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         Time from;
  70         Time to;
  71
  72         string line;
  73         int line_number = 0;
  74
  75         while (true) {
  76                 optional<string> line = get_line ();
  77                 if (!line) {
  78                         break;
  79                 }
  80
  81                 trim_right_if (*line, boost::is_any_of ("\n\r"));
  82
  83                 if (
  84                         line->length() >= 3 &&
  85                         static_cast<unsigned char> (line.get()[0]) == 0xef &&
  86                         static_cast<unsigned char> (line.get()[1]) == 0xbb &&
  87                         static_cast<unsigned char> (line.get()[2]) == 0xbf
  88                         ) {
  89
  90                         /* Skip Unicode byte order mark */
  91                         line = line->substr (3);
  92                 }
  93
  94                 switch (state) {
  95                 case COUNTER:
  96                 {
  97                         if (line->empty ()) {
  98                                 /* a blank line at the start is ok */
  99                                 break;
 100                         }
 101
 102                         state = METADATA;
 103                 }
 104                 break;
 105                 case METADATA:
 106                 {
 107                         vector<string> p;
 108                         boost::algorithm::split (p, *line, boost::algorithm::is_any_of (" "));
 109                         if (p.size() != 3 && p.size() != 7) {
 110                                 throw SubripError (*line, "a time/position line");
 111                         }
 112
 113                         from = convert_time (p[0]);
 114                         to = convert_time (p[2]);
 115
 116                         /* XXX: should not ignore coordinate specifications */
 117
 118                         state = CONTENT;
 119                         break;
 120                 }
 121                 case CONTENT:
 122                         if (line->empty ()) {
 123                                 state = COUNTER;
 124                                 line_number = 0;
 125                         } else {
 126                                 convert_line (*line, line_number, from, to);
 127                                 line_number++;
 128                         }
 129                         break;
 130                 }
 131         }
 132 }
 133
 134 Time
 135 SubripReader::convert_time (string t)
 136 {
 137         vector<string> a;
 138         boost::algorithm::split (a, t, boost::is_any_of (":"));
 139         if (a.size() != 3) {
 140                 throw SubripError (t, "time in the format h:m:s,ms");
 141         }
 142
 143         vector<string> b;
 144         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 145
 146         return Time::from_hms (
 147                 lexical_cast<int> (a[0]),
 148                 lexical_cast<int> (a[1]),
 149                 lexical_cast<int> (b[0]),
 150                 lexical_cast<int> (b[1])
 151                 );
 152 }
 153
 154 void
 155 SubripReader::convert_line (string t, int line_number, Time from, Time to)
 156 {
 157         enum {
 158                 TEXT,
 159                 TAG
 160         } state = TEXT;
 161
 162         string tag;
 163
 164         RawSubtitle p;
 165         p.font = "Arial";
 166         p.font_size.set_points (48);
 167         p.from = from;
 168         p.to = to;
 169         p.vertical_position.line = line_number;
 170         /* XXX: arbitrary */
 171         p.vertical_position.lines = 32;
 172         p.vertical_position.reference = TOP_OF_SUBTITLE;
 173
 174         list<Colour> colours;
 175         colours.push_back (Colour (1, 1, 1));
 176
 177         /* XXX: missing <font> support */
 178         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 179            not work, I think.
 180         */
 181
 182         for (size_t i = 0; i < t.size(); ++i) {
 183                 switch (state) {
 184                 case TEXT:
 185                         if (t[i] == '<' || t[i] == '{') {
 186                                 state = TAG;
 187                         } else {
 188                                 p.text += t[i];
 189                         }
 190                         break;
 191                 case TAG:
 192                         if (t[i] == '>' || t[i] == '}') {
 193                                 if (tag == "b") {
 194                                         maybe_content (p);
 195                                         p.bold = true;
 196                                 } else if (tag == "/b") {
 197                                         maybe_content (p);
 198                                         p.bold = false;
 199                                 } else if (tag == "i") {
 200                                         maybe_content (p);
 201                                         p.italic = true;
 202                                 } else if (tag == "/i") {
 203                                         maybe_content (p);
 204                                         p.italic = false;
 205                                 } else if (tag == "u") {
 206                                         maybe_content (p);
 207                                         p.underline = true;
 208                                 } else if (tag == "/u") {
 209                                         maybe_content (p);
 210                                         p.underline = false;
 211                                 } else if (boost::starts_with (tag, "font")) {
 212                                         maybe_content (p);
 213                                         boost::regex re (".*color=\"#([0123456789abcdef]+)\"");
 214                                         boost::smatch match;
 215                                         if (boost::regex_search (tag, match, re) && string (match[1]).size() == 6) {
 216                                                 p.colour = Colour::from_rgb_hex (match[1]);
 217                                                 colours.push_back (p.colour);
 218                                         }
 219                                 } else if (tag == "/font") {
 220                                         maybe_content (p);
 221                                         colours.pop_back ();
 222                                         p.colour = colours.back ();
 223                                 }
 224                                 tag.clear ();
 225                                 state = TEXT;
 226                         } else {
 227                                 tag += t[i];
 228                         }
 229                         break;
 230                 }
 231         }
 232
 233         maybe_content (p);
 234 }
 235
 236 void
 237 SubripReader::maybe_content (RawSubtitle& p)
 238 {
 239         if (!p.text.empty ()) {
 240                 _subs.push_back (p);
 241                 p.text.clear ();
 242         }
 243 }