Router: route patterns multi wildcards support.

This commit is contained in:
Axel Duch
2020-07-04 03:24:07 +01:00
parent 9bd6baefae
commit a9a21f6fe4
3 changed files with 280 additions and 183 deletions

View File

@@ -1346,15 +1346,8 @@ static nxt_int_t
nxt_conf_vldt_match_pattern(nxt_conf_validation_t *vldt, nxt_conf_vldt_match_pattern(nxt_conf_validation_t *vldt,
nxt_conf_value_t *value) nxt_conf_value_t *value)
{ {
u_char ch;
nxt_str_t pattern; nxt_str_t pattern;
nxt_uint_t i, first, last; nxt_uint_t i, first;
enum {
sw_none,
sw_side,
sw_middle
} state;
if (nxt_conf_type(value) != NXT_CONF_STRING) { if (nxt_conf_type(value) != NXT_CONF_STRING) {
return nxt_conf_vldt_error(vldt, "The \"match\" patterns for \"host\", " 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] == '!'); first = (pattern.start[0] == '!');
last = pattern.length - 1;
state = sw_none;
for (i = first; i != pattern.length; i++) { for (i = first; i < pattern.length; i++) {
if (pattern.start[i] == '*' && pattern.start[i + 1] == '*') {
ch = pattern.start[i]; return nxt_conf_vldt_error(vldt, "The \"match\" pattern must "
"not contain double \"*\" markers.");
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.");
} }
} }

View File

@@ -26,7 +26,6 @@ typedef enum {
typedef enum { typedef enum {
NXT_HTTP_ROUTE_PATTERN_EXACT = 0, NXT_HTTP_ROUTE_PATTERN_EXACT = 0,
NXT_HTTP_ROUTE_PATTERN_BEGIN, NXT_HTTP_ROUTE_PATTERN_BEGIN,
NXT_HTTP_ROUTE_PATTERN_MIDDLE,
NXT_HTTP_ROUTE_PATTERN_END, NXT_HTTP_ROUTE_PATTERN_END,
NXT_HTTP_ROUTE_PATTERN_SUBSTRING, NXT_HTTP_ROUTE_PATTERN_SUBSTRING,
} nxt_http_route_pattern_type_t; } nxt_http_route_pattern_type_t;
@@ -70,13 +69,16 @@ typedef struct {
typedef struct { typedef struct {
u_char *start1; u_char *start;
u_char *start2; uint32_t length;
uint32_t length1;
uint32_t length2;
uint32_t min_length;
nxt_http_route_pattern_type_t type:8; 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;
uint8_t case_sensitive; /* 1 bit */ uint8_t case_sensitive; /* 1 bit */
uint8_t negative; /* 1 bit */ uint8_t negative; /* 1 bit */
uint8_t any; /* 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); nxt_http_route_encoding_t encoding);
static nxt_int_t nxt_http_route_decode_str(nxt_str_t *str, static nxt_int_t nxt_http_route_decode_str(nxt_str_t *str,
nxt_http_route_encoding_t encoding); 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); nxt_http_route_pattern_case_t pattern_case);
static nxt_int_t nxt_http_route_resolve(nxt_task_t *task, 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_pattern_case_t pattern_case,
nxt_http_route_encoding_t encoding) nxt_http_route_encoding_t encoding)
{ {
u_char *start; u_char c, *p, *end;
nxt_str_t test, test2; nxt_str_t test, tmp;
nxt_int_t ret; nxt_int_t ret;
nxt_uint_t n, length; nxt_array_t *slices;
nxt_http_route_pattern_type_t type; nxt_http_route_pattern_type_t type;
nxt_http_route_pattern_slice_t *slice;
type = NXT_HTTP_ROUTE_PATTERN_EXACT; type = NXT_HTTP_ROUTE_PATTERN_EXACT;
nxt_conf_get_string(cv, &test); 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->negative = 0;
pattern->any = 1; pattern->any = 1;
pattern->min_length = 0; pattern->min_length = 0;
if (test.length != 0) { if (test.length != 0 && test.start[0] == '!') {
test.start++;
test.length--;
if (test.start[0] == '!') { pattern->negative = 1;
test.start++; pattern->any = 0;
test.length--;
pattern->negative = 1; if (test.length == 0) {
pattern->any = 0; return NXT_OK;
}
if (test.length != 0) {
if (test.start[0] == '*') {
test.start++;
test.length--;
if (test.length != 0) {
if (test.start[test.length - 1] == '*') {
test.length--;
type = NXT_HTTP_ROUTE_PATTERN_SUBSTRING;
} else {
type = NXT_HTTP_ROUTE_PATTERN_END;
}
} else {
type = NXT_HTTP_ROUTE_PATTERN_BEGIN;
}
} else if (test.start[test.length - 1] == '*') {
test.length--;
type = NXT_HTTP_ROUTE_PATTERN_BEGIN;
} else {
length = test.length - 1;
for (n = 1; n < length; n++) {
if (test.start[n] != '*') {
continue;
}
test.length = n;
test2.start = &test.start[n + 1];
test2.length = length - n;
ret = nxt_http_route_decode_str(&test2, encoding);
if (nxt_slow_path(ret != NXT_OK)) {
return ret;
}
type = NXT_HTTP_ROUTE_PATTERN_MIDDLE;
pattern->length2 = test2.length;
pattern->min_length += test2.length;
start = nxt_http_route_pattern_copy(mp, &test2,
pattern_case);
if (nxt_slow_path(start == NULL)) {
return NXT_ERROR;
}
pattern->start2 = start;
break;
}
}
ret = nxt_http_route_decode_str(&test, encoding);
if (nxt_slow_path(ret != NXT_OK)) {
return ret;
}
} }
} }
pattern->type = type; if (test.length == 0) {
pattern->min_length += test.length; slice = nxt_array_add(slices);
pattern->length1 = test.length; if (nxt_slow_path(slice == NULL)) {
return NXT_ERROR;
}
start = nxt_http_route_pattern_copy(mp, &test, pattern_case); slice->type = NXT_HTTP_ROUTE_PATTERN_EXACT;
if (nxt_slow_path(start == NULL)) { slice->start = NULL;
return NXT_ERROR; slice->length = 0;
return NXT_OK;
} }
pattern->start1 = start; if (test.start[0] == '*') {
/* 'type' is no longer 'EXACT', assume 'END'. */
type = NXT_HTTP_ROUTE_PATTERN_END;
test.start++;
test.length--;
}
if (type == NXT_HTTP_ROUTE_PATTERN_EXACT && test.length != 0) {
tmp.start = test.start;
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;
} else {
/* '*' found - BEGIN pattern. */
tmp.length = p - test.start;
type = NXT_HTTP_ROUTE_PATTERN_BEGIN;
test.start = p + 1;
test.length -= tmp.length + 1;
}
ret = nxt_http_route_pattern_slice(slices, &tmp, type, encoding,
pattern_case);
if (nxt_slow_path(ret != NXT_OK)) {
return ret;
}
pattern->min_length += tmp.length;
}
end = test.start + test.length;
if (test.length != 0 && end[-1] != '*') {
p = end - 1;
while (p != test.start) {
c = *p--;
if (c == '*') {
p += 2;
break;
}
}
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;
}
tmp.start = test.start;
tmp.length = 0;
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;
}
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; 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 * static nxt_int_t
nxt_http_route_pattern_copy(nxt_mp_t *mp, nxt_str_t *test, 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) 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)) { if (nxt_slow_path(start == NULL)) {
return start; return NXT_ERROR;
} }
switch (pattern_case) { switch (pattern_case) {
@@ -1211,7 +1286,16 @@ nxt_http_route_pattern_copy(nxt_mp_t *mp, nxt_str_t *test,
break; 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 static nxt_int_t
nxt_http_route_scheme(nxt_http_request_t *r, nxt_http_route_rule_t *rule) 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); tls = (r->tls != NULL);
return (tls == https); 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, nxt_http_route_pattern(nxt_http_request_t *r, nxt_http_route_pattern_t *pattern,
u_char *start, size_t length) u_char *start, size_t length)
{ {
u_char *p, *end, *test; u_char *p, *end, *test;
size_t test_length; size_t test_length;
nxt_int_t ret; uint32_t i;
nxt_array_t *pattern_slices;
nxt_http_route_pattern_slice_t *pattern_slice;
if (length < pattern->min_length) { if (length < pattern->min_length) {
return 0; return 0;
} }
test = pattern->start1; pattern_slices = pattern->pattern_slices;
test_length = pattern->length1; 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;
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));
case NXT_HTTP_ROUTE_PATTERN_BEGIN:
if (nxt_http_route_memcmp(start, test, test_length,
pattern->case_sensitive))
{
break;
}
case NXT_HTTP_ROUTE_PATTERN_EXACT:
if (length != test_length) {
return 0; return 0;
case NXT_HTTP_ROUTE_PATTERN_END:
p = start + length - test_length;
if (nxt_http_route_memcmp(p, test, test_length,
pattern->case_sensitive))
{
break;
}
return 0;
case NXT_HTTP_ROUTE_PATTERN_SUBSTRING:
end = start + length;
if (pattern->case_sensitive) {
p = nxt_memstrn(start, end, (char *) test, test_length);
} else {
p = nxt_memcasestrn(start, end, (char *) test, test_length);
}
if (p == NULL) {
return 0;
}
} }
break;
case NXT_HTTP_ROUTE_PATTERN_BEGIN:
break;
case NXT_HTTP_ROUTE_PATTERN_MIDDLE:
ret = nxt_http_route_memcmp(start, test, test_length,
pattern->case_sensitive);
if (!ret) {
return ret;
}
test = pattern->start2;
test_length = pattern->length2;
/* Fall through. */
case NXT_HTTP_ROUTE_PATTERN_END:
start += length - test_length;
break;
case NXT_HTTP_ROUTE_PATTERN_SUBSTRING:
end = start + length;
if (pattern->case_sensitive) {
p = nxt_memstrn(start, end, (char *) test, test_length);
} else {
p = nxt_memcasestrn(start, end, (char *) test, test_length);
}
return (p != NULL);
} }
return nxt_http_route_memcmp(start, test, test_length, return 1;
pattern->case_sensitive);
} }

View File

@@ -115,10 +115,41 @@ class TestRouting(TestApplicationProto):
def test_routes_match_invalid(self): def test_routes_match_invalid(self):
self.route_match_invalid({"method": "**"}) self.route_match_invalid({"method": "**"})
self.route_match_invalid({"method": "blah**"})
self.route_match_invalid({"host": "*blah*blah"}) def test_routes_match_valid(self):
self.route_match_invalid({"host": "blah*blah*blah"}) self.route_match({"method": "blah*"})
self.route_match_invalid({"host": "blah*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): def test_routes_match_wildcard_middle(self):
self.route_match({"host": "ex*le"}) 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'], 200, '/blah')
self.assertEqual(self.get(url='/BLAH')['status'], 404, '/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 test_routes_pass_encode(self):
def check_pass(path, name): def check_pass(path, name):
self.assertIn( self.assertIn(
@@ -1362,7 +1402,6 @@ class TestRouting(TestApplicationProto):
self.route_match_invalid({"arguments": ["var"]}) self.route_match_invalid({"arguments": ["var"]})
self.route_match_invalid({"arguments": [{"var1": {}}]}) self.route_match_invalid({"arguments": [{"var1": {}}]})
self.route_match_invalid({"arguments": {"": "bar"}}) 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": "%"}})
self.route_match_invalid({"arguments": {"foo": "%1G"}}) self.route_match_invalid({"arguments": {"foo": "%1G"}})
self.route_match_invalid({"arguments": {"%": "bar"}}) self.route_match_invalid({"arguments": {"%": "bar"}})