citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * $Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $
   3  *
   4  * Bring external RSS feeds into rooms.
   5  *
   6  */
   7
   8 #include <stdlib.h>
   9 #include <unistd.h>
  10 #include <stdio.h>
  11
  12 #if TIME_WITH_SYS_TIME
  13 # include <sys/time.h>
  14 # include <time.h>
  15 #else
  16 # if HAVE_SYS_TIME_H
  17 #  include <sys/time.h>
  18 # else
  19 #  include <time.h>
  20 # endif
  21 #endif
  22
  23 #include <ctype.h>
  24 #include <string.h>
  25 #include <errno.h>
  26 #include <sys/types.h>
  27 #include <sys/stat.h>
  28 #include <expat.h>
  29 #include <curl/curl.h>
  30 #include <libcitadel.h>
  31 #include "citadel.h"
  32 #include "server.h"
  33 #include "citserver.h"
  34 #include "support.h"
  35 #include "config.h"
  36 #include "threads.h"
  37 #include "room_ops.h"
  38 #include "ctdl_module.h"
  39 #include "clientsocket.h"
  40 #include "msgbase.h"
  41 #include "parsedate.h"
  42 #include "database.h"
  43 #include "citadel_dirs.h"
  44 #include "md5.h"
  45
  46
  47 struct rssnetcfg {
  48         struct rssnetcfg *next;
  49         char url[256];
  50         char *rooms;
  51 };
  52
  53 struct rss_item {
  54         char *chardata;
  55         int chardata_len;
  56         char *roomlist;
  57         int done_parsing;
  58         char *guid;
  59         char *title;
  60         char *link;
  61         char *description;
  62         time_t pubdate;
  63         char channel_title[256];
  64         int item_tag_nesting;
  65         char *author_or_creator;
  66 };
  67
  68 struct rssnetcfg *rnclist = NULL;
  69
  70
  71 /*
  72  * Commit a fetched and parsed RSS item to disk
  73  */
  74 void rss_save_item(struct rss_item *ri) {
  75
  76         struct MD5Context md5context;
  77         u_char rawdigest[MD5_DIGEST_LEN];
  78         int i;
  79         char utmsgid[SIZ];
  80         struct cdbdata *cdbut;
  81         struct UseTable ut;
  82         struct CtdlMessage *msg;
  83         struct recptypes *recp = NULL;
  84         int msglen = 0;
  85
  86         recp = (struct recptypes *) malloc(sizeof(struct recptypes));
  87         if (recp == NULL) return;
  88         memset(recp, 0, sizeof(struct recptypes));
  89         recp->recp_room = strdup(ri->roomlist);
  90         recp->num_room = num_tokens(ri->roomlist, '|');
  91         recp->recptypes_magic = RECPTYPES_MAGIC;
  92
  93         /* Construct a GUID to use in the S_USETABLE table.
  94          * If one is not present in the item itself, make one up.
  95          */
  96         if (ri->guid != NULL) {
  97                 snprintf(utmsgid, sizeof utmsgid, "rss/%s", ri->guid);
  98         }
  99         else {
 100                 MD5Init(&md5context);
 101                 if (ri->title != NULL) {
 102                         MD5Update(&md5context, (unsigned char*)ri->title, strlen(ri->title));
 103                 }
 104                 if (ri->link != NULL) {
 105                         MD5Update(&md5context, (unsigned char*)ri->link, strlen(ri->link));
 106                 }
 107                 MD5Final(rawdigest, &md5context);
 108                 for (i=0; i<MD5_DIGEST_LEN; i++) {
 109                         sprintf(&utmsgid[i*2], "%02X", (unsigned char) (rawdigest[i] & 0xff));
 110                         utmsgid[i*2] = tolower(utmsgid[i*2]);
 111                         utmsgid[(i*2)+1] = tolower(utmsgid[(i*2)+1]);
 112                 }
 113                 strcat(utmsgid, "_rss2ctdl");
 114         }
 115
 116         /* Find out if we've already seen this item */
 117         cdbut = cdb_fetch(CDB_USETABLE, utmsgid, strlen(utmsgid));
 118         if (cdbut != NULL) {
 119                 /* Item has already been seen */
 120                 CtdlLogPrintf(CTDL_DEBUG, "%s has already been seen\n", utmsgid);
 121                 cdb_free(cdbut);
 122
 123                 /* rewrite the record anyway, to update the timestamp */
 124                 strcpy(ut.ut_msgid, utmsgid);
 125                 ut.ut_timestamp = time(NULL);
 126                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 127         }
 128         else {
 129                 /* Item has not been seen, so save it. */
 130
 131                 if (ri->description == NULL) ri->description = strdup("");
 132                 for (i=strlen(ri->description); i>=0; --i) {
 133                         if (isspace(ri->description[i])) {
 134                                 ri->description[i] = ' ';
 135                         }
 136                 }
 137
 138                 msg = malloc(sizeof(struct CtdlMessage));
 139                 memset(msg, 0, sizeof(struct CtdlMessage));
 140                 msg->cm_magic = CTDLMESSAGE_MAGIC;
 141                 msg->cm_anon_type = MES_NORMAL;
 142                 msg->cm_format_type = FMT_RFC822;
 143
 144                 if (ri->author_or_creator != NULL) {
 145                         msg->cm_fields['A'] = html_to_ascii(ri->author_or_creator,
 146                                 strlen(ri->author_or_creator), 512, 0);
 147                 }
 148                 else {
 149                         msg->cm_fields['A'] = strdup("rss");
 150                 }
 151
 152                 msg->cm_fields['N'] = strdup(NODENAME);
 153                 if (ri->title != NULL) {
 154                         msg->cm_fields['U'] = html_to_ascii(ri->title, strlen(ri->title), 512, 0);
 155                 }
 156                 msg->cm_fields['T'] = malloc(64);
 157                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 158                 if (ri->channel_title != NULL) {
 159                         if (!IsEmptyStr(ri->channel_title)) {
 160                                 msg->cm_fields['O'] = strdup(ri->channel_title);
 161                         }
 162                 }
 163
 164                 msglen = 1024 + strlen(ri->link) + strlen(ri->description) ;
 165                 msg->cm_fields['M'] = malloc(msglen);
 166                 snprintf(msg->cm_fields['M'], msglen,
 167                         "Content-type: text/html\r\n\r\n"
 168                         "<html><body>\n"
 169                         "%s<br><br>\n"
 170                         "<a href=\"%s\">%s</a>\n"
 171                         "</body></html>\n"
 172                         ,
 173                         ri->description,
 174                         ri->link, ri->link
 175                 );
 176
 177                 CtdlSubmitMsg(msg, recp, NULL, 0);
 178                 CtdlFreeMessage(msg);
 179
 180                 /* write the uidl to the use table so we don't store this item again */
 181                 strcpy(ut.ut_msgid, utmsgid);
 182                 ut.ut_timestamp = time(NULL);
 183                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 184         }
 185         free_recipients(recp);
 186 }
 187
 188
 189
 190 /*
 191  * Convert an RDF/RSS datestamp into a time_t
 192  */
 193 time_t rdf_parsedate(char *p)
 194 {
 195         struct tm tm;
 196         time_t t = 0;
 197
 198         if (!p) return 0L;
 199         if (strlen(p) < 10) return 0L;
 200
 201         memset(&tm, 0, sizeof tm);
 202
 203         /* YYYY-MM-DDTHH:MM format...
 204          */
 205         if ( (p[4] == '-') && (p[7] == '-') ) {
 206                 tm.tm_year = atoi(&p[0]) - 1900;
 207                 tm.tm_mon = atoi(&p[5]) - 1;
 208                 tm.tm_mday = atoi(&p[8]);
 209                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 210                         tm.tm_hour = atoi(&p[11]);
 211                         tm.tm_min = atoi(&p[14]);
 212                 }
 213                 return mktime(&tm);
 214         }
 215
 216         /* hmm... try RFC822 date stamp format */
 217
 218         t = parsedate(p);
 219         if (t > 0) return(t);
 220
 221         /* yeesh.  ok, just return the current date and time. */
 222         return(time(NULL));
 223 }
 224
 225
 226
 227 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
 228         struct rss_item *ri = (struct rss_item *) data;
 229         char el[256];
 230         char *sep = NULL;
 231
 232         /* Axe the namespace, we don't care about it */
 233         safestrncpy(el, supplied_el, sizeof el);
 234         while (sep = strchr(el, ':'), sep) {
 235                 strcpy(el, ++sep);
 236         }
 237
 238         if (!strcasecmp(el, "item")) {
 239                 ++ri->item_tag_nesting;
 240
 241                 /* Initialize the feed item data structure */
 242                 if (ri->guid != NULL) free(ri->guid);
 243                 ri->guid = NULL;
 244                 if (ri->title != NULL) free(ri->title);
 245                 ri->title = NULL;
 246                 if (ri->link != NULL) free(ri->link);
 247                 ri->link = NULL;
 248                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 249                 ri->author_or_creator = NULL;
 250                 if (ri->description != NULL) free(ri->description);
 251                 ri->description = NULL;
 252
 253                 /* Throw away any existing character data */
 254                 if (ri->chardata_len > 0) {
 255                         free(ri->chardata);
 256                         ri->chardata = 0;
 257                         ri->chardata_len = 0;
 258                 }
 259         }
 260
 261
 262
 263 }
 264
 265 void rss_xml_end(void *data, const char *supplied_el) {
 266         struct rss_item *ri = (struct rss_item *) data;
 267         char el[256];
 268         char *sep = NULL;
 269
 270         /* Axe the namespace, we don't care about it */
 271         safestrncpy(el, supplied_el, sizeof el);
 272         while (sep = strchr(el, ':'), sep) {
 273                 strcpy(el, ++sep);
 274         }
 275
 276         if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
 277                 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
 278                 striplt(ri->channel_title);
 279         }
 280
 281         if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
 282                 if (ri->guid != NULL) free(ri->guid);
 283                 striplt(ri->chardata);
 284                 ri->guid = strdup(ri->chardata);
 285         }
 286
 287         if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
 288                 if (ri->title != NULL) free(ri->title);
 289                 striplt(ri->chardata);
 290                 ri->title = strdup(ri->chardata);
 291         }
 292
 293         if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
 294                 if (ri->link != NULL) free(ri->link);
 295                 striplt(ri->chardata);
 296                 ri->link = strdup(ri->chardata);
 297         }
 298
 299         if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
 300                 if (ri->description != NULL) free(ri->description);
 301                 ri->description = strdup(ri->chardata);
 302         }
 303
 304         if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
 305                 striplt(ri->chardata);
 306                 ri->pubdate = rdf_parsedate(ri->chardata);
 307         }
 308
 309         if ( ((!strcasecmp(el, "author")) || (!strcasecmp(el, "creator"))) && (ri->chardata != NULL) ) {
 310                 if (ri->author_or_creator != NULL) free(ri->title);
 311                 striplt(ri->chardata);
 312                 ri->author_or_creator = strdup(ri->chardata);
 313         }
 314
 315         if (!strcasecmp(el, "item")) {
 316                 --ri->item_tag_nesting;
 317                 rss_save_item(ri);
 318         }
 319
 320         if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
 321                 CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 322                 ri->done_parsing = 1;
 323         }
 324
 325         if (ri->chardata_len > 0) {
 326                 free(ri->chardata);
 327                 ri->chardata = 0;
 328                 ri->chardata_len = 0;
 329         }
 330
 331 }
 332
 333
 334 /*
 335  * This callback stores up the data which appears in between tags.
 336  */
 337 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
 338         struct rss_item *ri = (struct rss_item *) data;
 339         int old_len;
 340         int new_len;
 341         char *new_buffer;
 342
 343         old_len = ri->chardata_len;
 344         new_len = old_len + len;
 345         new_buffer = realloc(ri->chardata, new_len + 1);
 346         if (new_buffer != NULL) {
 347                 memcpy(&new_buffer[old_len], s, len);
 348                 new_buffer[new_len] = 0;
 349                 ri->chardata = new_buffer;
 350                 ri->chardata_len = new_len;
 351         }
 352 }
 353
 354
 355
 356 /*
 357  * Callback function for passing libcurl's output to expat for parsing
 358  */
 359 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
 360 {
 361         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
 362         return (size*nmemb);
 363 }
 364
 365
 366
 367 /*
 368  * Begin a feed parse
 369  */
 370 void rss_do_fetching(char *url, char *rooms) {
 371         struct rss_item ri;
 372         XML_Parser xp;
 373
 374         CURL *curl;
 375         CURLcode res;
 376         char errmsg[1024] = "";
 377
 378         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
 379
 380         curl = curl_easy_init();
 381         if (!curl) {
 382                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
 383                 return;
 384         }
 385
 386         xp = XML_ParserCreateNS("UTF-8", ':');
 387         if (!xp) {
 388                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
 389                 curl_easy_cleanup(curl);
 390                 return;
 391         }
 392
 393         curl_easy_setopt(curl, CURLOPT_URL, url);
 394         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 395         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 396         curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
 397         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 398         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 399         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 400         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 401         if (!IsEmptyStr(config.c_ip_addr)) {
 402                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 403         }
 404
 405         memset(&ri, 0, sizeof(struct rss_item));
 406         ri.roomlist = rooms;
 407         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
 408         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
 409         XML_SetUserData(xp, &ri);
 410
 411         if (CtdlThreadCheckStop())
 412         {
 413                 XML_ParserFree(xp);
 414                 curl_easy_cleanup(curl);
 415                 return;
 416         }
 417
 418         if (CtdlThreadCheckStop())
 419                 goto shutdown ;
 420
 421         res = curl_easy_perform(curl);
 422         if (res) {
 423                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
 424         }
 425
 426         if (CtdlThreadCheckStop())
 427                 goto shutdown ;
 428
 429         if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
 430
 431
 432 shutdown:
 433         curl_easy_cleanup(curl);
 434         XML_ParserFree(xp);
 435
 436         /* Free the feed item data structure */
 437         if (ri.guid != NULL) free(ri.guid);
 438         ri.guid = NULL;
 439         if (ri.title != NULL) free(ri.title);
 440         ri.title = NULL;
 441         if (ri.link != NULL) free(ri.link);
 442         ri.link = NULL;
 443         if (ri.author_or_creator != NULL) free(ri.author_or_creator);
 444         ri.author_or_creator = NULL;
 445         if (ri.description != NULL) free(ri.description);
 446         ri.description = NULL;
 447         if (ri.chardata_len > 0) {
 448                 free(ri.chardata);
 449                 ri.chardata = 0;
 450                 ri.chardata_len = 0;
 451         }
 452 }
 453
 454
 455 /*
 456  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 457  */
 458 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 459 {
 460         char filename[PATH_MAX];
 461         char buf[1024];
 462         char instr[32];
 463         FILE *fp;
 464         char feedurl[256];
 465         struct rssnetcfg *rncptr = NULL;
 466         struct rssnetcfg *use_this_rncptr = NULL;
 467         int len = 0;
 468         char *ptr = NULL;
 469
 470         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
 471
 472         if (CtdlThreadCheckStop())
 473                 return;
 474
 475         /* Only do net processing for rooms that have netconfigs */
 476         fp = fopen(filename, "r");
 477         if (fp == NULL) {
 478                 return;
 479         }
 480
 481         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
 482                 buf[strlen(buf)-1] = 0;
 483
 484                 extract_token(instr, buf, 0, '|', sizeof instr);
 485                 if (!strcasecmp(instr, "rssclient")) {
 486
 487                         use_this_rncptr = NULL;
 488
 489                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
 490
 491                         /* If any other rooms have requested the same feed, then we will just add this
 492                          * room to the target list for that client request.
 493                          */
 494                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
 495                                 if (!strcmp(rncptr->url, feedurl)) {
 496                                         use_this_rncptr = rncptr;
 497                                 }
 498                         }
 499
 500                         /* Otherwise create a new client request */
 501                         if (use_this_rncptr == NULL) {
 502                                 rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
 503                                 if (rncptr != NULL) {
 504                                         rncptr->next = rnclist;
 505                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
 506                                         rncptr->rooms = NULL;
 507                                         rnclist = rncptr;
 508                                         use_this_rncptr = rncptr;
 509                                 }
 510                         }
 511
 512                         /* Add the room name to the request */
 513                         if (use_this_rncptr != NULL) {
 514                                 if (use_this_rncptr->rooms == NULL) {
 515                                         rncptr->rooms = strdup(qrbuf->QRname);
 516                                 }
 517                                 else {
 518                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
 519                                         ptr = realloc(use_this_rncptr->rooms, len);
 520                                         if (ptr != NULL) {
 521                                                 strcat(ptr, "|");
 522                                                 strcat(ptr, qrbuf->QRname);
 523                                                 use_this_rncptr->rooms = ptr;
 524                                         }
 525                                 }
 526                         }
 527                 }
 528
 529         }
 530
 531         fclose(fp);
 532
 533 }
 534
 535 /*
 536  * Scan for rooms that have RSS client requests configured
 537  */
 538 void *rssclient_scan(void *args) {
 539         static time_t last_run = 0L;
 540         static int doing_rssclient = 0;
 541         struct rssnetcfg *rptr = NULL;
 542         struct CitContext rssclientCC;
 543
 544         /* Give this thread its own private CitContext */
 545         CtdlFillSystemContext(&rssclientCC, "rssclient");
 546         citthread_setspecific(MyConKey, (void *)&rssclientCC );
 547
 548         CtdlThreadAllocTSD();
 549
 550         /*
 551          * This is a simple concurrency check to make sure only one rssclient run
 552          * is done at a time.  We could do this with a mutex, but since we
 553          * don't really require extremely fine granularity here, we'll do it
 554          * with a static variable instead.
 555          */
 556         if (doing_rssclient) return NULL;
 557         doing_rssclient = 1;
 558
 559         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
 560         ForEachRoom(rssclient_scan_room, NULL);
 561
 562         while (rnclist != NULL && !CtdlThreadCheckStop()) {
 563                 rss_do_fetching(rnclist->url, rnclist->rooms);
 564                 rptr = rnclist;
 565                 rnclist = rnclist->next;
 566                 if (rptr->rooms != NULL) free(rptr->rooms);
 567                 free(rptr);
 568         }
 569
 570         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
 571         last_run = time(NULL);
 572         doing_rssclient = 0;
 573         if (!CtdlThreadCheckStop())
 574                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
 575         else
 576                 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
 577         return NULL;
 578 }
 579
 580
 581 CTDL_MODULE_INIT(rssclient)
 582 {
 583         if (threading)
 584         {
 585                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
 586                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
 587         }
 588         /* return our Subversion id for the Log */
 589         return "$Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $";
 590 }