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