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