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                         }
 153                 }
 154         }
 155
 156         string name;
 157         optional<string> font_name;
 158         int font_size; ///< points
 159         Colour primary_colour;
 160         /** outline colour */
 161         optional<Colour> back_colour;
 162         bool bold;
 163         bool italic;
 164         bool underline;
 165         optional<Effect> effect;
 166         HorizontalReference horizontal_reference;
 167         VerticalReference vertical_reference;
 168         int vertical_margin;
 169
 170 private:
 171         Colour colour (string c) const
 172         {
 173                 if (c.length() > 0 && c[0] == '&') {
 174                         /* &Hbbggrr or &Haabbggrr */
 175                         return h_colour (c);
 176                 } else {
 177                         /* integer */
 178                         int i = raw_convert<int>(c);
 179                         return Colour (
 180                                 ((i & 0x0000ff) >>  0) / 255.0,
 181                                 ((i & 0x00ff00) >>  8) / 255.0,
 182                                 ((i & 0xff0000) >> 16) / 255.0
 183                                 );
 184                 }
 185         }
 186 };
 187
 188 Time
 189 SSAReader::parse_time (string t) const
 190 {
 191         vector<string> bits;
 192         split (bits, t, is_any_of (":."));
 193         SUB_ASSERT (bits.size() == 4);
 194         return Time::from_hms (
 195                 raw_convert<int> (bits[0]),
 196                 raw_convert<int> (bits[1]),
 197                 raw_convert<int> (bits[2]),
 198                 raw_convert<int> (bits[3]) * 10
 199                 );
 200 }
 201
 202 void
 203 SSAReader::parse_style (RawSubtitle& sub, string style, int play_res_x, int play_res_y)
 204 {
 205         if (style == "\\i1") {
 206                 sub.italic = true;
 207         } else if (style == "\\i0" || style == "\\i") {
 208                 sub.italic = false;
 209         } else if (style == "\\b1") {
 210                 sub.bold = true;
 211         } else if (style == "\\b0") {
 212                 sub.bold = false;
 213         } else if (style == "\\u1") {
 214                 sub.underline = true;
 215         } else if (style == "\\u0") {
 216                 sub.underline = false;
 217         } else if (style == "\\an1") {
 218                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 219                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 220         } else if (style == "\\an2") {
 221                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 222                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 223         } else if (style == "\\an3") {
 224                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 225                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 226         } else if (style == "\\an4") {
 227                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 228                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 229         } else if (style == "\\an5") {
 230                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 231                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 232         } else if (style == "\\an6") {
 233                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 234                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 235         } else if (style == "\\an7") {
 236                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 237                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 238         } else if (style == "\\an8") {
 239                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 240                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 241         } else if (style == "\\an9") {
 242                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 243                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 244         } else if (boost::starts_with(style, "\\pos")) {
 245                 vector<string> bits;
 246                 boost::algorithm::split (bits, style, boost::is_any_of("(,"));
 247                 SUB_ASSERT (bits.size() == 3);
 248                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 249                 sub.horizontal_position.proportional = raw_convert<float>(bits[1]) / play_res_x;
 250                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 251                 sub.vertical_position.proportional = raw_convert<float>(bits[2]) / play_res_y;
 252         } else if (boost::starts_with(style, "\\fs")) {
 253                 SUB_ASSERT (style.length() > 3);
 254                 sub.font_size.set_proportional(raw_convert<float>(style.substr(3)) / play_res_y);
 255         } else if (boost::starts_with(style, "\\c")) {
 256                 /* \c&Hbbggrr& */
 257                 if (style.length() <= 2) {
 258                         throw SSAError(String::compose("Badly formatted colour tag %1", style));
 259                 }
 260                 sub.colour = h_colour (style.substr(2, style.length() - 3));
 261         }
 262 }
 263
 264 /** @param base RawSubtitle filled in with any required common values.
 265  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 266  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 267  */
 268 vector<RawSubtitle>
 269 SSAReader::parse_line (RawSubtitle base, string line, int play_res_x, int play_res_y)
 270 {
 271         enum {
 272                 TEXT,
 273                 STYLE,
 274                 BACKSLASH
 275         } state = TEXT;
 276
 277         vector<RawSubtitle> subs;
 278         RawSubtitle current = base;
 279         string style;
 280
 281         if (!current.vertical_position.reference) {
 282                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 283         }
 284
 285         /* Any vertical_position that is set in base (and therefore current) is a margin, which
 286          * we need to ignore if we end up vertically centering this subtitle.
 287          * Clear out vertical_position from current; we'll re-add it from base later
 288          * if required.
 289          */
 290         current.vertical_position.proportional = 0;
 291
 292         /* We must have a font size, as there could be a margin specified
 293            in pixels and in that case we must know how big the subtitle
 294            lines are to work out the position on screen.
 295         */
 296         if (!current.font_size.proportional()) {
 297                 current.font_size.set_proportional(72.0 / play_res_y);
 298         }
 299
 300         /* Count the number of line breaks */
 301         int line_breaks = 0;
 302         if (line.length() > 0) {
 303                 for (size_t i = 0; i < line.length() - 1; ++i) {
 304                         if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 305                                 ++line_breaks;
 306                         }
 307                 }
 308         }
 309
 310         /* There are vague indications that with ASS 1 point should equal 1 pixel */
 311         double const line_size = current.font_size.proportional(play_res_y) * 1.2;
 312
 313         for (size_t i = 0; i < line.length(); ++i) {
 314                 char const c = line[i];
 315                 switch (state) {
 316                 case TEXT:
 317                         if (c == '{') {
 318                                 state = STYLE;
 319                         } else if (c == '\\') {
 320                                 state = BACKSLASH;
 321                         } else if (c != '\r' && c != '\n') {
 322                                 current.text += c;
 323                         }
 324                         break;
 325                 case STYLE:
 326                         if (c == '}' || c == '\\') {
 327                                 if (!current.text.empty ()) {
 328                                         subs.push_back (current);
 329                                         current.text = "";
 330                                 }
 331                                 parse_style (current, style, play_res_x, play_res_y);
 332                                 style = "";
 333                         }
 334
 335                         if (c == '}') {
 336                                 state = TEXT;
 337                         } else {
 338                                 style += c;
 339                         }
 340                         break;
 341                 case BACKSLASH:
 342                         if (c == 'n' || c == 'N') {
 343                                 if (!current.text.empty ()) {
 344                                         subs.push_back (current);
 345                                         current.text = "";
 346                                 }
 347                                 /* Move down one line (1.2 times the font size) */
 348                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 349                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 350                                 } else {
 351                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 352                                 }
 353                         }
 354                         state = TEXT;
 355                         break;
 356                 }
 357         }
 358
 359         if (!current.text.empty ()) {
 360                 subs.push_back (current);
 361         }
 362
 363         /* Now we definitely know the vertical position reference we can finish off the position */
 364         for (auto& sub: subs) {
 365                 switch (sub.vertical_position.reference.get()) {
 366                 case TOP_OF_SCREEN:
 367                 case TOP_OF_SUBTITLE:
 368                         /* Just re-add any margins we came in with */
 369                         sub.vertical_position.proportional = sub.vertical_position.proportional.get() + base.vertical_position.proportional.get_value_or(0);
 370                         break;
 371                 case VERTICAL_CENTRE_OF_SCREEN:
 372                         /* Margins are ignored, but we need to centre */
 373                         sub.vertical_position.proportional = sub.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 374                         break;
 375                 case BOTTOM_OF_SCREEN:
 376                         /* Re-add margins and account for each line */
 377                         sub.vertical_position.proportional =
 378                                 sub.vertical_position.proportional.get()
 379                                 + base.vertical_position.proportional.get_value_or(0)
 380                                 + line_breaks * line_size;
 381                         break;
 382                 }
 383         }
 384
 385         return subs;
 386 }
 387
 388 void
 389 SSAReader::read (function<optional<string> ()> get_line)
 390 {
 391         enum {
 392                 INFO,
 393                 STYLES,
 394                 EVENTS
 395         } part = INFO;
 396
 397         int play_res_x = 288;
 398         int play_res_y = 288;
 399         map<string, Style> styles;
 400         string style_format_line;
 401         vector<string> event_format;
 402
 403         while (true) {
 404                 optional<string> line = get_line ();
 405                 if (!line) {
 406                         break;
 407                 }
 408
 409                 trim (*line);
 410                 remove_unicode_bom (line);
 411
 412                 if (starts_with (*line, ";") || line->empty ()) {
 413                         continue;
 414                 }
 415
 416                 if (starts_with (*line, "[")) {
 417                         /* Section heading */
 418                         if (line.get() == "[Script Info]") {
 419                                 part = INFO;
 420                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 421                                 part = STYLES;
 422                         } else if (line.get() == "[Events]") {
 423                                 part = EVENTS;
 424                         }
 425                         continue;
 426                 }
 427
 428                 size_t const colon = line->find (":");
 429                 SUB_ASSERT (colon != string::npos);
 430                 string const type = line->substr (0, colon);
 431                 string body = line->substr (colon + 1);
 432                 trim (body);
 433
 434                 switch (part) {
 435                 case INFO:
 436                         if (type == "PlayResX") {
 437                                 play_res_x = raw_convert<int> (body);
 438                         } else if (type == "PlayResY") {
 439                                 play_res_y = raw_convert<int> (body);
 440                         }
 441                         break;
 442                 case STYLES:
 443                         if (type == "Format") {
 444                                 style_format_line = body;
 445                         } else if (type == "Style") {
 446                                 SUB_ASSERT (!style_format_line.empty ());
 447                                 Style s (style_format_line, body);
 448                                 styles[s.name] = s;
 449                         }
 450                         break;
 451                 case EVENTS:
 452                         if (type == "Format") {
 453                                 split (event_format, body, is_any_of (","));
 454                                 for (auto& i: event_format) {
 455                                         trim (i);
 456                                 }
 457                         } else if (type == "Dialogue") {
 458                                 SUB_ASSERT (!event_format.empty ());
 459                                 vector<string> event;
 460                                 split (event, body, is_any_of (","));
 461
 462                                 /* There may be commas in the subtitle part; reassemble any extra parts
 463                                    from when we just split it.
 464                                 */
 465                                 while (event.size() > event_format.size()) {
 466                                         string const ex = event.back ();
 467                                         event.pop_back ();
 468                                         event.back() += "," + ex;
 469                                 }
 470
 471                                 SUB_ASSERT (!event.empty());
 472                                 SUB_ASSERT (event_format.size() == event.size());
 473
 474                                 RawSubtitle sub;
 475
 476                                 for (size_t i = 0; i < event.size(); ++i) {
 477                                         trim (event[i]);
 478                                         if (event_format[i] == "Start") {
 479                                                 sub.from = parse_time (event[i]);
 480                                         } else if (event_format[i] == "End") {
 481                                                 sub.to = parse_time (event[i]);
 482                                         } else if (event_format[i] == "Style") {
 483                                                 /* libass trims leading '*'s from style names, commenting that
 484                                                    "they seem to mean literally nothing".  Go figure...
 485                                                 */
 486                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 487                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 488                                                 Style style = styles[event[i]];
 489                                                 sub.font = style.font_name;
 490                                                 sub.font_size = FontSize::from_proportional(static_cast<float>(style.font_size) / play_res_y);
 491                                                 sub.colour = style.primary_colour;
 492                                                 sub.effect_colour = style.back_colour;
 493                                                 sub.bold = style.bold;
 494                                                 sub.italic = style.italic;
 495                                                 sub.underline = style.underline;
 496                                                 sub.effect = style.effect;
 497                                                 sub.horizontal_position.reference = style.horizontal_reference;
 498                                                 sub.vertical_position.reference = style.vertical_reference;
 499                                                 if (sub.vertical_position.reference != sub::VERTICAL_CENTRE_OF_SCREEN) {
 500                                                         sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 501                                                 }
 502                                         } else if (event_format[i] == "MarginV") {
 503                                                 if (event[i] != "0" && sub.vertical_position.reference != sub::VERTICAL_CENTRE_OF_SCREEN) {
 504                                                         /* Override the style if its non-zero */
 505                                                         sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 506                                                 }
 507                                         } else if (event_format[i] == "Text") {
 508                                                 for (auto j: parse_line (sub, event[i], play_res_x, play_res_y)) {
 509                                                         _subs.push_back (j);
 510                                                 }
 511                                         }
 512                                 }
 513                         }
 514                 }
 515
 516         }
 517 }