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