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