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                         striplt(msg->cm_fields['A']);
 148                 }
 149                 else {
 150                         msg->cm_fields['A'] = strdup("rss");
 151                 }
 152
 153                 msg->cm_fields['N'] = strdup(NODENAME);
 154                 if (ri->title != NULL) {
 155                         msg->cm_fields['U'] = html_to_ascii(ri->title, strlen(ri->title), 512, 0);
 156                         striplt(msg->cm_fields['U']);
 157                 }
 158                 msg->cm_fields['T'] = malloc(64);
 159                 snprintf(msg->cm_fields['T'], 64, "%ld", ri->pubdate);
 160                 if (ri->channel_title != NULL) {
 161                         if (!IsEmptyStr(ri->channel_title)) {
 162                                 msg->cm_fields['O'] = strdup(ri->channel_title);
 163                         }
 164                 }
 165
 166                 msglen = 1024 + strlen(ri->link) + strlen(ri->description) ;
 167                 msg->cm_fields['M'] = malloc(msglen);
 168                 snprintf(msg->cm_fields['M'], msglen,
 169                         "Content-type: text/html\r\n\r\n"
 170                         "<html><body>\n"
 171                         "%s<br><br>\n"
 172                         "<a href=\"%s\">%s</a>\n"
 173                         "</body></html>\n"
 174                         ,
 175                         ri->description,
 176                         ri->link, ri->link
 177                 );
 178
 179                 CtdlSubmitMsg(msg, recp, NULL, 0);
 180                 CtdlFreeMessage(msg);
 181
 182                 /* write the uidl to the use table so we don't store this item again */
 183                 strcpy(ut.ut_msgid, utmsgid);
 184                 ut.ut_timestamp = time(NULL);
 185                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 186         }
 187         free_recipients(recp);
 188 }
 189
 190
 191
 192 /*
 193  * Convert an RDF/RSS datestamp into a time_t
 194  */
 195 time_t rdf_parsedate(char *p)
 196 {
 197         struct tm tm;
 198         time_t t = 0;
 199
 200         if (!p) return 0L;
 201         if (strlen(p) < 10) return 0L;
 202
 203         memset(&tm, 0, sizeof tm);
 204
 205         /* YYYY-MM-DDTHH:MM format...
 206          */
 207         if ( (p[4] == '-') && (p[7] == '-') ) {
 208                 tm.tm_year = atoi(&p[0]) - 1900;
 209                 tm.tm_mon = atoi(&p[5]) - 1;
 210                 tm.tm_mday = atoi(&p[8]);
 211                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 212                         tm.tm_hour = atoi(&p[11]);
 213                         tm.tm_min = atoi(&p[14]);
 214                 }
 215                 return mktime(&tm);
 216         }
 217
 218         /* hmm... try RFC822 date stamp format */
 219
 220         t = parsedate(p);
 221         if (t > 0) return(t);
 222
 223         /* yeesh.  ok, just return the current date and time. */
 224         return(time(NULL));
 225 }
 226
 227
 228
 229 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
 230         struct rss_item *ri = (struct rss_item *) data;
 231         char el[256];
 232         char *sep = NULL;
 233
 234         /* Axe the namespace, we don't care about it */
 235         safestrncpy(el, supplied_el, sizeof el);
 236         while (sep = strchr(el, ':'), sep) {
 237                 strcpy(el, ++sep);
 238         }
 239
 240         if (!strcasecmp(el, "item")) {
 241                 ++ri->item_tag_nesting;
 242
 243                 /* Initialize the feed item data structure */
 244                 if (ri->guid != NULL) free(ri->guid);
 245                 ri->guid = NULL;
 246                 if (ri->title != NULL) free(ri->title);
 247                 ri->title = NULL;
 248                 if (ri->link != NULL) free(ri->link);
 249                 ri->link = NULL;
 250                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 251                 ri->author_or_creator = NULL;
 252                 if (ri->description != NULL) free(ri->description);
 253                 ri->description = NULL;
 254
 255                 /* Throw away any existing character data */
 256                 if (ri->chardata_len > 0) {
 257                         free(ri->chardata);
 258                         ri->chardata = 0;
 259                         ri->chardata_len = 0;
 260                 }
 261         }
 262
 263
 264
 265 }
 266
 267 void rss_xml_end(void *data, const char *supplied_el) {
 268         struct rss_item *ri = (struct rss_item *) data;
 269         char el[256];
 270         char *sep = NULL;
 271
 272         /* Axe the namespace, we don't care about it */
 273         safestrncpy(el, supplied_el, sizeof el);
 274         while (sep = strchr(el, ':'), sep) {
 275                 strcpy(el, ++sep);
 276         }
 277
 278         if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
 279                 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
 280                 striplt(ri->channel_title);
 281         }
 282
 283         if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
 284                 if (ri->guid != NULL) free(ri->guid);
 285                 striplt(ri->chardata);
 286                 ri->guid = strdup(ri->chardata);
 287         }
 288
 289         if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
 290                 if (ri->title != NULL) free(ri->title);
 291                 striplt(ri->chardata);
 292                 ri->title = strdup(ri->chardata);
 293         }
 294
 295         if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
 296                 if (ri->link != NULL) free(ri->link);
 297                 striplt(ri->chardata);
 298                 ri->link = strdup(ri->chardata);
 299         }
 300
 301         if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
 302                 if (ri->description != NULL) free(ri->description);
 303                 ri->description = strdup(ri->chardata);
 304         }
 305
 306         if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
 307                 striplt(ri->chardata);
 308                 ri->pubdate = rdf_parsedate(ri->chardata);
 309         }
 310
 311         if ( ((!strcasecmp(el, "author")) || (!strcasecmp(el, "creator"))) && (ri->chardata != NULL) ) {
 312                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 313                 striplt(ri->chardata);
 314                 ri->author_or_creator = strdup(ri->chardata);
 315         }
 316
 317         if (!strcasecmp(el, "item")) {
 318                 --ri->item_tag_nesting;
 319                 rss_save_item(ri);
 320         }
 321
 322         if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
 323                 CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 324                 ri->done_parsing = 1;
 325         }
 326
 327         if (ri->chardata_len > 0) {
 328                 free(ri->chardata);
 329                 ri->chardata = 0;
 330                 ri->chardata_len = 0;
 331         }
 332
 333 }
 334
 335
 336 /*
 337  * This callback stores up the data which appears in between tags.
 338  */
 339 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
 340         struct rss_item *ri = (struct rss_item *) data;
 341         int old_len;
 342         int new_len;
 343         char *new_buffer;
 344
 345         old_len = ri->chardata_len;
 346         new_len = old_len + len;
 347         new_buffer = realloc(ri->chardata, new_len + 1);
 348         if (new_buffer != NULL) {
 349                 memcpy(&new_buffer[old_len], s, len);
 350                 new_buffer[new_len] = 0;
 351                 ri->chardata = new_buffer;
 352                 ri->chardata_len = new_len;
 353         }
 354 }
 355
 356
 357
 358 /*
 359  * Callback function for passing libcurl's output to expat for parsing
 360  */
 361 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
 362 {
 363         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
 364         return (size*nmemb);
 365 }
 366
 367
 368
 369 /*
 370  * Begin a feed parse
 371  */
 372 void rss_do_fetching(char *url, char *rooms) {
 373         struct rss_item ri;
 374         XML_Parser xp;
 375
 376         CURL *curl;
 377         CURLcode res;
 378         char errmsg[1024] = "";
 379
 380         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
 381
 382         curl = curl_easy_init();
 383         if (!curl) {
 384                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
 385                 return;
 386         }
 387
 388         xp = XML_ParserCreateNS("UTF-8", ':');
 389         if (!xp) {
 390                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
 391                 curl_easy_cleanup(curl);
 392                 return;
 393         }
 394
 395         curl_easy_setopt(curl, CURLOPT_URL, url);
 396         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 397         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 398         curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
 399         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 400         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 401         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 402         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 403         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
 404         if (!IsEmptyStr(config.c_ip_addr)) {
 405                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 406         }
 407
 408         memset(&ri, 0, sizeof(struct rss_item));
 409         ri.roomlist = rooms;
 410         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
 411         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
 412         XML_SetUserData(xp, &ri);
 413
 414         if (CtdlThreadCheckStop())
 415         {
 416                 XML_ParserFree(xp);
 417                 curl_easy_cleanup(curl);
 418                 return;
 419         }
 420
 421         if (CtdlThreadCheckStop())
 422                 goto shutdown ;
 423
 424         res = curl_easy_perform(curl);
 425         if (res) {
 426                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
 427         }
 428
 429         if (CtdlThreadCheckStop())
 430                 goto shutdown ;
 431
 432         if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
 433
 434
 435 shutdown:
 436         curl_easy_cleanup(curl);
 437         XML_ParserFree(xp);
 438
 439         /* Free the feed item data structure */
 440         if (ri.guid != NULL) free(ri.guid);
 441         ri.guid = NULL;
 442         if (ri.title != NULL) free(ri.title);
 443         ri.title = NULL;
 444         if (ri.link != NULL) free(ri.link);
 445         ri.link = NULL;
 446         if (ri.author_or_creator != NULL) free(ri.author_or_creator);
 447         ri.author_or_creator = NULL;
 448         if (ri.description != NULL) free(ri.description);
 449         ri.description = NULL;
 450         if (ri.chardata_len > 0) {
 451                 free(ri.chardata);
 452                 ri.chardata = 0;
 453                 ri.chardata_len = 0;
 454         }
 455 }
 456
 457
 458 /*
 459  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 460  */
 461 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 462 {
 463         char filename[PATH_MAX];
 464         char buf[1024];
 465         char instr[32];
 466         FILE *fp;
 467         char feedurl[256];
 468         struct rssnetcfg *rncptr = NULL;
 469         struct rssnetcfg *use_this_rncptr = NULL;
 470         int len = 0;
 471         char *ptr = NULL;
 472
 473         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
 474
 475         if (CtdlThreadCheckStop())
 476                 return;
 477
 478         /* Only do net processing for rooms that have netconfigs */
 479         fp = fopen(filename, "r");
 480         if (fp == NULL) {
 481                 return;
 482         }
 483
 484         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
 485                 buf[strlen(buf)-1] = 0;
 486
 487                 extract_token(instr, buf, 0, '|', sizeof instr);
 488                 if (!strcasecmp(instr, "rssclient")) {
 489
 490                         use_this_rncptr = NULL;
 491
 492                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
 493
 494                         /* If any other rooms have requested the same feed, then we will just add this
 495                          * room to the target list for that client request.
 496                          */
 497                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
 498                                 if (!strcmp(rncptr->url, feedurl)) {
 499                                         use_this_rncptr = rncptr;
 500                                 }
 501                         }
 502
 503                         /* Otherwise create a new client request */
 504                         if (use_this_rncptr == NULL) {
 505                                 rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
 506                                 if (rncptr != NULL) {
 507                                         rncptr->next = rnclist;
 508                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
 509                                         rncptr->rooms = NULL;
 510                                         rnclist = rncptr;
 511                                         use_this_rncptr = rncptr;
 512                                 }
 513                         }
 514
 515                         /* Add the room name to the request */
 516                         if (use_this_rncptr != NULL) {
 517                                 if (use_this_rncptr->rooms == NULL) {
 518                                         rncptr->rooms = strdup(qrbuf->QRname);
 519                                 }
 520                                 else {
 521                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
 522                                         ptr = realloc(use_this_rncptr->rooms, len);
 523                                         if (ptr != NULL) {
 524                                                 strcat(ptr, "|");
 525                                                 strcat(ptr, qrbuf->QRname);
 526                                                 use_this_rncptr->rooms = ptr;
 527                                         }
 528                                 }
 529                         }
 530                 }
 531
 532         }
 533
 534         fclose(fp);
 535
 536 }
 537
 538 /*
 539  * Scan for rooms that have RSS client requests configured
 540  */
 541 void *rssclient_scan(void *args) {
 542         static time_t last_run = 0L;
 543         static int doing_rssclient = 0;
 544         struct rssnetcfg *rptr = NULL;
 545         struct CitContext rssclientCC;
 546
 547         /* Give this thread its own private CitContext */
 548         CtdlFillSystemContext(&rssclientCC, "rssclient");
 549         citthread_setspecific(MyConKey, (void *)&rssclientCC );
 550
 551         CtdlThreadAllocTSD();
 552
 553         /*
 554          * This is a simple concurrency check to make sure only one rssclient run
 555          * is done at a time.  We could do this with a mutex, but since we
 556          * don't really require extremely fine granularity here, we'll do it
 557          * with a static variable instead.
 558          */
 559         if (doing_rssclient) return NULL;
 560         doing_rssclient = 1;
 561
 562         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
 563         ForEachRoom(rssclient_scan_room, NULL);
 564
 565         while (rnclist != NULL && !CtdlThreadCheckStop()) {
 566                 rss_do_fetching(rnclist->url, rnclist->rooms);
 567                 rptr = rnclist;
 568                 rnclist = rnclist->next;
 569                 if (rptr->rooms != NULL) free(rptr->rooms);
 570                 free(rptr);
 571         }
 572
 573         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
 574         last_run = time(NULL);
 575         doing_rssclient = 0;
 576         if (!CtdlThreadCheckStop())
 577                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
 578         else
 579                 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
 580         return NULL;
 581 }
 582
 583
 584 CTDL_MODULE_INIT(rssclient)
 585 {
 586         if (threading)
 587         {
 588                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
 589                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
 590         }
 591         /* return our Subversion id for the Log */
 592         return "$Id: serv_rssclient.c 5652 2007-10-29 20:14:48Z ajc $";
 593 }