cannam@226: /*
cannam@226:   Copyright 2011-2014 David Robillard <http://drobilla.net>
cannam@226: 
cannam@226:   Permission to use, copy, modify, and/or distribute this software for any
cannam@226:   purpose with or without fee is hereby granted, provided that the above
cannam@226:   copyright notice and this permission notice appear in all copies.
cannam@226: 
cannam@226:   THIS SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
cannam@226:   WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
cannam@226:   MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
cannam@226:   ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
cannam@226:   WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
cannam@226:   ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
cannam@226:   OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
cannam@226: */
cannam@226: 
cannam@226: #include "serd_internal.h"
cannam@226: 
cannam@226: #include <stdlib.h>
cannam@226: #include <string.h>
cannam@226: 
cannam@226: // #define URI_DEBUG 1
cannam@226: 
cannam@226: SERD_API
cannam@226: const uint8_t*
cannam@226: serd_uri_to_path(const uint8_t* uri)
cannam@226: {
cannam@226: 	const uint8_t* path = uri;
cannam@226: 	if (!is_windows_path(uri) && serd_uri_string_has_scheme(uri)) {
cannam@226: 		if (strncmp((const char*)uri, "file:", 5)) {
cannam@226: 			fprintf(stderr, "Non-file URI `%s'\n", uri);
cannam@226: 			return NULL;
cannam@226: 		} else if (!strncmp((const char*)uri, "file://localhost/", 17)) {
cannam@226: 			path = uri + 16;
cannam@226: 		} else if (!strncmp((const char*)uri, "file://", 7)) {
cannam@226: 			path = uri + 7;
cannam@226: 		} else {
cannam@226: 			fprintf(stderr, "Invalid file URI `%s'\n", uri);
cannam@226: 			return NULL;
cannam@226: 		}
cannam@226: 		if (is_windows_path(path + 1)) {
cannam@226: 			++path;  // Special case for terrible Windows file URIs
cannam@226: 		}
cannam@226: 	}
cannam@226: 	return path;
cannam@226: }
cannam@226: 
cannam@226: SERD_API
cannam@226: uint8_t*
cannam@226: serd_file_uri_parse(const uint8_t* uri, uint8_t** hostname)
cannam@226: {
cannam@226: 	const uint8_t* path = uri;
cannam@226: 	if (hostname) {
cannam@226: 		*hostname = NULL;
cannam@226: 	}
cannam@226: 	if (!strncmp((const char*)uri, "file://", 7)) {
cannam@226: 		const uint8_t* auth = uri + 7;
cannam@226: 		if (*auth == '/') {  // No hostname
cannam@226: 			path = auth;
cannam@226: 		} else {  // Has hostname
cannam@226: 			if (!(path = (const uint8_t*)strchr((const char*)auth, '/'))) {
cannam@226: 				return NULL;
cannam@226: 			}
cannam@226: 			if (hostname) {
cannam@226: 				*hostname = (uint8_t*)calloc(1, path - auth + 1);
cannam@226: 				memcpy(*hostname, auth, path - auth);
cannam@226: 			}
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	if (is_windows_path(path + 1)) {
cannam@226: 		++path;
cannam@226: 	}
cannam@226: 
cannam@226: 	SerdChunk chunk = { NULL, 0 };
cannam@226: 	for (const uint8_t* s = path; *s; ++s) {
cannam@226: 		if (*s == '%') {
cannam@226: 			if (*(s + 1) == '%') {
cannam@226: 				serd_chunk_sink("%", 1, &chunk);
cannam@226: 				++s;
cannam@226: 			} else if (is_digit(*(s + 1)) && is_digit(*(s + 2))) {
cannam@226: 				const uint8_t code[3] = { *(s + 1), *(s + 2), 0 };
cannam@226: 				uint32_t num;
cannam@226: 				sscanf((const char*)code, "%X", &num);
cannam@226: 				const uint8_t c = num;
cannam@226: 				serd_chunk_sink(&c, 1, &chunk);
cannam@226: 				s += 2;
cannam@226: 			} else {
cannam@226: 				s += 2;  // Junk escape, ignore
cannam@226: 			}
cannam@226: 		} else {
cannam@226: 			serd_chunk_sink(s, 1, &chunk);
cannam@226: 		}
cannam@226: 	}
cannam@226: 	return serd_chunk_sink_finish(&chunk);
cannam@226: }
cannam@226: 
cannam@226: SERD_API
cannam@226: bool
cannam@226: serd_uri_string_has_scheme(const uint8_t* utf8)
cannam@226: {
cannam@226: 	// RFC3986: scheme ::= ALPHA *( ALPHA / DIGIT / "+" / "-" / "." )
cannam@226: 	if (!utf8 || !is_alpha(utf8[0])) {
cannam@226: 		return false;  // Invalid scheme initial character, URI is relative
cannam@226: 	}
cannam@226: 	for (uint8_t c; (c = *++utf8) != '\0';) {
cannam@226: 		switch (c) {
cannam@226: 		case ':':
cannam@226: 			return true;  // End of scheme
cannam@226: 		case '+': case '-': case '.':
cannam@226: 			break;  // Valid scheme character, continue
cannam@226: 		default:
cannam@226: 			if (!is_alpha(c) && !is_digit(c)) {
cannam@226: 				return false;  // Invalid scheme character
cannam@226: 			}
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	return false;
cannam@226: }
cannam@226: 
cannam@226: #ifdef URI_DEBUG
cannam@226: static void
cannam@226: serd_uri_dump(const SerdURI* uri, FILE* file)
cannam@226: {
cannam@226: #define PRINT_PART(range, name) \
cannam@226: 	if (range.buf) { \
cannam@226: 		fprintf(stderr, "  " name " = "); \
cannam@226: 		fwrite((range).buf, 1, (range).len, stderr); \
cannam@226: 		fprintf(stderr, "\n"); \
cannam@226: 	}
cannam@226: 
cannam@226: 	PRINT_PART(uri->scheme,    "scheme   ");
cannam@226: 	PRINT_PART(uri->authority, "authority");
cannam@226: 	PRINT_PART(uri->path_base, "path_base");
cannam@226: 	PRINT_PART(uri->path,      "path     ");
cannam@226: 	PRINT_PART(uri->query,     "query    ");
cannam@226: 	PRINT_PART(uri->fragment,  "fragment ");
cannam@226: }
cannam@226: #endif
cannam@226: 
cannam@226: SERD_API
cannam@226: SerdStatus
cannam@226: serd_uri_parse(const uint8_t* utf8, SerdURI* uri)
cannam@226: {
cannam@226: 	*uri = SERD_URI_NULL;
cannam@226: 
cannam@226: 	const uint8_t* ptr = utf8;
cannam@226: 
cannam@226: 	/* See http://tools.ietf.org/html/rfc3986#section-3
cannam@226: 	   URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ]
cannam@226: 	*/
cannam@226: 
cannam@226: 	/* S3.1: scheme ::= ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) */
cannam@226: 	if (is_alpha(*ptr)) {
cannam@226: 		for (uint8_t c = *++ptr; true; c = *++ptr) {
cannam@226: 			switch (c) {
cannam@226: 			case '\0': case '/': case '?': case '#':
cannam@226: 				ptr = utf8;
cannam@226: 				goto path;  // Relative URI (starts with path by definition)
cannam@226: 			case ':':
cannam@226: 				uri->scheme.buf = utf8;
cannam@226: 				uri->scheme.len = (ptr++) - utf8;
cannam@226: 				goto maybe_authority;  // URI with scheme
cannam@226: 			case '+': case '-': case '.':
cannam@226: 				continue;
cannam@226: 			default:
cannam@226: 				if (is_alpha(c) || is_digit(c)) {
cannam@226: 					continue;
cannam@226: 				}
cannam@226: 			}
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	/* S3.2: The authority component is preceded by a double slash ("//")
cannam@226: 	   and is terminated by the next slash ("/"), question mark ("?"),
cannam@226: 	   or number sign ("#") character, or by the end of the URI.
cannam@226: 	*/
cannam@226: maybe_authority:
cannam@226: 	if (*ptr == '/' && *(ptr + 1) == '/') {
cannam@226: 		ptr += 2;
cannam@226: 		uri->authority.buf = ptr;
cannam@226: 		for (uint8_t c; (c = *ptr) != '\0'; ++ptr) {
cannam@226: 			switch (c) {
cannam@226: 			case '/': goto path;
cannam@226: 			case '?': goto query;
cannam@226: 			case '#': goto fragment;
cannam@226: 			default:
cannam@226: 				++uri->authority.len;
cannam@226: 			}
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	/* RFC3986 S3.3: The path is terminated by the first question mark ("?")
cannam@226: 	   or number sign ("#") character, or by the end of the URI.
cannam@226: 	*/
cannam@226: path:
cannam@226: 	switch (*ptr) {
cannam@226: 	case '?':  goto query;
cannam@226: 	case '#':  goto fragment;
cannam@226: 	case '\0': goto end;
cannam@226: 	default:  break;
cannam@226: 	}
cannam@226: 	uri->path.buf = ptr;
cannam@226: 	uri->path.len = 0;
cannam@226: 	for (uint8_t c; (c = *ptr) != '\0'; ++ptr) {
cannam@226: 		switch (c) {
cannam@226: 		case '?': goto query;
cannam@226: 		case '#': goto fragment;
cannam@226: 		default:
cannam@226: 			++uri->path.len;
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	/* RFC3986 S3.4: The query component is indicated by the first question
cannam@226: 	   mark ("?") character and terminated by a number sign ("#") character
cannam@226: 	   or by the end of the URI.
cannam@226: 	*/
cannam@226: query:
cannam@226: 	if (*ptr == '?') {
cannam@226: 		uri->query.buf = ++ptr;
cannam@226: 		for (uint8_t c; (c = *ptr) != '\0'; ++ptr) {
cannam@226: 			switch (c) {
cannam@226: 			case '#':
cannam@226: 				goto fragment;
cannam@226: 			default:
cannam@226: 				++uri->query.len;
cannam@226: 			}
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	/* RFC3986 S3.5: A fragment identifier component is indicated by the
cannam@226: 	   presence of a number sign ("#") character and terminated by the end
cannam@226: 	   of the URI.
cannam@226: 	*/
cannam@226: fragment:
cannam@226: 	if (*ptr == '#') {
cannam@226: 		uri->fragment.buf = ptr;
cannam@226: 		while (*ptr++ != '\0') {
cannam@226: 			++uri->fragment.len;
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: end:
cannam@226: 	#ifdef URI_DEBUG
cannam@226: 	fprintf(stderr, "PARSE URI <%s>\n", utf8);
cannam@226: 	serd_uri_dump(uri, stderr);
cannam@226: 	fprintf(stderr, "\n");
cannam@226: 	#endif
cannam@226: 
cannam@226: 	return SERD_SUCCESS;
cannam@226: }
cannam@226: 
cannam@226: /**
cannam@226:    Remove leading dot components from `path`.
cannam@226:    See http://tools.ietf.org/html/rfc3986#section-5.2.3
cannam@226:    @param up Set to the number of up-references (e.g. "../") trimmed
cannam@226:    @return A pointer to the new start of `path`
cannam@226: */
cannam@226: static const uint8_t*
cannam@226: remove_dot_segments(const uint8_t* path, size_t len, size_t* up)
cannam@226: {
cannam@226: 	const uint8_t*       begin = path;
cannam@226: 	const uint8_t* const end   = path + len;
cannam@226: 
cannam@226: 	*up = 0;
cannam@226: 	while (begin < end) {
cannam@226: 		switch (begin[0]) {
cannam@226: 		case '.':
cannam@226: 			switch (begin[1]) {
cannam@226: 			case '/':
cannam@226: 				begin += 2;  // Chop leading "./"
cannam@226: 				break;
cannam@226: 			case '.':
cannam@226: 				switch (begin[2]) {
cannam@226: 				case '\0':
cannam@226: 					++*up;
cannam@226: 					begin += 2;  // Chop input ".."
cannam@226: 					break;
cannam@226: 				case '/':
cannam@226: 					++*up;
cannam@226: 					begin += 3;  // Chop leading "../"
cannam@226: 					break;
cannam@226: 				default:
cannam@226: 					return begin;
cannam@226: 				}
cannam@226: 				break;
cannam@226: 			case '\0':
cannam@226: 				++begin;  // Chop input "." (and fall-through)
cannam@226: 			default:
cannam@226: 				return begin;
cannam@226: 			}
cannam@226: 			break;
cannam@226: 		case '/':
cannam@226: 			switch (begin[1]) {
cannam@226: 			case '.':
cannam@226: 				switch (begin[2]) {
cannam@226: 				case '/':
cannam@226: 					begin += 2;  // Leading "/./" => "/"
cannam@226: 					break;
cannam@226: 				case '.':
cannam@226: 					switch (begin[3]) {
cannam@226: 					case '/':
cannam@226: 						++*up;
cannam@226: 						begin += 3;  // Leading "/../" => "/"
cannam@226: 					}
cannam@226: 					break;
cannam@226: 				default:
cannam@226: 					return begin;
cannam@226: 				}
cannam@226: 			}  // else fall through
cannam@226: 		default:
cannam@226: 			return begin;  // Finished chopping dot components
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	return begin;
cannam@226: }
cannam@226: 
cannam@226: /// Merge `base` and `path` in-place
cannam@226: static void
cannam@226: merge(SerdChunk* base, SerdChunk* path)
cannam@226: {
cannam@226: 	size_t         up;
cannam@226: 	const uint8_t* begin = remove_dot_segments(path->buf, path->len, &up);
cannam@226: 	const uint8_t* end   = path->buf + path->len;
cannam@226: 
cannam@226: 	if (base->len) {
cannam@226: 		// Find the up'th last slash
cannam@226: 		const uint8_t* base_last = (base->buf + base->len - 1);
cannam@226: 		++up;
cannam@226: 		do {
cannam@226: 			if (*base_last == '/') {
cannam@226: 				--up;
cannam@226: 			}
cannam@226: 		} while (up > 0 && (--base_last > base->buf));
cannam@226: 
cannam@226: 		// Set path prefix
cannam@226: 		base->len = base_last - base->buf + 1;
cannam@226: 	}
cannam@226: 
cannam@226: 	// Set path suffix
cannam@226: 	path->buf = begin;
cannam@226: 	path->len = end - begin;
cannam@226: }
cannam@226: 
cannam@226: /// See http://tools.ietf.org/html/rfc3986#section-5.2.2
cannam@226: SERD_API
cannam@226: void
cannam@226: serd_uri_resolve(const SerdURI* r, const SerdURI* base, SerdURI* t)
cannam@226: {
cannam@226: 	if (!base->scheme.len) {
cannam@226: 		*t = *r;  // Don't resolve against non-absolute URIs
cannam@226: 		return;
cannam@226: 	}
cannam@226: 
cannam@226: 	t->path_base.buf = NULL;
cannam@226: 	t->path_base.len = 0;
cannam@226: 	if (r->scheme.len) {
cannam@226: 		*t = *r;
cannam@226: 	} else {
cannam@226: 		if (r->authority.len) {
cannam@226: 			t->authority = r->authority;
cannam@226: 			t->path      = r->path;
cannam@226: 			t->query     = r->query;
cannam@226: 		} else {
cannam@226: 			t->path = r->path;
cannam@226: 			if (!r->path.len) {
cannam@226: 				t->path_base = base->path;
cannam@226: 				if (r->query.len) {
cannam@226: 					t->query = r->query;
cannam@226: 				} else {
cannam@226: 					t->query = base->query;
cannam@226: 				}
cannam@226: 			} else {
cannam@226: 				if (r->path.buf[0] != '/') {
cannam@226: 					t->path_base = base->path;
cannam@226: 				}
cannam@226: 				merge(&t->path_base, &t->path);
cannam@226: 				t->query = r->query;
cannam@226: 			}
cannam@226: 			t->authority = base->authority;
cannam@226: 		}
cannam@226: 		t->scheme   = base->scheme;
cannam@226: 		t->fragment = r->fragment;
cannam@226: 	}
cannam@226: 
cannam@226: 	#ifdef URI_DEBUG
cannam@226: 	fprintf(stderr, "## RESOLVE URI\n# BASE\n");
cannam@226: 	serd_uri_dump(base, stderr);
cannam@226: 	fprintf(stderr, "# URI\n");
cannam@226: 	serd_uri_dump(r, stderr);
cannam@226: 	fprintf(stderr, "# RESULT\n");
cannam@226: 	serd_uri_dump(t, stderr);
cannam@226: 	fprintf(stderr, "\n");
cannam@226: 	#endif
cannam@226: }
cannam@226: 
cannam@226: /** Write the path of `uri` starting at index `i` */
cannam@226: static size_t
cannam@226: write_path_tail(SerdSink sink, void* stream, const SerdURI* uri, size_t i)
cannam@226: {
cannam@226: 	size_t len = 0;
cannam@226: 	if (i < uri->path_base.len) {
cannam@226: 		len += sink(uri->path_base.buf + i, uri->path_base.len - i, stream);
cannam@226: 	}
cannam@226: 	if (uri->path.buf) {
cannam@226: 		if (i < uri->path_base.len) {
cannam@226: 			len += sink(uri->path.buf, uri->path.len, stream);
cannam@226: 		} else {
cannam@226: 			const size_t j = (i - uri->path_base.len);
cannam@226: 			len += sink(uri->path.buf + j, uri->path.len - j, stream);
cannam@226: 		}
cannam@226: 	}
cannam@226: 	return len;
cannam@226: }
cannam@226: 
cannam@226: /** Write the path of `uri` relative to the path of `base`. */
cannam@226: static size_t
cannam@226: write_rel_path(SerdSink       sink,
cannam@226:                void*          stream,
cannam@226:                const SerdURI* uri,
cannam@226:                const SerdURI* base)
cannam@226: {
cannam@226: 	const size_t path_len = uri_path_len(uri);
cannam@226: 	const size_t base_len = uri_path_len(base);
cannam@226: 	const size_t min_len  = (path_len < base_len) ? path_len : base_len;
cannam@226: 
cannam@226: 	// Find the last separator common to both paths
cannam@226: 	size_t last_shared_sep = 0;
cannam@226: 	size_t i               = 0;
cannam@226: 	for (; i < min_len && uri_path_at(uri, i) == uri_path_at(base, i); ++i) {
cannam@226: 		if (uri_path_at(uri, i) == '/') {
cannam@226: 			last_shared_sep = i;
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	if (i == path_len && i == base_len) {  // Paths are identical
cannam@226: 		return 0;
cannam@226: 	} else if (last_shared_sep == 0) {  // No common components
cannam@226: 		return write_path_tail(sink, stream, uri, 0);
cannam@226: 	}
cannam@226: 
cannam@226: 	// Find the number of up references ("..") required
cannam@226: 	size_t up = 0;
cannam@226: 	for (size_t s = last_shared_sep + 1; s < base_len; ++s) {
cannam@226: 		if (uri_path_at(base, s) == '/') {
cannam@226: 			++up;
cannam@226: 		}
cannam@226: 	}
cannam@226: 
cannam@226: 	// Write up references
cannam@226: 	size_t len = 0;
cannam@226: 	for (size_t u = 0; u < up; ++u) {
cannam@226: 		len += sink("../", 3, stream);
cannam@226: 	}
cannam@226: 
cannam@226: 	// Write suffix
cannam@226: 	return len += write_path_tail(sink, stream, uri, last_shared_sep + 1);
cannam@226: }
cannam@226: 
cannam@226: /// See http://tools.ietf.org/html/rfc3986#section-5.3
cannam@226: SERD_API
cannam@226: size_t
cannam@226: serd_uri_serialise_relative(const SerdURI* uri,
cannam@226:                             const SerdURI* base,
cannam@226:                             const SerdURI* root,
cannam@226:                             SerdSink       sink,
cannam@226:                             void*          stream)
cannam@226: {
cannam@226: 	size_t     len      = 0;
cannam@226: 	const bool relative = uri_is_under(uri, root ? root : base);
cannam@226: 	if (relative) {
cannam@226: 		len = write_rel_path(sink, stream, uri, base);
cannam@226: 	}
cannam@226: 	if (!relative || (!len && base->query.buf)) {
cannam@226: 		if (uri->scheme.buf) {
cannam@226: 			len += sink(uri->scheme.buf, uri->scheme.len, stream);
cannam@226: 			len += sink(":", 1, stream);
cannam@226: 		}
cannam@226: 		if (uri->authority.buf) {
cannam@226: 			len += sink("//", 2, stream);
cannam@226: 			len += sink(uri->authority.buf, uri->authority.len, stream);
cannam@226: 		}
cannam@226: 		len += write_path_tail(sink, stream, uri, 0);
cannam@226: 	}
cannam@226: 	if (uri->query.buf) {
cannam@226: 		len += sink("?", 1, stream);
cannam@226: 		len += sink(uri->query.buf, uri->query.len, stream);
cannam@226: 	}
cannam@226: 	if (uri->fragment.buf) {
cannam@226: 		// Note uri->fragment.buf includes the leading `#'
cannam@226: 		len += sink(uri->fragment.buf, uri->fragment.len, stream);
cannam@226: 	}
cannam@226: 	return len;
cannam@226: }
cannam@226: 
cannam@226: /// See http://tools.ietf.org/html/rfc3986#section-5.3
cannam@226: SERD_API
cannam@226: size_t
cannam@226: serd_uri_serialise(const SerdURI* uri, SerdSink sink, void* stream)
cannam@226: {
cannam@226: 	return serd_uri_serialise_relative(uri, NULL, NULL, sink, stream);
cannam@226: }