From 60c2a50ed9a0efc8bbbdbeb48b4a96a1eb80f946 Mon Sep 17 00:00:00 2001
From: default <nobody@localhost>
Date: Mon, 11 Mar 2024 06:00:21 +0100
Subject: [PATCH] New function content_check().

---
 activitypub.c |  7 ++++++-
 data.c        | 40 ++++++++++++++++++++++++++++++++++++++++
 snac.h        |  2 ++
 3 files changed, 48 insertions(+), 1 deletion(-)

diff --git a/activitypub.c b/activitypub.c
index 1976012..22a12fa 100644
--- a/activitypub.c
+++ b/activitypub.c
@@ -1921,10 +1921,15 @@ int process_input_message(snac *snac, xs_dict *msg, xs_dict *req)
                 snac_debug(snac, 0, xs_fmt("dropped reply %s to hidden post %s", id, in_reply_to));
             }
             else {
+                if (content_check("filter_reject.txt", object)) {
+                    snac_log(snac, xs_fmt("rejected by content %s", id));
+                    return 1;
+                }
+
                 timeline_request(snac, &in_reply_to, &wrk, 0);
 
                 if (timeline_add(snac, id, object)) {
-                    snac_log(snac, xs_fmt("new 'Note' %s %s", actor, id));
+                    snac_log(snac, xs_fmt("new '%s' %s %s", utype, actor, id));
                     do_notify = 1;
                 }
 
diff --git a/data.c b/data.c
index 4d9824a..67536fd 100644
--- a/data.c
+++ b/data.c
@@ -9,6 +9,7 @@
 #include "xs_glob.h"
 #include "xs_set.h"
 #include "xs_time.h"
+#include "xs_regex.h"
 
 #include "snac.h"
 
@@ -2006,6 +2007,45 @@ int instance_unblock(const char *instance)
 }
 
 
+/** content filtering **/
+
+int content_check(const char *file, const xs_dict *msg)
+/* checks if message content matches any of the regexes in file */
+{
+    xs *fn = xs_fmt("%s/%s", srv_basedir, file);
+    FILE *f;
+    int r = 0;
+    char *v = xs_dict_get(msg, "content");
+
+    if (xs_type(v) == XSTYPE_STRING && *v) {
+        if ((f = fopen(fn, "r")) != NULL) {
+            srv_debug(1, xs_fmt("content_check: loading regexes from %s", fn));
+
+            xs *c = xs_regex_replace(v, "<[^>]+>", " ");
+            c = xs_regex_replace_i(c, " {2,}", " ");
+            c = xs_tolower_i(c);
+
+            while (!r && !feof(f)) {
+                xs *rx = xs_strip_i(xs_readline(f));
+
+                if (*rx) {
+                    xs *l = xs_regex_select_n(c, rx, 1);
+
+                    if (xs_list_len(l)) {
+                        srv_debug(1, xs_fmt("content_check: match for '%s'", rx));
+                        r = 1;
+                    }
+                }
+            }
+
+            fclose(f);
+        }
+    }
+
+    return r;
+}
+
+
 /** notifications **/
 
 xs_str *notify_check_time(snac *snac, int reset)
diff --git a/snac.h b/snac.h
index 1afbfc7..f2bcedb 100644
--- a/snac.h
+++ b/snac.h
@@ -204,6 +204,8 @@ int is_instance_blocked(const char *instance);
 int instance_block(const char *instance);
 int instance_unblock(const char *instance);
 
+int content_check(const char *file, const xs_dict *msg);
+
 void enqueue_input(snac *snac, const xs_dict *msg, const xs_dict *req, int retries);
 void enqueue_shared_input(const xs_dict *msg, const xs_dict *req, int retries);
 void enqueue_output_raw(const char *keyid, const char *seckey,