commit bc5d0d4
default
·
2023-11-20 17:33:24 +0000 UTC
parent b68ed66
Replaced encode_html_strict() with xs_html_encode().
8 files changed,
+273,
-26
M
Makefile
+4,
-3
1@@ -37,9 +37,9 @@ activitypub.o: activitypub.c xs.h xs_json.h xs_curl.h xs_mime.h \
2 xs_openssl.h xs_regex.h xs_time.h xs_set.h xs_match.h snac.h
3 data.o: data.c xs.h xs_hex.h xs_io.h xs_json.h xs_openssl.h xs_glob.h \
4 xs_set.h xs_time.h snac.h
5-format.o: format.c xs.h xs_regex.h xs_mime.h snac.h
6+format.o: format.c xs.h xs_regex.h xs_mime.h xs_html.h snac.h
7 html.o: html.c xs.h xs_io.h xs_json.h xs_regex.h xs_set.h xs_openssl.h \
8- xs_time.h xs_mime.h xs_match.h snac.h
9+ xs_time.h xs_mime.h xs_match.h xs_html.h snac.h
10 http.o: http.c xs.h xs_io.h xs_openssl.h xs_curl.h xs_time.h xs_json.h \
11 snac.h
12 httpd.o: httpd.c xs.h xs_io.h xs_json.h xs_socket.h xs_httpd.h xs_mime.h \
13@@ -50,7 +50,8 @@ mastoapi.o: mastoapi.c xs.h xs_hex.h xs_openssl.h xs_json.h xs_io.h \
14 snac.h
15 snac.o: snac.c xs.h xs_hex.h xs_io.h xs_unicode.h xs_json.h xs_curl.h \
16 xs_openssl.h xs_socket.h xs_url.h xs_httpd.h xs_mime.h xs_regex.h \
17- xs_set.h xs_time.h xs_glob.h xs_random.h xs_match.h xs_fcgi.h snac.h
18+ xs_set.h xs_time.h xs_glob.h xs_random.h xs_match.h xs_fcgi.h xs_html.h \
19+ snac.h
20 upgrade.o: upgrade.c xs.h xs_io.h xs_json.h xs_glob.h snac.h
21 utils.o: utils.c xs.h xs_io.h xs_json.h xs_time.h xs_openssl.h \
22 xs_random.h xs_glob.h snac.h
M
format.c
+2,
-14
1@@ -4,6 +4,7 @@
2 #include "xs.h"
3 #include "xs_regex.h"
4 #include "xs_mime.h"
5+#include "xs_html.h"
6
7 #include "snac.h"
8
9@@ -260,23 +261,10 @@ xs_str *sanitize(const char *content)
10 }
11
12
13-xs_str *encode_html_strict(const char *str)
14-/* escapes html characters */
15-{
16- xs_str *encoded = xs_replace(str, "&", "&");
17- encoded = xs_replace_i(encoded, "<", "<");
18- encoded = xs_replace_i(encoded, ">", ">");
19- encoded = xs_replace_i(encoded, "\"", """);
20- encoded = xs_replace_i(encoded, "'", "'");
21-
22- return encoded;
23-}
24-
25-
26 xs_str *encode_html(const char *str)
27 /* escapes html characters */
28 {
29- xs_str *encoded = encode_html_strict(str);
30+ xs_str *encoded = xs_html_encode((char *)str);
31
32 /* Restore only <br>. Probably safe. Let's hope nothing goes wrong with this. */
33 encoded = xs_replace_i(encoded, "<br>", "<br>");
M
html.c
+6,
-5
1@@ -10,6 +10,7 @@
2 #include "xs_time.h"
3 #include "xs_mime.h"
4 #include "xs_match.h"
5+#include "xs_html.h"
6
7 #include "snac.h"
8
9@@ -2137,10 +2138,10 @@ int html_get_handler(const xs_dict *req, const char *q_path,
10 xs *bio = not_really_markdown(xs_dict_get(snac.config, "bio"), NULL);
11 char *p, *v;
12
13- xs *es1 = encode_html_strict(xs_dict_get(snac.config, "name"));
14- xs *es2 = encode_html_strict(snac.uid);
15- xs *es3 = encode_html_strict(xs_dict_get(srv_config, "host"));
16- xs *es4 = encode_html_strict(bio);
17+ xs *es1 = xs_html_encode(xs_dict_get(snac.config, "name"));
18+ xs *es2 = xs_html_encode(snac.uid);
19+ xs *es3 = xs_html_encode(xs_dict_get(srv_config, "host"));
20+ xs *es4 = xs_html_encode(bio);
21 rss = xs_fmt(
22 "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n"
23 "<rss version=\"0.91\">\n"
24@@ -2168,7 +2169,7 @@ int html_get_handler(const xs_dict *req, const char *q_path,
25 if (!xs_startswith(id, snac.actor))
26 continue;
27
28- xs *content = encode_html_strict(xs_dict_get(msg, "content"));
29+ xs *content = xs_html_encode(xs_dict_get(msg, "content"));
30
31 // We SHOULD only use sanitized one for description.
32 // So, only encode for feed title, while the description just keep it sanitized as is.
M
snac.c
+1,
-0
1@@ -21,6 +21,7 @@
2 #include "xs_random.h"
3 #include "xs_match.h"
4 #include "xs_fcgi.h"
5+#include "xs_html.h"
6
7 #include "snac.h"
8
M
snac.h
+1,
-2
1@@ -1,7 +1,7 @@
2 /* snac - A simple, minimalistic ActivityPub instance */
3 /* copyright (c) 2022 - 2023 grunfink et al. / MIT license */
4
5-#define VERSION "2.43"
6+#define VERSION "2.44-dev"
7
8 #define USER_AGENT "snac/" VERSION
9
10@@ -266,7 +266,6 @@ int activitypub_post_handler(const xs_dict *req, const char *q_path,
11
12 xs_str *not_really_markdown(const char *content, xs_list **attach);
13 xs_str *sanitize(const char *content);
14-xs_str *encode_html_strict(const char *str);
15 xs_str *encode_html(const char *str);
16
17 xs_str *html_timeline(snac *user, const xs_list *list, int local,
M
xs.h
+18,
-1
1@@ -62,7 +62,8 @@ xs_str *xs_str_new(const char *str);
2 xs_str *xs_str_new_sz(const char *mem, int sz);
3 xs_str *xs_str_wrap_i(const char *prefix, xs_str *str, const char *suffix);
4 #define xs_str_prepend_i(str, prefix) xs_str_wrap_i(prefix, str, NULL)
5-#define xs_str_cat(str, suffix) xs_str_wrap_i(NULL, str, suffix)
6+xs_str *_xs_str_cat(xs_str *str, const char *strs[]);
7+#define xs_str_cat(str, ...) _xs_str_cat(str, (const char *[]){ __VA_ARGS__, NULL })
8 xs_str *xs_replace_in(xs_str *str, const char *sfrom, const char *sto, int times);
9 #define xs_replace_i(str, sfrom, sto) xs_replace_in(str, sfrom, sto, XS_ALL)
10 #define xs_replace(str, sfrom, sto) xs_replace_in(xs_dup(str), sfrom, sto, XS_ALL)
11@@ -451,6 +452,22 @@ xs_str *xs_str_wrap_i(const char *prefix, xs_str *str, const char *suffix)
12 }
13
14
15+xs_str *_xs_str_cat(xs_str *str, const char *strs[])
16+/* concatenates all strings after str */
17+{
18+ int o = strlen(str);
19+
20+ while (*strs) {
21+ int sz = strlen(*strs);
22+ str = xs_insert_m(str, o, *strs, sz);
23+ o += sz;
24+ strs++;
25+ }
26+
27+ return str;
28+}
29+
30+
31 xs_str *xs_replace_in(xs_str *str, const char *sfrom, const char *sto, int times)
32 /* replaces inline all sfrom with sto */
33 {
+240,
-0
1@@ -0,0 +1,240 @@
2+/* copyright (c) 2022 - 2023 grunfink et al. / MIT license */
3+
4+#ifndef _XS_HTML_H
5+
6+#define _XS_HTML_H
7+
8+typedef struct xs_html xs_html;
9+
10+xs_str *xs_html_encode(char *str);
11+
12+xs_html *xs_html_attr(char *key, char *value);
13+xs_html *xs_html_text(char *content);
14+xs_html *xs_html_raw(char *content);
15+
16+xs_html *xs_html_add(xs_html *tag, xs_html *data);
17+
18+xs_html *_xs_html_tag(char *tag, xs_html *var[]);
19+#define xs_html_tag(tag, ...) _xs_html_tag(tag, (xs_html *[]) { __VA_ARGS__, NULL })
20+xs_html *_xs_html_sctag(char *tag, xs_html *var[]);
21+#define xs_html_sctag(tag, ...) _xs_html_sctag(tag, (xs_html *[]) { __VA_ARGS__, NULL })
22+xs_str *_xs_html_render(xs_html *h, xs_str *s);
23+#define xs_html_render(h) _xs_html_render(h, xs_str_new(NULL))
24+
25+#ifdef XS_IMPLEMENTATION
26+
27+typedef enum {
28+ XS_HTML_TAG,
29+ XS_HTML_SCTAG,
30+ XS_HTML_ATTR,
31+ XS_HTML_TEXT
32+} xs_html_type;
33+
34+struct xs_html {
35+ xs_html_type type;
36+ xs_str *content;
37+ xs_html *f_attr;
38+ xs_html *l_attr;
39+ xs_html *f_tag;
40+ xs_html *l_tag;
41+ xs_html *next;
42+};
43+
44+xs_str *xs_html_encode(char *str)
45+/* encodes str using HTML entities */
46+{
47+ xs_str *s = xs_str_new(NULL);
48+ int o = 0;
49+ char *e = str + strlen(str);
50+
51+ for (;;) {
52+ char *ec = "<>\"'&"; /* characters to escape */
53+ char *q = e;
54+ int z;
55+
56+ /* find the nearest happening of a char */
57+ while (*ec) {
58+ char *m = memchr(str, *ec++, q - str);
59+ if (m)
60+ q = m;
61+ }
62+
63+ /* copy string to here */
64+ z = q - str;
65+ s = xs_insert_m(s, o, str, z);
66+ o += z;
67+
68+ /* if q points to the end, nothing more to do */
69+ if (q == e)
70+ break;
71+
72+ /* insert the escaped char */
73+ char tmp[8];
74+ snprintf(tmp, sizeof(tmp), "&#%d;", *q);
75+
76+ z = strlen(tmp);
77+ s = xs_insert_m(s, o, tmp, z);
78+ o += z;
79+
80+ str = q + 1;
81+ }
82+
83+ return s;
84+}
85+
86+
87+#define XS_HTML_NEW() memset(xs_realloc(NULL, sizeof(xs_html)), '\0', sizeof(xs_html))
88+
89+xs_html *xs_html_attr(char *key, char *value)
90+/* creates an HTML block with an attribute */
91+{
92+ xs_html *a = XS_HTML_NEW();
93+
94+ a->type = XS_HTML_ATTR;
95+
96+ if (value) {
97+ xs *ev = xs_html_encode(value);
98+ a->content = xs_fmt("%s=\"%s\"", key, ev);
99+ }
100+ else
101+ a->content = xs_dup(key);
102+
103+ return a;
104+}
105+
106+
107+xs_html *xs_html_text(char *content)
108+/* creates an HTML block of text, escaping it previously */
109+{
110+ xs_html *a = XS_HTML_NEW();
111+
112+ a->type = XS_HTML_TEXT;
113+ a->content = xs_html_encode(content);
114+
115+ return a;
116+}
117+
118+
119+xs_html *xs_html_raw(char *content)
120+/* creates an HTML block without escaping (for pre-formatted HTML, comments, etc) */
121+{
122+ xs_html *a = XS_HTML_NEW();
123+
124+ a->type = XS_HTML_TEXT;
125+ a->content = xs_dup(content);
126+
127+ return a;
128+}
129+
130+
131+xs_html *xs_html_add(xs_html *tag, xs_html *data)
132+/* add data (attrs, tags or text) to a tag */
133+{
134+ xs_html **first;
135+ xs_html **last;
136+
137+ if (data->type == XS_HTML_ATTR) {
138+ first = &tag->f_attr;
139+ last = &tag->l_attr;
140+ }
141+ else {
142+ first = &tag->f_tag;
143+ last = &tag->l_tag;
144+ }
145+
146+ if (*first == NULL)
147+ *first = data;
148+
149+ if (*last != NULL)
150+ (*last)->next = data;
151+
152+ *last = data;
153+
154+ return tag;
155+}
156+
157+
158+static xs_html *_xs_html_tag_t(xs_html_type type, char *tag, xs_html *var[])
159+/* creates a tag with a variable list of attributes and subtags */
160+{
161+ xs_html *a = XS_HTML_NEW();
162+
163+ a->type = type;
164+ a->content = xs_dup(tag);
165+
166+ while (*var)
167+ xs_html_add(a, *var++);
168+
169+ return a;
170+}
171+
172+
173+xs_html *_xs_html_tag(char *tag, xs_html *var[])
174+{
175+ return _xs_html_tag_t(XS_HTML_TAG, tag, var);
176+}
177+
178+
179+xs_html *_xs_html_sctag(char *tag, xs_html *var[])
180+{
181+ return _xs_html_tag_t(XS_HTML_SCTAG, tag, var);
182+}
183+
184+
185+xs_str *_xs_html_render(xs_html *h, xs_str *s)
186+/* renders the tag and its subtags */
187+{
188+ xs_html *st;
189+
190+ switch (h->type) {
191+ case XS_HTML_TAG:
192+ case XS_HTML_SCTAG:
193+ s = xs_str_cat(s, "<", h->content);
194+
195+ /* render the attributes */
196+ st = h->f_attr;
197+ while (st) {
198+ xs_html *nst = st->next;
199+ s = _xs_html_render(st, s);
200+ st = nst;
201+ }
202+
203+ if (h->type == XS_HTML_SCTAG) {
204+ /* self-closing tags should not have subtags */
205+ s = xs_str_cat(s, "/>");
206+ }
207+ else {
208+ s = xs_str_cat(s, ">");
209+
210+ /* render the subtags */
211+ st = h->f_tag;
212+ while (st) {
213+ xs_html *nst = st->next;
214+ s = _xs_html_render(st, s);
215+ st = nst;
216+ }
217+
218+ s = xs_str_cat(s, "</", h->content, ">");
219+ }
220+
221+ break;
222+
223+ case XS_HTML_ATTR:
224+ s = xs_str_cat(s, " ", h->content);
225+ break;
226+
227+ case XS_HTML_TEXT:
228+ s = xs_str_cat(s, h->content);
229+ break;
230+ }
231+
232+ xs_free(h->content);
233+ xs_free(h);
234+
235+ return s;
236+}
237+
238+
239+#endif /* XS_IMPLEMENTATION */
240+
241+#endif /* _XS_HTML_H */
+1,
-1
1@@ -1 +1 @@
2-/* 416f5ffa99ecd4a3ec25d273b986d3d99dc92d22 */
3+/* 63beb583926bb5dfec89e1d694172cc887614460 2023-11-19T19:51:05+01:00 */