import regex functions to parse request

utils.c

@@ -1,5 +1,8 @@
+#include <sys/types.h>
+
 #include <err.h>
 #include <errno.h>
+#include <regex.h>
 #include <stdarg.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -115,3 +118,15 @@ print_file(FILE *fd)
 		datasent += fwrite(buffer, 1, nread, stdout);
 	return datasent;
 }
+
+void
+getsubexp(const char *str, regmatch_t m, char *dst)
+{
+	size_t len = 0;
+
+	if (m.rm_eo > m.rm_so) { /* skip empty substring */
+		len = m.rm_eo - m.rm_so;
+		memcpy(dst, str + m.rm_so, len);
+		dst[len] = '\0';
+	}
+}

utils.h

@@ -1,3 +1,4 @@
+void		getsubexp(const char *, regmatch_t, char *);
 void		echdir   (const char *);
 void		epledge(const char *, const char *);
 void		eunveil(const char *, const char *);

vger.c

@@ -9,6 +9,7 @@
 #include <fcntl.h>
 #include <limits.h>
 #include <pwd.h>
+#include <regex.h>
 #include <stdarg.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -503,3 +504,54 @@ get_query(char *path, char *query, size_t querysiz)
         }
 	return query;
 }
+
+void
+split_request(const char *request, char *hostname, char *path, char *query)
+{
+	regex_t greg = {0};		/* compiled gemini regex */
+	regmatch_t *match = {0};	/* matches founds */
+	size_t nmatch = 4;		/* number of substrings to look for */
+	char buf[BUFSIZ] = {'\0'};	/* to handle error messages */
+	int ret = 0;
+
+	const char *gemini_regex = "^gemini://+([^/|^\?]*)/*([^\?]*)[\?]?(.*)$";
+		/*
+		 * ^gemini://+ : in case of gemini:///
+		 * 1: hostname
+		 *	([^/|^\?]*) :
+		 *		catch everything, stop when / or ? is found
+		 *		then skip multiple /
+		 * 2: path
+		 *	([^\?]*) :
+		 *		catch everything and stop at ? if any
+		 * 3 : query
+		 *	[\?]?(.*)$:
+		 *		catch everything after ? if any
+		 */
+
+	if ((ret = regcomp(&greg, gemini_regex, REG_EXTENDED)) != 0) {
+		regerror(ret, &greg, buf, sizeof(buf));
+		regfree(&greg);
+		status(50, "Internal server error");
+		stop(EXIT_FAILURE, "%s", buf);
+	}
+
+	if ((ret = regexec(&greg, request, nmatch, match, 0)) != 0) {
+		regerror(ret, &greg, buf, sizeof(buf));
+		regfree(&greg);
+		status(59, "Malformed request");
+		stop(EXIT_FAILURE, "Malformed request, error:%s", buf);
+	}
+
+	/* one may want to check the return of getsubexp
+	 * and change memcpy to strlcpy
+	 * to make sure we didn't try to copy too long
+	 * and that string isn't trunkated.
+	 * It is unlikely to happen since dest string are as long as request
+	 */
+	getsubexp(request, match[1], hostname);
+	getsubexp(request, match[2], path);
+	getsubexp(request, match[3], query);
+
+	regfree(&greg);
+}

vger.h

@@ -31,6 +31,7 @@ char *		get_query(char *, char *, size_t);
 void		status(const int, const char *, ...);
 void		strip_trailing_slash(char *);
 int 		uridecode (char *);
+void		split_request(const char *, char *, char *, char *);
 void		stop(const int, const char *, ...);
 
 #endif // vger_h_INCLUDED