src/subrip_reader.cc

   1 /*
   2     Copyright (C) 2014-2020 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 /** @file  src/subrip_reader.cc
  21  *  @brief SubripReader class.
  22  */
  23
  24
  25 #include "compose.hpp"
  26 #include "exceptions.h"
  27 #include "raw_convert.h"
  28 #include "ssa_reader.h"
  29 #include "sub_assert.h"
  30 #include "subrip_reader.h"
  31 #include "util.h"
  32 #include <boost/algorithm/string.hpp>
  33 #include <boost/algorithm/string_regex.hpp>
  34 #include <boost/bind.hpp>
  35 #include <boost/lexical_cast.hpp>
  36 #include <boost/regex.hpp>
  37 #include <cstdio>
  38 #include <iostream>
  39 #include <vector>
  40
  41
  42 using std::string;
  43 using std::vector;
  44 using std::cout;
  45 using std::hex;
  46 using boost::lexical_cast;
  47 using boost::to_upper;
  48 using boost::optional;
  49 using boost::function;
  50 using boost::algorithm::replace_all;
  51 using namespace sub;
  52
  53 /** @param s Subtitle string encoded in UTF-8 */
  54 SubripReader::SubripReader (string s)
  55 {
  56         this->read (boost::bind(&get_line_string, &s));
  57 }
  58
  59 /** @param f Subtitle file encoded in UTF-8 */
  60 SubripReader::SubripReader (FILE* f)
  61 {
  62         this->read (boost::bind (&get_line_file, f));
  63 }
  64
  65 void
  66 SubripReader::read (function<optional<string> ()> get_line)
  67 {
  68         enum {
  69                 COUNTER,
  70                 METADATA,
  71                 CONTENT
  72         } state = COUNTER;
  73
  74         RawSubtitle rs;
  75
  76         rs.vertical_position.line = 0;
  77         rs.vertical_position.reference = TOP_OF_SUBTITLE;
  78
  79         while (true) {
  80                 auto line = get_line ();
  81                 if (!line) {
  82                         break;
  83                 }
  84
  85                 trim_right_if (*line, boost::is_any_of ("\n\r"));
  86                 remove_unicode_bom (line);
  87
  88                 /* Keep some history in case there is an error to report */
  89                 _context.push_back (*line);
  90                 if (_context.size() > 5) {
  91                         _context.pop_front ();
  92                 }
  93
  94                 switch (state) {
  95                 case COUNTER:
  96                 {
  97                         if (line->empty ()) {
  98                                 /* a blank line at the start is ok */
  99                                 break;
 100                         }
 101
 102                         state = METADATA;
 103
 104                         /* Reset stuff that should not persist across separate subtitles */
 105                         rs.bold = false;
 106                         rs.italic = false;
 107                         rs.underline = false;
 108                         rs.vertical_position.line = 0;
 109                         rs.vertical_position.reference = TOP_OF_SUBTITLE;
 110                 }
 111                 break;
 112                 case METADATA:
 113                 {
 114                         vector<string> p;
 115
 116                         /* Further trim this line, removing spaces from the end */
 117                         trim_right_if (*line, boost::is_any_of (" "));
 118
 119                         boost::algorithm::split (p, *line, boost::algorithm::is_any_of (" "), boost::token_compress_on);
 120                         if (p.size() != 3 && p.size() != 7) {
 121                                 for (int i = 0; i < 2; ++i) {
 122                                         optional<string> ex = get_line ();
 123                                         if (ex) {
 124                                                 _context.push_back (*ex);
 125                                         }
 126                                 }
 127                                 throw SubripError (*line, "a time/position line", _context);
 128                         }
 129
 130                         string expected;
 131                         auto from = convert_time(p[0], ",", &expected);
 132                         if (!from) {
 133                                 throw SubripError(p[0], expected, _context);
 134                         }
 135                         rs.from = *from;
 136
 137                         auto to = convert_time(p[2], ",", &expected);
 138                         if (!to) {
 139                                 throw SubripError(p[2], expected, _context);
 140                         }
 141                         rs.to = *to;
 142
 143                         /* XXX: should not ignore coordinate specifications */
 144
 145                         state = CONTENT;
 146                         break;
 147                 }
 148                 case CONTENT:
 149                         if (line->empty ()) {
 150                                 state = COUNTER;
 151                         } else {
 152                                 vector<string> sub_lines;
 153                                 /* Split up this line on unicode "LINE SEPARATOR".  This feels hacky but also
 154                                  * the least unpleasant place to do it.
 155                                  */
 156                                 boost::algorithm::split_regex(sub_lines, *line, boost::regex("\xe2\x80\xa8"));
 157                                 for (auto sub_line: sub_lines) {
 158                                         convert_line(sub_line, rs);
 159                                         rs.vertical_position.line = rs.vertical_position.line.get() + 1;
 160                                         rs.text.clear();
 161                                 }
 162                         }
 163                         break;
 164                 }
 165         }
 166 }
 167
 168 optional<Time>
 169 SubripReader::convert_time(string t, string milliseconds_separator, string* expected)
 170 {
 171         auto report_expected = [expected](string const& s) {
 172                 if (expected) {
 173                         *expected = s;
 174                 }
 175         };
 176
 177         vector<string> a;
 178         boost::algorithm::split (a, t, boost::is_any_of (":"));
 179         if (a.size() != 3) {
 180                 report_expected("time in the format h:m:s,ms");
 181                 return {};
 182         }
 183
 184         vector<string> b;
 185         boost::algorithm::split(b, a[2], boost::is_any_of(milliseconds_separator));
 186         if (b.size() != 2) {
 187                 report_expected(String::compose("time in the format h:m:s%1ms", milliseconds_separator));
 188                 return {};
 189         }
 190
 191         int h, m, s, ms;
 192
 193         try {
 194                 h = lexical_cast<int>(a[0]);
 195         } catch (boost::bad_lexical_cast &) {
 196                 report_expected("integer hour value");
 197                 return {};
 198         }
 199
 200         try {
 201                 m = lexical_cast<int>(a[1]);
 202         } catch (boost::bad_lexical_cast &) {
 203                 report_expected("integer minute value");
 204                 return {};
 205         }
 206
 207         try {
 208                 s = lexical_cast<int>(b[0]);
 209         } catch (boost::bad_lexical_cast &) {
 210                 report_expected("integer second value");
 211                 return {};
 212         }
 213
 214         try {
 215                 ms = lexical_cast<int>(b[1]);
 216         } catch (boost::bad_lexical_cast &) {
 217                 report_expected("integer millisecond value");
 218                 return {};
 219         }
 220
 221         return Time::from_hms (h, m, s, ms);
 222 }
 223
 224 void
 225 SubripReader::convert_line (string t, RawSubtitle& p)
 226 {
 227         vector<Colour> colours;
 228         colours.push_back (Colour (1, 1, 1));
 229
 230         auto has_next = [](string line, size_t& index, string s) {
 231                 boost::to_lower(s);
 232                 auto next = line.substr(index, s.size());
 233                 boost::to_lower(next);
 234                 if (next != s) {
 235                         return false;
 236                 }
 237
 238                 index += s.size();
 239                 return true;
 240         };
 241
 242         size_t i = 0;
 243         while (i < t.size()) {
 244                 if (has_next(t, i, "<b>") || has_next(t, i, "{b}")) {
 245                         maybe_content (p);
 246                         p.bold = true;
 247                 } else if (has_next(t, i, "</b>") || has_next(t, i, "{/b}")) {
 248                         maybe_content (p);
 249                         p.bold = false;
 250                 } else if (has_next(t, i, "<i>") || has_next(t, i, "{i}")) {
 251                         maybe_content (p);
 252                         p.italic = true;
 253                 } else if (has_next(t, i, "</i>") || has_next(t, i, "{/i}")) {
 254                         maybe_content (p);
 255                         p.italic = false;
 256                 } else if (has_next(t, i, "<u>") || has_next(t, i, "{u}")) {
 257                         maybe_content (p);
 258                         p.underline = true;
 259                 } else if (has_next(t, i, "</u>") || has_next(t, i, "{/u}")) {
 260                         maybe_content (p);
 261                         p.underline = false;
 262                 } else if (has_next(t, i, "<font") || has_next(t, i, "<Font")) {
 263                         maybe_content (p);
 264                         boost::regex re (".*color=\"?#([[:xdigit:]]+)\"?");
 265                         boost::smatch match;
 266                         string tag;
 267                         while (i < t.size() && t[i] != '>') {
 268                                 tag += t[i];
 269                                 ++i;
 270                         }
 271                         ++i;
 272                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 273                                 p.colour = Colour::from_rgb_hex (match[1]);
 274                                 colours.push_back (p.colour);
 275                         } else {
 276                                 re = boost::regex (
 277                                         ".*color=\"rgba\\("
 278                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 279                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 280                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*,"
 281                                         "[[:space:]]*([[:digit:]]+)[[:space:]]*"
 282                                         "\\)\""
 283                                         );
 284                                 if (boost::regex_search (tag, match, re) && match.size() == 5) {
 285                                         p.colour.r = raw_convert<int>(string(match[1])) / 255.0;
 286                                         p.colour.g = raw_convert<int>(string(match[2])) / 255.0;
 287                                         p.colour.b = raw_convert<int>(string(match[3])) / 255.0;
 288                                         colours.push_back (p.colour);
 289                                 } else {
 290                                         throw SubripError (tag, "a colour in the format #rrggbb or rgba(rr,gg,bb,aa)", _context);
 291                                 }
 292                         }
 293                 } else if (has_next(t, i, "</font>")) {
 294                         maybe_content (p);
 295                         SUB_ASSERT (!colours.empty());
 296                         colours.pop_back ();
 297                         p.colour = colours.back ();
 298                 } else if (has_next(t, i, "{\\")) {
 299                         string ssa = "\\";
 300                         while (i < t.size() && t[i] != '}') {
 301                                 ssa += t[i];
 302                                 ++i;
 303                         }
 304                         ++i;
 305                         SSAReader::parse_style (p, ssa, 288, 288, Colour(1, 1, 1));
 306                 } else {
 307                         p.text += t[i];
 308                         ++i;
 309                 }
 310         }
 311
 312         /* Strip Unicode U+202B (right-to-left embedding) as sometimes it is rendered
 313            as a missing character.  This may be a hack.
 314         */
 315         replace_all (p.text, "\xe2\x80\xab", "");
 316
 317         maybe_content (p);
 318 }
 319
 320 /* Push p into _subs if it has some text, and clear the text out of p */
 321 void
 322 SubripReader::maybe_content (RawSubtitle& p)
 323 {
 324         if (!p.text.empty ()) {
 325                 _subs.push_back (p);
 326                 p.text.clear ();
 327         }
 328 }