ali@475: /*
ali@475:  * Copyright (C) 2016  J. Ali Harlow <ali@juiblex.co.uk>
ali@475:  *
ali@475:  * This program is free software; you can redistribute it and/or modify
ali@475:  * it under the terms of the GNU General Public License as published by
ali@475:  * the Free Software Foundation; either version 2 of the License, or
ali@475:  * (at your option) any later version.
ali@475:  *
ali@475:  * This program is distributed in the hope that it will be useful,
ali@475:  * but WITHOUT ANY WARRANTY; without even the implied warranty of
ali@475:  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
ali@475:  * GNU General Public License for more details.
ali@475:  *
ali@475:  * You should have received a copy of the GNU General Public License along
ali@475:  * with this program; if not, write to the Free Software Foundation, Inc.,
ali@475:  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
ali@475:  */
ali@475: 
ali@475: #include "config.h"
ali@475: 
ali@475: #undef DEBUG
ali@475: 
ali@475: #include <stdlib.h>
ali@475: #include <string.h>
ali@475: #include "razor.h"
ali@475: #include "types/types.h"
ali@475: #include "razor-internal.h"
ali@475: #include "uri.h"
ali@475: 
ali@475: /*
ali@475:  * Following RFC 3986 § 3.
ali@475:  * Note that we don't validate queries or fragments.
ali@475:  */
ali@475: 
ali@475: #define strdup0(s)			((s) ? strdup(s) : NULL)
ali@475: 
ali@475: #define string_str(str)			((char *)(str)->data)
ali@475: 
ali@475: #define string_init(str)		do { \
ali@475: 						char *_p; \
ali@475: 						array_init(str); \
ali@475: 						_p = array_add(str, 1); \
ali@475: 						*_p = '\0'; \
ali@475: 					} while(0)
ali@475: 
ali@475: #define string_append_len(str, s, len)	do { \
ali@475: 						char *_p; \
ali@475: 						_p = array_add(str, len); \
ali@475: 						_p--; \
ali@475: 						strncpy(_p, s, len); \
ali@475: 						_p[(len)] = '\0'; \
ali@475: 					} while(0)
ali@475: 
ali@475: #define string_append(str, s)		string_append_len(str, s, strlen(s))
ali@475: 
ali@475: #define string_truncate_at(str, s)	do { \
ali@475: 						int _len; \
ali@475: 						_len = (s) - \
ali@475: 						       (char *)(str)->data; \
ali@475: 						*(s) = '\0'; \
ali@475: 						(str)->size = _len + 1; \
ali@475: 					} while(0)
ali@475: 
ali@475: 
ali@475: static const char *skip_uri_scheme(const char *uri)
ali@475: {
ali@475: 	/*
ali@475: 	 * RFC 3986 defines scheme as:
ali@475: 	 *	scheme      = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
ali@475: 	 */
ali@475: 	if (*uri >= 'a' && *uri <= 'z' || *uri >= 'A' && *uri <= 'Z') {
ali@475: 		do {
ali@475: 		    uri++;
ali@475: 		} while (is_alnum(*uri) || *uri == '+' || *uri == '-' ||
ali@475: 			 *uri == '.');
ali@475: 		if (*uri == ':')
ali@475: 			return uri;
ali@475: 	}
ali@475: 	return NULL;
ali@475: }
ali@475: 
ali@475: static char *razor_strndup(const char *s, size_t n)
ali@475: {
ali@475: 	char *result;
ali@475: 
ali@475: 	if (memchr(s, '\0', n))
ali@475: 		result = strdup(s);
ali@475: 	else {
ali@475: 		result = malloc(n + 1);
ali@475: 		memcpy(result, s, n);
ali@475: 		result[n] = '\0';
ali@475: 	}
ali@475: 
ali@475: 	return result;
ali@475: }
ali@475: 
ali@475: #if 0
ali@475: /*
ali@475:  * Return the (possibly decoded) pchar or 0 on end-of-string or -1 on error
ali@475:  */
ali@475: static int pchar_get_char_validated(const char *p)
ali@475: {
ali@475: 	int c;
ali@475: 
ali@475: 	if (p[0]=='\0')
ali@475: 		c = 0;
ali@475: 	else if (p[0]=='%') {
ali@475: 		if (xdigit_value(p[1]) < 0)
ali@475: 			return -1;
ali@475: 		c = xdigit_value(p[1]) * 16;
ali@475: 		if (xdigit_value(p[2]) < 0)
ali@475: 			return -1;
ali@475: 		c += xdigit_value(p[2]);
ali@475: 	} else if (p[0] >= 'a' && p[0] <= 'z' || p[0] >= 'A' && p[0] <= 'Z' || 
ali@475: 		   p[0] >= '0' && p[0] <= '9' ||
ali@475: 		   strchr("-._~!$&'()*+,;=:@", p[0]))
ali@475: 		c = p[0];
ali@475: 	else
ali@475: 		c = -1;
ali@475: 
ali@475: 	return c;
ali@475: }
ali@475: #endif
ali@475: 
ali@475: /*
ali@475:  * Verify the percent encoding. All '%' characters must be followed by
ali@475:  * exactly two hexadecimal digits.
ali@475:  */
ali@475: static int pct_encoding_validate(const char *s)
ali@475: {
ali@475: 	while (*s) {
ali@475: 		if (*s == '%') {
ali@475: 			if (xdigit_value(s[1]) < 0 || xdigit_value(s[2]) < 0)
ali@475: 				return -1;
ali@475: 			s += 2;
ali@475: 		}
ali@475: 
ali@475: 		s++;
ali@475: 	}
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: static char *pct_encoding_normalize(char *s)
ali@475: {
ali@475: 	char *retval, *p;
ali@475: 	int c;
ali@475: 
ali@475: 	if (!s)
ali@475: 		return NULL;
ali@475: 
ali@475: 	p = retval = malloc(strlen(s) + 1);
ali@475: 
ali@475: 	while (*s) {
ali@475: 		if (*s == '%') {
ali@475: 			c = pchar_get_char(s);
ali@475: 			if (is_unreserved(c))
ali@475: 				*p++ = c;
ali@475: 			else {
ali@475: 				*p++ = '%';
ali@475: 				*p++ = "0123456789ABCDEF"[c/16];
ali@475: 				*p++ = "0123456789ABCDEF"[c%16];
ali@475: 			}
ali@475: 			pchar_next_char(s);
ali@475: 		} else
ali@475: 			*p++ = *s++;
ali@475: 	}
ali@475: 
ali@475: 	*p++ = '\0';
ali@475: 
ali@475: 	return realloc(retval, p - retval);
ali@475: }
ali@475: 
ali@475: static int validate_userinfo(const char *userinfo, struct razor_error **error)
ali@475: {
ali@475: 	const char *s;
ali@475: 
ali@475: 	for (s = userinfo; *s; s++) {
ali@475: 		if (!is_unreserved(*s) && *s != '%' && !is_sub_delim(*s)
ali@475: 		    && *s != ':') {
ali@475: 			razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 					RAZOR_GENERAL_ERROR_BAD_URI, userinfo,
ali@475: 					"Invalid URI userinfo");
ali@475: 			return -1;
ali@475: 		}
ali@475: 	}
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: static int validate_reg_name(const char *reg_name)
ali@475: {
ali@475: 	const char *s;
ali@475: 
ali@475: 	for (s = reg_name; *s; s++) {
ali@475: 		if (!is_unreserved(*s) && *s != '%' && !is_sub_delim(*s))
ali@475: 			return -1;
ali@475: 	}
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: static int validate_ipv4address(const char *s, int length)
ali@475: {
ali@475: 	int count = 0, digits, octet;
ali@475: 
ali@475: 	for (;;) {
ali@475: 		if (!length)
ali@475: 			return -1;
ali@475: 
ali@475: 		if (*s == '0') {
ali@475: 			digits = 1;
ali@475: 			octet = 0;
ali@475: 		} else {
ali@475: 			if (*s < '1' || *s > '9')
ali@475: 				return -1;
ali@475: 
ali@475: 			octet = *s - '0';
ali@475: 
ali@475: 			for (digits = 1; digits < length; digits++) {
ali@475: 				if (s[digits] >= '0' && s[digits] <= '9') {
ali@475: 					octet *= 10;
ali@475: 					octet += s[digits] - '0';
ali@475: 					if (octet > 255)
ali@475: 						return -1;
ali@475: 				} else
ali@475: 					break;
ali@475: 			}
ali@475: 		}
ali@475: 
ali@475: 		s += digits;
ali@475: 		length -= digits;
ali@475: 
ali@475: 		if (++count == 4)
ali@475: 			break;
ali@475: 
ali@475: 		if (length < 1 || *s != '.')
ali@475: 			return -1;
ali@475: 
ali@475: 		s++;
ali@475: 		length--;
ali@475: 	}
ali@475: 
ali@475: 	return length ? -1 : 0;
ali@475: }
ali@475: 
ali@475: static int count_ipv6_pieces(const char **s, int *length)
ali@475: {
ali@475: 	int count, digits;
ali@475: 
ali@475: 	for (digits = 0; digits < 4 && digits < *length; digits++) {
ali@475: 		if (!is_xdigit((*s)[digits]))
ali@475: 			break;
ali@475: 	}
ali@475: 
ali@475: 	if (!digits)
ali@475: 		return 0;
ali@475: 
ali@475: 	(*s) += digits;
ali@475: 	(*length) -= digits;
ali@475: 	count = 1;
ali@475: 
ali@475: 	if (*length && **s == ':') {
ali@475: 		(*s)++;
ali@475: 		(*length)--;
ali@475: 		count += count_ipv6_pieces(s, length);
ali@475: 		if (count == 1) {
ali@475: 			(*s)--;
ali@475: 			(*length)++;
ali@475: 		}
ali@475: 	}
ali@475: 
ali@475: 	return count;
ali@475: }
ali@475: 
ali@475: static int validate_ip_literal(const char *ip_literal, int length)
ali@475: {
ali@475: 	const char *s, *dot;
ali@475: 	int len, no_pieces, elide;
ali@475: 
ali@475: 	if (length >= 4 && ip_literal[0] == 'v') {
ali@475: 		/* IPvFuture */
ali@475: 		dot = strchr(ip_literal + 2, '.');
ali@475: 		if (!dot || dot >= ip_literal + length)
ali@475: 			return -1;
ali@475: 		for (s = ip_literal + 1; s < dot; s++) {
ali@475: 			if (!is_xdigit(*s))
ali@475: 				return -1;
ali@475: 		}
ali@475: 		for (s = dot + 1; s < ip_literal + length; s++) {
ali@475: 			if (!is_unreserved(*s) && !is_sub_delim(*s) && *s != ':')
ali@475: 				return -1;
ali@475: 		}
ali@475: 	} else {
ali@475: 		/* IPv6address */
ali@475: 		s = ip_literal;
ali@475: 		len = length;
ali@475: 		no_pieces = count_ipv6_pieces(&s, &len);
ali@475: 
ali@475: 		if (len > 1 && s[0] == ':' && s[1] == ':') {
ali@475: 			s += 2;
ali@475: 			len -= 2;
ali@475: 			elide = 1;
ali@475: 			no_pieces += count_ipv6_pieces(&s, &len);
ali@475: 		} else
ali@475: 			elide = 0;
ali@475: 
ali@475: 		if (!validate_ipv4address(s, len))
ali@475: 			no_pieces += 2;
ali@475: 		else if (len)
ali@475: 			return -1;
ali@475: 
ali@475: 		if (no_pieces > 8 || no_pieces == 8 && elide || no_pieces < 1)
ali@475: 			return -1;
ali@475: 	}
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: static int validate_host(const char *host, struct razor_error **error)
ali@475: {
ali@475: 	int retval;
ali@475: 
ali@475: 	if (host[0] == '[' && host[strlen(host) - 1] == ']')
ali@475: 		retval = validate_ip_literal(host + 1, strlen(host) - 2);
ali@475: 	else {
ali@475: 		retval = validate_ipv4address(host, strlen(host));
ali@475: 		if (retval < 0)
ali@475: 			retval = validate_reg_name(host);
ali@475: 	}
ali@475: 
ali@475: 	if (retval)
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, host,
ali@475: 				"Invalid URI host");
ali@475: 
ali@475: 	return retval;
ali@475: }
ali@475: 
ali@475: static char *strdown(char *s)
ali@475: {
ali@475: 	while (*s) {
ali@475: 		if (*s >= 'A' && *s <= 'Z') {
ali@475: 			*s -= 'A';
ali@475: 			*s += 'a';
ali@475: 		}
ali@475: 		s++;
ali@475: 	}
ali@475: 
ali@475: 	return s;
ali@475: }
ali@475: 
ali@475: static int razor_uri_parse_authority(struct razor_uri *ru,
ali@475: 				     const char *authority, int length,
ali@475: 				     struct razor_error **error)
ali@475: {
ali@475: 	const char *s, *auth = authority;
ali@475: 	char *userinfo, *port, *host;
ali@475: 
ali@475: 	s = strchr(auth, '@');
ali@475: 	if (s && s < auth + length) {
ali@475: 		userinfo = razor_strndup(auth, s - auth);
ali@475: 		s++;
ali@475: 		length -= s - auth;
ali@475: 		auth = s;
ali@475: 
ali@475: 		if (validate_userinfo(userinfo, error)) {
ali@475: 			free(userinfo);
ali@475: 			return -1;
ali@475: 		}
ali@475: 	} else
ali@475: 		userinfo = NULL;
ali@475: 
ali@475: 	s = strchr(auth, ':');
ali@475: 	if (s && s < auth + length) {
ali@475: 		s++;
ali@475: 		port = razor_strndup(s, length - (s - auth));
ali@475: 		s--;
ali@475: 		length = s - auth;
ali@475: 
ali@475: 		if (strspn(port, "0123456789") != strlen(port)) {
ali@475: 			razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 					RAZOR_GENERAL_ERROR_BAD_URI, port,
ali@475: 					"Invalid URI port");
ali@475: 			free(userinfo);
ali@475: 			free(port);
ali@475: 			return -1;
ali@475: 		}
ali@475: 	} else
ali@475: 		port = NULL;
ali@475: 
ali@475: 	host = razor_strndup(auth, length);
ali@475: 
ali@475: 	if (validate_host(host, error)) {
ali@475: 		free(userinfo);
ali@475: 		free(port);
ali@475: 		free(host);
ali@475: 		return -1;
ali@475: 	}
ali@475: 
ali@475: 	ru->userinfo = userinfo;
ali@475: 	ru->port = port;
ali@475: 	ru->host = host;
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: /*
ali@475:  * Parse either a hier-part or a relative-part
ali@475:  */
ali@475: static int razor_uri_parse_part(struct razor_uri *ru, const char *part,
ali@475: 				int relative_part, struct razor_error **error)
ali@475: {
ali@475: 	const char *s, *hp = part;
ali@475: 	char *path, *p;
ali@475: 	int noscheme = 0;
ali@475: 
ali@475: 	if (hp[0] == '/' && hp[1] == '/') {
ali@475: 		hp += 2;
ali@475: 		s = strpbrk(hp, "/?#");
ali@475: 		if (!s)
ali@475: 			s = hp + strlen(hp);
ali@475: 		if (razor_uri_parse_authority(ru, hp, s - hp, error) < 0)
ali@475: 			return -1;
ali@475: 		hp = s;
ali@475: 	} else {
ali@475: 		ru->userinfo = NULL;
ali@475: 		ru->host = NULL;
ali@475: 		ru->port = NULL;
ali@475: 	}
ali@475: 
ali@475: 	if (!*hp) {
ali@475: 		/* path-empty */
ali@475: 		ru->path = strdup("");
ali@475: 		return 0;
ali@475: 	} else if (*hp == '/') {
ali@475: 		/* path-absolute */
ali@475: 		p = path = malloc(strlen(hp) + 1);
ali@475: 		*p++ = '/';
ali@475: 		hp++;
ali@475: 		if (!*hp) {
ali@475: 			*p++ = '\0';
ali@475: 			ru->path = realloc(path, p - path);
ali@475: 			return 0;
ali@475: 		}
ali@475: 	} else if (!ru->host) {
ali@475: 		/* path-rootless or path-noscheme */
ali@475: 		noscheme = relative_part;
ali@475: 		p = path = malloc(strlen(hp) + 1);
ali@475: 	} else {
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, part,
ali@475: 				relative_part ? "Invalid URI relative part" :
ali@475: 				"Invalid URI hierarchical part");
ali@475: 		return -1;
ali@475: 	}
ali@475: 
ali@475: 	if (!is_pchar(*hp) || noscheme && *hp == ':') {
ali@475: 		free(path);
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, part,
ali@475: 				"Invalid character in URI path");
ali@475: 		return -1;
ali@475: 	}
ali@475: 	*p++ = *hp++;
ali@475: 
ali@475: 	while (*hp) {
ali@475: 		if (*hp == '/')
ali@475: 			noscheme = 0;
ali@475: 		else if (!is_pchar(*hp) || noscheme && *hp == ':') {
ali@475: 			free(path);
ali@475: 			razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 					RAZOR_GENERAL_ERROR_BAD_URI, part,
ali@475: 					"Invalid character in URI path");
ali@475: 			return -1;
ali@475: 		}
ali@475: 		*p++ = *hp++;
ali@475: 	}
ali@475: 
ali@475: 	*p++ = '\0';
ali@475: 
ali@475: 	ru->path = realloc(path, p - path);
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: void razor_uri_destroy(struct razor_uri *ru)
ali@475: {
ali@475: 	free(ru->scheme);
ali@475: 	free(ru->userinfo);
ali@475: 	free(ru->host);
ali@475: 	free(ru->port);
ali@475: 	free(ru->path);
ali@475: 	free(ru->query);
ali@475: 	free(ru->fragment);
ali@475: }
ali@475: 
ali@475: int razor_uri_parse_uri(struct razor_uri *ru, const char *uri, int absolute,
ali@475: 			struct razor_error **error)
ali@475: {
ali@475: 	int r;
ali@475: 	const char *s;
ali@475: 	char *hier_part;
ali@475: 
ali@475: 	if (pct_encoding_validate(uri) < 0) {
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, uri,
ali@475: 				"Invalid percent encoding");
ali@475: 		return -1;
ali@475: 	}
ali@475: 
ali@475: 	memset(ru, 0, sizeof(*ru));
ali@475: 
ali@475: 	s = skip_uri_scheme(uri);
ali@475: 	if (!s) {
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, uri,
ali@475: 				"Invalid URI scheme");
ali@475: 		return -1;
ali@475: 	}
ali@475: 	ru->scheme = razor_strndup(uri, s - uri);
ali@475: 	uri = s + 1;
ali@475: 
ali@475: 	s = strchr(uri, '?');
ali@475: 	if (!s)
ali@475: 		s = strchr(uri, '#');
ali@475: 	if (!s)
ali@475: 		s = uri + strlen(uri);
ali@475: 	hier_part = razor_strndup(uri, s - uri);
ali@475: 	uri = s;
ali@475: 
ali@475: 	r = razor_uri_parse_part(ru, hier_part, 0, error);
ali@475: 	free(hier_part);
ali@475: 	if (r) {
ali@475: 		razor_uri_destroy(ru);
ali@475: 		return -1;
ali@475: 	}
ali@475: 
ali@475: 	if (*uri != '?')
ali@475: 		ru->query = NULL;
ali@475: 	else {
ali@475: 		uri++;
ali@475: 		s = strchr(uri, '#');
ali@475: 		if (!s)
ali@475: 			s = uri + strlen(uri);
ali@475: 		ru->query = razor_strndup(uri, s - uri);
ali@475: 		uri = s;
ali@475: 	}
ali@475: 
ali@475: 	if (*uri != '#')
ali@475: 		ru->fragment = NULL;
ali@475: 	else if (absolute) {
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, uri,
ali@475: 				"Fragments are not allowed in absolute URIs");
ali@475: 		razor_uri_destroy(ru);
ali@475: 		return -1;
ali@475: 	} else {
ali@475: 		uri++;
ali@475: 		ru->fragment = strdup(uri);
ali@475: 	}
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: int razor_uri_parse_relative_ref(struct razor_uri *ru, const char *uri,
ali@475: 				 struct razor_error **error)
ali@475: {
ali@475: 	int r;
ali@475: 	const char *s;
ali@475: 	char *relative_part;
ali@475: 
ali@475: 	if (pct_encoding_validate(uri) < 0) {
ali@475: 		razor_set_error(error, RAZOR_GENERAL_ERROR,
ali@475: 				RAZOR_GENERAL_ERROR_BAD_URI, uri,
ali@475: 				"Invalid percent encoding");
ali@475: 		return -1;
ali@475: 	}
ali@475: 
ali@475: 	memset(ru, 0, sizeof(*ru));
ali@475: 
ali@475: 	s = strchr(uri, '?');
ali@475: 	if (!s)
ali@475: 		s = strchr(uri, '#');
ali@475: 	if (!s)
ali@475: 		s = uri + strlen(uri);
ali@475: 	relative_part = razor_strndup(uri, s - uri);
ali@475: 	uri = s;
ali@475: 
ali@475: 	r = razor_uri_parse_part(ru, relative_part, 1, error);
ali@475: 	free(relative_part);
ali@475: 	if (r)
ali@475: 		return -1;
ali@475: 
ali@475: 	if (*uri == '?') {
ali@475: 		uri++;
ali@475: 		s = strchr(uri, '#');
ali@475: 		if (!s)
ali@475: 			s = uri + strlen(uri);
ali@475: 		ru->query = razor_strndup(uri, s - uri);
ali@475: 		uri = s;
ali@475: 	} else
ali@475: 		ru->query = NULL;
ali@475: 
ali@475: 	if (*uri == '#') {
ali@475: 		uri++;
ali@475: 		ru->fragment = strdup(uri);
ali@475: 	} else
ali@475: 		ru->fragment = NULL;
ali@475: 
ali@475: 	return 0;
ali@475: }
ali@475: 
ali@475: int razor_uri_parse(struct razor_uri *ru, const char *uri,
ali@475: 		    struct razor_error **error)
ali@475: {
ali@475: 	struct razor_error *tmp_error = NULL;
ali@475: 	int r;
ali@475: 
ali@475: 	r = razor_uri_parse_uri(ru, uri, 0, &tmp_error);
ali@475: 	if (r < 0) {
ali@475: 		r = razor_uri_parse_relative_ref(ru, uri, NULL);
ali@475: 		if (r < 0)
ali@475: 			razor_propagate_error(error, tmp_error, NULL);
ali@475: 		else
ali@475: 			razor_error_free(tmp_error);
ali@475: 	}
ali@475: 
ali@475: 	return r;
ali@475: }
ali@475: 
ali@475: /*
ali@475:  * Following RFC 3986 § 5.2.4
ali@475:  */
ali@475: static char *remove_dot_segments(const char *path)
ali@475: {
ali@475: 	struct array output;
ali@475: 	char *input, *in, *s, *t;
ali@475: 	const char *step;
ali@475: 
ali@475: #ifdef DEBUG
ali@475: 	fprintf(stderr, "STEP   OUTPUT BUFFER         INPUT BUFFER\n");
ali@475: #endif
ali@475: 
ali@475: 	input = strdup(path);
ali@475: 	in = input;
ali@475: 	string_init(&output);
ali@475: 
ali@475: #ifdef DEBUG
ali@475: 	fprintf(stderr, " 1 :   %-21s %s\n", string_str(&output), in);
ali@475: #endif
ali@475: 
ali@475: 	while (*in) {
ali@475: 		if (str_has_prefix(in, "../")) {
ali@475: 			step = "2A";
ali@475: 			in += 3;
ali@475: 		} else if (str_has_prefix(in, "./")) {
ali@475: 			step = "2A";
ali@475: 			in += 2;
ali@475: 		} else if (str_has_prefix(in, "/./")) {
ali@475: 			step = "2B";
ali@475: 			in += 2;
ali@475: 		} else if (!strcmp(in, "/.")) {
ali@475: 			step = "2B";
ali@475: 			in++;
ali@475: 			*in = '/';
ali@475: 		} else if (str_has_prefix(in, "/../")) {
ali@475: 			step = "2C";
ali@475: 			in += 3;
ali@475: 			s = strrchr(string_str(&output), '/');
ali@475: 			if (!s)
ali@475: 				s = string_str(&output);
ali@475: 			string_truncate_at(&output, s);
ali@475: 		} else if (!strcmp(in, "/..")) {
ali@475: 			step = "2C";
ali@475: 			in += 2;
ali@475: 			*in = '/';
ali@475: 			s = strrchr(string_str(&output), '/');
ali@475: 			if (!s)
ali@475: 				s = string_str(&output);
ali@475: 			string_truncate_at(&output, s);
ali@475: 		} else if (!strcmp(in, ".") || !strcmp(in, "..")) {
ali@475: 			step = "2D";
ali@475: 			in += strlen(in);
ali@475: 		} else {
ali@475: 			step = "2E";
ali@475: 			t = strchr(in + 1, '/');
ali@475: 			if (!t)
ali@475: 				t = in + strlen(in);
ali@475: 			string_append_len(&output, in, t - in);
ali@475: 			in = t;
ali@475: 		}
ali@475: #ifdef DEBUG
ali@475: 		fprintf(stderr, " %s:   %-21s %s\n", step, string_str(&output),
ali@475: 			in);
ali@475: #endif
ali@475: 	}
ali@475: 
ali@475: 	free(input);
ali@475: 	return string_str(&output);
ali@475: }
ali@475: 
ali@475: 
ali@475: /*
ali@475:  * Following RFC 3986 § 6.2.2
ali@475:  */
ali@475: void razor_uri_normalize(struct razor_uri *ru)
ali@475: {
ali@475: 	char *s;
ali@475: 
ali@475: 	strdown(ru->scheme);
ali@475: 	if (ru->host)
ali@475: 		strdown(ru->host);
ali@475: 
ali@475: 	s = pct_encoding_normalize(ru->userinfo);
ali@475: 	free(ru->userinfo);
ali@475: 	ru->userinfo = s;
ali@475: 
ali@475: 	s = pct_encoding_normalize(ru->host);
ali@475: 	free(ru->host);
ali@475: 	ru->host = s;
ali@475: 
ali@475: 	s = pct_encoding_normalize(ru->path);
ali@475: 	free(ru->path);
ali@475: 	ru->path = s;
ali@475: 
ali@475: 	s = pct_encoding_normalize(ru->query);
ali@475: 	free(ru->query);
ali@475: 	ru->query = s;
ali@475: 
ali@475: 	s = pct_encoding_normalize(ru->fragment);
ali@475: 	free(ru->fragment);
ali@475: 	ru->fragment = s;
ali@475: 
ali@475: 	s = remove_dot_segments(ru->path);
ali@475: 	free(ru->path);
ali@475: 	ru->path = s;
ali@475: }
ali@475: 
ali@475: char *razor_uri_get_authority(const struct razor_uri *ru)
ali@475: {
ali@475: 	char *result, *r;
ali@475: 	int len = 1;
ali@475: 
ali@475: 	if (ru->host) {
ali@475: 		if (ru->userinfo)
ali@475: 			len += strlen(ru->userinfo) + 1;
ali@475: 		len += strlen(ru->host);
ali@475: 		if (ru->port)
ali@475: 			len += strlen(ru->port) + 1;
ali@475: 	} else
ali@475: 		return NULL;
ali@475: 
ali@475: 	r = result = malloc(len);
ali@475: 
ali@475: 	if (ru->userinfo) {
ali@475: 		strcpy(r, ru->userinfo);
ali@475: 		r += strlen(r);
ali@475: 		*r++ = '@';
ali@475: 	}
ali@475: 
ali@475: 	strcpy(r, ru->host);
ali@475: 	r += strlen(r);
ali@475: 
ali@475: 	if (ru->port) {
ali@475: 		*r++ = ':';
ali@475: 		strcpy(r, ru->port);
ali@475: 	}
ali@475: 
ali@475: 	return result;
ali@475: }
ali@475: 
ali@475: /*
ali@475:  * Following RFC 3986 § 5.3
ali@475:  */
ali@475: char *razor_uri_recompose(const struct razor_uri *ru)
ali@475: {
ali@475: 	char *authority, *result, *r;
ali@475: 	int len = 1;
ali@475: 
ali@475: 	authority = razor_uri_get_authority(ru);
ali@475: 
ali@475: 	if (ru->scheme)
ali@475: 		len += strlen(ru->scheme) + 1;
ali@475: 	if (authority)
ali@475: 		len += strlen(authority) + 2;
ali@475: 	len += strlen(ru->path);
ali@475: 	if (ru->query)
ali@475: 		len += strlen(ru->query) + 1;
ali@475: 	if (ru->fragment)
ali@475: 		len += strlen(ru->fragment) + 1;
ali@475: 
ali@475: 	r = result = malloc(len);
ali@475: 
ali@475: 	if (ru->scheme) {
ali@475: 		strcpy(r, ru->scheme);
ali@475: 		r += strlen(r);
ali@475: 		*r++ = ':';
ali@475: 	}
ali@475: 
ali@475: 	if (authority) {
ali@475: 		*r++ = '/';
ali@475: 		*r++ = '/';
ali@475: 		strcpy(r, authority);
ali@475: 		free(authority);
ali@475: 		r += strlen(r);
ali@475: 	}
ali@475: 
ali@475: 	strcpy(r, ru->path);
ali@475: 	r += strlen(r);
ali@475: 
ali@475: 	if (ru->query) {
ali@475: 		*r++ = '?';
ali@475: 		strcpy(r, ru->query);
ali@475: 		r += strlen(r);
ali@475: 	}
ali@475: 
ali@475: 	if (ru->fragment) {
ali@475: 		*r++ = '#';
ali@475: 		strcpy(r, ru->fragment);
ali@475: 	}
ali@475: 
ali@475: 	return result;
ali@475: }
ali@475: 
ali@475: /*
ali@475:  * Following RFC 3986 § 5.2.3
ali@475:  */
ali@475: static char *merge_paths(const struct razor_uri *base,const struct razor_uri *R)
ali@475: {
ali@475: 	char *s, *t, *path;
ali@475: 
ali@475: 	if (base->host && !*base->path)
ali@475: 		path = razor_concat("/", R->path, NULL);
ali@475: 	else {
ali@475: 		s = strrchr(base->path, '/');
ali@475: 		if (s) {
ali@475: 			t = razor_strndup(base->path, s + 1 - base->path);
ali@475: 			path = razor_concat(t, R->path, NULL);
ali@475: 			free(t);
ali@475: 		} else
ali@475: 			path = strdup(R->path);
ali@475: 	}
ali@475: 
ali@475: 	return path;
ali@475: }
ali@475: 
ali@475: /*
ali@475:  * Following RFC 3986 § 5.2
ali@475:  */
ali@475: void razor_uri_resolve(struct razor_uri *T, const struct razor_uri *base,
ali@475: 		       const struct razor_uri *R)
ali@475: {
ali@475: 	char *s;
ali@475: 
ali@475: 	if (R->scheme) {
ali@475: 		T->scheme = strdup(R->scheme);
ali@475: 		T->userinfo = strdup0(R->userinfo);
ali@475: 		T->host = strdup0(R->host);
ali@475: 		T->port = strdup0(R->port);
ali@475: 		T->path = remove_dot_segments(R->path);
ali@475: 		T->query = strdup0(R->query);
ali@475: 	} else {
ali@475: 		if (R->host) {
ali@475: 			T->userinfo = strdup0(R->userinfo);
ali@475: 			T->host = strdup0(R->host);
ali@475: 			T->port = strdup0(R->port);
ali@475: 			T->path = remove_dot_segments(R->path);
ali@475: 			T->query = strdup0(R->query);
ali@475: 		} else {
ali@475: 			if (!*R->path) {
ali@475: 				T->path = strdup(base->path);
ali@475: 				if (R->query)
ali@475: 					T->query = strdup(R->query);
ali@475: 				else
ali@475: 					T->query = strdup0(base->query);
ali@475: 			} else {
ali@475: 				if (*R->path == '/')
ali@475: 					T->path = remove_dot_segments(R->path);
ali@475: 				else {
ali@475: 					s = merge_paths(base, R);
ali@475: 					T->path = remove_dot_segments(s);
ali@475: 					free(s);
ali@475: 				}
ali@475: 				T->query = strdup0(R->query);
ali@475: 			}
ali@475: 			T->userinfo = strdup0(base->userinfo);
ali@475: 			T->host = strdup0(base->host);
ali@475: 			T->port = strdup0(base->port);
ali@475: 		}
ali@475: 		T->scheme = strdup(base->scheme);
ali@475: 	}
ali@475: 	T->fragment = strdup0(R->fragment);
ali@475: }
ali@475: 
ali@475: /*
ali@475:  * This differs from razor_uri_resolve() both in the types of its arguments
ali@475:  * and in the fact that it takes a root URI rather than a base URI. The base
ali@475:  * URI is determined by appending a slash to the root URI (if it doesn't
ali@475:  * already end in a slash). Finally, uri can be explicitly marked as either
ali@475:  * relative (ie., a relative-ref) or not (ie., a URI). This is important as
ali@475:  * otherwise "c:/xxx" could be interpreted as a URI in the "c" scheme.
ali@475:  */
ali@475: char *razor_resolve_uri_root(const char *root_uri, const char *uri,
ali@475: 			     int is_relative, struct razor_error **error)
ali@475: {
ali@475: 	int r;
ali@475:         char *base_uri, *s, *result;
ali@475: 	struct razor_uri ru, base, file;
ali@475: 
ali@475: 	if (!root_uri || !*root_uri)
ali@475: 		root_uri = "file:/";
ali@475: 
ali@475: 	if (root_uri[strlen(root_uri) - 1] == '/')
ali@475: 		base_uri = strdup(root_uri);
ali@475: 	else
ali@475: 		base_uri = razor_concat(root_uri, "/", NULL);
ali@475: 
ali@475: 	r = razor_uri_parse_uri(&base, base_uri, 1, error);
ali@475: 	free(base_uri);
ali@475: 	if (r)
ali@475: 		return NULL;
ali@475: 
ali@475: 	if (is_relative > 0) {
ali@475: 		/*
ali@475: 		 * We can't use razor_uri_parse_relative_ref() to parse
ali@475: 		 * uri in case it starts with a segment that includes a
ali@475: 		 * colon. Thus we use this kludge.
ali@475: 		 */
ali@475: 		s = razor_concat("scheme:", uri, NULL);
ali@475: 		r = razor_uri_parse_uri(&file, s, 0, error);
ali@475: 		free(s);
ali@475: 		if (!r) {
ali@475: 			free(file.scheme);
ali@475: 			file.scheme = NULL;
ali@475: 		}
ali@475: 	}
ali@475: 	else if (!is_relative)
ali@475: 		r = razor_uri_parse_uri(&file, uri, 0, error);
ali@475: 	else
ali@475: 		r = razor_uri_parse(&file, uri, error);
ali@475: 	if (r) {
ali@475: 		razor_uri_destroy(&base);
ali@475: 		return NULL;
ali@475: 	}
ali@475: 
ali@475: 	razor_uri_resolve(&ru, &base, &file);
ali@475: 
ali@475: 	razor_uri_destroy(&base);
ali@475: 	razor_uri_destroy(&file);
ali@475: 
ali@475: 	result = razor_uri_recompose(&ru);
ali@475: 
ali@475: 	razor_uri_destroy(&ru);
ali@475: 
ali@475: 	return result;
ali@475: }