commit b783f28

New function rss_to_timeline().

  1@@ -5,6 +5,8 @@
  2 #include "xs_html.h"
  3 #include "xs_regex.h"
  4 #include "xs_time.h"
  5+#include "xs_match.h"
  6+#include "xs_curl.h"
  7 
  8 #include "snac.h"
  9 
 10@@ -103,3 +105,125 @@ xs_str *rss_from_timeline(snac *user, const xs_list *timeline,
 11 
 12     return xs_html_render_s(rss, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n");
 13 }
 14+
 15+
 16+void rss_to_timeline(snac *user, const char *url)
 17+/* reads an RSS and inserts all ActivityPub posts into the user's timeline */
 18+{
 19+    xs *hdrs = xs_dict_new();
 20+    hdrs = xs_dict_set(hdrs, "accept",     "application/rss+xml");
 21+    hdrs = xs_dict_set(hdrs, "user-agent", USER_AGENT);
 22+
 23+    xs *payload = NULL;
 24+    int status;
 25+    int p_size;
 26+
 27+    xs *rsp = xs_http_request("GET", url, hdrs, NULL, 0, &status, &payload, &p_size, 0);
 28+
 29+    if (!valid_status(status) || !xs_is_string(payload))
 30+        return;
 31+
 32+    /* not an RSS? done */
 33+    const char *ctype = xs_dict_get(rsp, "content-type");
 34+    if (!xs_is_string(ctype) || xs_str_in(ctype, "application/rss+xml") == -1)
 35+        return;
 36+
 37+    snac_log(user, xs_fmt("parsing RSS %s", url));
 38+
 39+    /* yes, parsing is done with regexes (now I have two problems blah blah blah) */
 40+    xs *links = xs_regex_select(payload, "<link>[^<]+</link>");
 41+    const char *link;
 42+
 43+    xs_list_foreach(links, link) {
 44+        xs *l = xs_replace(link, "<link>", "");
 45+        char *p = strchr(l, '<');
 46+
 47+        if (p == NULL)
 48+            continue;
 49+        *p = '\0';
 50+
 51+        /* skip this same URL */
 52+        if (strcmp(l, url) == 0)
 53+            continue;
 54+
 55+        snac_debug(user, 1, xs_fmt("RSS link: %s", l));
 56+
 57+        if (timeline_here(user, l)) {
 58+            snac_debug(user, 1, xs_fmt("RSS entry already in timeline %s", l));
 59+            continue;
 60+        }
 61+
 62+        /* special trick for Mastodon: convert from the alternate format */
 63+        if (strchr(l, '@') != NULL) {
 64+            xs *l2 = xs_split(l, "/");
 65+
 66+            if (xs_list_len(l2) == 5) {
 67+                const char *uid = xs_list_get(l2, 3);
 68+                if (*uid == '@') {
 69+                    xs *guessed_id = xs_fmt("https:/" "/%s/users/%s/statuses/%s",
 70+                        xs_list_get(l2, 2), uid + 1, xs_list_get(l2, -1));
 71+
 72+                    if (timeline_here(user, guessed_id)) {
 73+                        snac_debug(user, 1, xs_fmt("RSS entry already in timeline (alt) %s", guessed_id));
 74+                        continue;
 75+                    }
 76+                }
 77+            }
 78+        }
 79+
 80+        xs *obj = NULL;
 81+
 82+        if (!valid_status(object_get(l, &obj))) {
 83+            /* object is not here: bring it */
 84+            if (!valid_status(activitypub_request(user, l, &obj)))
 85+                continue;
 86+        }
 87+
 88+        if (xs_is_dict(obj)) {
 89+            const char *id      = xs_dict_get(obj, "id");
 90+            const char *type    = xs_dict_get(obj, "type");
 91+            const char *attr_to = get_atto(obj);
 92+
 93+            if (!xs_is_string(id) || !xs_is_string(type) || !xs_is_string(attr_to))
 94+                continue;
 95+
 96+            if (!xs_match(type, POSTLIKE_OBJECT_TYPE))
 97+                continue;
 98+
 99+            if (timeline_here(user, id)) {
100+                snac_debug(user, 1, xs_fmt("RSS entry already in timeline (id) %s", id));
101+                continue;
102+            }
103+
104+            if (!valid_status(actor_request(user, attr_to, NULL)))
105+                continue;
106+
107+            timeline_add(user, id, obj);
108+        }
109+    }
110+}
111+
112+
113+void rss_process(void)
114+/* parses all RSS from all users */
115+{
116+    xs *list = user_list();
117+    const char *uid;
118+
119+    xs_list_foreach(list, uid) {
120+        snac user;
121+
122+        if (user_open(&user, uid)) {
123+            const xs_list *rss = xs_dict_get(user.config, "rss");
124+
125+            if (xs_is_list(rss)) {
126+                const char *url;
127+
128+                xs_list_foreach(rss, url)
129+                    rss_to_timeline(&user, url);
130+            }
131+
132+            user_free(&user);
133+        }
134+    }
135+}

1@@ -463,3 +463,5 @@ const char *lang_str(const char *str, const snac *user);
2 
3 xs_str *rss_from_timeline(snac *user, const xs_list *timeline,
4                         const char *title, const char *link, const char *desc);
5+void rss_to_timeline(snac *user, const char *url);
6+void rss_process(void);