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