Implement URL rewrites with regex capture groups

6 files changed, 151 insertions, 28 deletions

diff --git a/doc/gmnisrvini.scd b/doc/gmnisrvini.scd
index 5b936e5..78c80be 100644
--- a/doc/gmnisrvini.scd
+++ b/doc/gmnisrvini.scd
@@ -91,6 +91,24 @@ Within each routing section, the following keys are used to configure how
 	*[example.org:/foo]* with the root set to /srv/gemini,
 	/srv/gemini/foo/bar.txt will be served.
 
+*rewrite*
+	If regular expression routing is used, the rewrite directive may be used
+	to rewrite the URL path component before proceeding. The URL will be set
+	to the value of the rewrite expression. If *\\N* appears in the rewrite
+	value, where *N* is a number, that capture group will be substituted for
+	*\\N*. If *\\{name}* appears, where *name* is a named capture group, it
+	will be substituted.
+
+	Example:
+
+	```
+	[localhost~^/([a-zA-Z]+)\.(?<extension>png|jpg)$]
+	root=./root
+	rewrite=/images/\1.\{extension}
+	```
+
+	This will rewrite a request for /example.png to /images/example.png.
+
 *index*
 	Configures the name of the index file which shall be served in the event
 	that a request for this host does not include the filename part.
diff --git a/include/config.h b/include/config.h
index cf9f2a3..fc7e9fb 100644
--- a/include/config.h
+++ b/include/config.h
@@ -26,6 +26,7 @@ struct gmnisrv_route {
 
 	char *root;
 	char *index;
+	char *rewrite;
 	bool autoindex;
 	bool cgi;
 
diff --git a/include/ini.h b/include/ini.h
index 3757b83..81556cd 100644
--- a/include/ini.h
+++ b/include/ini.h
@@ -55,16 +55,6 @@ int ini_parse_stream(ini_reader reader, void* stream, ini_handler handler,
 #define INI_ALLOW_BOM 1
 #endif
 
-/* Nonzero to allow inline comments (with valid inline comment characters
-   specified by INI_INLINE_COMMENT_PREFIXES). Set to 0 to turn off and match
-   Python 3.2+ configparser behaviour. */
-#ifndef INI_ALLOW_INLINE_COMMENTS
-#define INI_ALLOW_INLINE_COMMENTS 1
-#endif
-#ifndef INI_INLINE_COMMENT_PREFIXES
-#define INI_INLINE_COMMENT_PREFIXES ";"
-#endif
-
 /* Nonzero to use stack, zero to use heap (malloc/free). */
 #ifndef INI_USE_STACK
 #define INI_USE_STACK 1
diff --git a/src/config.c b/src/config.c
index ec767da..662d02e 100644
--- a/src/config.c
+++ b/src/config.c
@@ -213,6 +213,7 @@ conf_ini_handler(void *user, const char *section,
 		char *name;
 		char **value;
 	} route_strvars[] = {
+		{ "rewrite", &route->rewrite },
 		{ "root", &route->root },
 		{ "index", &route->index },
 	};
@@ -228,6 +229,11 @@ conf_ini_handler(void *user, const char *section,
 		if (strcmp(route_strvars[i].name, name) != 0) {
 			continue;
 		}
+		if (strcmp(route_strvars[i].name, "rewrite") == 0
+				&& routing != ROUTE_REGEX) {
+			fprintf(stderr, "rewrite directives are only valid for regex routes\n");
+			return 0;
+		}
 		*route_strvars[i].value = strdup(value);
 		return 1;
 	}
@@ -315,9 +321,10 @@ config_finish(struct gmnisrv_config *conf)
 			}
 
 			struct gmnisrv_route *rnext = route->next;
-			free(route->spec);
-			free(route->root);
 			free(route->index);
+			free(route->rewrite);
+			free(route->root);
+			free(route->spec);
 			free(route);
 			route = rnext;
 		}
diff --git a/src/ini.c b/src/ini.c
index 88ff0d1..a001101 100644
--- a/src/ini.c
+++ b/src/ini.c
@@ -45,18 +45,9 @@ static char* lskip(const char* s)
    be prefixed by a whitespace character to register as a comment. */
 static char* find_chars_or_comment(const char* s, const char* chars)
 {
-#if INI_ALLOW_INLINE_COMMENTS
-    int was_space = 0;
-    while (*s && (!chars || !strchr(chars, *s)) &&
-           !(was_space && strchr(INI_INLINE_COMMENT_PREFIXES, *s))) {
-        was_space = isspace((unsigned char)(*s));
-        s++;
-    }
-#else
     while (*s && (!chars || !strchr(chars, *s))) {
         s++;
     }
-#endif
     return (char*)s;
 }
 
@@ -123,7 +114,7 @@ int ini_parse_stream(ini_reader reader, void* stream, ini_handler handler,
 #endif
         else if (*start == '[') {
             /* A "[section]" line */
-            end = find_chars_or_comment(start + 1, "]");
+            end = strrchr(start + 1, ']');
             if (*end == ']') {
                 *end = '\0';
                 strncpy0(section, start + 1, sizeof(section));
@@ -141,11 +132,6 @@ int ini_parse_stream(ini_reader reader, void* stream, ini_handler handler,
                 *end = '\0';
                 name = rstrip(start);
                 value = lskip(end + 1);
-#if INI_ALLOW_INLINE_COMMENTS
-                end = find_chars_or_comment(value, NULL);
-                if (*end)
-                    *end = '\0';
-#endif
                 rstrip(value);
 
                 /* Valid name[=:]value pair found, call handler */
diff --git a/src/serve.c b/src/serve.c
index 6d9d87d..74bb979 100644
--- a/src/serve.c
+++ b/src/serve.c
@@ -197,6 +197,126 @@ serve_cgi(struct gmnisrv_client *client, const char *path,
 	}
 }
 
+static char *
+ensure_buf(char *buf, size_t *sz, size_t desired)
+{
+	while (*sz < desired) {
+		*sz = *sz * 2;
+		char *new = realloc(buf, *sz);
+		assert(new);
+		buf = new;
+	}
+	return buf;
+}
+
+static int
+get_group(struct gmnisrv_route *route, const char *name, int ncapture)
+{
+	const char *groupname = lre_get_groupnames(route->regex);
+	for (int i = 0; i < ncapture; groupname += strlen(groupname) + 1, ++i) {
+		if (strcmp(groupname, name) == 0) {
+			return i;
+		}
+	}
+	return -1;
+}
+
+static char *
+process_rewrites(struct gmnisrv_route *route, const char *path,
+		uint8_t **capture, int ncapture)
+{
+	if (!route->rewrite) {
+		return strdup(path);
+	}
+
+	size_t new_sz = strlen(path) * 2;
+	size_t new_ln = 0;
+	char *new = malloc(new_sz);
+	*new = '\0';
+
+	char *temp = strdup(route->rewrite);
+	char *rewrite = temp;
+	char *next;
+	do {
+		next = strchr(rewrite, '\\');
+
+		size_t len;
+		if (next) {
+			len = next - rewrite;
+			*next = '\0';
+		} else {
+			len = strlen(rewrite);
+		}
+
+		ensure_buf(new, &new_sz, new_ln + len);
+		strcat(new, rewrite);
+		new_ln += len;
+
+		if (!next) {
+			break;
+		}
+
+		int group;
+		char *endptr;
+		switch (next[1]) {
+		case '\0':
+			server_error("Misconfigured rewrite rule for route %s: expected capture group identifier",
+					route->spec);
+			return strdup(path);
+		case '{':;
+			char *rbrace = strchr(&next[1], '}');
+			if (!rbrace) {
+				server_error("Misconfigured rewrite rule for route %s: expected capture group terminator '}'", route->spec);
+				return strdup(path);
+			}
+			*rbrace = '\0';
+			group = get_group(route, &next[2], ncapture);
+			if (group == -1) {
+				server_error("Misconfigured rewrite rule for route %s: unknown capture group '%s'", route->spec, &next[2]);
+				return strdup(path);
+			}
+			++group;
+			endptr = &rbrace[1];
+			goto subgroup;
+		case '0':
+		case '1':
+		case '2':
+		case '3':
+		case '4':
+		case '5':
+		case '6':
+		case '7':
+		case '8':
+		case '9':;
+			group = (int)strtoul(&next[1], &endptr, 10);
+			if (group >= ncapture) {
+				server_error("Misconfigured rewrite rule for route %s: unknown capture group %d\n", group);
+				return strdup(path);
+			}
+subgroup:;
+			fprintf(stderr, "replace group %d\n", group);
+			char *start = (char *)capture[group * 2],
+				*end = (char *)capture[group * 2 + 1];
+
+			char c = *end;
+			*end = '\0';
+			len = strlen(start);
+			ensure_buf(new, &new_sz, new_ln + len);
+			strcat(new, start);
+			new_ln += len;
+
+			fprintf(stderr, "+%s = %s\n", start, new);
+			rewrite = endptr;
+			*end = c;
+			break;
+		}
+	} while (next);
+
+	free(temp);
+	fprintf(stderr, "rewritten: %s\n", new);
+	return new;
+}
+
 static bool
 route_match(struct gmnisrv_route *route, const char *path, char **revised)
 {
@@ -227,7 +347,8 @@ route_match(struct gmnisrv_route *route, const char *path, char **revised)
 			free(capture);
 			return false;
 		}
-		*revised = strdup(path);
+		*revised = process_rewrites(route, path, capture, ncapture);
+		free(capture);
 		return true;
 	}