citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * $Id$
   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                 if (ri->link == NULL)
 166                         ri->link = strdup("");
 167                 msglen += 1024 + strlen(ri->link) + strlen(ri->description) ;
 168                 msg->cm_fields['M'] = malloc(msglen);
 169                 snprintf(msg->cm_fields['M'], msglen,
 170                         "Content-type: text/html; charset=\"UTF-8\"\r\n\r\n"
 171                         "<html><body>\n"
 172                         "%s<br><br>\n"
 173                         "<a href=\"%s\">%s</a>\n"
 174                         "</body></html>\n"
 175                         ,
 176                         ri->description,
 177                         ri->link, ri->link
 178                 );
 179
 180                 CtdlSubmitMsg(msg, recp, NULL, 0);
 181                 CtdlFreeMessage(msg);
 182
 183                 /* write the uidl to the use table so we don't store this item again */
 184                 strcpy(ut.ut_msgid, utmsgid);
 185                 ut.ut_timestamp = time(NULL);
 186                 cdb_store(CDB_USETABLE, utmsgid, strlen(utmsgid), &ut, sizeof(struct UseTable) );
 187         }
 188         free_recipients(recp);
 189 }
 190
 191
 192
 193 /*
 194  * Convert an RDF/RSS datestamp into a time_t
 195  */
 196 time_t rdf_parsedate(char *p)
 197 {
 198         struct tm tm;
 199         time_t t = 0;
 200
 201         if (!p) return 0L;
 202         if (strlen(p) < 10) return 0L;
 203
 204         memset(&tm, 0, sizeof tm);
 205
 206         /* YYYY-MM-DDTHH:MM format...
 207          */
 208         if ( (p[4] == '-') && (p[7] == '-') ) {
 209                 tm.tm_year = atoi(&p[0]) - 1900;
 210                 tm.tm_mon = atoi(&p[5]) - 1;
 211                 tm.tm_mday = atoi(&p[8]);
 212                 if ( (p[10] == 'T') && (p[13] == ':') ) {
 213                         tm.tm_hour = atoi(&p[11]);
 214                         tm.tm_min = atoi(&p[14]);
 215                 }
 216                 return mktime(&tm);
 217         }
 218
 219         /* hmm... try RFC822 date stamp format */
 220
 221         t = parsedate(p);
 222         if (t > 0) return(t);
 223
 224         /* yeesh.  ok, just return the current date and time. */
 225         return(time(NULL));
 226 }
 227
 228
 229
 230 void rss_xml_start(void *data, const char *supplied_el, const char **attr) {
 231         struct rss_item *ri = (struct rss_item *) data;
 232         char el[256];
 233         char *sep = NULL;
 234
 235         /* Axe the namespace, we don't care about it */
 236         safestrncpy(el, supplied_el, sizeof el);
 237         while (sep = strchr(el, ':'), sep) {
 238                 strcpy(el, ++sep);
 239         }
 240
 241         if (!strcasecmp(el, "item")) {
 242                 ++ri->item_tag_nesting;
 243
 244                 /* Initialize the feed item data structure */
 245                 if (ri->guid != NULL) free(ri->guid);
 246                 ri->guid = NULL;
 247                 if (ri->title != NULL) free(ri->title);
 248                 ri->title = NULL;
 249                 if (ri->link != NULL) free(ri->link);
 250                 ri->link = NULL;
 251                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 252                 ri->author_or_creator = NULL;
 253                 if (ri->description != NULL) free(ri->description);
 254                 ri->description = NULL;
 255
 256                 /* Throw away any existing character data */
 257                 if (ri->chardata_len > 0) {
 258                         free(ri->chardata);
 259                         ri->chardata = 0;
 260                         ri->chardata_len = 0;
 261                 }
 262         }
 263
 264
 265
 266 }
 267
 268 void rss_xml_end(void *data, const char *supplied_el) {
 269         struct rss_item *ri = (struct rss_item *) data;
 270         char el[256];
 271         char *sep = NULL;
 272
 273         /* Axe the namespace, we don't care about it */
 274         safestrncpy(el, supplied_el, sizeof el);
 275         while (sep = strchr(el, ':'), sep) {
 276                 strcpy(el, ++sep);
 277         }
 278
 279         if ( (!strcasecmp(el, "title")) && (ri->item_tag_nesting == 0) && (ri->chardata != NULL) ) {
 280                 safestrncpy(ri->channel_title, ri->chardata, sizeof ri->channel_title);
 281                 striplt(ri->channel_title);
 282         }
 283
 284         if ( (!strcasecmp(el, "guid")) && (ri->chardata != NULL) ) {
 285                 if (ri->guid != NULL) free(ri->guid);
 286                 striplt(ri->chardata);
 287                 ri->guid = strdup(ri->chardata);
 288         }
 289
 290         if ( (!strcasecmp(el, "title")) && (ri->chardata != NULL) ) {
 291                 if (ri->title != NULL) free(ri->title);
 292                 striplt(ri->chardata);
 293                 ri->title = strdup(ri->chardata);
 294         }
 295
 296         if ( (!strcasecmp(el, "link")) && (ri->chardata != NULL) ) {
 297                 if (ri->link != NULL) free(ri->link);
 298                 striplt(ri->chardata);
 299                 ri->link = strdup(ri->chardata);
 300         }
 301
 302         if ( (!strcasecmp(el, "description")) && (ri->chardata != NULL) ) {
 303                 if (ri->description != NULL) free(ri->description);
 304                 ri->description = strdup(ri->chardata);
 305         }
 306
 307         if ( ((!strcasecmp(el, "pubdate")) || (!strcasecmp(el, "date"))) && (ri->chardata != NULL) ) {
 308                 striplt(ri->chardata);
 309                 ri->pubdate = rdf_parsedate(ri->chardata);
 310         }
 311
 312         if ( ((!strcasecmp(el, "author")) || (!strcasecmp(el, "creator"))) && (ri->chardata != NULL) ) {
 313                 if (ri->author_or_creator != NULL) free(ri->author_or_creator);
 314                 striplt(ri->chardata);
 315                 ri->author_or_creator = strdup(ri->chardata);
 316         }
 317
 318         if (!strcasecmp(el, "item")) {
 319                 --ri->item_tag_nesting;
 320                 rss_save_item(ri);
 321         }
 322
 323         if ( (!strcasecmp(el, "rss")) || (!strcasecmp(el, "rdf")) ) {
 324                 CtdlLogPrintf(CTDL_DEBUG, "End of feed detected.  Closing parser.\n");
 325                 ri->done_parsing = 1;
 326         }
 327
 328         if (ri->chardata_len > 0) {
 329                 free(ri->chardata);
 330                 ri->chardata = 0;
 331                 ri->chardata_len = 0;
 332         }
 333
 334 }
 335
 336
 337 /*
 338  * This callback stores up the data which appears in between tags.
 339  */
 340 void rss_xml_chardata(void *data, const XML_Char *s, int len) {
 341         struct rss_item *ri = (struct rss_item *) data;
 342         int old_len;
 343         int new_len;
 344         char *new_buffer;
 345
 346         old_len = ri->chardata_len;
 347         new_len = old_len + len;
 348         new_buffer = realloc(ri->chardata, new_len + 1);
 349         if (new_buffer != NULL) {
 350                 memcpy(&new_buffer[old_len], s, len);
 351                 new_buffer[new_len] = 0;
 352                 ri->chardata = new_buffer;
 353                 ri->chardata_len = new_len;
 354         }
 355 }
 356
 357
 358
 359 /*
 360  * Callback function for passing libcurl's output to expat for parsing
 361  */
 362 size_t rss_libcurl_callback(void *ptr, size_t size, size_t nmemb, void *stream)
 363 {
 364         XML_Parse((XML_Parser)stream, ptr, (size * nmemb), 0);
 365         return (size*nmemb);
 366 }
 367
 368
 369
 370 /*
 371  * Begin a feed parse
 372  */
 373 void rss_do_fetching(char *url, char *rooms) {
 374         struct rss_item ri;
 375         XML_Parser xp;
 376
 377         CURL *curl;
 378         CURLcode res;
 379         char errmsg[1024] = "";
 380
 381         CtdlLogPrintf(CTDL_DEBUG, "Fetching RSS feed <%s>\n", url);
 382
 383         curl = curl_easy_init();
 384         if (!curl) {
 385                 CtdlLogPrintf(CTDL_ALERT, "Unable to initialize libcurl.\n");
 386                 return;
 387         }
 388
 389         xp = XML_ParserCreateNS("UTF-8", ':');
 390         if (!xp) {
 391                 CtdlLogPrintf(CTDL_ALERT, "Cannot create XML parser!\n");
 392                 curl_easy_cleanup(curl);
 393                 return;
 394         }
 395
 396         curl_easy_setopt(curl, CURLOPT_URL, url);
 397         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, 0);
 398         curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, 0);
 399         curl_easy_setopt(curl, CURLOPT_WRITEDATA, xp);
 400         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, rss_libcurl_callback);
 401         curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, errmsg);
 402         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
 403 #ifdef CURLOPT_HTTP_CONTENT_DECODING
 404         curl_easy_setopt(curl, CURLOPT_HTTP_CONTENT_DECODING, 1);
 405         curl_easy_setopt(curl, CURLOPT_ENCODING, "");
 406 #endif
 407         curl_easy_setopt(curl, CURLOPT_USERAGENT, CITADEL);
 408         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 180);           /* die after 180 seconds */
 409         if (!IsEmptyStr(config.c_ip_addr)) {
 410                 curl_easy_setopt(curl, CURLOPT_INTERFACE, config.c_ip_addr);
 411         }
 412
 413         memset(&ri, 0, sizeof(struct rss_item));
 414         ri.roomlist = rooms;
 415         XML_SetElementHandler(xp, rss_xml_start, rss_xml_end);
 416         XML_SetCharacterDataHandler(xp, rss_xml_chardata);
 417         XML_SetUserData(xp, &ri);
 418
 419         if (CtdlThreadCheckStop())
 420         {
 421                 XML_ParserFree(xp);
 422                 curl_easy_cleanup(curl);
 423                 return;
 424         }
 425
 426         if (CtdlThreadCheckStop())
 427                 goto shutdown ;
 428
 429         res = curl_easy_perform(curl);
 430         if (res) {
 431                 CtdlLogPrintf(CTDL_ALERT, "libcurl error %d: %s\n", res, errmsg);
 432         }
 433
 434         if (CtdlThreadCheckStop())
 435                 goto shutdown ;
 436
 437         if (ri.done_parsing == 0) XML_Parse(xp, "", 0, 1);
 438
 439
 440 shutdown:
 441         curl_easy_cleanup(curl);
 442         XML_ParserFree(xp);
 443
 444         /* Free the feed item data structure */
 445         if (ri.guid != NULL) free(ri.guid);
 446         ri.guid = NULL;
 447         if (ri.title != NULL) free(ri.title);
 448         ri.title = NULL;
 449         if (ri.link != NULL) free(ri.link);
 450         ri.link = NULL;
 451         if (ri.author_or_creator != NULL) free(ri.author_or_creator);
 452         ri.author_or_creator = NULL;
 453         if (ri.description != NULL) free(ri.description);
 454         ri.description = NULL;
 455         if (ri.chardata_len > 0) {
 456                 free(ri.chardata);
 457                 ri.chardata = 0;
 458                 ri.chardata_len = 0;
 459         }
 460 }
 461
 462
 463 /*
 464  * Scan a room's netconfig to determine whether it is requesting any RSS feeds
 465  */
 466 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 467 {
 468         char filename[PATH_MAX];
 469         char buf[1024];
 470         char instr[32];
 471         FILE *fp;
 472         char feedurl[256];
 473         struct rssnetcfg *rncptr = NULL;
 474         struct rssnetcfg *use_this_rncptr = NULL;
 475         int len = 0;
 476         char *ptr = NULL;
 477
 478         assoc_file_name(filename, sizeof filename, qrbuf, ctdl_netcfg_dir);
 479
 480         if (CtdlThreadCheckStop())
 481                 return;
 482
 483         /* Only do net processing for rooms that have netconfigs */
 484         fp = fopen(filename, "r");
 485         if (fp == NULL) {
 486                 return;
 487         }
 488
 489         while (fgets(buf, sizeof buf, fp) != NULL && !CtdlThreadCheckStop()) {
 490                 buf[strlen(buf)-1] = 0;
 491
 492                 extract_token(instr, buf, 0, '|', sizeof instr);
 493                 if (!strcasecmp(instr, "rssclient")) {
 494
 495                         use_this_rncptr = NULL;
 496
 497                         extract_token(feedurl, buf, 1, '|', sizeof feedurl);
 498
 499                         /* If any other rooms have requested the same feed, then we will just add this
 500                          * room to the target list for that client request.
 501                          */
 502                         for (rncptr=rnclist; rncptr!=NULL; rncptr=rncptr->next) {
 503                                 if (!strcmp(rncptr->url, feedurl)) {
 504                                         use_this_rncptr = rncptr;
 505                                 }
 506                         }
 507
 508                         /* Otherwise create a new client request */
 509                         if (use_this_rncptr == NULL) {
 510                                 rncptr = (struct rssnetcfg *) malloc(sizeof(struct rssnetcfg));
 511                                 if (rncptr != NULL) {
 512                                         rncptr->next = rnclist;
 513                                         safestrncpy(rncptr->url, feedurl, sizeof rncptr->url);
 514                                         rncptr->rooms = NULL;
 515                                         rnclist = rncptr;
 516                                         use_this_rncptr = rncptr;
 517                                 }
 518                         }
 519
 520                         /* Add the room name to the request */
 521                         if (use_this_rncptr != NULL) {
 522                                 if (use_this_rncptr->rooms == NULL) {
 523                                         rncptr->rooms = strdup(qrbuf->QRname);
 524                                 }
 525                                 else {
 526                                         len = strlen(use_this_rncptr->rooms) + strlen(qrbuf->QRname) + 5;
 527                                         ptr = realloc(use_this_rncptr->rooms, len);
 528                                         if (ptr != NULL) {
 529                                                 strcat(ptr, "|");
 530                                                 strcat(ptr, qrbuf->QRname);
 531                                                 use_this_rncptr->rooms = ptr;
 532                                         }
 533                                 }
 534                         }
 535                 }
 536
 537         }
 538
 539         fclose(fp);
 540
 541 }
 542
 543 /*
 544  * Scan for rooms that have RSS client requests configured
 545  */
 546 void *rssclient_scan(void *args) {
 547         static time_t last_run = 0L;
 548         static int doing_rssclient = 0;
 549         struct rssnetcfg *rptr = NULL;
 550         struct CitContext rssclientCC;
 551
 552         /* Give this thread its own private CitContext */
 553         CtdlFillSystemContext(&rssclientCC, "rssclient");
 554         citthread_setspecific(MyConKey, (void *)&rssclientCC );
 555
 556         CtdlThreadAllocTSD();
 557
 558         /*
 559          * This is a simple concurrency check to make sure only one rssclient run
 560          * is done at a time.  We could do this with a mutex, but since we
 561          * don't really require extremely fine granularity here, we'll do it
 562          * with a static variable instead.
 563          */
 564         if (doing_rssclient) return NULL;
 565         doing_rssclient = 1;
 566
 567         CtdlLogPrintf(CTDL_DEBUG, "rssclient started\n");
 568         ForEachRoom(rssclient_scan_room, NULL);
 569
 570         while (rnclist != NULL && !CtdlThreadCheckStop()) {
 571                 rss_do_fetching(rnclist->url, rnclist->rooms);
 572                 rptr = rnclist;
 573                 rnclist = rnclist->next;
 574                 if (rptr->rooms != NULL) free(rptr->rooms);
 575                 free(rptr);
 576         }
 577
 578         CtdlLogPrintf(CTDL_DEBUG, "rssclient ended\n");
 579         last_run = time(NULL);
 580         doing_rssclient = 0;
 581         if (!CtdlThreadCheckStop())
 582                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, last_run + config.c_net_freq);
 583         else
 584                 CtdlLogPrintf(CTDL_DEBUG, "rssclient: Task STOPPED.\n");
 585         return NULL;
 586 }
 587
 588
 589 CTDL_MODULE_INIT(rssclient)
 590 {
 591         if (threading)
 592         {
 593                 CtdlLogPrintf(CTDL_INFO, "%s\n", curl_version());
 594                 CtdlThreadSchedule ("RSS Client", CTDLTHREAD_BIGSTACK, rssclient_scan, NULL, 0);
 595         }
 596         /* return our Subversion id for the Log */
 597         return "$Id$";
 598 }