1/* snac - A simple, minimalistic ActivityPub instance */
  2/* copyright (c) 2025 - 2026 grunfink et al. / MIT license */
  3
  4#include "xs.h"
  5#include "xs_html.h"
  6#include "xs_regex.h"
  7#include "xs_time.h"
  8#include "xs_match.h"
  9#include "xs_curl.h"
 10#include "xs_openssl.h"
 11#include "xs_json.h"
 12#include "xs_http.h"
 13#include "xs_unicode.h"
 14
 15#include "snac.h"
 16
 17xs_str *rss_from_timeline(snac *user, const xs_list *timeline,
 18                        const char *title, const char *link, const char *desc)
 19/* converts a timeline to rss */
 20{
 21    xs_html *rss = xs_html_tag("rss",
 22        xs_html_attr("xmlns:content", "http:/" "/purl.org/rss/1.0/modules/content/"),
 23        xs_html_attr("version",       "2.0"),
 24        xs_html_attr("xmlns:atom",    "http:/" "/www.w3.org/2005/Atom"));
 25
 26    xs_html *channel = xs_html_tag("channel",
 27        xs_html_tag("title",
 28            xs_html_text(title)),
 29        xs_html_tag("language",
 30            xs_html_text("en")),
 31        xs_html_tag("link",
 32            xs_html_text(link)),
 33        xs_html_sctag("atom:link",
 34            xs_html_attr("href", link),
 35            xs_html_attr("rel", "self"),
 36            xs_html_attr("type", "application/rss+xml")),
 37        xs_html_tag("generator",
 38            xs_html_text(USER_AGENT)),
 39        xs_html_tag("description",
 40            xs_html_text(desc)));
 41
 42    xs_html_add(rss, channel);
 43
 44    int cnt = 0;
 45    const char *v;
 46
 47    xs_list_foreach(timeline, v) {
 48        xs *msg = NULL;
 49
 50        if (user) {
 51            if (!valid_status(timeline_get_by_md5(user, v, &msg)))
 52                continue;
 53        }
 54        else {
 55            if (!valid_status(object_get_by_md5(v, &msg)))
 56                continue;
 57        }
 58
 59        const char *id = xs_dict_get(msg, "id");
 60        const char *content = xs_dict_get(msg, "content");
 61        const char *published = xs_dict_get(msg, "published");
 62
 63        if (user && !is_msg_mine(user, id))
 64            continue;
 65
 66        if (!id || !content || !published)
 67            continue;
 68
 69        if (!is_msg_public(msg))
 70            continue;
 71
 72        /* create a title with the first line of the content */
 73        xs *title = xs_replace(content, "<br>", "\n");
 74        title = xs_regex_replace_i(title, "<[^>]+>", " ");
 75        title = xs_regex_replace_i(title, "&[^;]+;", " ");
 76        int i;
 77
 78        for (i = 0; title[i] && title[i] != '\n' && i < 50; ) {
 79            const char *p = &title[i];
 80            unsigned int cp = xs_utf8_dec(&p);
 81            int n = p - title;
 82            if (cp == 0xfffd || n > 50)
 83                break;
 84            i = n;
 85        }
 86
 87        if (title[i] != '\0') {
 88            title[i] = '\0';
 89            title = xs_str_cat(title, "...");
 90        }
 91
 92        title = xs_strip_i(title);
 93
 94        /* convert the date */
 95        time_t t = xs_parse_iso_date(published, 0);
 96        xs *rss_date = xs_str_utctime(t, "%a, %d %b %Y %T +0000");
 97
 98        /* if it's the first one, add it to the header */
 99        if (cnt == 0)
100            xs_html_add(channel,
101                xs_html_tag("lastBuildDate",
102                    xs_html_text(rss_date)));
103
104        xs_html_add(channel,
105            xs_html_tag("item",
106                xs_html_tag("title",
107                    xs_html_text(title)),
108                xs_html_tag("link",
109                    xs_html_text(id)),
110                xs_html_tag("guid",
111                    xs_html_text(id)),
112                xs_html_tag("pubDate",
113                    xs_html_text(rss_date)),
114                xs_html_tag("description",
115                    xs_html_text(content))));
116
117        cnt++;
118    }
119
120    return xs_html_render_s(rss, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
121}
122
123
124void rss_to_timeline(snac *user, const char *url)
125/* reads an RSS and inserts all ActivityPub posts into the user's timeline */
126{
127    if (!xs_startswith(url, "https:/") && !xs_startswith(url, "http:/"))
128        return;
129
130    xs *hdrs = xs_dict_new();
131    hdrs = xs_dict_set(hdrs, "accept",     "application/rss+xml");
132    hdrs = xs_dict_set(hdrs, "user-agent", USER_AGENT);
133
134    /* get the RSS metadata */
135    xs *md5 = xs_md5_hex(url, strlen(url));
136    xs *rss_md_fn = xs_fmt("%s/rss", user->basedir);
137    mkdirx(rss_md_fn);
138    rss_md_fn = xs_str_cat(rss_md_fn, "/", md5, ".json");
139
140    xs *rss_md = NULL;
141    const char *etag = NULL;
142
143    FILE *f;
144    if ((f = fopen(rss_md_fn, "r")) != NULL) {
145        rss_md = xs_json_load(f);
146        fclose(f);
147
148        etag = xs_dict_get(rss_md, "etag");
149
150        if (xs_is_string(etag))
151            hdrs = xs_dict_set(hdrs, "if-none-match", etag);
152    }
153
154    if (rss_md == NULL)
155        rss_md = xs_dict_new();
156
157    xs *payload = NULL;
158    int status;
159    int p_size;
160
161    xs *rsp = xs_http_request("GET", url, hdrs, NULL, 0, &status, &payload, &p_size, 0);
162
163    snac_log(user, xs_fmt("parsing RSS %s %d", url, status));
164
165    if (!valid_status(status) || !xs_is_string(payload))
166        return;
167
168    /* not an RSS? done */
169    const char *ctype = xs_dict_get(rsp, "content-type");
170    if (!xs_is_string(ctype) || xs_str_in(ctype, "application/rss+xml") == -1)
171        return;
172
173    /* yes, parsing is done with regexes (now I have two problems blah blah blah) */
174    xs *links = xs_regex_select(payload, "<link>[^<]+</link>");
175    const char *link;
176
177    xs_list_foreach(links, link) {
178        xs *l = xs_replace(link, "<link>", "");
179        char *p = strchr(l, '<');
180
181        if (p == NULL)
182            continue;
183        *p = '\0';
184
185        /* skip this same URL */
186        if (strcmp(l, url) == 0)
187            continue;
188
189        /* skip crap */
190        if (!xs_startswith(l, "https:/") && !xs_startswith(l, "http:/"))
191            continue;
192
193        snac_debug(user, 1, xs_fmt("RSS link: %s", l));
194
195        if (timeline_here(user, l)) {
196            snac_debug(user, 1, xs_fmt("RSS entry already in timeline %s", l));
197            continue;
198        }
199
200        /* special trick for Mastodon: convert from the alternate format */
201        if (strchr(l, '@') != NULL) {
202            xs *l2 = xs_split(l, "/");
203
204            if (xs_list_len(l2) == 5) {
205                const char *uid = xs_list_get(l2, 3);
206                if (*uid == '@') {
207                    xs *guessed_id = xs_fmt("https:/" "/%s/users/%s/statuses/%s",
208                        xs_list_get(l2, 2), uid + 1, xs_list_get(l2, -1));
209
210                    if (timeline_here(user, guessed_id)) {
211                        snac_debug(user, 1, xs_fmt("RSS entry already in timeline (alt) %s", guessed_id));
212                        continue;
213                    }
214                }
215            }
216        }
217
218        xs *obj = NULL;
219
220        if (!valid_status(object_get(l, &obj))) {
221            /* object is not here: bring it */
222            if (!valid_status(activitypub_request(user, l, &obj)))
223                continue;
224        }
225
226        if (xs_is_dict(obj)) {
227            const char *id      = xs_dict_get(obj, "id");
228            const char *type    = xs_dict_get(obj, "type");
229            const char *attr_to = get_atto(obj);
230
231            if (!xs_is_string(id) || !xs_is_string(type) || !xs_is_string(attr_to))
232                continue;
233
234            if (!xs_match(type, POSTLIKE_OBJECT_TYPE))
235                continue;
236
237            if (timeline_here(user, id)) {
238                snac_debug(user, 1, xs_fmt("RSS entry already in timeline (id) %s", id));
239                continue;
240            }
241
242            enqueue_actor_refresh(user, attr_to, 0);
243
244            timeline_add(user, id, obj);
245
246            snac_log(user, xs_fmt("new '%s' (RSS) %s %s", type, attr_to, id));
247        }
248    }
249
250    /* update the RSS metadata */
251    etag = xs_dict_get(rsp, "etag");
252
253    if (xs_is_string(etag)) {
254        rss_md = xs_dict_set(rss_md, "etag", etag);
255        rss_md = xs_dict_set(rss_md, "url", url);
256        if ((f = fopen(rss_md_fn, "w")) != NULL) {
257            xs_json_dump(rss_md, 4, f);
258            fclose(f);
259        }
260    }
261}
262
263
264void rss_poll_hashtags(void)
265/* parses all RSS from all users */
266{
267    xs *list = user_list();
268    const char *uid;
269
270    xs_list_foreach(list, uid) {
271        snac user;
272
273        if (user_open(&user, uid)) {
274            const xs_list *rss = xs_dict_get(user.config, "followed_hashtags");
275
276            if (xs_is_list(rss)) {
277                const char *url;
278
279                xs_list_foreach(rss, url)
280                    rss_to_timeline(&user, url);
281            }
282
283            user_free(&user);
284        }
285    }
286}