util: improve/refactor URI parsing and formatting

Removed/rewritten the functions:
absuri, parseuri, and encodeuri() for percent-encoding.

The functions are now split separately with the following purpose:

- uri_format: format struct uri into a string.
- uri_hasscheme: quick check if a string is absolute or not.
- uri_makeabs: make a URI absolute using a base uri and the original URI.
- uri_parse: parse a string into a struct uri.

The following URLs are better parsed:

- URLs with extra "/"'s in the path prepended are kept as is, no "/" is added
  either for empty paths.
- URLs like "http://codemadness.org" are not changed to
  "http://codemadness.org/" anymore (paths are kept as is, unless they are
  non-empty and not start with "/").
- Paths are not percent-encoded anymore.
- URLs with userinfo field (username, password) are parsed.
  like: ftp://user:password@[2001:db8::7]:2121/rfc/rfc1808.txt
- Non-authoritive URLs like mailto:some@email.org, magnet URIs, ISBN URIs/urn,
  like: urn:isbn:0-395-36341-1 are allowed and parsed correctly.
- Both local (file:///) and non-local (file://) are supported.
- Specifying a base URL with a port will now only use it when the relative URL
  has no host and port set and follows RFC3986 5.2.2 more closely.
- Parsing numeric port: parse as signed long and check <= 0, empty port is
  allowed.
- Parsing URIs containing query, fragment, but no path separator (/) will now
  parse the component properly.

For sfeed:

- Parse the baseURI only once (no need to do it every time for making absolute
  URIs).
- If a link/enclosure is absolute already or if there is no base URL specified
  then just print the link directly. There have also been other small performance
  improvements related to handling URIs.

References:
- https://tools.ietf.org/html/rfc3986
  - Section "5.2.2. Transform References" have also been helpful.

1 files changed, 160 insertions, 124 deletions

diff --git a/util.c b/util.c
index 362fdd9..6aef1f3 100644
--- a/util.c
+++ b/util.c
@@ -7,167 +7,203 @@
 
 #include "util.h"
 
+/* check if string has a non-empty scheme / protocol part */
 int
-parseuri(const char *s, struct uri *u, int rel)
+uri_hasscheme(const char *s)
 {
-	const char *p = s, *b;
-	char *endptr = NULL;
+	const char *p = s;
+
+	for (; isalpha((unsigned char)*p) || isdigit((unsigned char)*p) ||
+		       *p == '+' || *p == '-' || *p == '.'; p++)
+		;
+	/* scheme, except if empty and starts with ":" then it is a path */
+	return (*p == ':' && p != s);
+}
+
+int
+uri_parse(const char *s, struct uri *u)
+{
+	const char *p = s;
+	char *endptr;
 	size_t i;
-	unsigned long l;
+	long l;
 
-	u->proto[0] = u->host[0] = u->path[0] = u->port[0] = '\0';
-	if (!*s)
-		return 0;
+	u->proto[0] = u->userinfo[0] = u->host[0] = u->port[0] = '\0';
+	u->path[0] = u->query[0] = u->fragment[0] = '\0';
 
-	/* prefix is "//", don't read protocol, skip to domain parsing */
-	if (!strncmp(p, "//", 2)) {
+	/* protocol-relative */
+	if (*p == '/' && *(p + 1) == '/') {
 		p += 2; /* skip "//" */
-	} else {
-		/* protocol part */
-		for (p = s; isalpha((unsigned char)*p) || isdigit((unsigned char)*p) ||
-			       *p == '+' || *p == '-' || *p == '.'; p++)
-			;
-		if (!strncmp(p, "://", 3)) {
-			if ((size_t)(p - s) >= sizeof(u->proto))
-				return -1; /* protocol too long */
-			memcpy(u->proto, s, p - s);
-			u->proto[p - s] = '\0';
+		goto parseauth;
+	}
+
+	/* scheme / protocol part */
+	for (; isalpha((unsigned char)*p) || isdigit((unsigned char)*p) ||
+		       *p == '+' || *p == '-' || *p == '.'; p++)
+		;
+	/* scheme, except if empty and starts with ":" then it is a path */
+	if (*p == ':' && p != s) {
+		if (*(p + 1) == '/' && *(p + 2) == '/')
 			p += 3; /* skip "://" */
-		} else {
-			p = s; /* no protocol format, set to start */
-			/* relative url: read rest as path, else as domain */
-			if (rel)
-				goto readpath;
-		}
+		else
+			p++; /* skip ":" */
+
+		if ((size_t)(p - s) >= sizeof(u->proto))
+			return -1; /* protocol too long */
+		memcpy(u->proto, s, p - s);
+		u->proto[p - s] = '\0';
+
+		if (*(p - 1) != '/')
+			goto parsepath;
+	} else {
+		p = s; /* no scheme format, reset to start */
+		goto parsepath;
+	}
+
+parseauth:
+	/* userinfo (username:password) */
+	i = strcspn(p, "@/?#");
+	if (p[i] == '@') {
+		if (i >= sizeof(u->userinfo))
+			return -1; /* userinfo too long */
+		memcpy(u->userinfo, p, i);
+		u->userinfo[i] = '\0';
+		p += i + 1;
 	}
+
 	/* IPv6 address */
 	if (*p == '[') {
-		/* bracket not found or host too long */
-		if (!(b = strchr(p, ']')) || (size_t)(b - p) < 3 ||
-		    (size_t)(b - p) >= sizeof(u->host))
+		/* bracket not found, host too short or too long */
+		i = strcspn(p, "]");
+		if (p[i] != ']' || i < 3)
 			return -1;
-		memcpy(u->host, p, b - p + 1);
-		u->host[b - p + 1] = '\0';
-		p = b + 1;
+		i++; /* including "]" */
 	} else {
 		/* domain / host part, skip until port, path or end. */
-		if ((i = strcspn(p, ":/")) >= sizeof(u->host))
-			return -1; /* host too long */
-		memcpy(u->host, p, i);
-		u->host[i] = '\0';
-		p = &p[i];
+		i = strcspn(p, ":/?#");
 	}
+	if (i >= sizeof(u->host))
+		return -1; /* host too long */
+	memcpy(u->host, p, i);
+	u->host[i] = '\0';
+	p += i;
+
 	/* port */
 	if (*p == ':') {
-		if ((i = strcspn(++p, "/")) >= sizeof(u->port))
+		p++;
+		if ((i = strcspn(p, "/?#")) >= sizeof(u->port))
 			return -1; /* port too long */
 		memcpy(u->port, p, i);
 		u->port[i] = '\0';
-		/* check for valid port: range 1 - 65535 */
+		/* check for valid port: range 1 - 65535, may be empty */
 		errno = 0;
-		l = strtoul(u->port, &endptr, 10);
-		if (errno || u->port[0] == '\0' || *endptr ||
-		    !l || l > 65535)
+		l = strtol(u->port, &endptr, 10);
+		if (i && (errno || *endptr || l <= 0 || l > 65535))
 			return -1;
-		p = &p[i];
+		p += i;
 	}
-readpath:
-	if (u->host[0]) {
-		p = &p[strspn(p, "/")];
-		strlcpy(u->path, "/", sizeof(u->path));
-	} else {
-		/* absolute uri must have a host specified */
-		if (!rel)
-			return -1;
-	}
-	/* treat truncation as an error */
-	if (strlcat(u->path, p, sizeof(u->path)) >= sizeof(u->path))
-		return -1;
-	return 0;
-}
 
-static int
-encodeuri(char *buf, size_t bufsiz, const char *s)
-{
-	static const char *table = "0123456789ABCDEF";
-	size_t i, b;
+parsepath:
+	/* path */
+	if ((i = strcspn(p, "?#")) >= sizeof(u->path))
+		return -1; /* path too long */
+	memcpy(u->path, p, i);
+	u->path[i] = '\0';
+	p += i;
 
-	for (i = 0, b = 0; s[i]; i++) {
-		if ((unsigned char)s[i] <= ' ' ||
-		    (unsigned char)s[i] >= 127) {
-			if (b + 3 >= bufsiz)
-				return -1;
-			buf[b++] = '%';
-			buf[b++] = table[((unsigned char)s[i] >> 4) & 15];
-			buf[b++] = table[(unsigned char)s[i] & 15];
-		} else if (b < bufsiz) {
-			buf[b++] = s[i];
-		} else {
-			return -1;
-		}
+	/* query */
+	if (*p == '?') {
+		p++;
+		if ((i = strcspn(p, "#")) >= sizeof(u->query))
+			return -1; /* query too long */
+		memcpy(u->query, p, i);
+		u->query[i] = '\0';
+		p += i;
+	}
+
+	/* fragment */
+	if (*p == '#') {
+		p++;
+		if ((i = strlen(p)) >= sizeof(u->fragment))
+			return -1; /* fragment too long */
+		memcpy(u->fragment, p, i);
+		u->fragment[i] = '\0';
 	}
-	if (b >= bufsiz)
-		return -1;
-	buf[b] = '\0';
 
 	return 0;
 }
 
-/* Get absolute uri; if `link` is relative use `base` to make it absolute.
- * the returned string in `buf` is uri encoded, see: encodeuri(). */
+/* Transform and try to make the URI `u` absolute using base URI `b` into `a`.
+   Follows some of the logic from "RFC 3986 - 5.2.2. Transform References".
+   Returns 0 on success, -1 on error or truncation. */
 int
-absuri(char *buf, size_t bufsiz, const char *link, const char *base)
+uri_makeabs(struct uri *a, struct uri *u, struct uri *b)
 {
-	struct uri ulink, ubase;
-	char tmp[4096], *host, *p, *port;
-	int c, r;
-	size_t i;
+	char *p;
+	int c;
 
-	buf[0] = '\0';
-	if (parseuri(base, &ubase, 0) == -1 ||
-	    parseuri(link, &ulink, 1) == -1 ||
-	    (!ulink.host[0] && !ubase.host[0]))
-		return -1;
+	strlcpy(a->fragment, u->fragment, sizeof(a->fragment));
 
-	if (!strncmp(link, "//", 2)) {
-		host = ulink.host;
-		port = ulink.port;
-	} else {
-		host = ulink.host[0] ? ulink.host : ubase.host;
-		port = ulink.port[0] ? ulink.port : ubase.port;
+	if (u->proto[0] || u->host[0]) {
+		strlcpy(a->proto, u->proto[0] ? u->proto : b->proto, sizeof(a->proto));
+		strlcpy(a->host, u->host, sizeof(a->host));
+		strlcpy(a->userinfo, u->userinfo, sizeof(a->userinfo));
+		strlcpy(a->host, u->host, sizeof(a->host));
+		strlcpy(a->port, u->port, sizeof(a->port));
+		strlcpy(a->path, u->path, sizeof(a->path));
+		strlcpy(a->query, u->query, sizeof(a->query));
+		return 0;
 	}
-	r = snprintf(tmp, sizeof(tmp), "%s://%s%s%s",
-		ulink.proto[0] ?
-			ulink.proto :
-			(ubase.proto[0] ? ubase.proto : "http"),
-		host,
-		port[0] ? ":" : "",
-		port);
-	if (r < 0 || (size_t)r >= sizeof(tmp))
-		return -1; /* error or truncation */
-
-	/* relative to root */
-	if (!ulink.host[0] && ulink.path[0] != '/') {
-		/* relative to base url path */
-		if (ulink.path[0]) {
-			if ((p = strrchr(ubase.path, '/'))) {
-				/* temporary null-terminate */
-				c = *(++p);
-				*p = '\0';
-				i = strlcat(tmp, ubase.path, sizeof(tmp));
-				*p = c; /* restore */
-				if (i >= sizeof(tmp))
-					return -1;
-			}
-		} else if (strlcat(tmp, ubase.path, sizeof(tmp)) >=
-		           sizeof(tmp)) {
-			return -1;
+
+	strlcpy(a->proto, b->proto, sizeof(a->proto));
+	strlcpy(a->host, b->host, sizeof(a->host));
+	strlcpy(a->userinfo, b->userinfo, sizeof(a->userinfo));
+	strlcpy(a->host, b->host, sizeof(a->host));
+	strlcpy(a->port, b->port, sizeof(a->port));
+
+	if (!u->path[0]) {
+		strlcpy(a->path, b->path, sizeof(a->path));
+	} else if (u->path[0] == '/') {
+		strlcpy(a->path, u->path, sizeof(a->path));
+	} else {
+		a->path[0] = (a->host[0] && b->path[0] != '/') ? '/' : '\0';
+		a->path[1] = '\0';
+
+		if ((p = strrchr(b->path, '/'))) {
+			c = *(++p);
+			*p = '\0'; /* temporary NUL-terminate */
+			if (strlcat(a->path, b->path, sizeof(a->path)) >= sizeof(a->path))
+				return -1;
+			*p = c; /* restore */
 		}
+		if (strlcat(a->path, u->path, sizeof(a->path)) >= sizeof(a->path))
+			return -1;
 	}
-	if (strlcat(tmp, ulink.path, sizeof(tmp)) >= sizeof(tmp))
-		return -1;
 
-	return encodeuri(buf, bufsiz, tmp);
+	if (u->path[0] || u->query[0])
+		strlcpy(a->query, u->query, sizeof(a->query));
+	else
+		strlcpy(a->query, b->query, sizeof(a->query));
+
+	return 0;
+}
+
+int
+uri_format(char *buf, size_t bufsiz, struct uri *u)
+{
+	return snprintf(buf, bufsiz, "%s%s%s%s%s%s%s%s%s%s%s%s",
+		u->proto,
+		u->userinfo[0] ? u->userinfo : "",
+		u->userinfo[0] ? "@" : "",
+		u->host,
+		u->port[0] ? ":" : "",
+		u->port,
+		u->host[0] && u->path[0] && u->path[0] != '/' ? "/" : "",
+		u->path,
+		u->query[0] ? "?" : "",
+		u->query,
+		u->fragment[0] ? "#" : "",
+		u->fragment);
 }
 
 /* Splits fields in the line buffer by replacing TAB separators with NUL ('\0')