citadel/modules/rssclient/serv_rssclient.c

   1 /*
   2  * Bring external RSS feeds into rooms.
   3  *
   4  * Copyright (c) 2007-2017 by the citadel.org team
   5  *
   6  * This program is open source software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License version 3.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.See the
  12  * GNU General Public License for more details.
  13  */
  14
  15 #include <stdlib.h>
  16 #include <unistd.h>
  17 #include <stdio.h>
  18
  19 #if TIME_WITH_SYS_TIME
  20 # include <sys/time.h>
  21 # include <time.h>
  22 #else
  23 # if HAVE_SYS_TIME_H
  24 #include <sys/time.h>
  25 # else
  26 #include <time.h>
  27 # endif
  28 #endif
  29
  30 #include <ctype.h>
  31 #include <string.h>
  32 #include <errno.h>
  33 #include <sys/types.h>
  34 #include <sys/stat.h>
  35 #include <expat.h>
  36 #include <curl/curl.h>
  37 #include <libcitadel.h>
  38 #include "citadel.h"
  39 #include "server.h"
  40 #include "citserver.h"
  41 #include "support.h"
  42 #include "config.h"
  43 #include "threads.h"
  44 #include "ctdl_module.h"
  45 #include "msgbase.h"
  46 #include "parsedate.h"
  47 #include "database.h"
  48 #include "citadel_dirs.h"
  49 #include "md5.h"
  50 #include "context.h"
  51 #include "internet_addressing.h"
  52
  53 struct rssroom {
  54         struct rssroom *next;
  55         char *room;
  56 };
  57
  58 struct rssurl {
  59         struct rssurl *next;
  60         char *url;
  61         struct rssroom *rooms;
  62 };
  63
  64 struct rssparser {
  65         StrBuf *CData;
  66         struct CtdlMessage *msg;
  67         char *link;
  68         char *description;
  69         char *item_id;
  70         struct rssroom *rooms;
  71 };
  72
  73 time_t last_run = 0L;
  74 struct CitContext rss_CC;
  75 struct rssurl *rsstodo = NULL;
  76
  77
  78 // This handler is called whenever an XML tag opens.
  79 //
  80 void rss_start_element(void *data, const char *el, const char **attribute)
  81 {
  82         struct rssparser *r = (struct rssparser *)data;
  83         int i;
  84
  85         if (
  86                 (!strcasecmp(el, "entry"))
  87                 || (!strcasecmp(el, "item"))
  88         ) {
  89                 // this is the start of a new item(rss) or entry(atom)
  90                 if (r->msg != NULL) {
  91                         CM_Free(r->msg);
  92                         r->msg = NULL;
  93                 }
  94                 r->msg = malloc(sizeof(struct CtdlMessage));
  95                 memset(r->msg, 0, sizeof(struct CtdlMessage));
  96                 r->msg->cm_magic = CTDLMESSAGE_MAGIC;
  97                 r->msg->cm_anon_type = MES_NORMAL;
  98                 r->msg->cm_format_type = FMT_RFC822;
  99         }
 100
 101         else if (!strcasecmp(el, "link")) {                     // atom feeds have the link as an attribute
 102                 for(i = 0; attribute[i]; i += 2) {
 103                         if (!strcasecmp(attribute[i], "href")) {
 104                                 if (r->link != NULL) {
 105                                         free(r->link);
 106                                         r->link = NULL;
 107                                 }
 108                                 r->link = strdup(attribute[i+1]);
 109                                 striplt(r->link);
 110                         }
 111                 }
 112         }
 113 }
 114
 115
 116 // This handler is called whenever an XML tag closes.
 117 //
 118 void rss_end_element(void *data, const char *el)
 119 {
 120         struct rssparser *r = (struct rssparser *)data;
 121
 122         if (                                                    // end of a new item(rss) or entry(atom)
 123                 (!strcasecmp(el, "entry"))
 124                 || (!strcasecmp(el, "item"))
 125         ) {
 126
 127                 if (r->msg != NULL) {                           // Save the message to the rooms
 128
 129                         // use the link as an item id if nothing else is available
 130                         if ((r->item_id == NULL) && (r->link != NULL)) {
 131                                 r->item_id = strdup(r->link);
 132                         }
 133
 134                         // check the use table
 135                         StrBuf *u = NewStrBuf();
 136                         StrBufAppendPrintf(u, "rss/%s", r->item_id);
 137                         int already_seen = CheckIfAlreadySeen(u);
 138                         FreeStrBuf(&u);
 139
 140                         if (already_seen == 0) {
 141
 142                                 // Compose the message text
 143                                 StrBuf *TheMessage = NewStrBuf();
 144                                 StrBufAppendPrintf(TheMessage,
 145                                         "Content-type: text/html\n\n"
 146                                         "\n\n"
 147                                         "<html><head></head><body>"
 148                                 );
 149
 150                                 if (r->description != NULL) {
 151                                         StrBufAppendPrintf(TheMessage, "%s<br><br>\r\n", r->description);
 152                                         free(r->description);
 153                                         r->description = NULL;
 154                                 }
 155
 156                                 if (r->link != NULL) {
 157                                         StrBufAppendPrintf(TheMessage, "<a href=\"%s\">%s</a>\r\n", r->link, r->link);
 158                                         free(r->link);
 159                                         r->link = NULL;
 160                                 }
 161
 162                                 StrBufAppendPrintf(TheMessage, "</body></html>\r\n");
 163                                 CM_SetField(r->msg, eMesageText, ChrPtr(TheMessage), StrLength(TheMessage));
 164                                 FreeStrBuf(&TheMessage);
 165
 166                                 if (CM_IsEmpty(r->msg, eAuthor)) {
 167                                         CM_SetField(r->msg, eAuthor, HKEY("rss"));
 168                                 }
 169
 170                                 if (CM_IsEmpty(r->msg, eTimestamp)) {
 171                                         CM_SetFieldLONG(r->msg, eTimestamp, time(NULL));
 172                                 }
 173
 174                                 // Save it to the room(s)
 175                                 struct rssroom *rr = NULL;
 176                                 long msgnum = (-1);
 177                                 for (rr=r->rooms; rr!=NULL; rr=rr->next) {
 178                                         if (rr == r->rooms) {
 179                                                 msgnum = CtdlSubmitMsg(r->msg, NULL, rr->room, 0);
 180                                         }
 181                                         else {
 182                                                 CtdlSaveMsgPointerInRoom(rr->room, msgnum, 0, NULL);
 183                                         }
 184                                         syslog(LOG_DEBUG, "Saved message %ld to %s", msgnum, rr->room);
 185                                 }
 186                         }
 187                         else {
 188                                 syslog(LOG_DEBUG, "%s was already seen", r->item_id);
 189                         }
 190
 191                         CM_Free(r->msg);
 192                         r->msg = NULL;
 193                 }
 194
 195                 if (r->item_id != NULL) {
 196                         free(r->item_id);
 197                         r->item_id = NULL;
 198                 }
 199         }
 200
 201         else if (!strcasecmp(el, "title")) {                    // item subject (rss and atom)
 202                 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eMsgSubject))) {
 203                         CM_SetField(r->msg, eMsgSubject, ChrPtr(r->CData), StrLength(r->CData));
 204                         striplt(r->msg->cm_fields[eMsgSubject]);
 205                 }
 206         }
 207
 208         else if (!strcasecmp(el, "author")) {                   // author of item (rss and maybe atom)
 209                 if ((r->msg != NULL) && (CM_IsEmpty(r->msg, eAuthor))) {
 210                         CM_SetField(r->msg, eAuthor, ChrPtr(r->CData), StrLength(r->CData));
 211                         striplt(r->msg->cm_fields[eAuthor]);
 212                 }
 213         }
 214
 215         else if (!strcasecmp(el, "pubdate")) {                  // date/time stamp (rss) Sat, 25 Feb 2017 14:28:01 EST
 216                 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
 217                         CM_SetFieldLONG(r->msg, eTimestamp, parsedate(ChrPtr(r->CData)));
 218                 }
 219         }
 220
 221         else if (!strcasecmp(el, "updated")) {                  // date/time stamp (atom) 2003-12-13T18:30:02Z
 222                 if ((r->msg)&&(r->msg->cm_fields[eTimestamp]==NULL)) {
 223                         struct tm t;
 224                         char zulu;
 225                         memset(&t, 0, sizeof t);
 226                         sscanf(ChrPtr(r->CData), "%d-%d-%dT%d:%d:%d%c", &t.tm_year, &t.tm_mon, &t.tm_mday, &t.tm_hour, &t.tm_min, &t.tm_sec, &zulu);
 227                         t.tm_year -= 1900;
 228                         t.tm_mon -= 1;
 229                         CM_SetFieldLONG(r->msg, eTimestamp, mktime(&t));
 230                 }
 231         }
 232
 233         else if (!strcasecmp(el, "link")) {                     // link to story (rss)
 234                 if (r->link != NULL) {
 235                         free(r->link);
 236                         r->link = NULL;
 237                 }
 238                 r->link = strdup(ChrPtr(r->CData));
 239                 striplt(r->link);
 240         }
 241
 242         else if (
 243                 (!strcasecmp(el, "guid"))                       // unique item id (rss)
 244                 || (!strcasecmp(el, "id"))                      // unique item id (atom)
 245         ) {
 246                 if (r->item_id != NULL) {
 247                         free(r->item_id);
 248                         r->item_id = NULL;
 249                 }
 250                 r->item_id = strdup(ChrPtr(r->CData));
 251                 striplt(r->item_id);
 252         }
 253
 254         else if (
 255                 (!strcasecmp(el, "description"))                // message text (rss)
 256                 || (!strcasecmp(el, "summary"))                 // message text (atom)
 257                 || (!strcasecmp(el, "content"))                 // message text (atom)
 258         ) {
 259                 if (r->description != NULL) {
 260                         free(r->description);
 261                         r->description = NULL;
 262                 }
 263                 r->description = strdup(ChrPtr(r->CData));
 264                 striplt(r->description);
 265         }
 266
 267         if (r->CData != NULL) {
 268                 FreeStrBuf(&r->CData);
 269                 r->CData = NULL;
 270         }
 271 }
 272
 273
 274 // This handler is called whenever data appears between opening and closing tags.
 275 //
 276 void rss_handle_data(void *data, const char *content, int length)
 277 {
 278         struct rssparser *r = (struct rssparser *)data;
 279
 280         if (r->CData == NULL) {
 281                 r->CData = NewStrBuf();
 282         }
 283
 284         StrBufAppendBufPlain(r->CData, content, length, 0);
 285 }
 286
 287
 288 // Feed has been downloaded, now parse it.
 289 //
 290 void rss_parse_feed(StrBuf *Feed, struct rssroom *rooms)
 291 {
 292         struct rssparser r;
 293
 294         memset(&r, 0, sizeof r);
 295         r.rooms = rooms;
 296         XML_Parser p = XML_ParserCreate("UTF-8");
 297         XML_SetElementHandler(p, rss_start_element, rss_end_element);
 298         XML_SetCharacterDataHandler(p, rss_handle_data);
 299         XML_SetUserData(p, (void *)&r);
 300         XML_Parse(p, ChrPtr(Feed), StrLength(Feed), XML_TRUE);
 301         XML_ParserFree(p);
 302 }
 303
 304
 305 // Add a feed/room pair into the todo list
 306 //
 307 void rssclient_push_todo(char *rssurl, char *roomname)
 308 {
 309         struct rssurl *r = NULL;
 310         struct rssurl *thisone = NULL;
 311         struct rssroom *newroom = NULL;
 312
 313         syslog(LOG_DEBUG, "rssclient_push_todo(%s, %s)", rssurl, roomname);
 314
 315         for (r=rsstodo; r!=NULL; r=r->next) {
 316                 if (!strcasecmp(r->url, rssurl)) {
 317                         thisone = r;
 318                 }
 319         }
 320         if (thisone == NULL) {
 321                 thisone = malloc(sizeof(struct rssurl));
 322                 thisone->url = strdup(rssurl);
 323                 thisone->rooms = NULL;
 324                 thisone->next = rsstodo;
 325                 rsstodo = thisone;
 326         }
 327
 328         newroom = malloc(sizeof(struct rssroom));
 329         newroom->room = strdup(roomname);
 330         newroom->next = thisone->rooms;
 331         thisone->rooms = newroom;
 332 }
 333
 334
 335 // pull one feed (possibly multiple rooms)
 336 //
 337 void rss_pull_one_feed(struct rssurl *url)
 338 {
 339         CURL *curl;
 340         CURLcode res;
 341         StrBuf *Downloaded = NULL;
 342
 343         syslog(LOG_DEBUG, "rss_pull_one_feed(%s)", url->url);
 344
 345         curl = curl_easy_init();
 346         if (!curl) {
 347                 return;
 348         }
 349
 350         Downloaded = NewStrBuf();
 351
 352         curl_easy_setopt(curl, CURLOPT_URL, url->url);
 353         curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1L);                     // Follow redirects
 354         curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, CurlFillStrBuf_callback); // What to do with downloaded data
 355         curl_easy_setopt(curl, CURLOPT_WRITEDATA, Downloaded);                  // Give it our StrBuf to work with
 356         curl_easy_setopt(curl, CURLOPT_TIMEOUT, 20L);                           // Time out after 20 seconds
 357         res = curl_easy_perform(curl);                                          // Perform the request
 358         if (res != CURLE_OK) {
 359                 syslog(LOG_WARNING, "Failed to load feed: %s", curl_easy_strerror(res));
 360         }
 361         curl_easy_cleanup(curl);
 362
 363         rss_parse_feed(Downloaded, url->rooms);                                 // parse the feed
 364         FreeStrBuf(&Downloaded);                                                // free the downloaded feed data
 365 }
 366
 367
 368 // We have a list, now download the feeds
 369 //
 370 void rss_pull_feeds(void)
 371 {
 372         struct rssurl *r;
 373         struct rssroom *rr;
 374
 375         while (rsstodo != NULL) {
 376                 rss_pull_one_feed(rsstodo);
 377                 r = rsstodo;
 378                 rsstodo = rsstodo->next;
 379                 while (r->rooms != NULL) {
 380                         rr = r->rooms;
 381                         r->rooms = r->rooms->next;
 382                         free(rr->room);
 383                         free(rr);
 384                 }
 385                 free(r->url);
 386                 free(r);
 387         }
 388 }
 389
 390
 391 // Scan a room's netconfig looking for RSS feed parsing requests
 392 //
 393 void rssclient_scan_room(struct ctdlroom *qrbuf, void *data)
 394 {
 395         char *serialized_config = NULL;
 396         int num_configs = 0;
 397         char cfgline[SIZ];
 398         int i = 0;
 399
 400         serialized_config = LoadRoomNetConfigFile(qrbuf->QRnumber);
 401         if (!serialized_config) {
 402                 return;
 403         }
 404
 405         num_configs = num_tokens(serialized_config, '\n');
 406         for (i=0; i<num_configs; ++i) {
 407                 extract_token(cfgline, serialized_config, i, '\n', sizeof cfgline);
 408                 if (!strncasecmp(cfgline, HKEY("rssclient|"))) {
 409                         strcpy(cfgline, &cfgline[10]);
 410                         char *vbar = strchr(cfgline, '|');
 411                         if (vbar != NULL) {
 412                                 *vbar = 0;
 413                         }
 414                         rssclient_push_todo(cfgline, qrbuf->QRname);
 415                 }
 416         }
 417
 418         free(serialized_config);
 419 }
 420
 421
 422 /*
 423  * Scan for rooms that have RSS client requests configured
 424  */
 425 void rssclient_scan(void) {
 426         time_t now = time(NULL);
 427
 428         /* Run no more than once every 15 minutes. */
 429         if ((now - last_run) < 900) {
 430                 syslog(LOG_DEBUG,
 431                         "Client: polling interval not yet reached; last run was %ldm%lds ago",
 432                         ((now - last_run) / 60),
 433                         ((now - last_run) % 60)
 434                 );
 435                 return;
 436         }
 437
 438         become_session(&rss_CC);
 439         syslog(LOG_DEBUG, "rssclient started");
 440         CtdlForEachRoom(rssclient_scan_room, NULL);
 441         rss_pull_feeds();
 442         syslog(LOG_DEBUG, "rssclient ended");
 443         last_run = time(NULL);
 444         return;
 445 }
 446
 447
 448 CTDL_MODULE_INIT(rssclient)
 449 {
 450         if (!threading)
 451         {
 452                 syslog(LOG_INFO, "%s", curl_version());
 453                 CtdlRegisterSessionHook(rssclient_scan, EVT_TIMER, PRIO_AGGR + 300);
 454         }
 455         else
 456         {
 457                 CtdlFillSystemContext(&rss_CC, "rssclient");
 458         }
 459         return "rssclient";
 460 }
 461