CDRIVER-2128 ensure regex options are sorted

Ensure that the regex option flags "ilmsux" are stored in sorted order,
even when passed to bson_append_regex.

Author: ajdavis

src/bson/bson-private.h

@@ -86,6 +86,8 @@ typedef struct {
 BSON_STATIC_ASSERT (sizeof (bson_impl_alloc_t) <= 128);
 
 
+#define BSON_REGEX_OPTIONS_SORTED "ilmsux"
+
 BSON_END_DECLS
 
 

src/bson/bson.c

@@ -1512,7 +1512,9 @@ bson_append_regex (bson_t *bson,
 {
    static const uint8_t type = BSON_TYPE_REGEX;
    uint32_t regex_len;
-   uint32_t options_len;
+   bson_string_t *options_sorted;
+   const char *c;
+   bool r;
 
    BSON_ASSERT (bson);
    BSON_ASSERT (key);
@@ -1530,21 +1532,31 @@ bson_append_regex (bson_t *bson,
    }
 
    regex_len = (int) strlen (regex) + 1;
-   options_len = (int) strlen (options) + 1;
+   options_sorted = bson_string_new (NULL);
 
-   return _bson_append (bson,
-                        5,
-                        (1 + key_length + 1 + regex_len + options_len),
-                        1,
-                        &type,
-                        key_length,
-                        key,
-                        1,
-                        &gZero,
-                        regex_len,
-                        regex,
-                        options_len,
-                        options);
+   for (c = BSON_REGEX_OPTIONS_SORTED; *c; c++) {
+      if (strchr (options, *c)) {
+         bson_string_append_c (options_sorted, *c);
+      }
+   }
+
+   r =  _bson_append (bson,
+                      5,
+                      (1 + key_length + 1 + regex_len + options_sorted->len),
+                      1,
+                      &type,
+                      key_length,
+                      key,
+                      1,
+                      &gZero,
+                      regex_len,
+                      regex,
+                      options_sorted->len + 1,
+                      options_sorted->str);
+
+   bson_string_free (options_sorted, true);
+
+   return r;
 }
 
 
@@ -2744,7 +2756,7 @@ _bson_as_json_visit_regex (const bson_iter_t *iter,
    bson_string_append (state->str, "\", \"options\" : \"");
 
    /* sort the options */
-   for (c = "ilmsux"; *c; c++) {
+   for (c = BSON_REGEX_OPTIONS_SORTED; *c; c++) {
       if (strchr (v_options, *c)) {
          bson_string_append_c (state->str, *c);
       }

tests/test-bcon-basic.c

@@ -193,10 +193,9 @@ test_regex (void)
    bson_init (&bcon);
    bson_init (&expected);
 
-   bson_append_regex (&expected, "foo", -1, "^foo|bar$", "i");
-
-   BCON_APPEND (&bcon, "foo", BCON_REGEX ("^foo|bar$", "i"));
-
+   /* option flags are sorted */
+   bson_append_regex (&expected, "foo", -1, "^foo|bar$", "mis");
+   BCON_APPEND (&bcon, "foo", BCON_REGEX ("^foo|bar$", "msi"));
    bson_eq_bson (&bcon, &expected);
 
    bson_destroy (&bcon);

tests/test-bson-corpus.c

@@ -70,8 +70,11 @@ github.com/mongodb/specifications/blob/master/source/bson-corpus/bson-corpus.rst
     * bson_to_canonical_extended_json(dB) = cE
     * bson_to_relaxed_extended_json(dB) = rE (if rE exists)
 
+* for dE input (if it exists):
+  * json_to_bson(dE) = cB (unless lossy)
+
 * for rE input (if it exists):
-    bson_to_relaxed_extended_json( json_to_bson(rE) ) = rE
+    bson_to_relaxed_extended_json(json_to_bson(rE)) = rE
 
  */
 static void
@@ -133,6 +136,11 @@ test_bson_corpus (test_bson_type_t *test)
       ASSERT_OR_PRINT (decode_dE, error);
       ASSERT_CMPJSON (bson_as_extended_json (decode_dE, NULL), test->cE);
 
+      if (!test->lossy) {
+         compare_data (
+            bson_get_data (decode_dE), decode_dE->len, test->cB, test->cB_len);
+      }
+
       bson_destroy (decode_dE);
    }
 

tests/test-json.c

@@ -841,6 +841,33 @@ test_bson_json_read_legacy_regex (void)
 }
 
 
+static void
+test_bson_json_read_regex_options_order (void)
+{
+   bson_t b;
+   bson_error_t error;
+   bool r;
+   const char *pattern;
+   const char *flags;
+
+   r = bson_init_from_json (
+      &b, "{\"a\": {\"$regex\": \"\", \"$options\": \"ism\"}}", -1, &error);
+   ASSERT_OR_PRINT (r, error);
+   BCON_EXTRACT (&b, "a", BCONE_REGEX (pattern, flags));
+   ASSERT_CMPSTR (flags, "ims");
+
+   bson_destroy (&b);
+
+   r = bson_init_from_json (
+      &b, "{\"a\": {\"$regex\": \"\", \"$options\": \"misl\"}}", -1, &error);
+   ASSERT_OR_PRINT (r, error);
+   BCON_EXTRACT (&b, "a", BCONE_REGEX (pattern, flags));
+   ASSERT_CMPSTR (flags, "ilms");
+
+   bson_destroy (&b);
+}
+
+
 static void
 test_bson_json_read_binary (void)
 {
@@ -2396,6 +2423,9 @@ test_json_install (TestSuite *suite)
       suite, "/bson/json/read/dbpointer", test_bson_json_read_dbpointer);
    TestSuite_Add (
       suite, "/bson/json/read/legacy_regex", test_bson_json_read_legacy_regex);
+   TestSuite_Add (suite,
+                  "/bson/json/read/regex_options_order",
+                  test_bson_json_read_regex_options_order);
    TestSuite_Add (suite, "/bson/json/read/binary", test_bson_json_read_binary);
    TestSuite_Add (suite,
                   "/bson/json/read/legacy_binary",