lib: uri-util: Always fully check the syntax of percent encoding while parsing URI...

diff --git a/src/lib/uri-util.c b/src/lib/uri-util.c
index 0cadc839a03e10da68b574082ea2d95f4369c7fa..d3ff1a98608fa53c767cdbba52369c680b5bb0a5 100644 (file)
--- a/src/lib/uri-util.c
+++ b/src/lib/uri-util.c
@@ -597,33 +597,37 @@ int uri_parse_slashslash_authority(struct uri_parser *parser,
 
 int uri_parse_path_segment(struct uri_parser *parser, const char **segment_r)
 {
-       const unsigned char *p = parser->cur;
+       const unsigned char *first = parser->cur;
+       int ret;
 
-       while (p < parser->end) {
-               if (*p == '%') {
-                       p++;
-                       continue;
+       while (parser->cur < parser->end) {
+               if (*parser->cur == '%') {
+                       unsigned char ch = 0;
+                       if ((ret=uri_parse_pct_encoded(parser, &ch)) < 0)
+                               return -1;
+                       if (ret > 0)
+                               continue;
                }
 
-               if ((*p & 0x80) != 0 || (_uri_char_lookup[*p] & CHAR_MASK_PCHAR) == 0)
+               if ((*parser->cur & 0x80) != 0 ||
+                       (_uri_char_lookup[*parser->cur] & CHAR_MASK_PCHAR) == 0)
                        break;
 
-               p++;
+               parser->cur++;
        }
 
-       if (p < parser->end &&
-               *p != '/' && *p != '?' && *p != '#' ) {
+       if (parser->cur < parser->end &&
+               *parser->cur != '/' && *parser->cur != '?' && *parser->cur != '#' ) {
                parser->error =
                        "Path component contains invalid character";
                return -1;
        }
 
-       if (p == parser->cur)
+       if (first == parser->cur)
                return 0;
 
        if (segment_r != NULL)
-               *segment_r = p_strdup_until(parser->pool, parser->cur, p);
-       parser->cur = p;
+               *segment_r = p_strdup_until(parser->pool, first, parser->cur);
        return 1;
 }
 
@@ -728,7 +732,8 @@ int uri_parse_path(struct uri_parser *parser,
 
 int uri_parse_query(struct uri_parser *parser, const char **query_r)
 {
-       const unsigned char *p = parser->cur;
+       const unsigned char *first = parser->cur;
+       int ret;
 
        /* RFC 3986:
         *
@@ -736,35 +741,39 @@ int uri_parse_query(struct uri_parser *parser, const char **query_r)
         * query         = *( pchar / "/" / "?" )
         * pchar         = unreserved / pct-encoded / sub-delims / ":" / "@"
         */
-       if (p >= parser->end || *p != '?')
+       if (parser->cur >= parser->end || *parser->cur != '?')
                return 0;
-       p++;
+       parser->cur++;
 
-       while (p < parser->end) {
-               if (*p == '%') {
-                       p++;
-                       continue;
+       while (parser->cur < parser->end) {
+               if (*parser->cur == '%') {
+                       unsigned char ch = 0;
+                       if ((ret=uri_parse_pct_encoded(parser, &ch)) < 0)
+                               return -1;
+                       if (ret > 0)
+                               continue;
                }
 
-               if ((*p & 0x80) != 0 || (_uri_char_lookup[*p] & CHAR_MASK_QCHAR) == 0)
+               if ((*parser->cur & 0x80) != 0 ||
+                       (_uri_char_lookup[*parser->cur] & CHAR_MASK_QCHAR) == 0)
                        break;
-               p++;
+               parser->cur++;
        }
 
-       if (p < parser->end && *p != '#') {
+       if (parser->cur < parser->end && *parser->cur != '#') {
                parser->error = "Query component contains invalid character";
                return -1;
        }
 
        if (query_r != NULL)
-               *query_r = p_strdup_until(parser->pool, parser->cur+1, p);
-       parser->cur = p;
+               *query_r = p_strdup_until(parser->pool, first+1, parser->cur);
        return 1;
 }
 
 int uri_parse_fragment(struct uri_parser *parser, const char **fragment_r)
 {
-       const unsigned char *p = parser->cur;
+       const unsigned char *first = parser->cur;
+       int ret;
 
        /* RFC 3986:
         *
@@ -773,29 +782,32 @@ int uri_parse_fragment(struct uri_parser *parser, const char **fragment_r)
         * pchar         = unreserved / pct-encoded / sub-delims / ":" / "@"
         */
 
-       if (p >= parser->end || *p != '#')
+       if (parser->cur >= parser->end || *parser->cur != '#')
                return 0;
-       p++;
+       parser->cur++;
 
-       while (p < parser->end) {
-               if (*p == '%') {
-                       p++;
-                       continue;
+       while (parser->cur < parser->end) {
+               if (*parser->cur == '%') {
+                       unsigned char ch = 0;
+                       if ((ret=uri_parse_pct_encoded(parser, &ch)) < 0)
+                               return -1;
+                       if (ret > 0)
+                               continue;
                }
 
-               if ((*p & 0x80) != 0 || (_uri_char_lookup[*p] & CHAR_MASK_QCHAR) == 0)
+               if ((*parser->cur & 0x80) != 0 ||
+                       (_uri_char_lookup[*parser->cur] & CHAR_MASK_QCHAR) == 0)
                        break;
-               p++;
+               parser->cur++;
        }
 
-       if (p < parser->end) {
+       if (parser->cur < parser->end) {
                parser->error = "Fragment component contains invalid character";
                return -1;
        }
 
        if (fragment_r != NULL)
-               *fragment_r = p_strdup_until(parser->pool, parser->cur+1, p);
-       parser->cur = p;
+               *fragment_r = p_strdup_until(parser->pool, first+1, parser->cur);
        return 1;
 }