src/subrip_reader.cc

   1 /*
   2     Copyright (C) 2014 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 #include "subrip_reader.h"
  21 #include "exceptions.h"
  22 #include <boost/algorithm/string.hpp>
  23 #include <boost/lexical_cast.hpp>
  24 #include <cstdio>
  25 #include <vector>
  26
  27 using std::string;
  28 using std::vector;
  29 using boost::lexical_cast;
  30 using namespace sub;
  31
  32 SubripReader::SubripReader (FILE* f)
  33 {
  34         enum {
  35                 COUNTER,
  36                 METADATA,
  37                 CONTENT
  38         } state = COUNTER;
  39
  40         char buffer[256];
  41
  42         TimePair from;
  43         TimePair to;
  44
  45         string line;
  46         int line_number = 0;
  47
  48         while (!feof (f)) {
  49                 char* r = fgets (buffer, sizeof (buffer), f);
  50                 if (r == 0 || feof (f)) {
  51                         break;
  52                 }
  53
  54                 line = string (buffer);
  55                 trim_right_if (line, boost::is_any_of ("\n\r"));
  56
  57                 switch (state) {
  58                 case COUNTER:
  59                 {
  60                         if (line.empty ()) {
  61                                 /* a blank line at the start is ok */
  62                                 break;
  63                         }
  64
  65                         state = METADATA;
  66                 }
  67                 break;
  68                 case METADATA:
  69                 {
  70                         vector<string> p;
  71                         boost::algorithm::split (p, line, boost::algorithm::is_any_of (" "));
  72                         if (p.size() != 3 && p.size() != 7) {
  73                                 throw SubripError (line, "a time/position line");
  74                         }
  75
  76                         from = convert_time (p[0]);
  77                         to = convert_time (p[2]);
  78
  79                         /* XXX: should not ignore coordinate specifications */
  80
  81                         state = CONTENT;
  82                         break;
  83                 }
  84                 case CONTENT:
  85                         if (line.empty ()) {
  86                                 /* XXX: I think this line_number should be set to some sensible value and TOP_OF_SUBTITLE
  87                                    should not be used.
  88                                 */
  89                                 state = COUNTER;
  90                                 line_number = 0;
  91                         } else {
  92                                 convert_line (line, line_number, from, to);
  93                                 line_number++;
  94                         }
  95                         break;
  96                 }
  97         }
  98 }
  99
 100 TimePair
 101 SubripReader::convert_time (string t)
 102 {
 103         vector<string> a;
 104         boost::algorithm::split (a, t, boost::is_any_of (":"));
 105         if (a.size() != 3) {
 106                 throw SubripError (t, "time in the format h:m:s,ms");
 107         }
 108
 109         vector<string> b;
 110         boost::algorithm::split (b, a[2], boost::is_any_of (","));
 111
 112         return TimePair (
 113                 MetricTime (
 114                         lexical_cast<int> (a[0]),
 115                         lexical_cast<int> (a[1]),
 116                         lexical_cast<int> (b[0]),
 117                         lexical_cast<int> (b[1])
 118                         )
 119                 );
 120 }
 121
 122 void
 123 SubripReader::convert_line (string t, int line_number, TimePair from, TimePair to)
 124 {
 125         enum {
 126                 TEXT,
 127                 TAG
 128         } state = TEXT;
 129
 130         string tag;
 131
 132         RawSubtitle p;
 133         p.font = "Arial";
 134         p.font_size.set_points (48);
 135         p.from = from;
 136         p.to = to;
 137         p.vertical_position.line = line_number;
 138         p.vertical_position.reference = TOP_OF_SUBTITLE;
 139
 140         /* XXX: missing <font> support */
 141         /* XXX: nesting of tags e.g. <b>foo<i>bar<b>baz</b>fred</i>jim</b> might
 142            not work, I think.
 143         */
 144
 145         for (size_t i = 0; i < t.size(); ++i) {
 146                 switch (state) {
 147                 case TEXT:
 148                         if (t[i] == '<' || t[i] == '{') {
 149                                 state = TAG;
 150                         } else {
 151                                 p.text += t[i];
 152                         }
 153                         break;
 154                 case TAG:
 155                         if (t[i] == '>' || t[i] == '}') {
 156                                 if (tag == "b") {
 157                                         maybe_content (p);
 158                                         p.bold = true;
 159                                 } else if (tag == "/b") {
 160                                         maybe_content (p);
 161                                         p.bold = false;
 162                                 } else if (tag == "i") {
 163                                         maybe_content (p);
 164                                         p.italic = true;
 165                                 } else if (tag == "/i") {
 166                                         maybe_content (p);
 167                                         p.italic = false;
 168                                 } else if (tag == "u") {
 169                                         maybe_content (p);
 170                                         p.underline = true;
 171                                 } else if (tag == "/u") {
 172                                         maybe_content (p);
 173                                         p.underline = false;
 174                                 }
 175                                 tag.clear ();
 176                                 state = TEXT;
 177                         } else {
 178                                 tag += t[i];
 179                         }
 180                         break;
 181                 }
 182         }
 183
 184         maybe_content (p);
 185 }
 186
 187 void
 188 SubripReader::maybe_content (RawSubtitle& p)
 189 {
 190         if (!p.text.empty ()) {
 191                 _subs.push_back (p);
 192                 p.text.clear ();
 193         }
 194 }