((i & 0xff0000) >> 16) / 255.0
 181                                 );
 182                 }
 183         }
 184 };
 185
 186 Time
 187 SSAReader::parse_time (string t) const
 188 {
 189         vector<string> bits;
 190         split (bits, t, is_any_of (":."));
 191         SUB_ASSERT (bits.size() == 4);
 192         return Time::from_hms (
 193                 raw_convert<int> (bits[0]),
 194                 raw_convert<int> (bits[1]),
 195                 raw_convert<int> (bits[2]),
 196                 raw_convert<int> (bits[3]) * 10
 197                 );
 198 }
 199
 200 void
 201 SSAReader::parse_style (RawSubtitle& sub, string style, int play_res_x, int play_res_y)
 202 {
 203         if (style == "\\i1") {
 204                 sub.italic = true;
 205         } else if (style == "\\i0" || style == "\\i") {
 206                 sub.italic = false;
 207         } else if (style == "\\b1") {
 208                 sub.bold = true;
 209         } else if (style == "\\b0") {
 210                 sub.bold = false;
 211         } else if (style == "\\u1") {
 212                 sub.underline = true;
 213         } else if (style == "\\u0") {
 214                 sub.underline = false;
 215         } else if (style == "\\an1") {
 216                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 217                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 218         } else if (style == "\\an2") {
 219                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 220                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 221         } else if (style == "\\an3") {
 222                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 223                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 224         } else if (style == "\\an4") {
 225                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 226                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 227         } else if (style == "\\an5") {
 228                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 229                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 230         } else if (style == "\\an6") {
 231                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 232                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 233         } else if (style == "\\an7") {
 234                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 235                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 236         } else if (style == "\\an8") {
 237                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 238                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 239         } else if (style == "\\an9") {
 240                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 241                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 242         } else if (boost::starts_with(style, "\\pos")) {
 243                 vector<string> bits;
 244                 boost::algorithm::split (bits, style, boost::is_any_of("(,"));
 245                 SUB_ASSERT (bits.size() == 3);
 246                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 247                 sub.horizontal_position.proportional = raw_convert<float>(bits[1]) / play_res_x;
 248                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 249                 sub.vertical_position.proportional = raw_convert<float>(bits[2]) / play_res_y;
 250         } else if (boost::starts_with(style, "\\fs")) {
 251                 SUB_ASSERT (style.length() > 3);
 252                 sub.font_size.set_points (raw_convert<int>(style.substr(3)));
 253         } else if (boost::starts_with(style, "\\c")) {
 254                 /* \c&Hbbggrr& */
 255                 if (style.length() <= 2) {
 256                         throw SSAError(String::compose("Badly formatted colour tag %1", style));
 257                 }
 258                 sub.colour = h_colour (style.substr(2, style.length() - 3));
 259         }
 260 }
 261
 262 /** @param base RawSubtitle filled in with any required common values.
 263  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 264  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 265  */
 266 vector<RawSubtitle>
 267 SSAReader::parse_line (RawSubtitle base, string line, int play_res_x, int play_res_y)
 268 {
 269         enum {
 270                 TEXT,
 271                 STYLE,
 272                 BACKSLASH
 273         } state = TEXT;
 274
 275         vector<RawSubtitle> subs;
 276         RawSubtitle current = base;
 277         string style;
 278
 279         if (!current.vertical_position.reference) {
 280                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 281         }
 282
 283         if (!current.vertical_position.proportional) {
 284                 current.vertical_position.proportional = 0;
 285         }
 286
 287         /* We must have a font size, as there could be a margin specified
 288            in pixels and in that case we must know how big the subtitle
 289            lines are to work out the position on screen.
 290         */
 291         if (!current.font_size.points()) {
 292                 current.font_size.set_points (72);
 293         }
 294
 295         /* Count the number of line breaks */
 296         int line_breaks = 0;
 297         if (line.length() > 0) {
 298                 for (size_t i = 0; i < line.length() - 1; ++i) {
 299                         if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 300                                 ++line_breaks;
 301                         }
 302                 }
 303         }
 304
 305         /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
 306         double const line_size = current.font_size.proportional(792) * 1.2;
 307
 308         /* Tweak vertical_position accordingly */
 309         switch (current.vertical_position.reference.get()) {
 310         case TOP_OF_SCREEN:
 311         case TOP_OF_SUBTITLE:
 312                 /* Nothing to do */
 313                 break;
 314         case VERTICAL_CENTRE_OF_SCREEN:
 315                 current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 316                 break;
 317         case BOTTOM_OF_SCREEN:
 318                 current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
 319                 break;
 320         }
 321
 322         for (size_t i = 0; i < line.length(); ++i) {
 323                 char const c = line[i];
 324                 switch (state) {
 325                 case TEXT:
 326                         if (c == '{') {
 327                                 state = STYLE;
 328                         } else if (c == '\\') {
 329                                 state = BACKSLASH;
 330                         } else if (c != '\r' && c != '\n') {
 331                                 current.text += c;
 332                         }
 333                         break;
 334                 case STYLE:
 335                         if (c == '}' || c == '\\') {
 336                                 if (!current.text.empty ()) {
 337                                         subs.push_back (current);
 338                                         current.text = "";
 339                                 }
 340                                 parse_style (current, style, play_res_x, play_res_y);
 341                                 style = "";
 342                         }
 343
 344                         if (c == '}') {
 345                                 state = TEXT;
 346                         } else {
 347                                 style += c;
 348                         }
 349                         break;
 350                 case BACKSLASH:
 351                         if (c == 'n' || c == 'N') {
 352                                 if (!current.text.empty ()) {
 353                                         subs.push_back (current);
 354                                         current.text = "";
 355                                 }
 356                                 /* Move down one line (1.2 times the font size) */
 357                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 358                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 359                                 } else {
 360                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 361                                 }
 362                         }
 363                         state = TEXT;
 364                         break;
 365                 }
 366         }
 367
 368         if (!current.text.empty ()) {
 369                 subs.push_back (current);
 370         }
 371
 372         return subs;
 373 }
 374
 375 void
 376 SSAReader::read (function<optional<string> ()> get_line)
 377 {
 378         enum {
 379                 INFO,
 380                 STYLES,
 381                 EVENTS
 382         } part = INFO;
 383
 384         int play_res_x = 288;
 385         int play_res_y = 288;
 386         map<string, Style> styles;
 387         string style_format_line;
 388         vector<string> event_format;
 389
 390         while (true) {
 391                 optional<string> line = get_line ();
 392                 if (!line) {
 393                         break;
 394                 }
 395
 396                 trim (*line);
 397                 remove_unicode_bom (line);
 398
 399                 if (starts_with (*line, ";") || line->empty ()) {
 400                         continue;
 401                 }
 402
 403                 if (starts_with (*line, "[")) {
 404                         /* Section heading */
 405                         if (line.get() == "[Script Info]") {
 406                                 part = INFO;
 407                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 408                                 part = STYLES;
 409                         } else if (line.get() == "[Events]") {
 410                                 part = EVENTS;
 411                         }
 412                         continue;
 413                 }
 414
 415                 size_t const colon = line->find (":");
 416                 SUB_ASSERT (colon != string::npos);
 417                 string const type = line->substr (0, colon);
 418                 string body = line->substr (colon + 1);
 419                 trim (body);
 420
 421                 switch (part) {
 422                 case INFO:
 423                         if (type == "PlayResX") {
 424                                 play_res_x = raw_convert<int> (body);
 425                         } else if (type == "PlayResY") {
 426                                 play_res_y = raw_convert<int> (body);
 427                         }
 428                         break;
 429                 case STYLES:
 430                         if (type == "Format") {
 431                                 style_format_line = body;
 432                         } else if (type == "Style") {
 433                                 SUB_ASSERT (!style_format_line.empty ());
 434                                 Style s (style_format_line, body);
 435                                 styles[s.name] = s;
 436                         }
 437                         break;
 438                 case EVENTS:
 439                         if (type == "Format") {
 440                                 split (event_format, body, is_any_of (","));
 441                                 for (auto& i: event_format) {
 442                                         trim (i);
 443                                 }
 444                         } else if (type == "Dialogue") {
 445                                 SUB_ASSERT (!event_format.empty ());
 446                                 vector<string> event;
 447                                 split (event, body, is_any_of (","));
 448
 449                                 /* There may be commas in the subtitle part; reassemble any extra parts
 450                                    from when we just split it.
 451                                 */
 452                                 while (event.size() > event_format.size()) {
 453                                         string const ex = event.back ();
 454                                         event.pop_back ();
 455                                         event.back() += "," + ex;
 456                                 }
 457
 458                                 SUB_ASSERT (!event.empty());
 459                                 SUB_ASSERT (event_format.size() == event.size());
 460
 461                                 RawSubtitle sub;
 462
 463                                 for (size_t i = 0; i < event.size(); ++i) {
 464                                         trim (event[i]);
 465                                         if (event_format[i] == "Start") {
 466                                                 sub.from = parse_time (event[i]);
 467                                         } else if (event_format[i] == "End") {
 468                                                 sub.to = parse_time (event[i]);
 469                                         } else if (event_format[i] == "Style") {
 470                                                 /* libass trims leading '*'s from style names, commenting that
 471                                                    "they seem to mean literally nothing".  Go figure...
 472                                                 */
 473                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 474                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 475                                                 Style style = styles[event[i]];
 476                                                 sub.font = style.font_name;
 477                                                 sub.font_size = FontSize::from_points (style.font_size);
 478                                                 sub.colour = style.primary_colour;
 479                                                 sub.effect_colour = style.back_colour;
 480                                                 sub.bold = style.bold;
 481                                                 sub.italic = style.italic;
 482                                                 sub.underline = style.underline;
 483                                                 sub.effect = style.effect;
 484                                                 sub.horizontal_position.reference = style.horizontal_reference;
 485                                                 sub.vertical_position.reference = style.vertical_reference;
 486                                                 sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 487                                         } else if (event_format[i] == "MarginV") {
 488                                                 sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 489                                         } else if (event_format[i] == "Text") {
 490                                                 for (auto j: parse_line (sub, event[i], play_res_x, play_res_y)) {
 491                                                         _subs.push_back (j);
 492                                                 }
 493                                         }
 494                                 }
 495                         }
 496                 }
 497
 498         }
 499 }