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 <boost/algorithm/string.hpp>
  26 #include <boost/bind.hpp>
  27 #include <boost/foreach.hpp>
  28 #include <sstream>
  29 #include <iostream>
  30 #include <vector>
  31
  32 using std::string;
  33 using std::stringstream;
  34 using std::vector;
  35 using std::map;
  36 using std::cout;
  37 using std::list;
  38 using boost::optional;
  39 using boost::function;
  40 using namespace boost::algorithm;
  41 using namespace sub;
  42
  43 /** @param s Subtitle string encoded in UTF-8 */
  44 SSAReader::SSAReader (string const & s)
  45 {
  46         stringstream str (s);
  47         this->read (boost::bind (&get_line_stringstream, &str));
  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 class Style
  57 {
  58 public:
  59         Style ()
  60                 : font_size (72)
  61                 , primary_colour (255, 255, 255)
  62                 , bold (false)
  63                 , italic (false)
  64                 , vertical_reference (BOTTOM_OF_SCREEN)
  65                 , vertical_margin (0)
  66         {}
  67
  68         Style (string format_line, string style_line)
  69                 : font_size (72)
  70                 , primary_colour (255, 255, 255)
  71                 , bold (false)
  72                 , italic (false)
  73                 , vertical_reference (BOTTOM_OF_SCREEN)
  74                 , vertical_margin (0)
  75         {
  76                 vector<string> keys;
  77                 split (keys, format_line, is_any_of (","));
  78                 vector<string> style;
  79                 split (style, style_line, is_any_of (","));
  80
  81                 SUB_ASSERT (!keys.empty());
  82                 SUB_ASSERT (!style.empty());
  83                 SUB_ASSERT (keys.size() == style.size());
  84
  85                 for (size_t i = 0; i < style.size(); ++i) {
  86                         trim (keys[i]);
  87                         trim (style[i]);
  88                         if (keys[i] == "Name") {
  89                                 name = style[i];
  90                         } else if (keys[i] == "Fontname") {
  91                                 font_name = style[i];
  92                         } else if (keys[i] == "Fontsize") {
  93                                 font_size = raw_convert<int> (style[i]);
  94                         } else if (keys[i] == "PrimaryColour") {
  95                                 primary_colour = colour (raw_convert<int> (style[i]));
  96                         } else if (keys[i] == "BackColour") {
  97                                 back_colour = colour (raw_convert<int> (style[i]));
  98                         } else if (keys[i] == "Bold") {
  99                                 bold = style[i] == "-1";
 100                         } else if (keys[i] == "Italic") {
 101                                 italic = style[i] == "-1";
 102                         } else if (keys[i] == "BorderStyle") {
 103                                 if (style[i] == "1") {
 104                                         effect = SHADOW;
 105                                 }
 106                         } else if (keys[i] == "Alignment") {
 107                                 /* These values from libass' source code */
 108                                 switch (raw_convert<int> (style[i]) & 12) {
 109                                 case 4:
 110                                         vertical_reference = TOP_OF_SCREEN;
 111                                         break;
 112                                 case 8:
 113                                         vertical_reference = CENTRE_OF_SCREEN;
 114                                         break;
 115                                 case 0:
 116                                         vertical_reference = BOTTOM_OF_SCREEN;
 117                                         break;
 118                                 }
 119                         } else if (keys[i] == "MarginV") {
 120                                 vertical_margin = raw_convert<int> (style[i]);
 121                         }
 122                 }
 123         }
 124
 125         string name;
 126         optional<string> font_name;
 127         int font_size;
 128         Colour primary_colour;
 129         /** outline colour */
 130         optional<Colour> back_colour;
 131         bool bold;
 132         bool italic;
 133         optional<Effect> effect;
 134         VerticalReference vertical_reference;
 135         int vertical_margin;
 136
 137 private:
 138         Colour colour (int c) const
 139         {
 140                 return Colour (
 141                         ((c & 0x0000ff) >>  0) / 255.0,
 142                         ((c & 0x00ff00) >>  8) / 255.0,
 143                         ((c & 0xff0000) >> 16) / 255.0
 144                         );
 145         }
 146 };
 147
 148 Time
 149 SSAReader::parse_time (string t) const
 150 {
 151         vector<string> bits;
 152         split (bits, t, is_any_of (":."));
 153         SUB_ASSERT (bits.size() == 4);
 154         return Time::from_hms (
 155                 raw_convert<int> (bits[0]),
 156                 raw_convert<int> (bits[1]),
 157                 raw_convert<int> (bits[2]),
 158                 raw_convert<int> (bits[3]) * 10
 159                 );
 160 }
 161
 162 /** @param base RawSubtitle filled in with any required common values.
 163  *  @param line SSA line string (i.e. just the subtitle, possibly with embedded stuff)
 164  *  @return List of RawSubtitles to represent line with vertical reference TOP_OF_SUBTITLE.
 165  */
 166 list<RawSubtitle>
 167 SSAReader::parse_line (RawSubtitle base, string line)
 168 {
 169         enum {
 170                 TEXT,
 171                 STYLE,
 172                 BACKSLASH
 173         } state = TEXT;
 174
 175         list<RawSubtitle> subs;
 176         RawSubtitle current = base;
 177         string style;
 178
 179         if (!current.vertical_position.reference) {
 180                 current.vertical_position.reference = BOTTOM_OF_SCREEN;
 181         }
 182
 183         if (!current.vertical_position.proportional) {
 184                 current.vertical_position.proportional = 0;
 185         }
 186
 187         /* We must have a font size, as there could be a margin specified
 188            in pixels and in that case we must know how big the subtitle
 189            lines are to work out the position on screen.
 190         */
 191         if (!current.font_size.points()) {
 192                 current.font_size.set_points (72);
 193         }
 194
 195         /* Count the number of line breaks */
 196         int line_breaks = 0;
 197         for (size_t i = 0; i < line.length() - 1; ++i) {
 198                 if (line[i] == '\\' && (line[i+1] == 'n' || line[i+1] == 'N')) {
 199                         ++line_breaks;
 200                 }
 201         }
 202
 203         /* Imagine that the screen is 792 points (i.e. 11 inches) high (as with DCP) */
 204         double const line_size = current.font_size.proportional(792) * 1.2;
 205
 206         /* Tweak vertical_position accordingly */
 207         switch (current.vertical_position.reference.get()) {
 208         case TOP_OF_SCREEN:
 209         case TOP_OF_SUBTITLE:
 210                 /* Nothing to do */
 211                 break;
 212         case CENTRE_OF_SCREEN:
 213                 current.vertical_position.proportional = current.vertical_position.proportional.get() - ((line_breaks + 1) * line_size) / 2;
 214                 break;
 215         case BOTTOM_OF_SCREEN:
 216                 current.vertical_position.proportional = current.vertical_position.proportional.get() + line_breaks * line_size;
 217                 break;
 218         }
 219
 220         for (size_t i = 0; i < line.length(); ++i) {
 221                 char const c = line[i];
 222                 switch (state) {
 223                 case TEXT:
 224                         if (c == '{') {
 225                                 state = STYLE;
 226                         } else if (c == '\\') {
 227                                 state = BACKSLASH;
 228                         } else if (c != '\r' && c != '\n') {
 229                                 current.text += c;
 230                         }
 231                         break;
 232                 case STYLE:
 233                         if (c == '}') {
 234                                 if (!current.text.empty ()) {
 235                                         subs.push_back (current);
 236                                         current.text = "";
 237                                 }
 238                                 if (style == "\\i1") {
 239                                         current.italic = true;
 240                                 } else if (style == "\\i0" || style == "\\i") {
 241                                         current.italic = false;
 242                                 } else if (style == "\\b1") {
 243                                         current.bold = true;
 244                                 } else if (style == "\\b0") {
 245                                         current.bold = false;
 246                                 } else if (style == "\\an1" || style == "\\an2" || style == "\\an3") {
 247                                         current.vertical_position.reference = sub::BOTTOM_OF_SCREEN;
 248                                 } else if (style == "\\an4" || style == "\\an5" || style == "\\an6") {
 249                                         current.vertical_position.reference = sub::CENTRE_OF_SCREEN;
 250                                 } else if (style == "\\an7" || style == "\\an8" || style == "\\an9") {
 251                                         current.vertical_position.reference = sub::TOP_OF_SCREEN;
 252                                 }
 253                                 style = "";
 254                                 state = TEXT;
 255                         } else {
 256                                 style += c;
 257                         }
 258                         break;
 259                 case BACKSLASH:
 260                         if (c == 'n' || c == 'N') {
 261                                 if (!current.text.empty ()) {
 262                                         subs.push_back (current);
 263                                         current.text = "";
 264                                 }
 265                                 /* Move down one line (1.2 times the font size) */
 266                                 if (current.vertical_position.reference.get() == BOTTOM_OF_SCREEN) {
 267                                         current.vertical_position.proportional = current.vertical_position.proportional.get() - line_size;
 268                                 } else {
 269                                         current.vertical_position.proportional = current.vertical_position.proportional.get() + line_size;
 270                                 }
 271                         }
 272                         state = TEXT;
 273                         break;
 274                 }
 275         }
 276
 277         if (!current.text.empty ()) {
 278                 subs.push_back (current);
 279         }
 280
 281         return subs;
 282 }
 283
 284 void
 285 SSAReader::read (function<optional<string> ()> get_line)
 286 {
 287         enum {
 288                 INFO,
 289                 STYLES,
 290                 EVENTS
 291         } part = INFO;
 292
 293         int play_res_y = 288;
 294         map<string, Style> styles;
 295         string style_format_line;
 296         vector<string> event_format;
 297
 298         while (true) {
 299                 optional<string> line = get_line ();
 300                 if (!line) {
 301                         break;
 302                 }
 303
 304                 trim (*line);
 305                 remove_unicode_bom (line);
 306
 307                 if (starts_with (*line, ";") || line->empty ()) {
 308                         continue;
 309                 }
 310
 311                 if (starts_with (*line, "[")) {
 312                         /* Section heading */
 313                         if (line.get() == "[Script Info]") {
 314                                 part = INFO;
 315                         } else if (line.get() == "[V4 Styles]" || line.get() == "[V4+ Styles]") {
 316                                 part = STYLES;
 317                         } else if (line.get() == "[Events]") {
 318                                 part = EVENTS;
 319                         }
 320                         continue;
 321                 }
 322
 323                 size_t const colon = line->find (":");
 324                 SUB_ASSERT (colon != string::npos);
 325                 string const type = line->substr (0, colon);
 326                 string body = line->substr (colon + 1);
 327                 trim (body);
 328
 329                 switch (part) {
 330                 case INFO:
 331                         if (type == "PlayResY") {
 332                                 play_res_y = raw_convert<int> (body);
 333                         }
 334                         break;
 335                 case STYLES:
 336                         if (type == "Format") {
 337                                 style_format_line = body;
 338                         } else if (type == "Style") {
 339                                 SUB_ASSERT (!style_format_line.empty ());
 340                                 Style s (style_format_line, body);
 341                                 styles[s.name] = s;
 342                         }
 343                         break;
 344                 case EVENTS:
 345                         if (type == "Format") {
 346                                 split (event_format, body, is_any_of (","));
 347                                 BOOST_FOREACH (string& i, event_format) {
 348                                         trim (i);
 349                                 }
 350                         } else if (type == "Dialogue") {
 351                                 SUB_ASSERT (!event_format.empty ());
 352                                 vector<string> event;
 353                                 split (event, body, is_any_of (","));
 354
 355                                 /* There may be commas in the subtitle part; reassemble any extra parts
 356                                    from when we just split it.
 357                                 */
 358                                 while (event.size() > event_format.size()) {
 359                                         string const ex = event.back ();
 360                                         event.pop_back ();
 361                                         event.back() += "," + ex;
 362                                 }
 363
 364                                 SUB_ASSERT (!event.empty());
 365                                 SUB_ASSERT (event_format.size() == event.size());
 366
 367                                 RawSubtitle sub;
 368
 369                                 for (size_t i = 0; i < event.size(); ++i) {
 370                                         trim (event[i]);
 371                                         if (event_format[i] == "Start") {
 372                                                 sub.from = parse_time (event[i]);
 373                                         } else if (event_format[i] == "End") {
 374                                                 sub.to = parse_time (event[i]);
 375                                         } else if (event_format[i] == "Style") {
 376                                                 /* libass trims leading '*'s from style names, commenting that
 377                                                    "they seem to mean literally nothing".  Go figure...
 378                                                 */
 379                                                 trim_left_if (event[i], boost::is_any_of ("*"));
 380                                                 SUB_ASSERT (styles.find(event[i]) != styles.end());
 381                                                 Style style = styles[event[i]];
 382                                                 sub.font = style.font_name;
 383                                                 sub.font_size = FontSize::from_points (style.font_size);
 384                                                 sub.colour = style.primary_colour;
 385                                                 sub.effect_colour = style.back_colour;
 386                                                 sub.bold = style.bold;
 387                                                 sub.italic = style.italic;
 388                                                 sub.effect = style.effect;
 389                                                 sub.vertical_position.reference = style.vertical_reference;
 390                                                 sub.vertical_position.proportional = float(style.vertical_margin) / play_res_y;
 391                                         } else if (event_format[i] == "MarginV") {
 392                                                 sub.vertical_position.proportional = raw_convert<float>(event[i]) / play_res_y;
 393                                         } else if (event_format[i] == "Text") {
 394                                                 BOOST_FOREACH (sub::RawSubtitle j, parse_line (sub, event[i])) {
 395                                                         _subs.push_back (j);
 396                                                 }
 397                                         }
 398                                 }
 399                         }
 400                 }
 401
 402         }
 403 }