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                                 if (style[i] == "7" || style[i] == "8" || style[i] == "9") {
 129                                         vertical_reference = TOP_OF_SCREEN;
 130                                 } else if (style[i] == "4" || style[i] == "5" || style[i] == "6") {
 131                                         vertical_reference = VERTICAL_CENTRE_OF_SCREEN;
 132                                 } else {
 133                                         vertical_reference = BOTTOM_OF_SCREEN;
 134                                 }
 135                                 if (style[i] == "1" || style[i] == "4" || style[i] == "7") {
 136                                         horizontal_reference = LEFT_OF_SCREEN;
 137                                 } else if (style[i] == "3" || style[i] == "6" || style[i] == "9") {
 138                                         horizontal_reference = RIGHT_OF_SCREEN;
 139                                 } else {
 140                                         horizontal_reference = HORIZONTAL_CENTRE_OF_SCREEN;
 141                                 }
 142                         } else if (keys[i] == "MarginV") {
 143                                 vertical_margin = raw_convert<int> (style[i]);
 144                         }
 145                 }
 146         }
 147
 148         string name;
 149         optional<string> font_name;
 150         int font_size;
 151         Colour primary_colour;
 152         /** outline colour */
 153         optional<Colour> back_colour;
 154         bool bold;
 155         bool italic;
 156         bool underline;
 157         optional<Effect> effect;
 158         HorizontalReference horizontal_reference;
 159         VerticalReference vertical_reference;
 160         int vertical_margin;
 161
 162 private:
 163         Colour colour (string c) const
 164         {
 165                 if (c.length() > 0 && c[0] == '&') {
 166                         /* &Hbbggrr or &Haabbggrr */
 167                         return h_colour (c);
 168                 } else {
 169                         /* integer */
 170                         int i = raw_convert<int>(c);
 171                         return Colour (
 172                                 ((i & 0x0000ff) >>  0) / 255.0,
 173                                 ((i & 0x00ff00) >>  8) / 255.0,
 174                                 ((i & 0xff0000) >> 16) / 255.0
 175                                 );
 176                 }
 177         }
 178 };
 179
 180 Time
 181 SSAReader::parse_time (string t) const
 182 {
 183         vector<string> bits;
 184         split (bits, t, is_any_of (":."));
 185         SUB_ASSERT (bits.size() == 4);
 186         return Time::from_hms (
 187                 raw_convert<int> (bits[0]),
 188                 raw_convert<int> (bits[1]),
 189                 raw_convert<int> (bits[2]),
 190                 raw_convert<int> (bits[3]) * 10
 191                 );
 192 }
 193
 194 void
 195 SSAReader::parse_style (RawSubtitle& sub, string style, int play_res_x, int play_res_y)
 196 {
 197         if (style == "\\i1") {
 198                 sub.italic = true;
 199         } else if (style == "\\i0" || style == "\\i") {
 200                 sub.italic = false;
 201         } else if (style == "\\b1") {
 202                 sub.bold = true;
 203         } else if (style == "\\b0") {
 204                 sub.bold = false;
 205         } else if (style == "\\u1") {
 206                 sub.underline = true;
 207         } else if (style == "\\u0") {
 208                 sub.underline = false;
 209         } else if (style == "\\an1") {
 210                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 211                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 212         } else if (style == "\\an2") {
 213                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 214                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 215         } else if (style == "\\an3") {
 216                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 217                 sub.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 218         } else if (style == "\\an4") {
 219                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 220                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 221         } else if (style == "\\an5") {
 222                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 223                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 224         } else if (style == "\\an6") {
 225                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 226                 sub.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 227         } else if (style == "\\an7") {
 228                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 229                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 230         } else if (style == "\\an8") {
 231                 sub.horizontal_position.reference = sub::HORIZONTAL_CENTRE_OF_SCREEN;
 232                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 233         } else if (style == "\\an9") {
 234                 sub.horizontal_position.reference = sub::RIGHT_OF_SCREEN;
 235                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 236         } else if (boost::starts_with(style, "\\pos")) {
 237                 vector<string> bits;
 238                 boost::algorithm::split (bits, style, boost::is_any_of("(,"));
 239                 SUB_ASSERT (bits.size() == 3);
 240                 sub.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 241                 sub.horizontal_position.proportional = raw_convert<float>(bits[1]) / play_res_x;
 242                 sub.vertical_position.reference = sub::TOP_OF_SCREEN;
 243                 sub.vertical_position.proportional = raw_convert<float>(bits[2]) / play_res_y;
 244         } else if (boost::starts_with(style, "\\fs")) {
 245                 SUB_ASSERT (style.length() > 3);
 246                 sub.font_size.set_points (raw_convert<int>(style.substr(3)));
 247         } else if (boost::starts_with(style, "\\c")) {
 248                 /* \c&Hbbggrr& */
 249                 if (style.length() <= 2) {
 250                         throw SSAError(String::compose("Badly formatted colour tag %1", style));
 251                 }
 252                 sub.colour = h_colour (style.substr(2, style.length() - 3));
 253         }
 254 }
 255
 256 /** @param base RawSubtitle filled in with any required common values.
 257  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 258  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 259  */
 260 vector<RawSubtitle>
 261 SSAReader::parse_line (RawSubtitle base, string line, int play_res_x, int play_res_y)
 262 {
 263         enum {
 264                 TEXT,
 265                 STYLE,
 266                 BACKSLASH
 267         } state = TEXT;
 268
 269         vector<RawSubtitle> subs;
 270         RawSubtitle current = base;
 271         string style;
 272
 273         if (!current.vertical_position.reference) {
 274                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 275         }
 276
 277         if (!current.vertical_position.proportional) {
 278                 current.vertical_position.proportional = 0;
 279         }
 280
 281         /* We must have a font size, as there could be a margin specified
 282            in pixels and in that case we must know how big the subtitle
 283            lines are to work out the position on screen.
 284         */
 285         if (!current.font_size.points()) {
 286                 current.font_size.set_points (72);
 287         }
 288
 289         /* Count the number of line breaks */
 290         int line_breaks = 0;
 291         if (line.length() > 0) {
 292                 for (size_t i = 0; i < line.length() - 1; ++i) {
 293                         if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 294                                 ++line_breaks;
 295                         }
 296                 }
 297         }
 298
 299         /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
 300         double const line_size = current.font_size.proportional(792) * 1.2;
 301
 302         /* Tweak vertical_position accordingly */
 303         switch (current.vertical_position.reference.get()) {
 304         case TOP_OF_SCREEN:
 305         case TOP_OF_SUBTITLE:
 306                 /* Nothing to do */
 307                 break;
 308         case VERTICAL_CENTRE_OF_SCREEN:
 309                 current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 310                 break;
 311         case BOTTOM_OF_SCREEN:
 312                 current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
 313                 break;
 314         }
 315
 316         for (size_t i = 0; i < line.length(); ++i) {
 317                 char const c = line[i];
 318                 switch (state) {
 319                 case TEXT:
 320                         if (c == '{') {
 321                                 state = STYLE;
 322                         } else if (c == '\\') {
 323                                 state = BACKSLASH;
 324                         } else if (c != '\r' && c != '\n') {
 325                                 current.text += c;
 326                         }
 327                         break;
 328                 case STYLE:
 329                         if (c == '}' || c == '\\') {
 330                                 if (!current.text.empty ()) {
 331                                         subs.push_back (current);
 332                                         current.text = "";
 333                                 }
 334                                 parse_style (current, style, play_res_x, play_res_y);
 335                                 style = "";
 336                         }
 337
 338                         if (c == '}') {
 339                                 state = TEXT;
 340                         } else {
 341                                 style += c;
 342                         }
 343                         break;
 344                 case BACKSLASH:
 345                         if (c == 'n' || c == 'N') {
 346                                 if (!current.text.empty ()) {
 347                                         subs.push_back (current);
 348                                         current.text = "";
 349                                 }
 350                                 /* Move down one line (1.2 times the font size) */
 351                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 352                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 353                                 } else {
 354                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 355                                 }
 356                         }
 357                         state = TEXT;
 358                         break;
 359                 }
 360         }
 361
 362         if (!current.text.empty ()) {
 363                 subs.push_back (current);
 364         }
 365
 366         return subs;
 367 }
 368
 369 void
 370 SSAReader::read (function<optional<string> ()> get_line)
 371 {
 372         enum {
 373                 INFO,
 374                 STYLES,
 375                 EVENTS
 376         } part = INFO;
 377
 378         int play_res_x = 288;
 379         int play_res_y = 288;
 380         map<string, Style> styles;
 381         string style_format_line;
 382         vector<string> event_format;
 383
 384         while (true) {
 385                 optional<string> line = get_line ();
 386                 if (!line) {
 387                         break;
 388                 }
 389
 390                 trim (*line);
 391                 remove_unicode_bom (line);
 392
 393                 if (starts_with (*line, ";") || line->empty ()) {
 394                         continue;
 395                 }
 396
 397                 if (starts_with (*line, "[")) {
 398                         /* Section heading */
 399                         if (line.get() == "[Script Info]") {
 400                                 part = INFO;
 401                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 402                                 part = STYLES;
 403                         } else if (line.get() == "[Events]") {
 404                                 part = EVENTS;
 405                         }
 406                         continue;
 407                 }
 408
 409                 size_t const colon = line->find (":");
 410                 SUB_ASSERT (colon != string::npos);
 411                 string const type = line->substr (0, colon);
 412                 string body = line->substr (colon + 1);
 413                 trim (body);
 414
 415                 switch (part) {
 416                 case INFO:
 417                         if (type == "PlayResX") {
 418                                 play_res_x = raw_convert<int> (body);
 419                         } else if (type == "PlayResY") {
 420                                 play_res_y = raw_convert<int> (body);
 421                         }
 422                         break;
 423                 case STYLES:
 424                         if (type == "Format") {
 425                                 style_format_line = body;
 426                         } else if (type == "Style") {
 427                                 SUB_ASSERT (!style_format_line.empty ());
 428                                 Style s (style_format_line, body);
 429                                 styles[s.name] = s;
 430                         }
 431                         break;
 432                 case EVENTS:
 433                         if (type == "Format") {
 434                                 split (event_format, body, is_any_of (","));
 435                                 for (auto& i: event_format) {
 436                                         trim (i);
 437                                 }
 438                         } else if (type == "Dialogue") {
 439                                 SUB_ASSERT (!event_format.empty ());
 440                                 vector<string> event;
 441                                 split (event, body, is_any_of (","));
 442
 443                                 /* There may be commas in the subtitle part; reassemble any extra parts
 444                                    from when we just split it.
 445                                 */
 446                                 while (event.size() > event_format.size()) {
 447                                         string const ex = event.back ();
 448                                         event.pop_back ();
 449                                         event.back() += "," + ex;
 450                                 }
 451
 452                                 SUB_ASSERT (!event.empty());
 453                                 SUB_ASSERT (event_format.size() == event.size());
 454
 455                                 RawSubtitle sub;
 456
 457                                 for (size_t i = 0; i < event.size(); ++i) {
 458                                         trim (event[i]);
 459                                         if (event_format[i] == "Start") {
 460                                                 sub.from = parse_time (event[i]);
 461                                         } else if (event_format[i] == "End") {
 462                                                 sub.to = parse_time (event[i]);
 463                                         } else if (event_format[i] == "Style") {
 464                                                 /* libass trims leading '*'s from style names, commenting that
 465                                                    "they seem to mean literally nothing".  Go figure...
 466                                                 */
 467                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 468                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 469                                                 Style style = styles[event[i]];
 470                                                 sub.font = style.font_name;
 471                                                 sub.font_size = FontSize::from_points (style.font_size);
 472                                                 sub.colour = style.primary_colour;
 473                                                 sub.effect_colour = style.back_colour;
 474                                                 sub.bold = style.bold;
 475                                                 sub.italic = style.italic;
 476                                                 sub.underline = style.underline;
 477                                                 sub.effect = style.effect;
 478                                                 sub.horizontal_position.reference = style.horizontal_reference;
 479                                                 sub.vertical_position.reference = style.vertical_reference;
 480                                                 sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 481                                         } else if (event_format[i] == "MarginV") {
 482                                                 sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 483                                         } else if (event_format[i] == "Text") {
 484                                                 for (auto j: parse_line (sub, event[i], play_res_x, play_res_y)) {
 485                                                         _subs.push_back (j);
 486                                                 }
 487                                         }
 488                                 }
 489                         }
 490                 }
 491
 492         }
 493 }