vcs.maemo.org Git - cinaest/blob - src/backends/google/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 Cinema {
  25         public string name;
  26         public string address;
  27         public string phone;
  28
  29         public Cinema (string _name) {
  30                 name = _name;
  31         }
  32 }
  33
  34 public class GoogleMovie {
  35         public string title;
  36         public int rating;
  37         public string secondary;
  38         public Cinema cinema;
  39         public string runtime;
  40         public string fsk;
  41         public string showtimes;
  42 }
  43
  44 public class GoogleParser : Object {
  45         char *current;
  46         Cinema last_cinema;
  47         public string location;
  48         string _title;
  49         PatternSpec pattern;
  50
  51         public delegate void ReceiveMovie (GoogleMovie movie);
  52         public ReceiveMovie _get_callback;
  53
  54         public int next_tag_offset () {
  55                 int i = -1;
  56                 while (current[++i] != '<' && current[i] != 0);
  57                 return i;
  58         }
  59
  60         public void next_tag () {
  61                 if (current[0] == 0)
  62                         return;
  63                 current += next_tag_offset ();
  64         }
  65
  66         public void finish_tag () {
  67                 while (current[0] != '>' && current[0] != 0)
  68                         current++;
  69                 if (current[0] == '>')
  70                         current++;
  71         }
  72
  73         public weak string parse_tag (bool finish = true) throws Error {
  74                 weak string tag;
  75                 next_tag ();
  76                 int i = 1;
  77                 while (current[++i].isalnum ());
  78                 if (current[i] == 0)
  79                         throw new ParserError.EOF ("EOF in tag");
  80                 if (current[i] == '>')
  81                         finish = false;
  82                 current[i] = 0;
  83                 tag = (string) (current + 1);
  84                 current += i + 1;
  85                 if (finish)
  86                         finish_tag ();
  87                 return tag;
  88         }
  89
  90         public void expect_tag (string tag) throws Error {
  91                 var found = parse_tag (true);
  92                 if (tag != found) {
  93                         throw new ParserError.WRONG_TAG ("Wrong tag \"%s\", expected \"%s\"",
  94                                                          found, tag);
  95                 }
  96         }
  97
  98         public string parse_text () {
  99                 string text = ((string) current).ndup (next_tag_offset ());
 100                 next_tag ();
 101                 return text;
 102         }
 103
 104         public void parse_attribute (string _attr, out string value) {
 105                 string attr;
 106                 if (current[0] == 0)
 107                         return;
 108                 int i = -1;
 109                 while (current[++i] != '=' && current[i] != '>' && current[i] != 0) {
 110
 111                 }
 112                 attr = ((string) current).ndup (i);
 113                 current += i;
 114                 if (current[0] == 0)
 115                         return;
 116                 current++;
 117                 i = -1;
 118                 while (!current[++i].isspace () && current[i] != '>' && current[i] != 0) {
 119                         if (current[i] == '"')
 120                                 while (current[++i] != '"' && current[i] != 0);
 121                 }
 122                 if (attr == _attr) {
 123                         if (current[0] == '"')
 124                                 value = ((string) current).substring (1, i - 2);
 125                         else
 126                                 value = ((string) current).ndup (i);
 127                 }
 128                 current += i;
 129         }
 130
 131         public void skip_whitespace () {
 132                 if (current[0] == 0)
 133                         return;
 134                 int i = -1;
 135                 while (current[++i].isspace () && current[i] != 0);
 136                 current += i;
 137         }
 138
 139         public string? parse_tag_attribute (string tag, string attribute) throws Error {
 140                 var found = parse_tag (false);
 141                 if (tag != found) {
 142                         throw new ParserError.WRONG_TAG ("Wrong tag \"%s\", expected \"%s\"",
 143                                                          found, tag);
 144                 }
 145
 146                 string? value = null;
 147                 skip_whitespace ();
 148                 while (current[0] != '>' && current[0] != 0) {
 149                         parse_attribute (attribute, out value);
 150                         skip_whitespace ();
 151                 }
 152                 // Skip the closing '>' bracket
 153                 if (current[0] != 0)
 154                         current++;
 155
 156                 return value;
 157         }
 158
 159         public string unescape_unicode (string s) {
 160                 string result = "";
 161                 int i, j;
 162                 long l = s.length;
 163
 164                 for (i = 0; i < l; i++) {
 165                         if (s[i] == '&' && s[i + 1] == '#') {
 166                                 for (j = i + 2; j < l; j++) {
 167                                         if (!s[j].isdigit ())
 168                                                 break;
 169                                         if (s[j] == ';')
 170                                                 break;
 171                                 }
 172                                 if (s[j] == ';') {
 173                                         int codepoint = s.substring (i + 2, j - i - 2).to_int ();
 174                                         char[] buf = new char[6];
 175                                         ((unichar) codepoint).to_utf8 ((string) buf);
 176                                         result += (string) buf;
 177                                         i = j;
 178                                         continue;
 179                                 }
 180                         }
 181                         if (s.offset (i).has_prefix ("&amp;")) {
 182                                 result += "&";
 183                                 i += 4;
 184                                 continue;
 185                         }
 186                         if (s.offset (i).has_prefix ("&quot;")) {
 187                                 result += "\"";
 188                                 i += 5;
 189                                 continue;
 190                         }
 191                         result += s.substring (i, 1);
 192                 }
 193
 194                 return result;
 195         }
 196
 197         public void parse_movie () throws Error {
 198                 expect_tag ("div"); // class=movie
 199                 expect_tag ("div"); // class=name
 200                 expect_tag ("a"); // href="/movies?near=city&amp;mid=..."
 201                 expect_tag ("span"); // dir=ltr
 202                 var title = unescape_unicode (convert (parse_text (), -1, "utf-8", "iso-8859-1")); // FIXME
 203                 expect_tag ("/span");
 204                 expect_tag ("/a");
 205                 expect_tag ("/div");
 206                 expect_tag ("span"); // class=info
 207                 string[] runtime_and_fsk = {};
 208                 double rating = 0.0;
 209                 var tag = parse_tag ();
 210                 if (tag == "a") {
 211                         // Trailer
 212                         expect_tag ("/a");
 213                         tag = parse_tag ();
 214                 }
 215                 if (tag == "a") {
 216                         // IMDb
 217                         expect_tag ("/a");
 218                         tag = parse_tag ();
 219                 }
 220                 if (tag == "nobr") {
 221                         expect_tag ("nobr");
 222                         string rating_string = parse_tag_attribute ("img", "alt").offset (6); // "Rated " ->"0.0 out of 5.0"
 223                         rating = rating_string.to_double ();
 224                         expect_tag ("img");
 225                         expect_tag ("img");
 226                         expect_tag ("img");
 227                         expect_tag ("img");
 228                         expect_tag ("/nobr");
 229                         expect_tag ("/nobr");
 230                         runtime_and_fsk = parse_text ().replace ("&#8206;", "").offset (3).split (" - ");
 231                         if (parse_tag () == "a") {
 232                                 // Trailer
 233                                 expect_tag ("/a");
 234                                 if (parse_tag () == "a") {
 235                                         // IMDb link
 236                                         expect_tag ("/a");
 237                                         expect_tag ("/span");
 238                                 }
 239                         }
 240                 }
 241                 expect_tag ("div"); // class=times
 242                 var showtimes = parse_text ().replace ("&nbsp;", ",");
 243                 while (parse_tag () == "a") {
 244                         showtimes += parse_text () + ",";
 245                         expect_tag ("/a");
 246                 }
 247
 248                 if (pattern == null) {
 249                         if (!title.has_prefix (_title))
 250                                 return;
 251                 } else {
 252                         if (!pattern.match ((uint) title.length, title, null))
 253                                 return;
 254                 }
 255
 256                 var movie = new GoogleMovie ();
 257
 258                 movie.title = strip_tags (title).replace ("\"", "\\\"");
 259                 movie.rating = (int) (rating * 10);
 260
 261                 movie.cinema = last_cinema;
 262                 if (runtime_and_fsk.length >= 2) {
 263                         movie.runtime = runtime_and_fsk[0];
 264                         movie.fsk = runtime_and_fsk[1];
 265                 }
 266                 movie.showtimes = showtimes;
 267
 268                 // TODO - could be configurable by settings
 269                 if (movie.runtime != null)
 270                         movie.secondary = "%s - %s - %s".printf (movie.runtime, last_cinema.name, showtimes);
 271                 else
 272                         movie.secondary = "%s - %s".printf (last_cinema.name, showtimes);
 273
 274                 _get_callback (movie);
 275         }
 276
 277         // FIXME - this is specific for Germany
 278         private string strip_tags (string title) {
 279                 string tag_suffix = " (OmU)"; // original audio with subtitles
 280                 if (title.has_suffix (tag_suffix))
 281                         return title.substring (0, title.length - tag_suffix.length);
 282                 tag_suffix = " (OV)"; // original audio
 283                 if (title.has_suffix (tag_suffix))
 284                         return title.substring (0, title.length - tag_suffix.length);
 285                 return title.dup ();
 286         }
 287
 288         public void parse_cinema () throws Error {
 289                 expect_tag ("div"); // class=theater
 290                 expect_tag ("div"); // class=desc id=theater_...
 291                 expect_tag ("h2"); // class=name
 292                 expect_tag ("a"); // href="/movies?near=city&amp;tid=..."
 293                 expect_tag ("span"); // dir=ltr
 294                 var name = unescape_unicode (convert (parse_text (), -1, "utf-8", "iso-8859-1")); // FIXME
 295                 expect_tag ("/span");
 296                 expect_tag ("/a");
 297                 expect_tag ("/h2");
 298                 expect_tag ("div"); // class=info
 299                 var address_and_phone = parse_text ().replace ("&nbsp;", " ").split (" - ");
 300                 string address = null;
 301                 string phone = null;
 302                 if (address_and_phone.length >= 2) {
 303                         address = address_and_phone[0];
 304                         phone = address_and_phone[1].replace (" ", "").replace ("-", "");
 305                 }
 306                 expect_tag ("a"); // target=_top
 307                 expect_tag ("/a");
 308                 expect_tag ("/div");
 309                 expect_tag ("/div");
 310
 311                 last_cinema = new Cinema (name);
 312                 last_cinema.address = address;
 313                 last_cinema.phone = phone;
 314         }
 315
 316         public int parse (ref char[] buf) throws Error {
 317                 int movies = 0;
 318
 319                 current = buf;
 320                 next_tag ();
 321                 while (location == null && current[0] != 0) {
 322                         int i = 1;
 323                         while (current[i++] != '>');
 324                         if (((string) current).has_prefix ("<a href=\"/movies?near=")) {
 325                                 string href = parse_tag_attribute ("a", "href");
 326                                 char* p = (char*) href.offset (13); // skip "/movies?near="
 327                                 int j = -1;
 328
 329                                 while (p[++j] != '&' && p[j] != 0);
 330                                 p[0] = p[0].toupper ();
 331                                 location = ((string) p).ndup (j);
 332                         }
 333                         current += i;
 334                         next_tag ();
 335                 }
 336                 while (current[0] != 0) {
 337                         int i = 1;
 338                         while (current[i++] != '>');
 339                         if (((string) current).has_prefix ("<div class=movie>")) {
 340                                 parse_movie ();
 341                                 movies++;
 342                         } else if (((string) current).has_prefix("<div class=theater>")) {
 343                                 parse_cinema ();
 344                         } else {
 345                                 current += i;
 346                         }
 347                         next_tag ();
 348                 }
 349
 350                 return movies;
 351         }
 352
 353         public async int query (string title, string? location, ReceiveMovie callback, Cancellable? cancellable = null) {
 354                 _get_callback = callback;
 355                 _title = title;
 356                 if (title.chr(title.length, '*') != null) {
 357                         pattern = new PatternSpec (title);
 358                 } else {
 359                         pattern = null;
 360                 }
 361                 try {
 362                         // TODO - use google.de in Germany, also provides genres
 363                         string uri = "http://google.com/movies";
 364                         if (location != null && location != "")
 365                                 uri += "?near=" + location;
 366
 367                         stdout.printf ("GET: %s\n", uri);
 368
 369                         File file = File.new_for_uri (uri);
 370                         InputStream stream = yield file.read_async (Priority.DEFAULT_IDLE, null);
 371
 372                         char[] buf = new char[256*1024];
 373                         size_t nread;
 374                         size_t total = 0;
 375                         while (total < 256*1024) {
 376                                 nread = yield stream.read_async ((char *)buf + total, 256*1024 - total, Priority.DEFAULT_IDLE, cancellable);
 377                                 total += nread;
 378                                 if (cancellable.is_cancelled ())
 379                                         return 0;
 380                                 if (nread == 0)
 381                                         break;
 382                         }
 383                         buf[total] = 0;
 384                         return parse (ref buf);
 385                 } catch (Error e) {
 386                         stderr.printf ("Error: %s\n", e.message);
 387                 }
 388
 389                 return 0;
 390         }
 391 }