git.maemo.org Git - cinaest/blob - src/plugins/google-parser.vala

   1 /* This file is part of Cinaest.
   2  *
   3  * Copyright (C) 2009 Philipp Zabel
   4  *
   5  * Cinaest is free software: you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation, either version 3 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * Cinaest is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with Cinaest. If not, see <http://www.gnu.org/licenses/>.
  17  */
  18
  19 errordomain ParserError {
  20         WRONG_TAG,
  21         EOF
  22 }
  23
  24 public class GoogleMovie : Movie {
  25         public string cinema;
  26         public string runtime;
  27         public string fsk;
  28         public string showtimes;
  29 }
  30
  31 public class GoogleParser : Object {
  32         private MovieSource.ReceiveMovieFunction _get_callback;
  33         char *current;
  34         string cinema_name;
  35         MovieFilter _filter;
  36         PatternSpec pattern;
  37
  38         public int next_tag_offset () {
  39                 int i = -1;
  40                 while (current[++i] != '<' && current[i] != 0);
  41                 return i;
  42         }
  43
  44         public void next_tag () {
  45                 if (current[0] == 0)
  46                         return;
  47                 current += next_tag_offset ();
  48         }
  49
  50         public void finish_tag () {
  51                 while (current[0] != '>' && current[0] != 0)
  52                         current++;
  53                 if (current[0] == '>')
  54                         current++;
  55         }
  56
  57         public weak string parse_tag (bool finish = true) throws Error {
  58                 weak string tag;
  59                 next_tag ();
  60                 int i = 1;
  61                 while (current[++i].isalnum ());
  62                 if (current[i] == 0)
  63                         throw new ParserError.EOF ("EOF in tag");
  64                 if (current[i] == '>')
  65                         finish = false;
  66                 current[i] = 0;
  67                 tag = (string) (current + 1);
  68                 current += i + 1;
  69                 if (finish)
  70                         finish_tag ();
  71                 return tag;
  72         }
  73
  74         public void expect_tag (string tag) throws Error {
  75                 var found = parse_tag (true);
  76                 if (tag != found) {
  77                         throw new ParserError.WRONG_TAG ("Wrong tag \"%s\", expected \"%s\"",
  78                                                          found, tag);
  79                 }
  80         }
  81
  82         public string parse_text () {
  83                 string text = ((string) current).ndup (next_tag_offset ());
  84                 next_tag ();
  85                 return text;
  86         }
  87
  88         public void parse_attribute (string _attr, out string value) {
  89                 string attr;
  90                 if (current[0] == 0)
  91                         return;
  92                 int i = -1;
  93                 while (current[++i] != '=' && current[i] != '>' && current[i] != 0) {
  94
  95                 }
  96                 attr = ((string) current).ndup (i);
  97                 current += i;
  98                 if (current[0] == 0)
  99                         return;
 100                 current++;
 101                 i = -1;
 102                 while (!current[++i].isspace () && current[i] != '>' && current[i] != 0) {
 103                         if (current[i] == '"')
 104                                 while (current[++i] != '"' && current[i] != 0);
 105                 }
 106                 if (attr == _attr) {
 107                         if (current[0] == '"')
 108                                 value = ((string) current).substring (1, i - 2);
 109                         else
 110                                 value = ((string) current).ndup (i);
 111                 }
 112                 current += i;
 113         }
 114
 115         public void skip_whitespace () {
 116                 if (current[0] == 0)
 117                         return;
 118                 int i = -1;
 119                 while (current[++i].isspace () && current[i] != 0);
 120                 current += i;
 121         }
 122
 123         public string? parse_tag_attribute (string tag, string attribute) throws Error {
 124                 var found = parse_tag (false);
 125                 if (tag != found) {
 126                         throw new ParserError.WRONG_TAG ("Wrong tag \"%s\", expected \"%s\"",
 127                                                          found, tag);
 128                 }
 129
 130                 string? value = null;
 131                 skip_whitespace ();
 132                 while (current[0] != '>' && current[0] != 0) {
 133                         parse_attribute (attribute, out value);
 134                         skip_whitespace ();
 135                 }
 136                 // Skip the closing '>' bracket
 137                 if (current[0] != 0)
 138                         current++;
 139
 140                 return value;
 141         }
 142
 143         public void parse_movie () throws Error {
 144                 expect_tag ("a"); // rating
 145                 expect_tag ("nobr");
 146                 expect_tag ("nobr");
 147                 weak string rating_string = parse_tag_attribute ("img", "alt").offset (6); // "Rated " ->"0.0 out of 5.0"
 148                 double rating = rating_string.to_double ();
 149
 150                 expect_tag ("img");
 151                 expect_tag ("img");
 152                 expect_tag ("img");
 153                 expect_tag ("img");
 154                 expect_tag ("/nobr");
 155                 expect_tag ("/nobr");
 156                 expect_tag ("br");
 157                 expect_tag ("nobr");
 158                 expect_tag ("/nobr");
 159                 expect_tag ("/a");
 160                 expect_tag ("/font");
 161                 expect_tag ("/td");
 162                 expect_tag ("td");
 163                 expect_tag ("font");
 164                 expect_tag ("a"); // <a href="/movies?near=city&mid=hexnumber"> --> link
 165                 expect_tag ("b");
 166                 var title = convert (parse_text ().replace ("&#39;", "'"), -1, "utf-8", "iso-8859-1"); // FIXME
 167                 expect_tag ("/b");
 168                 expect_tag ("/a");
 169                 expect_tag ("br");
 170                 var runtime_and_fsk = parse_text ().replace ("&nbsp;", " ").replace ("&#8206;", "").split (" - ");
 171
 172                 var showtimes = "";
 173                 if (parse_tag () == "br") {
 174                         showtimes = parse_text ().replace ("&nbsp;", ",");
 175                         expect_tag ("/font");
 176                 }
 177
 178                 if (pattern == null) {
 179                         if (!title.has_prefix (_filter.title))
 180                                 return;
 181                 } else {
 182                         if (!pattern.match ((uint) title.length, title, null))
 183                                 return;
 184                 }
 185
 186                 var movie = new GoogleMovie ();
 187
 188                 movie.title = strip_tags (title);
 189                 movie.year = 0;
 190                 movie.rating = (int) (rating * 10);
 191
 192                 movie.cinema = cinema_name;
 193                 if (runtime_and_fsk.length >= 2) {
 194                         movie.runtime = runtime_and_fsk[0];
 195                         movie.fsk = runtime_and_fsk[1];
 196                 }
 197                 movie.showtimes = showtimes;
 198
 199                 // TODO - could be configurable by settings
 200                 if (movie.runtime != null)
 201                         movie.secondary = "%s - %s - %s".printf (movie.runtime, cinema_name, showtimes);
 202                 else
 203                         movie.secondary = "%s - %s".printf (cinema_name, showtimes);
 204
 205                 _get_callback (movie);
 206         }
 207
 208         // FIXME - this is specific for Germany
 209         private string strip_tags (string title) {
 210                 string tag_suffix = " (OmU)"; // original audio with subtitles
 211                 if (title.has_suffix (tag_suffix))
 212                         return title.substring (0, title.length - tag_suffix.length);
 213                 tag_suffix = " (OV)"; // original audio
 214                 if (title.has_suffix (tag_suffix))
 215                         return title.substring (0, title.length - tag_suffix.length);
 216                 return title.dup ();
 217         }
 218
 219         public void parse_cinema () throws Error {
 220                 expect_tag ("tr");
 221                 expect_tag ("td");
 222                 expect_tag ("a"); // --> link
 223                 expect_tag ("b");
 224                 string name = convert (parse_text ().replace ("&amp;", "&"), -1, "utf-8", "iso-8859-1"); // FIXME
 225                 expect_tag ("/b");
 226                 expect_tag ("/a");
 227                 expect_tag ("br");
 228                 expect_tag ("font");
 229                 string address = parse_text ().replace ("&nbsp;", " ");
 230                 expect_tag ("a"); // --> map
 231                 expect_tag ("/a");
 232                 expect_tag ("/font");
 233                 expect_tag ("/td");
 234                 expect_tag ("/tr");
 235
 236                 cinema_name = name;
 237                 // FIXME - store cinema address for movie detail window
 238         }
 239
 240         public void parse (ref char[] buf) throws Error {
 241                 current = buf;
 242                 next_tag ();
 243                 while (current[0] != 0) {
 244                         int i = 1;
 245                         while (current[i++] != '>');
 246                         if (((string) current).has_prefix ("<a href=\"/movies/reviews?cid="))
 247                                 parse_movie ();
 248                         else if (((string) current).has_prefix("<tr valign=top><td colspan=4><a href=\"/movies?near="))
 249                                 parse_cinema ();
 250                         else
 251                                 current += i;
 252                         next_tag ();
 253                 }
 254         }
 255
 256         public GoogleParser (MovieFilter filter, string? location, MovieSource.ReceiveMovieFunction callback) {
 257                 _get_callback = callback;
 258                 _filter = filter;
 259                 if (filter.title.chr(filter.title.length, '*') != null) {
 260                         pattern = new PatternSpec (filter.title);
 261                 } else {
 262                         pattern = null;
 263                 }
 264                 try {
 265                         // TODO - use google.de in Germany, also provides genres
 266                         string uri = "http://google.com/movies";
 267                         if (location != null && location != "")
 268                                 uri += "?near=" + location;
 269                         File file = File.new_for_uri (uri);
 270                         InputStream stream = file.read (null);
 271
 272                         char[] buf = new char[256*1024];
 273                         size_t nread;
 274                         bool ok = stream.read_all (buf, buf.length, out nread, null);
 275
 276                         buf[nread] = 0;
 277                         parse (ref buf);
 278                 } catch (Error e) {
 279                         stderr.printf ("Error: %s\n", e.message);
 280                 }
 281         }
 282 }