src/ssa_reader.cc

   1 /*
   2     Copyright (C) 2016 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 <locked_sstream.h>
  27 #include <boost/algorithm/string.hpp>
  28 #include <boost/bind.hpp>
  29 #include <boost/foreach.hpp>
  30 #include <iostream>
  31 #include <vector>
  32 #include <cstdio>
  33
  34 using std::string;
  35 using std::vector;
  36 using std::map;
  37 using std::cout;
  38 using std::list;
  39 using boost::optional;
  40 using boost::function;
  41 using namespace boost::algorithm;
  42 using namespace sub;
  43
  44 /** @param s Subtitle string encoded in UTF-8 */
  45 SSAReader::SSAReader (string const & s)
  46 {
  47         locked_stringstream str (s);
  48         this->read (boost::bind (&get_line_stringstream, &str));
  49 }
  50
  51 /** @param f Subtitle file encoded in UTF-8 */
  52 SSAReader::SSAReader (FILE* f)
  53 {
  54         this->read (boost::bind (&get_line_file, f));
  55 }
  56
  57 class Style
  58 {
  59 public:
  60         Style ()
  61                 : font_size (72)
  62                 , primary_colour (255, 255, 255)
  63                 , bold (false)
  64                 , italic (false)
  65                 , underline (false)
  66                 , horizontal_reference (HORIZONTAL_CENTRE_OF_SCREEN)
  67                 , vertical_reference (BOTTOM_OF_SCREEN)
  68                 , vertical_margin (0)
  69         {}
  70
  71         Style (string format_line, string style_line)
  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                 vector<string> keys;
  82                 split (keys, format_line, boost::is_any_of (","));
  83                 vector<string> style;
  84                 split (style, style_line, boost::is_any_of (","));
  85
  86                 SUB_ASSERT (!keys.empty());
  87                 SUB_ASSERT (!style.empty());
  88                 SUB_ASSERT (keys.size() == style.size());
  89
  90                 for (size_t i = 0; i < style.size(); ++i) {
  91                         trim (keys[i]);
  92                         trim (style[i]);
  93                         if (keys[i] == "Name") {
  94                                 name = style[i];
  95                         } else if (keys[i] == "Fontname") {
  96                                 font_name = style[i];
  97                         } else if (keys[i] == "Fontsize") {
  98                                 font_size = raw_convert<int> (style[i]);
  99                         } else if (keys[i] == "PrimaryColour") {
 100                                 primary_colour = colour (raw_convert<int> (style[i]));
 101                         } else if (keys[i] == "BackColour") {
 102                                 back_colour = colour (raw_convert<int> (style[i]));
 103                         } else if (keys[i] == "Bold") {
 104                                 bold = style[i] == "-1";
 105                         } else if (keys[i] == "Italic") {
 106                                 italic = style[i] == "-1";
 107                         } else if (keys[i] == "Underline") {
 108                                 underline = style[i] == "-1";
 109                         } else if (keys[i] == "BorderStyle") {
 110                                 if (style[i] == "1") {
 111                                         effect = SHADOW;
 112                                 }
 113                         } else if (keys[i] == "Alignment") {
 114                                 /* These values from libass' source code */
 115                                 switch ((raw_convert<int> (style[i]) - 1) % 3) {
 116                                 case 0:
 117                                         horizontal_reference = LEFT_OF_SCREEN;
 118                                         break;
 119                                 case 1:
 120                                         horizontal_reference = HORIZONTAL_CENTRE_OF_SCREEN;
 121                                         break;
 122                                 case 2:
 123                                         horizontal_reference = RIGHT_OF_SCREEN;
 124                                         break;
 125                                 }
 126                                 switch (raw_convert<int> (style[i]) & 12) {
 127                                 case 4:
 128                                         vertical_reference = TOP_OF_SCREEN;
 129                                         break;
 130                                 case 8:
 131                                         vertical_reference = VERTICAL_CENTRE_OF_SCREEN;
 132                                         break;
 133                                 case 0:
 134                                         vertical_reference = BOTTOM_OF_SCREEN;
 135                                         break;
 136                                 }
 137                         } else if (keys[i] == "MarginV") {
 138                                 vertical_margin = raw_convert<int> (style[i]);
 139                         }
 140                 }
 141         }
 142
 143         string name;
 144         optional<string> font_name;
 145         int font_size;
 146         Colour primary_colour;
 147         /** outline colour */
 148         optional<Colour> back_colour;
 149         bool bold;
 150         bool italic;
 151         bool underline;
 152         optional<Effect> effect;
 153         HorizontalReference horizontal_reference;
 154         VerticalReference vertical_reference;
 155         int vertical_margin;
 156
 157 private:
 158         Colour colour (int c) const
 159         {
 160                 return Colour (
 161                         ((c & 0x0000ff) >>  0) / 255.0,
 162                         ((c & 0x00ff00) >>  8) / 255.0,
 163                         ((c & 0xff0000) >> 16) / 255.0
 164                         );
 165         }
 166 };
 167
 168 Time
 169 SSAReader::parse_time (string t) const
 170 {
 171         vector<string> bits;
 172         split (bits, t, is_any_of (":."));
 173         SUB_ASSERT (bits.size() == 4);
 174         return Time::from_hms (
 175                 raw_convert<int> (bits[0]),
 176                 raw_convert<int> (bits[1]),
 177                 raw_convert<int> (bits[2]),
 178                 raw_convert<int> (bits[3]) * 10
 179                 );
 180 }
 181
 182 /** @param base RawSubtitle filled in with any required common values.
 183  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 184  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 185  */
 186 list<RawSubtitle>
 187 SSAReader::parse_line (RawSubtitle base, string line, int play_res_x, int play_res_y)
 188 {
 189         enum {
 190                 TEXT,
 191                 STYLE,
 192                 BACKSLASH
 193         } state = TEXT;
 194
 195         list<RawSubtitle> subs;
 196         RawSubtitle current = base;
 197         string style;
 198
 199         if (!current.vertical_position.reference) {
 200                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 201         }
 202
 203         if (!current.vertical_position.proportional) {
 204                 current.vertical_position.proportional = 0;
 205         }
 206
 207         /* We must have a font size, as there could be a margin specified
 208            in pixels and in that case we must know how big the subtitle
 209            lines are to work out the position on screen.
 210         */
 211         if (!current.font_size.points()) {
 212                 current.font_size.set_points (72);
 213         }
 214
 215         /* Count the number of line breaks */
 216         int line_breaks = 0;
 217         if (line.length() > 0) {
 218                 for (size_t i = 0; i < line.length() - 1; ++i) {
 219                         if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 220                                 ++line_breaks;
 221                         }
 222                 }
 223         }
 224
 225         /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
 226         double const line_size = current.font_size.proportional(792) * 1.2;
 227
 228         /* Tweak vertical_position accordingly */
 229         switch (current.vertical_position.reference.get()) {
 230         case TOP_OF_SCREEN:
 231         case TOP_OF_SUBTITLE:
 232                 /* Nothing to do */
 233                 break;
 234         case VERTICAL_CENTRE_OF_SCREEN:
 235                 current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 236                 break;
 237         case BOTTOM_OF_SCREEN:
 238                 current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
 239                 break;
 240         }
 241
 242         for (size_t i = 0; i < line.length(); ++i) {
 243                 char const c = line[i];
 244                 switch (state) {
 245                 case TEXT:
 246                         if (c == '{') {
 247                                 state = STYLE;
 248                         } else if (c == '\\') {
 249                                 state = BACKSLASH;
 250                         } else if (c != '\r' && c != '\n') {
 251                                 current.text += c;
 252                         }
 253                         break;
 254                 case STYLE:
 255                         if (c == '}' || c == '\\') {
 256                                 if (!current.text.empty ()) {
 257                                         subs.push_back (current);
 258                                         current.text = "";
 259                                 }
 260                                 if (style == "\\i1") {
 261                                         current.italic = true;
 262                                 } else if (style == "\\i0" || style == "\\i") {
 263                                         current.italic = false;
 264                                 } else if (style == "\\b1") {
 265                                         current.bold = true;
 266                                 } else if (style == "\\b0") {
 267                                         current.bold = false;
 268                                 } else if (style == "\\u1") {
 269                                         current.underline = true;
 270                                 } else if (style == "\\u0") {
 271                                         current.underline = false;
 272                                 } else if (style == "\\an1" || style == "\\an2" || style == "\\an3") {
 273                                         current.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 274                                 } else if (style == "\\an4" || style == "\\an5" || style == "\\an6") {
 275                                         current.vertical_position.reference = sub::VERTICAL_CENTRE_OF_SCREEN;
 276                                 } else if (style == "\\an7" || style == "\\an8" || style == "\\an9") {
 277                                         current.vertical_position.reference = sub::TOP_OF_SCREEN;
 278                                 } else if (boost::starts_with(style, "\\pos")) {
 279                                         vector<string> bits;
 280                                         boost::algorithm::split (bits, style, boost::is_any_of("(,"));
 281                                         SUB_ASSERT (bits.size() == 3);
 282                                         current.horizontal_position.reference = sub::LEFT_OF_SCREEN;
 283                                         current.horizontal_position.proportional = raw_convert<float>(bits[1]) / play_res_x;
 284                                         current.vertical_position.reference = sub::TOP_OF_SCREEN;
 285                                         current.vertical_position.proportional = raw_convert<float>(bits[2]) / play_res_y;
 286                                 } else if (boost::starts_with(style, "\\fs")) {
 287                                         SUB_ASSERT (style.length() > 3);
 288                                         current.font_size.set_points (raw_convert<int>(style.substr(3)));
 289                                 } else if (boost::starts_with(style, "\\c")) {
 290                                         /* \c&Hbbggrr& */
 291                                         if (style.length() != 11 || style[2] != '&' || style[3] != 'H' || style[10] != '&') {
 292                                                 throw SSAError(String::compose("Badly formatted colour tag %1", style));
 293                                         }
 294                                         int ir, ig, ib;
 295                                         if (sscanf(style.c_str() + 4, "%2x%2x%2x", &ib, &ig, &ir) < 3) {
 296                                                 throw SSAError(String::compose("Badly formatted colour tag %1", style));
 297                                         }
 298                                         current.colour = sub::Colour(ir / 255.0, ig / 255.0, ib / 255.0);
 299                                 }
 300                                 style = "";
 301                         }
 302
 303                         if (c == '}') {
 304                                 state = TEXT;
 305                         } else {
 306                                 style += c;
 307                         }
 308                         break;
 309                 case BACKSLASH:
 310                         if (c == 'n' || c == 'N') {
 311                                 if (!current.text.empty ()) {
 312                                         subs.push_back (current);
 313                                         current.text = "";
 314                                 }
 315                                 /* Move down one line (1.2 times the font size) */
 316                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 317                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 318                                 } else {
 319                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 320                                 }
 321                         }
 322                         state = TEXT;
 323                         break;
 324                 }
 325         }
 326
 327         if (!current.text.empty ()) {
 328                 subs.push_back (current);
 329         }
 330
 331         return subs;
 332 }
 333
 334 void
 335 SSAReader::read (function<optional<string> ()> get_line)
 336 {
 337         enum {
 338                 INFO,
 339                 STYLES,
 340                 EVENTS
 341         } part = INFO;
 342
 343         int play_res_x = 288;
 344         int play_res_y = 288;
 345         map<string, Style> styles;
 346         string style_format_line;
 347         vector<string> event_format;
 348
 349         while (true) {
 350                 optional<string> line = get_line ();
 351                 if (!line) {
 352                         break;
 353                 }
 354
 355                 trim (*line);
 356                 remove_unicode_bom (line);
 357
 358                 if (starts_with (*line, ";") || line->empty ()) {
 359                         continue;
 360                 }
 361
 362                 if (starts_with (*line, "[")) {
 363                         /* Section heading */
 364                         if (line.get() == "[Script Info]") {
 365                                 part = INFO;
 366                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 367                                 part = STYLES;
 368                         } else if (line.get() == "[Events]") {
 369                                 part = EVENTS;
 370                         }
 371                         continue;
 372                 }
 373
 374                 size_t const colon = line->find (":");
 375                 SUB_ASSERT (colon != string::npos);
 376                 string const type = line->substr (0, colon);
 377                 string body = line->substr (colon + 1);
 378                 trim (body);
 379
 380                 switch (part) {
 381                 case INFO:
 382                         if (type == "PlayResX") {
 383                                 play_res_x = raw_convert<int> (body);
 384                         } else if (type == "PlayResY") {
 385                                 play_res_y = raw_convert<int> (body);
 386                         }
 387                         break;
 388                 case STYLES:
 389                         if (type == "Format") {
 390                                 style_format_line = body;
 391                         } else if (type == "Style") {
 392                                 SUB_ASSERT (!style_format_line.empty ());
 393                                 Style s (style_format_line, body);
 394                                 styles[s.name] = s;
 395                         }
 396                         break;
 397                 case EVENTS:
 398                         if (type == "Format") {
 399                                 split (event_format, body, is_any_of (","));
 400                                 BOOST_FOREACH (string& i, event_format) {
 401                                         trim (i);
 402                                 }
 403                         } else if (type == "Dialogue") {
 404                                 SUB_ASSERT (!event_format.empty ());
 405                                 vector<string> event;
 406                                 split (event, body, is_any_of (","));
 407
 408                                 /* There may be commas in the subtitle part; reassemble any extra parts
 409                                    from when we just split it.
 410                                 */
 411                                 while (event.size() > event_format.size()) {
 412                                         string const ex = event.back ();
 413                                         event.pop_back ();
 414                                         event.back() += "," + ex;
 415                                 }
 416
 417                                 SUB_ASSERT (!event.empty());
 418                                 SUB_ASSERT (event_format.size() == event.size());
 419
 420                                 RawSubtitle sub;
 421
 422                                 for (size_t i = 0; i < event.size(); ++i) {
 423                                         trim (event[i]);
 424                                         if (event_format[i] == "Start") {
 425                                                 sub.from = parse_time (event[i]);
 426                                         } else if (event_format[i] == "End") {
 427                                                 sub.to = parse_time (event[i]);
 428                                         } else if (event_format[i] == "Style") {
 429                                                 /* libass trims leading '*'s from style names, commenting that
 430                                                    "they seem to mean literally nothing".  Go figure...
 431                                                 */
 432                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 433                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 434                                                 Style style = styles[event[i]];
 435                                                 sub.font = style.font_name;
 436                                                 sub.font_size = FontSize::from_points (style.font_size);
 437                                                 sub.colour = style.primary_colour;
 438                                                 sub.effect_colour = style.back_colour;
 439                                                 sub.bold = style.bold;
 440                                                 sub.italic = style.italic;
 441                                                 sub.underline = style.underline;
 442                                                 sub.effect = style.effect;
 443                                                 sub.horizontal_position.reference = style.horizontal_reference;
 444                                                 sub.vertical_position.reference = style.vertical_reference;
 445                                                 sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 446                                         } else if (event_format[i] == "MarginV") {
 447                                                 sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 448                                         } else if (event_format[i] == "Text") {
 449                                                 BOOST_FOREACH (sub::RawSubtitle j, parse_line (sub, event[i], play_res_x, play_res_y)) {
 450                                                         _subs.push_back (j);
 451                                                 }
 452                                         }
 453                                 }
 454                         }
 455                 }
 456
 457         }
 458 }