wasm/glob.match: fix default delimiter handling (#3296)

Also - use opa_value_iter in opa_glob_match - add test cases to `wasm-rego-test` - adapt existing test cases in `wasm-lib-test` Fixes #3294. Signed-off-by: Stephan Renatus <stephan.renatus@gmail.com>
open-policy-agent · Mar 19, 2021 · d9bbaf4 · d9bbaf4
1 parent 134ed9b
commit d9bbaf4
Show file tree

Hide file tree

Showing 6 changed files with 35 additions and 19 deletions.
diff --git a/test/wasm/assets/018_builtins.yaml b/test/wasm/assets/018_builtins.yaml
@@ -210,8 +210,17 @@ cases:
     query: net.cidr_intersects("192.168.1.0/25", "192.168.1.64/25", x)
     want_result: [{'x': true}]
   - note: glob.match built-in
-    query: glob.match("*:github:com", [":"], "api:github:com",x)
+    query: glob.match("*:github:com", [":"], "api:github:com", x)
     want_result: [{'x': true}]
+  - note: glob.match built-in, multiple delimiters
+    query: glob.match("*.github.com:foo", [".", ":"], "api.github.com:foo", x)
+    want_result: [{'x': true}]
+  - note: glob.match built-in delimiters default
+    query: glob.match("*.github.com", [], "api.github.com", x)
+    want_result: [{'x': true}]
+  - note: glob.match built-in delimiters default, negative
+    query: glob.match("*foo*", [], "5.0 foo/90", x)
+    want_result: [{'x': false}]
   - note: json.marshal built-in
     query: json.marshal("string",x)
     want_result: [{'x': '"string"'}]

diff --git a/wasm/src/glob-compiler.cc b/wasm/src/glob-compiler.cc
@@ -128,7 +128,7 @@ std::string glob_translate(const char *glob, size_t n, const std::vector<std::st
 
     if (delimiters.empty())
     {
-        single_mark = ".";
+        single_mark = "[^\\.]";
     } else {
         single_mark = "[^";
 

diff --git a/wasm/src/glob-parser.cc b/wasm/src/glob-parser.cc
@@ -6,7 +6,7 @@
 #include "unicode.h"
 
 // The following is a re-implementation of parser in
-// https://github.com/gobwas/glob/blob.
+// https://github.com/gobwas/glob.
 //
 // The MIT License (MIT)
 //

diff --git a/wasm/src/glob.cc b/wasm/src/glob.cc
@@ -59,20 +59,21 @@ opa_value *opa_glob_match(opa_value *pattern, opa_value *delimiters, opa_value *
     }
 
     opa_string_t *p = opa_cast_string(pattern);
-    opa_array_t *d = opa_cast_array(delimiters);
 
     std::vector<std::string> v;
 
-    for (int i = 0; i < d->len; i++)
+    opa_value *prev = NULL;
+    opa_value *curr = NULL;
+    while ((curr = opa_value_iter(delimiters, prev)) != NULL)
     {
-        if (opa_value_type(d->elems[i].v) != OPA_STRING)
+        opa_value *elem = opa_value_get(delimiters, curr);
+        if (opa_value_type(elem) != OPA_STRING)
         {
             return NULL;
         }
-
-        opa_string_t *s = opa_cast_string(d->elems[i].v);
-        std::string delimiter(s->v, s->len);
-        v.push_back(delimiter);
+        opa_string_t *s = opa_cast_string(elem);
+        v.push_back(std::string(s->v, s->len));
+        prev = curr;
     }
 
     glob_cache *c = cache();

diff --git a/wasm/tests/test-glob.cc b/wasm/tests/test-glob.cc
@@ -235,21 +235,25 @@ void test_glob_translate()
             v.push_back(delimiters[i]);                                 \
         }                                                               \
         glob_translate(pattern, strlen(pattern), v, &re2);              \
-        test(test_case, memcmp(re2.c_str(), expected, strlen(expected)) == 0); \
+        test_str_eq(test_case, expected, re2.c_str());                  \
         re2::RE2::Options options;                                      \
         options.set_log_errors(false);                                  \
         re2::RE2 compiled(std::string(re2.c_str(),strlen(re2.c_str())), options); \
         test(test_case, compiled.ok());                                 \
     }
 
-    TEST("glob/translate", "[a-z][!a-x]*cat*[h][!b]*eyes*", "^[a-z][^a-x].*cat.*[h][^b].*eyes.*$");
-    TEST("glob/translate", "https://*.google.*", "^https\\:\\/\\/.*\\.google\\..*$");
-    TEST("glob/translate", "{https://*.google.*,*yandex.*,*yahoo.*,*mail.ru}", "^(https\\:\\/\\/.*\\.google\\..*|.*yandex\\..*|.*yahoo\\..*|.*mail\\.ru)$");
-    TEST("glob/translate", "{https://*gobwas.com,http://exclude.gobwas.com}", "^(https\\:\\/\\/.*gobwas\\.com|http\\:\\/\\/exclude\\.gobwas\\.com)$");
-    TEST("glob/translate", "abc*", "^abc.*$");
-    TEST("glob/translate", "*def", "^.*def$");
-    TEST("glob/translate", "ab*ef", "^ab.*ef$");
+    TEST("glob/translate", "[a-z][!a-x]*cat*[h][!b]*eyes*", "^[a-z][^a-x][^\\.]*cat[^\\.]*[h][^b][^\\.]*eyes[^\\.]*$");
+    TEST("glob/translate", "https://*.google.*", "^https\\:\\/\\/[^\\.]*\\.google\\.[^\\.]*$");
+    TEST("glob/translate", "https://*.google.*", "^https\\:\\/\\/[^\\.]*\\.google\\.[^\\.]*$", "."); // "." is the default
+    TEST("glob/translate", "{https://*.google.*,*yandex.*,*yahoo.*,*mail.ru}", "^(https\\:\\/\\/[^\\.]*\\.google\\.[^\\.]*|[^\\.]*yandex\\.[^\\.]*|[^\\.]*yahoo\\.[^\\.]*|[^\\.]*mail\\.ru)$");
+    TEST("glob/translate", "{https://*gobwas.com,http://exclude.gobwas.com}", "^(https\\:\\/\\/[^\\.]*gobwas\\.com|http\\:\\/\\/exclude\\.gobwas\\.com)$");
+    TEST("glob/translate", "abc*", "^abc[^\\.]*$");
+    TEST("glob/translate", "*def", "^[^\\.]*def$");
+    TEST("glob/translate", "*def", "^[^\\.]*def$", "."); // "." is the default
+    TEST("glob/translate", "ab*ef", "^ab[^\\.]*ef$");
     TEST("glob/translate", "api.*.com", "^api\\.[^\\.\\,]*\\.com$", ".", ",");
-    TEST("glob/translate", "api.**.com", "^api\\..*\\.com$", ".", ",");
+    TEST("glob/translate", "api.**.com", "^api\\..*\\.com$");
+    TEST("glob/translate", "api.**.com", "^api\\..*\\.com$", "."); // "." is the default...
+    TEST("glob/translate", "api.**.com", "^api\\..*\\.com$", ".", ","); // and "," does not matter here
 #undef TEST
 }
diff --git a/wasm/tests/test.h b/wasm/tests/test.h
@@ -1,6 +1,8 @@
 #ifndef OPA_TEST_H
 #define OPA_TEST_H
 
+#include "str.h"
+
 #ifdef __cplusplus
 extern "C" {
 #endif