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 namespace sub;
  46
  47 /** @param s Subtitle string encoded in UTF-8 */
  48 SubripReader::SubripReader (string const & s)
  49 {
  50         locked_stringstream str (s);
  51         this->read (boost::bind (&get_line_stringstream, &str));
  52 }
  53
  54 /** @param f Subtitle file encoded in UTF-8 */
  55 SubripReader::SubripReader (FILE* f)
  56 {
  57         this->read (boost::bind (&get_line_file, f));
  58 }
  59
  60 void
  61 SubripReader::read (function<optional<string> ()> get_line)
  62 {
  63         enum {
  64                 COUNTER,
  65                 METADATA,
  66                 CONTENT
  67         } state = COUNTER;
  68
  69         RawSubtitle rs;
  70
  71         /* This reader extracts no information about where the subtitle
  72            should be on screen, so its reference is TOP_OF_SUBTITLE.
  73         */
  74         rs.vertical_position.line = 0;
  75         rs.vertical_position.reference = TOP_OF_SUBTITLE;
  76
  77         while (true) {
  78                 optional<string> line = get_line ();
  79                 if (!line) {
  80                         break;
  81                 }
  82
  83                 trim_right_if (*line, boost::is_any_of ("\n\r"));
  84                 remove_unicode_bom (line);
  85
  86                 /* Keep some history in case there is an error to report */
  87                 _context.push_back (*line);
  88                 if (_context.size() > 5) {
  89                         _context.pop_front ();
  90                 }
  91
  92                 switch (state) {
  93                 case COUNTER:
  94                 {
  95                         if (line->empty ()) {
  96                                 /* a blank line at the start is ok */
  97                                 break;
  98                         }
  99
 100                         state = METADATA;
 101
 102                         /* Reset stuff that should not persist across separate subtitles */
 103                         rs.bold = false;
 104                         rs.italic = false;
 105                         rs.underline = false;
 106                         rs.vertical_position.line = 0;
 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 (" "));
 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                                 convert_line (*line, rs);
 140                                 rs.vertical_position.line = rs.vertical_position.line.get() + 1;
 141                         }
 142                         break;
 143                 }
 144         }
 145 }
 146
 147 Time
 148 SubripReader::convert_time (string t)
 149 {
 150         vector<string> a;
 151         boost::algorithm::split (a, t, boost::is_any_of (":"));
 152         if (a.size() != 3) {
 153                 throw SubripError (t, "time in the format h:m:s,ms", _context);
 154         }
 155
 156         vector<string> b;
 157         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 158
 159         return Time::from_hms (
 160                 lexical_cast<int> (a[0]),
 161                 lexical_cast<int> (a[1]),
 162                 lexical_cast<int> (b[0]),
 163                 lexical_cast<int> (b[1])
 164                 );
 165 }
 166
 167 void
 168 SubripReader::convert_line (string t, RawSubtitle& p)
 169 {
 170         enum {
 171                 TEXT,
 172                 TAG
 173         } state = TEXT;
 174
 175         string tag;
 176
 177         list<Colour> colours;
 178         colours.push_back (Colour (1, 1, 1));
 179
 180         /* XXX: missing <font> support */
 181         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 182            not work, I think.
 183         */
 184
 185         for (size_t i = 0; i < t.size(); ++i) {
 186                 switch (state) {
 187                 case TEXT:
 188                         if (t[i] == '<' || t[i] == '{') {
 189                                 state = TAG;
 190                         } else {
 191                                 p.text += t[i];
 192                         }
 193                         break;
 194                 case TAG:
 195                         if (t[i] == '>' || t[i] == '}') {
 196                                 if (tag == "b") {
 197                                         maybe_content (p);
 198                                         p.bold = true;
 199                                 } else if (tag == "/b") {
 200                                         maybe_content (p);
 201                                         p.bold = false;
 202                                 } else if (tag == "i") {
 203                                         maybe_content (p);
 204                                         p.italic = true;
 205                                 } else if (tag == "/i") {
 206                                         maybe_content (p);
 207                                         p.italic = false;
 208                                 } else if (tag == "u") {
 209                                         maybe_content (p);
 210                                         p.underline = true;
 211                                 } else if (tag == "/u") {
 212                                         maybe_content (p);
 213                                         p.underline = false;
 214                                 } else if (boost::starts_with (tag, "font")) {
 215                                         maybe_content (p);
 216                                         boost::regex re (".*color=\"#([0123456789abcdef]+)\"");
 217                                         boost::smatch match;
 218                                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 219                                                 p.colour = Colour::from_rgb_hex (match[1]);
 220                                                 colours.push_back (p.colour);
 221                                         }
 222                                 } else if (tag == "/font") {
 223                                         maybe_content (p);
 224                                         colours.pop_back ();
 225                                         p.colour = colours.back ();
 226                                 }
 227                                 tag.clear ();
 228                                 state = TEXT;
 229                         } else {
 230                                 tag += t[i];
 231                         }
 232                         break;
 233                 }
 234         }
 235
 236         maybe_content (p);
 237 }
 238
 239 /* Push p into _subs if it has some text, and clear the text out of p */
 240 void
 241 SubripReader::maybe_content (RawSubtitle& p)
 242 {
 243         if (!p.text.empty ()) {
 244                 _subs.push_back (p);
 245                 p.text.clear ();
 246         }
 247 }