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