vcs.maemo.org Git - maevies/blob - examples/gmovies.c

   1 /* Queries Google movies for the theaters of a city and parses
   2  * the response using libxml2.
   3  */
   4
   5 #include <libxml/HTMLparser.h>
   6 #include <libxml/tree.h>
   7 #include <libsoup/soup.h>
   8 #include <glib.h>
   9 #include <stdio.h>
  10 #include <unistd.h>
  11
  12 #define BASE_URL "http://www.google.com/movies?near=%s"
  13
  14 typedef struct struct_theater
  15 {
  16         gchar* name;
  17         gchar* address;
  18         xmlNodePtr data; //points to the tree node with the theater info, including movies
  19 } Theater;
  20
  21
  22 typedef struct struct_showtime
  23 {
  24         gchar* time;
  25 } ShowTime;
  26
  27
  28 typedef struct struct_movie
  29 {
  30         gchar* title;
  31         int rating;
  32         gchar* info;
  33         GList* showTimes;
  34 } Movie;
  35
  36
  37 /** Search for a sibling node by the name of the sibling node */
  38 xmlNodePtr getSiblingByName(xmlNodePtr node, xmlChar* name, int nameLen)
  39 {
  40
  41         xmlNodePtr sibling = node->next;
  42         while((sibling != NULL) && (strncmp(sibling->name, name, nameLen) != 0)) {
  43                 sibling = sibling->next;
  44         }
  45
  46         return sibling;
  47
  48 }
  49
  50 /** Search a child node by its node name */
  51 xmlNodePtr getChildByName(xmlNodePtr node, xmlChar* name, int nameLen)
  52 {
  53         return getSiblingByName(node->children, name, nameLen);
  54 }
  55
  56 /** Search the first sibling node that has an attribute 'attr'
  57  * with the value 'attrValue' */
  58 xmlNodePtr getFirstSiblingByAttributeValue(
  59         xmlNodePtr sibling, xmlChar* attr, xmlChar * attrValue, int attrValueLen)
  60 {
  61         xmlNodePtr tempNode = sibling;
  62         xmlNodePtr result = NULL;
  63
  64         while ((tempNode != NULL) && (result == NULL)) {
  65                 xmlChar* value = xmlGetProp(tempNode, attr);
  66                 if (value != NULL) {
  67                         if (strncmp(value, attrValue, attrValueLen) == 0) {
  68                                 result = tempNode;
  69                         }
  70                         free(value);
  71                 }
  72                 tempNode = tempNode->next;
  73         }
  74
  75         return result;
  76 }
  77
  78 /** Search the first child node that has an attribute 'attr' with
  79  * value 'attrValue' */
  80 xmlNodePtr getFirstChildByAttributeValue(
  81         xmlNodePtr node, xmlChar* attr, xmlChar * attrValue, int attrValueLen)
  82 {
  83         return getFirstSiblingByAttributeValue(node->children, attr, attrValue, attrValueLen);
  84 }
  85
  86 /** Advances N sibling nodes in the node list */
  87 xmlNodePtr jumpXSiblings(xmlNodePtr node, int siblings)
  88 {
  89         xmlNodePtr r = node;
  90
  91         int i = 0;
  92         for(; i<siblings; i++) {
  93                 r = r->next;
  94         }
  95
  96         return r;
  97 }
  98
  99
 100 int childrenCount(xmlNodePtr node)
 101 {
 102         int i=0;
 103         xmlNodePtr nav = node->children;
 104         while(nav != NULL) {
 105                 i++;
 106                 nav = nav->next;
 107         }
 108
 109         return i;
 110 }
 111
 112
 113 /** Search the <div> with the results and returns it, or NULL
 114  * if it couldn't be found */
 115 xmlNodePtr getMovieResultsDiv(xmlNodePtr root)
 116 {
 117         //<body>
 118         xmlNodePtr body = getSiblingByName(root->children, "body", 4);
 119
 120         //<div id="results">
 121         xmlNodePtr tempNode = getFirstChildByAttributeValue(body, "id", "results", 8);
 122
 123         if (tempNode == NULL) {
 124                 //no results
 125                 return NULL;
 126         }
 127
 128         //<div id="movie_results">
 129         tempNode = getFirstChildByAttributeValue(tempNode, "id", "movie_results", 14);
 130
 131         if (tempNode == NULL) {
 132                 //no results
 133                 return NULL;
 134         }
 135
 136         //<div class="movie_results">
 137         tempNode = getFirstChildByAttributeValue(tempNode, "class", "movie_results", 14);
 138
 139         if (tempNode == NULL) {
 140                 //no results
 141                 return NULL;
 142         }
 143
 144         return tempNode;
 145
 146 }
 147
 148
 149 /** Parses the results and returns a list with all the theaters.
 150  * Theater info is parsed and returned as Theater 'objects", the movie
 151  * info is not parsed */
 152 GList* getTheaterList(xmlNodePtr movieResults)
 153 {
 154         //<div class="movie_results"><div class="theater"/><div class="theater"/>...
 155         xmlNodePtr nav = movieResults->children;
 156         xmlNodePtr tmp1, tmp2, tmp3 = NULL;
 157
 158         GList* resultList = NULL;
 159
 160         while(nav != NULL) {
 161                 tmp1 = getFirstSiblingByAttributeValue(nav, "class", "theater", 7);
 162
 163                 /*
 164                  *  <div class="theater">
 165                  *              <div class="desc">
 166                  *                      <div class="name"/><div class="info"/>
 167                  *              </div>
 168                  *              <div class="showtimes"/>
 169                  * </div>
 170                  */
 171
 172                 if (tmp1 != NULL) { //its theater data
 173                         Theater* t = malloc(sizeof(Theater));
 174                         tmp2 = getFirstChildByAttributeValue(tmp1, "class", "desc", 4);
 175                         if (tmp2 != NULL) { //has desc
 176                                 tmp3 = getFirstChildByAttributeValue(tmp2, "class", "name", 4);
 177                                 if ((tmp3 != NULL) && (tmp3->children != NULL) && (tmp3->children->children != NULL)) { //<div class="name"><a><span/>
 178                                         t->name = xmlNodeGetContent(tmp3->children->children);
 179                                 }
 180                                 tmp3 = getFirstChildByAttributeValue(tmp2, "class", "info", 4);
 181                                 if (tmp3 != NULL) { //<div class="info">
 182                                         t->address = xmlNodeGetContent(tmp3);
 183                                 }
 184                         }
 185                         t->data = tmp1;
 186                         resultList = g_list_append(resultList, t);
 187                 }
 188                 nav = nav->next;
 189         }
 190
 191         return resultList;
 192
 193 }
 194
 195 GList* getShowtimes(gchar* times)
 196 {
 197         GList* resultList = NULL;
 198
 199         gchar** timesArray = g_strsplit(times, " ", -1);
 200
 201         int i = 0;
 202         for(i=0; timesArray[i] != NULL; i++) {
 203                 ShowTime* st = malloc(sizeof(ShowTime));
 204                 st->time = g_strndup(timesArray[i], 5);
 205                 resultList = g_list_append(resultList, st);
 206         }
 207
 208         g_strfreev(timesArray);
 209
 210         return resultList;
 211 }
 212
 213
 214 GList* getMovieList(xmlNodePtr movieSideDiv)
 215 {
 216
 217         xmlNodePtr nav = movieSideDiv->children;
 218         xmlNodePtr tmp1, tmp2, tmp3 = NULL;
 219
 220         GList* resultList = NULL;
 221
 222         while(nav != NULL) {
 223                 tmp1 = getFirstSiblingByAttributeValue(nav, "class", "movie", 5);
 224                 if (tmp1 != NULL) { //is a movie
 225                         Movie* m = malloc(sizeof(Movie));
 226                         tmp2 = getFirstChildByAttributeValue(tmp1, "class", "name", 4);
 227                         if ((tmp2 != NULL) && (tmp2->children != NULL) && (tmp2->children->children != NULL)) { //<div class="name"><a><span/>
 228                                 m->title = xmlNodeGetContent(tmp2->children->children);
 229                         }
 230                         tmp2 = getFirstChildByAttributeValue(tmp1, "class", "info", 4);
 231                         if (tmp2 != NULL) {
 232                                 m->info = xmlNodeGetContent(tmp2);
 233                         }
 234                         tmp2 = getFirstChildByAttributeValue(tmp1, "class", "times", 5);
 235                         if (tmp2 != NULL) {
 236                                 gchar* showtimesStr = xmlNodeGetContent(tmp2);
 237                                 GList* showtimes = getShowtimes(showtimesStr);
 238                                 free(showtimesStr);
 239                                 m->showTimes = showtimes;
 240                         }
 241                         resultList = g_list_append(resultList, m);
 242                 }
 243
 244                 nav = nav->next;
 245         }
 246
 247         return resultList;
 248 }
 249
 250 GList* getTheaterMovies(Theater* t)
 251 {
 252
 253         /*
 254          * <div class="theater">
 255          *              <div class="desc"/>
 256          *              <div class="showtimes">
 257          *                      <div class="show_left">
 258          *                              <div class="movie/>
 259          *                              ...
 260          *                      </div>
 261          *                      <div class="show_right">
 262          *                              <div class="movie"/>
 263          *                              ...
 264          *                      </div>
 265          *              </div>
 266          * </div>
 267          */
 268
 269         GList* left = NULL;
 270         GList* right = NULL;
 271         xmlNodePtr showtimesDiv = getFirstChildByAttributeValue(t->data, "class", "showtimes", 9);
 272         if (showtimesDiv != NULL) {
 273
 274                 xmlNodePtr showLeft = getFirstChildByAttributeValue(showtimesDiv, "class", "show_left", 9);
 275                 if (showLeft != NULL)
 276                         left = getMovieList(showLeft);
 277
 278                 xmlNodePtr showRight = getFirstChildByAttributeValue(showtimesDiv, "class", "show_right", 10);
 279                 if (showRight != NULL)
 280                         right = getMovieList(showRight);
 281
 282                 return g_list_concat(left, right);
 283
 284         }
 285
 286
 287 }
 288
 289
 290 void deleteShowTime(ShowTime* st)
 291 {
 292         free(st->time);
 293         free(st);
 294 }
 295
 296
 297 void deleteShowTimeList(GList * showTimeList)
 298 {
 299         g_list_foreach(showTimeList, (GFunc) deleteShowTime, NULL);
 300         g_list_free(showTimeList);
 301 }
 302
 303
 304 void deleteMovie(Movie* m)
 305 {
 306         free(m->title);
 307         free(m->info);
 308         deleteShowTimeList(m->showTimes);
 309         free(m);
 310 }
 311
 312
 313 void deleteMovieList(GList* movieList)
 314 {
 315         g_list_foreach(movieList, (GFunc) deleteMovie, NULL);
 316         g_list_free(movieList);
 317 }
 318
 319
 320 void deleteTheater(Theater* t)
 321 {
 322         free(t->name);
 323         free(t->address);
 324         free(t);
 325 }
 326
 327 void deleteTheaterList(GList* theaterList)
 328 {
 329         g_list_foreach(theaterList, (GFunc) deleteTheater, NULL);
 330         g_list_free(theaterList);
 331 }
 332
 333
 334 void showTime(ShowTime* st, gpointer nothing)
 335 {
 336         printf("%s ", st->time);
 337 }
 338
 339 void showMovie(Movie* m, gpointer nothing)
 340 {
 341         printf("        Title = %s\n", m->title);
 342         printf("        Info = %s\n", m->info);
 343         printf("        Schedule = ");
 344         g_list_foreach(m->showTimes, (GFunc) showTime, NULL);
 345         printf("\n");
 346 }
 347
 348
 349 void showTheater(Theater * t, gpointer nothing)
 350 {
 351         printf("Name = %s\n", t->name);
 352         printf("Info = %s\n", t->address);
 353         GList* movieList = getTheaterMovies(t);
 354         g_list_foreach(movieList, (GFunc) showMovie, NULL);
 355         deleteMovieList(movieList);
 356         printf("\n");
 357 }
 358
 359 static char*
 360 get_query_uri (const char *city)
 361 {
 362         char *query_uri = g_strdup_printf (BASE_URL,
 363                         city);
 364         g_message ("%s\n", query_uri);
 365
 366         return query_uri;
 367 }
 368
 369 int main (int argc, char ** argv)
 370 {
 371
 372         if (argc != 2) {
 373                 printf("usage: gmovies city_name\n");
 374                 exit(-1);
 375         }
 376
 377         SoupSession *session;
 378         SoupMessage *message;
 379         char *query_uri;
 380         const gchar *payload;
 381         const char *city = argv[1];
 382         gssize len;
 383
 384         g_thread_init(NULL);
 385         g_type_init();
 386
 387         session = soup_session_async_new ();
 388         query_uri = get_query_uri (city);
 389         message = soup_message_new ("GET", query_uri);
 390
 391         soup_session_send_message (session, message);
 392
 393         payload = message->response_body->data;
 394         len = message->response_body->length;
 395
 396         htmlDocPtr doc = htmlReadMemory(payload, len, "http://movies.google.com", "UTF-8", HTML_PARSE_NOERROR | HTML_PARSE_NOWARNING);
 397
 398         xmlNodePtr root = xmlDocGetRootElement(doc);//html
 399
 400         //get the body node
 401         xmlNodePtr movieResults = getMovieResultsDiv(root);
 402
 403         if (movieResults == NULL) {
 404                 printf("NO RESULTS.\n");
 405                 exit(-1);
 406         }
 407
 408         GList* theaterList = getTheaterList(movieResults);
 409
 410         g_list_foreach(theaterList, (GFunc) showTheater, NULL);
 411
 412         deleteTheaterList(theaterList);
 413
 414         xmlFreeDoc(doc);
 415
 416         g_free(query_uri);
 417         g_object_unref(session);
 418
 419         exit(0);
 420
 421 }
 422