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 std::stringstream;
  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         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         rs.font = "Arial";
  71         rs.font_size.set_points (48);
  72         rs.vertical_position.line = 0;
  73         /* XXX: arbitrary */
  74         rs.vertical_position.lines = 32;
  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
  85                 if (
  86                         line->length() >= 3 &&
  87                         static_cast<unsigned char> (line.get()[0]) == 0xef &&
  88                         static_cast<unsigned char> (line.get()[1]) == 0xbb &&
  89                         static_cast<unsigned char> (line.get()[2]) == 0xbf
  90                         ) {
  91
  92                         /* Skip Unicode byte order mark */
  93                         line = line->substr (3);
  94                 }
  95
  96                 switch (state) {
  97                 case COUNTER:
  98                 {
  99                         if (line->empty ()) {
 100                                 /* a blank line at the start is ok */
 101                                 break;
 102                         }
 103
 104                         state = METADATA;
 105
 106                         /* Reset stuff that should not persist across separate subtitles */
 107                         rs.bold = false;
 108                         rs.italic = false;
 109                         rs.underline = false;
 110                         rs.vertical_position.line = 0;
 111                 }
 112                 break;
 113                 case METADATA:
 114                 {
 115                         vector<string> p;
 116
 117                         /* Further trim this line, removing spaces from the end */
 118                         trim_right_if (*line, boost::is_any_of (" "));
 119
 120                         boost::algorithm::split (p, *line, boost::algorithm::is_any_of (" "));
 121                         if (p.size() != 3 && p.size() != 7) {
 122                                 throw SubripError (*line, "a time/position line");
 123                         }
 124
 125                         rs.from = convert_time (p[0]);
 126                         rs.to = convert_time (p[2]);
 127
 128                         /* XXX: should not ignore coordinate specifications */
 129
 130                         state = CONTENT;
 131                         break;
 132                 }
 133                 case CONTENT:
 134                         if (line->empty ()) {
 135                                 state = COUNTER;
 136                         } else {
 137                                 convert_line (*line, rs);
 138                                 rs.vertical_position.line = rs.vertical_position.line.get() + 1;
 139                         }
 140                         break;
 141                 }
 142         }
 143 }
 144
 145 Time
 146 SubripReader::convert_time (string t)
 147 {
 148         vector<string> a;
 149         boost::algorithm::split (a, t, boost::is_any_of (":"));
 150         if (a.size() != 3) {
 151                 throw SubripError (t, "time in the format h:m:s,ms");
 152         }
 153
 154         vector<string> b;
 155         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 156
 157         return Time::from_hms (
 158                 lexical_cast<int> (a[0]),
 159                 lexical_cast<int> (a[1]),
 160                 lexical_cast<int> (b[0]),
 161                 lexical_cast<int> (b[1])
 162                 );
 163 }
 164
 165 void
 166 SubripReader::convert_line (string t, RawSubtitle& p)
 167 {
 168         enum {
 169                 TEXT,
 170                 TAG
 171         } state = TEXT;
 172
 173         string tag;
 174
 175         list<Colour> colours;
 176         colours.push_back (Colour (1, 1, 1));
 177
 178         /* XXX: missing <font> support */
 179         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 180            not work, I think.
 181         */
 182
 183         for (size_t i = 0; i < t.size(); ++i) {
 184                 switch (state) {
 185                 case TEXT:
 186                         if (t[i] == '<' || t[i] == '{') {
 187                                 state = TAG;
 188                         } else {
 189                                 p.text += t[i];
 190                         }
 191                         break;
 192                 case TAG:
 193                         if (t[i] == '>' || t[i] == '}') {
 194                                 if (tag == "b") {
 195                                         maybe_content (p);
 196                                         p.bold = true;
 197                                 } else if (tag == "/b") {
 198                                         maybe_content (p);
 199                                         p.bold = false;
 200                                 } else if (tag == "i") {
 201                                         maybe_content (p);
 202                                         p.italic = true;
 203                                 } else if (tag == "/i") {
 204                                         maybe_content (p);
 205                                         p.italic = false;
 206                                 } else if (tag == "u") {
 207                                         maybe_content (p);
 208                                         p.underline = true;
 209                                 } else if (tag == "/u") {
 210                                         maybe_content (p);
 211                                         p.underline = false;
 212                                 } else if (boost::starts_with (tag, "font")) {
 213                                         maybe_content (p);
 214                                         boost::regex re (".*color=\"#([0123456789abcdef]+)\"");
 215                                         boost::smatch match;
 216                                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 217                                                 p.colour = Colour::from_rgb_hex (match[1]);
 218                                                 colours.push_back (p.colour);
 219                                         }
 220                                 } else if (tag == "/font") {
 221                                         maybe_content (p);
 222                                         colours.pop_back ();
 223                                         p.colour = colours.back ();
 224                                 }
 225                                 tag.clear ();
 226                                 state = TEXT;
 227                         } else {
 228                                 tag += t[i];
 229                         }
 230                         break;
 231                 }
 232         }
 233
 234         maybe_content (p);
 235 }
 236
 237 /* Push p into _subs if it has some text, and clear the text out of p */
 238 void
 239 SubripReader::maybe_content (RawSubtitle& p)
 240 {
 241         if (!p.text.empty ()) {
 242                 _subs.push_back (p);
 243                 p.text.clear ();
 244         }
 245 }