diff options
| -rw-r--r-- | src/nxt_conf_validation.c | 44 | ||||
| -rw-r--r-- | src/nxt_http_route.c | 318 | ||||
| -rw-r--r-- | test/test_routing.py | 49 |
3 files changed, 254 insertions, 157 deletions
diff --git a/src/nxt_conf_validation.c b/src/nxt_conf_validation.c index c4f78608..f34712bd 100644 --- a/src/nxt_conf_validation.c +++ b/src/nxt_conf_validation.c @@ -1346,15 +1346,8 @@ static nxt_int_t nxt_conf_vldt_match_pattern(nxt_conf_validation_t *vldt, nxt_conf_value_t *value) { - u_char ch; nxt_str_t pattern; - nxt_uint_t i, first, last; - - enum { - sw_none, - sw_side, - sw_middle - } state; + nxt_uint_t i, first; if (nxt_conf_type(value) != NXT_CONF_STRING) { return nxt_conf_vldt_error(vldt, "The \"match\" patterns for \"host\", " @@ -1368,38 +1361,11 @@ nxt_conf_vldt_match_pattern(nxt_conf_validation_t *vldt, } first = (pattern.start[0] == '!'); - last = pattern.length - 1; - state = sw_none; - - for (i = first; i != pattern.length; i++) { - - ch = pattern.start[i]; - - if (ch != '*') { - continue; - } - - switch (state) { - case sw_none: - state = (i == first) ? sw_side : sw_middle; - break; - - case sw_side: - if (i == last) { - if (last - first != 1) { - break; - } - - return nxt_conf_vldt_error(vldt, "The \"match\" pattern must " - "not contain double \"*\" markers."); - } - - /* Fall through. */ - case sw_middle: - return nxt_conf_vldt_error(vldt, "The \"match\" patterns can " - "either contain \"*\" markers at " - "the sides or only one in the middle."); + for (i = first; i < pattern.length; i++) { + if (pattern.start[i] == '*' && pattern.start[i + 1] == '*') { + return nxt_conf_vldt_error(vldt, "The \"match\" pattern must " + "not contain double \"*\" markers."); } } diff --git a/src/nxt_http_route.c b/src/nxt_http_route.c index a8a6b181..66ab0fcd 100644 --- a/src/nxt_http_route.c +++ b/src/nxt_http_route.c @@ -26,7 +26,6 @@ typedef enum { typedef enum { NXT_HTTP_ROUTE_PATTERN_EXACT = 0, NXT_HTTP_ROUTE_PATTERN_BEGIN, - NXT_HTTP_ROUTE_PATTERN_MIDDLE, NXT_HTTP_ROUTE_PATTERN_END, NXT_HTTP_ROUTE_PATTERN_SUBSTRING, } nxt_http_route_pattern_type_t; @@ -70,13 +69,16 @@ typedef struct { typedef struct { - u_char *start1; - u_char *start2; - uint32_t length1; - uint32_t length2; + u_char *start; + uint32_t length; + nxt_http_route_pattern_type_t type:8; +} nxt_http_route_pattern_slice_t; + + +typedef struct { uint32_t min_length; + nxt_array_t *pattern_slices; - nxt_http_route_pattern_type_t type:8; uint8_t case_sensitive; /* 1 bit */ uint8_t negative; /* 1 bit */ uint8_t any; /* 1 bit */ @@ -209,7 +211,10 @@ static nxt_int_t nxt_http_route_pattern_create(nxt_task_t *task, nxt_mp_t *mp, nxt_http_route_encoding_t encoding); static nxt_int_t nxt_http_route_decode_str(nxt_str_t *str, nxt_http_route_encoding_t encoding); -static u_char *nxt_http_route_pattern_copy(nxt_mp_t *mp, nxt_str_t *test, +static nxt_int_t nxt_http_route_pattern_slice(nxt_array_t *slices, + nxt_str_t *test, + nxt_http_route_pattern_type_t type, + nxt_http_route_encoding_t encoding, nxt_http_route_pattern_case_t pattern_case); static nxt_int_t nxt_http_route_resolve(nxt_task_t *task, @@ -1044,103 +1049,163 @@ nxt_http_route_pattern_create(nxt_task_t *task, nxt_mp_t *mp, nxt_http_route_pattern_case_t pattern_case, nxt_http_route_encoding_t encoding) { - u_char *start; - nxt_str_t test, test2; - nxt_int_t ret; - nxt_uint_t n, length; - nxt_http_route_pattern_type_t type; + u_char c, *p, *end; + nxt_str_t test, tmp; + nxt_int_t ret; + nxt_array_t *slices; + nxt_http_route_pattern_type_t type; + + nxt_http_route_pattern_slice_t *slice; type = NXT_HTTP_ROUTE_PATTERN_EXACT; nxt_conf_get_string(cv, &test); + slices = nxt_array_create(mp, 1, sizeof(nxt_http_route_pattern_slice_t)); + if (nxt_slow_path(slices == NULL)) { + return NXT_ERROR; + } + + pattern->pattern_slices = slices; + pattern->negative = 0; pattern->any = 1; pattern->min_length = 0; - if (test.length != 0) { + if (test.length != 0 && test.start[0] == '!') { + test.start++; + test.length--; - if (test.start[0] == '!') { - test.start++; - test.length--; + pattern->negative = 1; + pattern->any = 0; - pattern->negative = 1; - pattern->any = 0; + if (test.length == 0) { + return NXT_OK; } + } - if (test.length != 0) { - if (test.start[0] == '*') { - test.start++; - test.length--; + if (test.length == 0) { + slice = nxt_array_add(slices); + if (nxt_slow_path(slice == NULL)) { + return NXT_ERROR; + } - if (test.length != 0) { - if (test.start[test.length - 1] == '*') { - test.length--; - type = NXT_HTTP_ROUTE_PATTERN_SUBSTRING; + slice->type = NXT_HTTP_ROUTE_PATTERN_EXACT; + slice->start = NULL; + slice->length = 0; - } else { - type = NXT_HTTP_ROUTE_PATTERN_END; - } + return NXT_OK; + } - } else { - type = NXT_HTTP_ROUTE_PATTERN_BEGIN; - } + if (test.start[0] == '*') { + /* 'type' is no longer 'EXACT', assume 'END'. */ + type = NXT_HTTP_ROUTE_PATTERN_END; + test.start++; + test.length--; + } - } else if (test.start[test.length - 1] == '*') { - test.length--; - type = NXT_HTTP_ROUTE_PATTERN_BEGIN; + if (type == NXT_HTTP_ROUTE_PATTERN_EXACT && test.length != 0) { + tmp.start = test.start; - } else { - length = test.length - 1; + p = nxt_memchr(test.start, '*', test.length); + + if (p == NULL) { + /* No '*' found - EXACT pattern. */ + tmp.length = test.length; + type = NXT_HTTP_ROUTE_PATTERN_EXACT; + + test.start += test.length; + test.length = 0; - for (n = 1; n < length; n++) { - if (test.start[n] != '*') { - continue; - } + } else { + /* '*' found - BEGIN pattern. */ + tmp.length = p - test.start; + type = NXT_HTTP_ROUTE_PATTERN_BEGIN; - test.length = n; + test.start = p + 1; + test.length -= tmp.length + 1; + } - test2.start = &test.start[n + 1]; - test2.length = length - n; + ret = nxt_http_route_pattern_slice(slices, &tmp, type, encoding, + pattern_case); + if (nxt_slow_path(ret != NXT_OK)) { + return ret; + } - ret = nxt_http_route_decode_str(&test2, encoding); - if (nxt_slow_path(ret != NXT_OK)) { - return ret; - } + pattern->min_length += tmp.length; + } - type = NXT_HTTP_ROUTE_PATTERN_MIDDLE; + end = test.start + test.length; - pattern->length2 = test2.length; - pattern->min_length += test2.length; + if (test.length != 0 && end[-1] != '*') { + p = end - 1; - start = nxt_http_route_pattern_copy(mp, &test2, - pattern_case); - if (nxt_slow_path(start == NULL)) { - return NXT_ERROR; - } + while (p != test.start) { + c = *p--; - pattern->start2 = start; - break; - } + if (c == '*') { + p += 2; + break; } + } - ret = nxt_http_route_decode_str(&test, encoding); - if (nxt_slow_path(ret != NXT_OK)) { - return ret; - } + tmp.start = p; + tmp.length = end - p; + + test.length -= tmp.length; + end = p; + + ret = nxt_http_route_pattern_slice(slices, &tmp, + NXT_HTTP_ROUTE_PATTERN_END, + encoding, pattern_case); + if (nxt_slow_path(ret != NXT_OK)) { + return ret; } + + pattern->min_length += tmp.length; } - pattern->type = type; - pattern->min_length += test.length; - pattern->length1 = test.length; + tmp.start = test.start; + tmp.length = 0; - start = nxt_http_route_pattern_copy(mp, &test, pattern_case); - if (nxt_slow_path(start == NULL)) { - return NXT_ERROR; + p = tmp.start; + + while (p != end) { + c = *p++; + + if (c != '*') { + tmp.length++; + continue; + } + + if (tmp.length == 0) { + tmp.start = p; + continue; + } + + ret = nxt_http_route_pattern_slice(slices, &tmp, + NXT_HTTP_ROUTE_PATTERN_SUBSTRING, + encoding, pattern_case); + if (nxt_slow_path(ret != NXT_OK)) { + return ret; + } + + pattern->min_length += tmp.length; + + tmp.start = p; + tmp.length = 0; } - pattern->start1 = start; + if (tmp.length != 0) { + ret = nxt_http_route_pattern_slice(slices, &tmp, + NXT_HTTP_ROUTE_PATTERN_SUBSTRING, + encoding, pattern_case); + if (nxt_slow_path(ret != NXT_OK)) { + return ret; + } + + pattern->min_length += tmp.length; + } return NXT_OK; } @@ -1185,15 +1250,25 @@ nxt_http_route_decode_str(nxt_str_t *str, nxt_http_route_encoding_t encoding) } -static u_char * -nxt_http_route_pattern_copy(nxt_mp_t *mp, nxt_str_t *test, +static nxt_int_t +nxt_http_route_pattern_slice(nxt_array_t *slices, + nxt_str_t *test, + nxt_http_route_pattern_type_t type, + nxt_http_route_encoding_t encoding, nxt_http_route_pattern_case_t pattern_case) { - u_char *start; + u_char *start; + nxt_int_t ret; + nxt_http_route_pattern_slice_t *slice; - start = nxt_mp_nget(mp, test->length); + ret = nxt_http_route_decode_str(test, encoding); + if (nxt_slow_path(ret != NXT_OK)) { + return ret; + } + + start = nxt_mp_nget(slices->mem_pool, test->length); if (nxt_slow_path(start == NULL)) { - return start; + return NXT_ERROR; } switch (pattern_case) { @@ -1211,7 +1286,16 @@ nxt_http_route_pattern_copy(nxt_mp_t *mp, nxt_str_t *test, break; } - return start; + slice = nxt_array_add(slices); + if (nxt_slow_path(slices == NULL)) { + return NXT_ERROR; + } + + slice->type = type; + slice->start = start; + slice->length = test->length; + + return NXT_OK; } @@ -2037,9 +2121,11 @@ nxt_http_route_test_argument(nxt_http_request_t *r, static nxt_int_t nxt_http_route_scheme(nxt_http_request_t *r, nxt_http_route_rule_t *rule) { - nxt_bool_t tls, https; + nxt_bool_t tls, https; + nxt_http_route_pattern_slice_t *pattern_slice; - https = (rule->pattern[0].length1 == nxt_length("https")); + pattern_slice = rule->pattern[0].pattern_slices->elts; + https = (pattern_slice->length == nxt_length("https")); tls = (r->tls != NULL); return (tls == https); @@ -2246,60 +2332,66 @@ static nxt_int_t nxt_http_route_pattern(nxt_http_request_t *r, nxt_http_route_pattern_t *pattern, u_char *start, size_t length) { - u_char *p, *end, *test; - size_t test_length; - nxt_int_t ret; + u_char *p, *end, *test; + size_t test_length; + uint32_t i; + nxt_array_t *pattern_slices; + nxt_http_route_pattern_slice_t *pattern_slice; if (length < pattern->min_length) { return 0; } - test = pattern->start1; - test_length = pattern->length1; + pattern_slices = pattern->pattern_slices; + pattern_slice = pattern_slices->elts; - switch (pattern->type) { + for (i = 0; i < pattern_slices->nelts; i++, pattern_slice++) { + test = pattern_slice->start; + test_length = pattern_slice->length; - case NXT_HTTP_ROUTE_PATTERN_EXACT: - if (length != test_length) { - return 0; - } + switch (pattern_slice->type) { + case NXT_HTTP_ROUTE_PATTERN_EXACT: + return ((length == pattern->min_length) && + nxt_http_route_memcmp(start, test, test_length, + pattern->case_sensitive)); - break; + case NXT_HTTP_ROUTE_PATTERN_BEGIN: + if (nxt_http_route_memcmp(start, test, test_length, + pattern->case_sensitive)) + { + break; + } - case NXT_HTTP_ROUTE_PATTERN_BEGIN: - break; + return 0; - case NXT_HTTP_ROUTE_PATTERN_MIDDLE: - ret = nxt_http_route_memcmp(start, test, test_length, - pattern->case_sensitive); - if (!ret) { - return ret; - } + case NXT_HTTP_ROUTE_PATTERN_END: + p = start + length - test_length; - test = pattern->start2; - test_length = pattern->length2; + if (nxt_http_route_memcmp(p, test, test_length, + pattern->case_sensitive)) + { + break; + } - /* Fall through. */ + return 0; - case NXT_HTTP_ROUTE_PATTERN_END: - start += length - test_length; - break; + case NXT_HTTP_ROUTE_PATTERN_SUBSTRING: + end = start + length; - case NXT_HTTP_ROUTE_PATTERN_SUBSTRING: - end = start + length; + if (pattern->case_sensitive) { + p = nxt_memstrn(start, end, (char *) test, test_length); - if (pattern->case_sensitive) { - p = nxt_memstrn(start, end, (char *) test, test_length); + } else { + p = nxt_memcasestrn(start, end, (char *) test, test_length); + } - } else { - p = nxt_memcasestrn(start, end, (char *) test, test_length); + if (p == NULL) { + return 0; + } } - - return (p != NULL); } - return nxt_http_route_memcmp(start, test, test_length, - pattern->case_sensitive); + return 1; } diff --git a/test/test_routing.py b/test/test_routing.py index 3cf4009c..8a196e88 100644 --- a/test/test_routing.py +++ b/test/test_routing.py @@ -115,10 +115,41 @@ class TestRouting(TestApplicationProto): def test_routes_match_invalid(self): self.route_match_invalid({"method": "**"}) - self.route_match_invalid({"method": "blah**"}) - self.route_match_invalid({"host": "*blah*blah"}) - self.route_match_invalid({"host": "blah*blah*blah"}) - self.route_match_invalid({"host": "blah*blah*"}) + + def test_routes_match_valid(self): + self.route_match({"method": "blah*"}) + self.route_match({"host": "*blah*blah"}) + self.route_match({"host": "blah*blah*blah"}) + self.route_match({"host": "blah*blah*"}) + + def test_routes_match_empty_exact(self): + self.route_match({"uri": ""}) + self.assertEqual(self.get()['status'], 404) + + self.route_match({"uri": "/"}) + self.assertEqual(self.get()['status'], 200) + self.assertEqual(self.get(url='/blah')['status'], 404) + + def test_routes_match_negative(self): + self.route_match({"uri": "!"}) + self.assertEqual(self.get()['status'], 404) + + self.route_match({"uri": "!/"}) + self.assertEqual(self.get()['status'], 404) + self.assertEqual(self.get(url='/blah')['status'], 200) + + self.route_match({"uri": "!*blah"}) + self.assertEqual(self.get()['status'], 200) + self.assertEqual(self.get(url='/bla')['status'], 200) + self.assertEqual(self.get(url='/blah')['status'], 404) + self.assertEqual(self.get(url='/blah1')['status'], 200) + + self.route_match({"uri": "!/blah*1*"}) + self.assertEqual(self.get()['status'], 200) + self.assertEqual(self.get(url='/blah')['status'], 200) + self.assertEqual(self.get(url='/blah1')['status'], 404) + self.assertEqual(self.get(url='/blah12')['status'], 404) + self.assertEqual(self.get(url='/blah2')['status'], 200) def test_routes_match_wildcard_middle(self): self.route_match({"host": "ex*le"}) @@ -181,6 +212,15 @@ class TestRouting(TestApplicationProto): self.assertEqual(self.get(url='/blah')['status'], 200, '/blah') self.assertEqual(self.get(url='/BLAH')['status'], 404, '/BLAH') + def test_routes_match_many_wildcard_substrings_case_sensitive(self): + self.route_match({"uri": "*a*B*c*"}) + + self.assertEqual(self.get(url='/blah-a-B-c-blah')['status'], 200) + self.assertEqual(self.get(url='/a-B-c')['status'], 200) + self.assertEqual(self.get(url='/aBc')['status'], 200) + self.assertEqual(self.get(url='/aBCaBbc')['status'], 200) + self.assertEqual(self.get(url='/ABc')['status'], 404) + def test_routes_pass_encode(self): def check_pass(path, name): self.assertIn( @@ -1362,7 +1402,6 @@ class TestRouting(TestApplicationProto): self.route_match_invalid({"arguments": ["var"]}) self.route_match_invalid({"arguments": [{"var1": {}}]}) self.route_match_invalid({"arguments": {"": "bar"}}) - self.route_match_invalid({"arguments": {"foo": "*ba*r"}}) self.route_match_invalid({"arguments": {"foo": "%"}}) self.route_match_invalid({"arguments": {"foo": "%1G"}}) self.route_match_invalid({"arguments": {"%": "bar"}}) |