src/subrip_reader.cc

   1 /*
   2     Copyright (C) 2014-2015 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 <locked_sstream.h>
  28 #include <boost/algorithm/string.hpp>
  29 #include <boost/lexical_cast.hpp>
  30 #include <boost/regex.hpp>
  31 #include <boost/bind.hpp>
  32 #include <cstdio>
  33 #include <vector>
  34 #include <iostream>
  35
  36 using std::string;
  37 using std::vector;
  38 using std::list;
  39 using std::cout;
  40 using std::hex;
  41 using boost::lexical_cast;
  42 using boost::to_upper;
  43 using boost::optional;
  44 using boost::function;
  45 using boost::algorithm::replace_all;
  46 using namespace sub;
  47
  48 /** @param s Subtitle string encoded in UTF-8 */
  49 SubripReader::SubripReader (string const & s)
  50 {
  51         locked_stringstream str (s);
  52         this->read (boost::bind (&get_line_stringstream, &str));
  53 }
  54
  55 /** @param f Subtitle file encoded in UTF-8 */
  56 SubripReader::SubripReader (FILE* f)
  57 {
  58         this->read (boost::bind (&get_line_file, f));
  59 }
  60
  61 void
  62 SubripReader::read (function<optional<string> ()> get_line)
  63 {
  64         enum {
  65                 COUNTER,
  66                 METADATA,
  67                 CONTENT
  68         } state = COUNTER;
  69
  70         RawSubtitle rs;
  71
  72         /* This reader extracts no information about where the subtitle
  73            should be on screen, so its reference is TOP_OF_SUBTITLE.
  74         */
  75         rs.vertical_position.line = 0;
  76         rs.vertical_position.reference = TOP_OF_SUBTITLE;
  77
  78         while (true) {
  79                 optional<string> line = get_line ();
  80                 if (!line) {
  81                         break;
  82                 }
  83
  84                 trim_right_if (*line, boost::is_any_of ("\n\r"));
  85                 remove_unicode_bom (line);
  86
  87                 /* Keep some history in case there is an error to report */
  88                 _context.push_back (*line);
  89                 if (_context.size() > 5) {
  90                         _context.pop_front ();
  91                 }
  92
  93                 switch (state) {
  94                 case COUNTER:
  95                 {
  96                         if (line->empty ()) {
  97                                 /* a blank line at the start is ok */
  98                                 break;
  99                         }
 100
 101                         state = METADATA;
 102
 103                         /* Reset stuff that should not persist across separate subtitles */
 104                         rs.bold = false;
 105                         rs.italic = false;
 106                         rs.underline = false;
 107                         rs.vertical_position.line = 0;
 108                 }
 109                 break;
 110                 case METADATA:
 111                 {
 112                         vector<string> p;
 113
 114                         /* Further trim this line, removing spaces from the end */
 115                         trim_right_if (*line, boost::is_any_of (" "));
 116
 117                         boost::algorithm::split (p, *line, boost::algorithm::is_any_of (" "), boost::token_compress_on);
 118                         if (p.size() != 3 && p.size() != 7) {
 119                                 for (int i = 0; i < 2; ++i) {
 120                                         optional<string> ex = get_line ();
 121                                         if (ex) {
 122                                                 _context.push_back (*ex);
 123                                         }
 124                                 }
 125                                 throw SubripError (*line, "a time/position line", _context);
 126                         }
 127
 128                         rs.from = convert_time (p[0]);
 129                         rs.to = convert_time (p[2]);
 130
 131                         /* XXX: should not ignore coordinate specifications */
 132
 133                         state = CONTENT;
 134                         break;
 135                 }
 136                 case CONTENT:
 137                         if (line->empty ()) {
 138                                 state = COUNTER;
 139                         } else {
 140                                 convert_line (*line, rs);
 141                                 rs.vertical_position.line = rs.vertical_position.line.get() + 1;
 142                         }
 143                         break;
 144                 }
 145         }
 146 }
 147
 148 Time
 149 SubripReader::convert_time (string t)
 150 {
 151         vector<string> a;
 152         boost::algorithm::split (a, t, boost::is_any_of (":"));
 153         if (a.size() != 3) {
 154                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 155         }
 156
 157         vector<string> b;
 158         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 159         if (b.size() != 2) {
 160                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 161         }
 162
 163         int h, m, s, ms;
 164
 165         try {
 166                 h = lexical_cast<int>(a[0]);
 167         } catch (boost::bad_lexical_cast &) {
 168                 throw SubripError (t, "integer hour value", _context);
 169         }
 170
 171         try {
 172                 m = lexical_cast<int>(a[1]);
 173         } catch (boost::bad_lexical_cast &) {
 174                 throw SubripError (t, "integer minute value", _context);
 175         }
 176
 177         try {
 178                 s = lexical_cast<int>(b[0]);
 179         } catch (boost::bad_lexical_cast &) {
 180                 throw SubripError (t, "integer second value", _context);
 181         }
 182
 183         try {
 184                 ms = lexical_cast<int>(b[1]);
 185         } catch (boost::bad_lexical_cast &) {
 186                 throw SubripError (t, "integer millisecond value", _context);
 187         }
 188
 189         return Time::from_hms (h, m, s, ms);
 190 }
 191
 192 void
 193 SubripReader::convert_line (string t, RawSubtitle& p)
 194 {
 195         enum {
 196                 TEXT,
 197                 TAG
 198         } state = TEXT;
 199
 200         string tag;
 201
 202         list<Colour> colours;
 203         colours.push_back (Colour (1, 1, 1));
 204
 205         /* XXX: missing <font> support */
 206         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 207            not work, I think.
 208         */
 209
 210         for (size_t i = 0; i < t.size(); ++i) {
 211                 switch (state) {
 212                 case TEXT:
 213                         if (t[i] == '<' || t[i] == '{') {
 214                                 state = TAG;
 215                         } else {
 216                                 p.text += t[i];
 217                         }
 218                         break;
 219                 case TAG:
 220                         if (t[i] == '>' || t[i] == '}') {
 221                                 if (tag == "b") {
 222                                         maybe_content (p);
 223                                         p.bold = true;
 224                                 } else if (tag == "/b") {
 225                                         maybe_content (p);
 226                                         p.bold = false;
 227                                 } else if (tag == "i") {
 228                                         maybe_content (p);
 229                                         p.italic = true;
 230                                 } else if (tag == "/i") {
 231                                         maybe_content (p);
 232                                         p.italic = false;
 233                                 } else if (tag == "u") {
 234                                         maybe_content (p);
 235                                         p.underline = true;
 236                                 } else if (tag == "/u") {
 237                                         maybe_content (p);
 238                                         p.underline = false;
 239                                 } else if (boost::starts_with (tag, "font")) {
 240                                         maybe_content (p);
 241                                         boost::regex re (".*color=\"#([0123456789abcdef]+)\"");
 242                                         boost::smatch match;
 243                                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 244                                                 p.colour = Colour::from_rgb_hex (match[1]);
 245                                                 colours.push_back (p.colour);
 246                                         }
 247                                 } else if (tag == "/font") {
 248                                         maybe_content (p);
 249                                         colours.pop_back ();
 250                                         p.colour = colours.back ();
 251                                 }
 252                                 tag.clear ();
 253                                 state = TEXT;
 254                         } else {
 255                                 tag += tolower (t[i]);
 256                         }
 257                         break;
 258                 }
 259         }
 260
 261         /* Strip Unicode U+202B (right-to-left embedding) as sometimes it is rendered
 262            as a missing character.  This may be a hack.
 263         */
 264         replace_all (p.text, "\xe2\x80\xab", "");
 265
 266         maybe_content (p);
 267 }
 268
 269 /* Push p into _subs if it has some text, and clear the text out of p */
 270 void
 271 SubripReader::maybe_content (RawSubtitle& p)
 272 {
 273         if (!p.text.empty ()) {
 274                 _subs.push_back (p);
 275                 p.text.clear ();
 276         }
 277 }