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 <cstdlib>
  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 boost::optional;
  37 using boost::function;
  38 using namespace boost::algorithm;
  39 #if BOOST_VERSION >= 106100
  40 using namespace boost::placeholders;
  41 #endif
  42 using namespace sub;
  43
  44 /** @param s Subtitle string encoded in UTF-8 */
  45 SSAReader::SSAReader (string s)
  46 {
  47         this->read (boost::bind(&get_line_string, &s));
  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 Colour
  57 h_colour (string s)
  58 {
  59         if (s.empty() || s[0] != '&' || s[1] != 'H') {
  60                 throw SSAError(String::compose("Badly formatted colour tag %1", s));
  61         }
  62
  63         auto start = s.c_str();
  64         auto const end = start + s.length();
  65         while (start < end && (*start == '&' || *start == 'H')) {
  66                 ++start;
  67         }
  68
  69         auto const colour = strtoll(start, nullptr, 16);
  70
  71         /* XXX: ignoring alpha channel here; note that 00 is opaque and FF is transparent */
  72         return sub::Colour(
  73                 ((colour & 0x000000ff) >> 0) / 255.0,
  74                 ((colour & 0x0000ff00) >> 8) / 255.0,
  75                 ((colour & 0x00ff0000) >> 16) / 255.0
  76                 );
  77 }
  78
  79 class Style
  80 {
  81 public:
  82         Style ()
  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
  93         Style (string format_line, string style_line)
  94                 : font_size (72)
  95                 , primary_colour (255, 255, 255)
  96                 , bold (false)
  97                 , italic (false)
  98                 , underline (false)
  99                 , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
 100                 , vertical_reference (BOTTOM_OF_SCREEN)
 101                 , vertical_margin (0)
 102         {
 103                 vector<string> keys;
 104                 split (keys, format_line, boost::is_any_of (","));
 105                 vector<string> style;
 106                 split (style, style_line, boost::is_any_of (","));
 107
 108                 SUB_ASSERT (!keys.empty());
 109                 SUB_ASSERT (!style.empty());
 110                 SUB_ASSERT (keys.size() == style.size());
 111
 112                 for (size_t i = 0; i < style.size(); ++i) {
 113                         trim (keys[i]);
 114                         trim (style[i]);
 115                         if (keys[i] == "Name") {
 116                                 name = style[i];
 117                         } else if (keys[i] == "Fontname") {
 118                                 font_name = style[i];
 119                         } else if (keys[i] == "Fontsize") {
 120                                 font_size = raw_convert<int> (style[i]);
 121                         } else if (keys[i] == "PrimaryColour") {
 122                                 primary_colour = colour (style[i]);
 123                         } else if (keys[i] == "BackColour") {
 124                                 back_colour = colour (style[i]);
 125                         } else if (keys[i] == "Bold") {
 126                                 bold = style[i] == "-1";
 127                         } else if (keys[i] == "Italic") {
 128                                 italic = style[i] == "-1";
 129                         } else if (keys[i] == "Underline") {
 130                                 underline = style[i] == "-1";
 131                         } else if (keys[i] == "BorderStyle") {
 132                                 if (style[i] == "1") {
 133                                         effect = SHADOW;
 134                                 }
 135                         } else if (keys[i] == "Alignment") {
 136                                 if (style[i] == "7" || style[i] == "8" || style[i] == "9") {
 137                                         vertical_reference = TOP_OF_SCREEN;
 138                                 } else if (style[i] == "4" || style[i] == "5" || style[i] == "6") {
 139                                         vertical_reference = VERTICAL_CENTRE_OF_SCREEN;
 140                                 } else {
 141                                         vertical_reference = BOTTOM_OF_SCREEN;
 142                                 }
 143                                 if (style[i] == "1" || style[i] == "4" || style[i] == "7") {
 144                                         horizontal_reference = LEFT_OF_SCREEN;
 145                                 } else if (style[i] == "3" || style[i] == "6" || style[i] == "9") {
 146                                         horizontal_reference = RIGHT_OF_SCREEN;
 147                                 } else {
 148                                         horizontal_reference = HORIZONTAL_CENTRE_OF_SCREEN;
 149                                 }
 150                         } else if (keys[i] == "MarginV") {
 151                                 vertical_margin = raw_convert<int> (style[i]);
 152                         } else if (keys[i] == "MarginL") {
 153                                 left_margin = raw_convert<int>(style[i]);
 154                         } else if (keys[i] == "MarginR") {
 155                                 right_margin = raw_convert<int>(style[i]);
 156                         }
 157                 }
 158         }
 159
 160         string name;
 161         optional<string> font_name;
 162         int font_size; ///< points
 163         Colour primary_colour;
 164         /** outline colour */
 165         optional<Colour> back_colour;
 166         bool bold;
 167         bool italic;
 168         bool underline;
 169         optional<Effect> effect;
 170         HorizontalReference horizontal_reference;
 171         VerticalReference vertical_reference;
 172         int vertical_margin;
 173         int left_margin = 0;
 174         int right_margin = 0;
 175
 176 private:
 177         Colour colour (string c) const
 178         {
 179                 if (c.length() > 0 && c[0] == '&') {
 180                         /* &Hbbggrr or &Haabbggrr */
 181                         return h_colour (c);
 182                 } else {
 183                         /* integer */
 184                         int i = raw_convert<int>(c);
 185                         return Colour (
 186                                 ((i & 0x0000ff) >>  0) / 255.0,
 187                                 ((i & 0x00ff00) >>  8) / 255.0,
 188                                 ((i & 0xff0000) >> 16) / 255.0
 189                                 );
 190                 }
 191         }
 192 };
 193
 194
 195 void
 196 SSAReader::Context::update_horizontal_position(RawSubtitle& sub) const
 197 {
 198         switch (sub.horizontal_position.reference) {
 199         case LEFT_OF_SCREEN:
 200                 sub.horizontal_position.proportional = static_cast<float>(left_margin) / play_res_x;
 201                 break;
 202         case HORIZONTAL_CENTRE_OF_SCREEN:
 203                 sub.horizontal_position.proportional = static_cast<float>(left_margin - right_margin) / (2 * play_res_x);
 204                 break;
 205         case RIGHT_OF_SCREEN:
 206                 sub.horizontal_position.proportional = static_cast<float>(right_margin) / play_res_x;
 207                 break;
 208         }
 209 }
 210
 211
 212 Time
 213 SSAReader::parse_time (string t) const
 214 {
 215         vector<string> bits;
 216         split (bits, t, is_any_of (":."));
 217         SUB_ASSERT (bits.size() == 4);
 218         return Time::from_hms (
 219                 raw_convert<int> (bits[0]),
 220                 raw_convert<int> (bits[1]),
 221                 raw_convert<int> (bits[2]),
 222                 raw_convert<int> (bits[3]) * 10
 223                 );
 224 }
 225
 226
 227 void
 228 SSAReader::parse_tag(RawSubtitle& sub, string tag, Context const& context)
 229 {
 230         if (tag == "\\i1") {
 231                 sub.italic = true;
 232         } else if (tag == "\\i0" || tag == "\\i") {
 233                 sub.italic = false;
 234         } else if (tag == "\\b1") {
 235                 sub.bold = true;
 236         } else if (tag == "\\b0") {
 237                 sub.bold = false;
 238         } else if (tag == "\\u1") {
 239                 sub.underline = true;
 240         } else if (tag == "\\u0") {
 241                 sub.underline = false;
 242         } else if (tag == "\\an1") {
 243                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 244                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 245                 context.update_horizontal_position(sub);
 246         } else if (tag == "\\an2") {
 247                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 248                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 249                 context.update_horizontal_position(sub);
 250         } else if (tag == "\\an3") {
 251                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 252                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 253                 context.update_horizontal_position(sub);
 254         } else if (tag == "\\an4") {
 255                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 256                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 257                 context.update_horizontal_position(sub);
 258         } else if (tag == "\\an5") {
 259                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 260                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 261                 context.update_horizontal_position(sub);
 262         } else if (tag == "\\an6") {
 263                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 264                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 265                 context.update_horizontal_position(sub);
 266         } else if (tag == "\\an7") {
 267                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 268                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 269                 context.update_horizontal_position(sub);
 270         } else if (tag == "\\an8") {
 271                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 272                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 273                 context.update_horizontal_position(sub);
 274         } else if (tag == "\\an9") {
 275                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 276                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 277                 context.update_horizontal_position(sub);
 278         } else if (boost::starts_with(tag, "\\pos")) {
 279                 vector<string> bits;
 280                 boost::algorithm::split (bits, tag, boost::is_any_of("(,"));
 281                 SUB_ASSERT (bits.size() == 3);
 282                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 283                 sub.horizontal_position.proportional = raw_convert<float>(bits[1]) / context.play_res_x;
 284                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 285                 sub.vertical_position.proportional = raw_convert<float>(bits[2]) / context.play_res_y;
 286         } else if (boost::starts_with(tag, "\\fs")) {
 287                 SUB_ASSERT (tag.length() > 3);
 288                 sub.font_size.set_proportional(raw_convert<float>(tag.substr(3)) / context.play_res_y);
 289         } else if (boost::starts_with(tag, "\\c")) {
 290                 /* \c&Hbbggrr& */
 291                 if (tag.length() > 2) {
 292                         sub.colour = h_colour(tag.substr(2, tag.length() - 3));
 293                 } else if (tag.length() == 2) {
 294                         sub.colour = context.primary_colour;
 295                 } else {
 296                         throw SSAError(String::compose("Badly formatted colour tag %1", tag));
 297                 }
 298         }
 299 }
 300
 301 /** @param base RawSubtitle filled in with any required common values.
 302  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 303  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 304  */
 305 vector<RawSubtitle>
 306 SSAReader::parse_line(RawSubtitle base, string line, Context const& context)
 307 {
 308         enum {
 309                 TEXT,
 310                 TAG,
 311                 BACKSLASH
 312         } state = TEXT;
 313
 314         vector<RawSubtitle> subs;
 315         RawSubtitle current = base;
 316         string tag;
 317
 318         if (!current.vertical_position.reference) {
 319                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 320         }
 321
 322         /* Any vertical_position that is set in base (and therefore current) is a margin, which
 323          * we need to ignore if we end up vertically centering this subtitle.
 324          * Clear out vertical_position from current; we'll re-add it from base later
 325          * if required.
 326          */
 327         current.vertical_position.proportional = 0;
 328
 329         context.update_horizontal_position(current);
 330
 331         /* We must have a font size, as there could be a margin specified
 332            in pixels and in that case we must know how big the subtitle
 333            lines are to work out the position on screen.
 334         */
 335         if (!current.font_size.proportional()) {
 336                 current.font_size.set_proportional(72.0 / context.play_res_y);
 337         }
 338
 339         /* Count the number of line breaks */
 340         int line_breaks = 0;
 341         if (line.length() > 0) {
 342                 for (size_t i = 0; i < line.length() - 1; ++i) {
 343                         if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 344                                 ++line_breaks;
 345                         }
 346                 }
 347         }
 348
 349         /* There are vague indications that with ASS 1 point should equal 1 pixel */
 350         double const line_size = current.font_size.proportional(context.play_res_y) * 1.2;
 351
 352         for (size_t i = 0; i < line.length(); ++i) {
 353                 char const c = line[i];
 354                 switch (state) {
 355                 case TEXT:
 356                         if (c == '{') {
 357                                 state = TAG;
 358                         } else if (c == '\\') {
 359                                 state = BACKSLASH;
 360                         } else if (c != '\r' && c != '\n') {
 361                                 current.text += c;
 362                         }
 363                         break;
 364                 case TAG:
 365                         if (c == '}' || c == '\\') {
 366                                 if (!current.text.empty ()) {
 367                                         subs.push_back (current);
 368                                         current.text = "";
 369                                 }
 370                                 parse_tag(current, tag, context);
 371                                 tag = "";
 372                         }
 373
 374                         if (c == '}') {
 375                                 state = TEXT;
 376                         } else {
 377                                 tag += c;
 378                         }
 379                         break;
 380                 case BACKSLASH:
 381                         if (c == 'n' || c == 'N') {
 382                                 if (!current.text.empty ()) {
 383                                         subs.push_back (current);
 384                                         current.text = "";
 385                                 }
 386                                 /* Move down one line (1.2 times the font size) */
 387                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 388                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 389                                 } else {
 390                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 391                                 }
 392                         }
 393                         state = TEXT;
 394                         break;
 395                 }
 396         }
 397
 398         if (!current.text.empty ()) {
 399                 subs.push_back (current);
 400         }
 401
 402         /* Now we definitely know the vertical position reference we can finish off the position */
 403         for (auto& sub: subs) {
 404                 switch (sub.vertical_position.reference.get()) {
 405                 case TOP_OF_SCREEN:
 406                 case TOP_OF_SUBTITLE:
 407                         /* Just re-add any margins we came in with */
 408                         sub.vertical_position.proportional = sub.vertical_position.proportional.get() + base.vertical_position.proportional.get_value_or(0);
 409                         break;
 410                 case VERTICAL_CENTRE_OF_SCREEN:
 411                         /* Margins are ignored, but we need to centre */
 412                         sub.vertical_position.proportional = sub.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 413                         break;
 414                 case BOTTOM_OF_SCREEN:
 415                         /* Re-add margins and account for each line */
 416                         sub.vertical_position.proportional =
 417                                 sub.vertical_position.proportional.get()
 418                                 + base.vertical_position.proportional.get_value_or(0)
 419                                 + line_breaks * line_size;
 420                         break;
 421                 }
 422         }
 423
 424         return subs;
 425 }
 426
 427 void
 428 SSAReader::read (function<optional<string> ()> get_line)
 429 {
 430         enum {
 431                 INFO,
 432                 STYLES,
 433                 EVENTS
 434         } part = INFO;
 435
 436         int play_res_x = 288;
 437         int play_res_y = 288;
 438         map<string, Style> styles;
 439         string style_format_line;
 440         vector<string> event_format;
 441
 442         while (true) {
 443                 optional<string> line = get_line ();
 444                 if (!line) {
 445                         break;
 446                 }
 447
 448                 trim (*line);
 449                 remove_unicode_bom (line);
 450
 451                 if (starts_with (*line, ";") || line->empty ()) {
 452                         continue;
 453                 }
 454
 455                 if (starts_with (*line, "[")) {
 456                         /* Section heading */
 457                         if (line.get() == "[Script Info]") {
 458                                 part = INFO;
 459                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 460                                 part = STYLES;
 461                         } else if (line.get() == "[Events]") {
 462                                 part = EVENTS;
 463                         }
 464                         continue;
 465                 }
 466
 467                 size_t const colon = line->find (":");
 468                 SUB_ASSERT (colon != string::npos);
 469                 string const type = line->substr (0, colon);
 470                 string body = line->substr (colon + 1);
 471                 trim (body);
 472
 473                 switch (part) {
 474                 case INFO:
 475                         if (type == "PlayResX") {
 476                                 play_res_x = raw_convert<int> (body);
 477                         } else if (type == "PlayResY") {
 478                                 play_res_y = raw_convert<int> (body);
 479                         }
 480                         break;
 481                 case STYLES:
 482                         if (type == "Format") {
 483                                 style_format_line = body;
 484                         } else if (type == "Style") {
 485                                 SUB_ASSERT (!style_format_line.empty ());
 486                                 Style s (style_format_line, body);
 487                                 styles[s.name] = s;
 488                         }
 489                         break;
 490                 case EVENTS:
 491                         if (type == "Format") {
 492                                 split (event_format, body, is_any_of (","));
 493                                 for (auto& i: event_format) {
 494                                         trim (i);
 495                                 }
 496                         } else if (type == "Dialogue") {
 497                                 SUB_ASSERT (!event_format.empty ());
 498                                 vector<string> event;
 499                                 split (event, body, is_any_of (","));
 500
 501                                 /* There may be commas in the subtitle part; reassemble any extra parts
 502                                    from when we just split it.
 503                                 */
 504                                 while (event.size() > event_format.size()) {
 505                                         string const ex = event.back ();
 506                                         event.pop_back ();
 507                                         event.back() += "," + ex;
 508                                 }
 509
 510                                 SUB_ASSERT (!event.empty());
 511                                 SUB_ASSERT (event_format.size() == event.size());
 512
 513                                 RawSubtitle sub;
 514                                 optional<Style> style;
 515                                 int left_margin = 0;
 516                                 int right_margin = 0;
 517
 518                                 for (size_t i = 0; i < event.size(); ++i) {
 519                                         trim (event[i]);
 520                                         if (event_format[i] == "Start") {
 521                                                 sub.from = parse_time (event[i]);
 522                                         } else if (event_format[i] == "End") {
 523                                                 sub.to = parse_time (event[i]);
 524                                         } else if (event_format[i] == "Style") {
 525                                                 /* libass trims leading '*'s from style names, commenting that
 526                                                    "they seem to mean literally nothing".  Go figure...
 527                                                 */
 528                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 529                                                 /* Use the specified style unless it's not defined, in which case use
 530                                                  * "Default" (if it exists).
 531                                                  */
 532                                                 if (styles.find(event[i]) != styles.end()) {
 533                                                         style = styles[event[i]];
 534                                                 } else if (styles.find("Default") != styles.end()) {
 535                                                         style = styles["Default"];
 536                                                 } else {
 537                                                         continue;
 538                                                 }
 539                                                 sub.font = style->font_name;
 540                                                 sub.font_size = FontSize::from_proportional(static_cast<float>(style->font_size) / play_res_y);
 541                                                 sub.colour = style->primary_colour;
 542                                                 sub.effect_colour = style->back_colour;
 543                                                 sub.bold = style->bold;
 544                                                 sub.italic = style->italic;
 545                                                 sub.underline = style->underline;
 546                                                 sub.effect = style->effect;
 547                                                 sub.horizontal_position.reference = style->horizontal_reference;
 548                                                 sub.vertical_position.reference = style->vertical_reference;
 549                                                 if (sub.vertical_position.reference != sub::VERTICAL_CENTRE_OF_SCREEN) {
 550                                                         sub.vertical_position.proportional = float(style->vertical_margin) / play_res_y;
 551                                                 }
 552                                                 left_margin = style->left_margin;
 553                                                 right_margin = style->right_margin;
 554                                         } else if (event_format[i] == "MarginV") {
 555                                                 if (event[i] != "0" && sub.vertical_position.reference != sub::VERTICAL_CENTRE_OF_SCREEN) {
 556                                                         /* Override the style if its non-zero */
 557                                                         sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 558                                                 }
 559                                         } else if (event_format[i] == "MarginL") {
 560                                                 if (event[i] != "0") {
 561                                                         left_margin = raw_convert<int>(event[i]);
 562                                                 }
 563                                         } else if (event_format[i] == "MarginR") {
 564                                                 if (event[i] != "0") {
 565                                                         right_margin = raw_convert<int>(event[i]);
 566                                                 }
 567                                         } else if (event_format[i] == "Text") {
 568                                                 auto context = Context(play_res_x, play_res_y, style ? style->primary_colour : Colour(1, 1, 1), left_margin, right_margin);
 569                                                 for (auto j: parse_line(sub, event[i], context)) {
 570                                                         _subs.push_back (j);
 571                                                 }
 572                                         }
 573                                 }
 574                         }
 575                 }
 576
 577         }
 578 }