Auto check keyword (#1922)

* Supports the detection keyword, name_label supports string

* add tests

* address Shylock's comment

* modify  to add reserved keyword

Co-authored-by: dutor <440396+dutor@users.noreply.github.com>

.linters/cpp/checkKeyword.py

@@ -0,0 +1,139 @@
+# signout--coding:utf-8--
+
+# Copyright (c) 2019 vesoft inc. All rights reserved.
+#
+# This source code is licensed under Apache 2.0 License,
+# attached with Common Clause Condition 1.0, found in the LICENSES directory.
+
+import os
+import sys
+import re
+
+
+PASER_FILE_PATH = 'src/parser/parser.yy'
+SCANNER_FILE_PATH = 'src/parser/scanner.lex'
+
+reserved_key_words = [
+    'KW_GO',
+    'KW_AS',
+    'KW_TO',
+    'KW_OR',
+    'KW_AND',
+    'KW_XOR',
+    'KW_USE',
+    'KW_SET',
+    'KW_FROM',
+    'KW_WHERE',
+    'KW_MATCH',
+    'KW_INSERT',
+    'KW_YIELD',
+    'KW_RETURN',
+    'KW_DESCRIBE',
+    'KW_DESC',
+    'KW_VERTEX',
+    'KW_EDGE',
+    'KW_EDGES',
+    'KW_UPDATE',
+    'KW_UPSERT',
+    'KW_WHEN',
+    'KW_DELETE',
+    'KW_FIND',
+    'KW_LOOKUP',
+    'KW_ALTER',
+    'KW_STEPS',
+    'KW_OVER',
+    'KW_UPTO',
+    'KW_REVERSELY',
+    'KW_INDEX',
+    'KW_INDEXES',
+    'KW_REBUILD',
+    'KW_INT',
+    'KW_BIGINT',
+    'KW_DOUBLE',
+    'KW_STRING',
+    'KW_BOOL',
+    'KW_TIMESTAMP',
+    'KW_TAG',
+    'KW_TAGS',
+    'KW_UNION',
+    'KW_INTERSECT',
+    'KW_MINUS',
+    'KW_NO',
+    'KW_OVERWRITE',
+    'KW_SHOW',
+    'KW_ADD',
+    'KW_CREATE',
+    'KW_DROP',
+    'KW_REMOVE',
+    'KW_IF',
+    'KW_NOT',
+    'KW_EXISTS',
+    'KW_WITH',
+    'KW_CHANGE',
+    'KW_GRANT',
+    'KW_REVOKE',
+    'KW_ON',
+    'KW_BY',
+    'KW_IN',
+    'KW_DOWNLOAD',
+    'KW_GET',
+    'KW_OF',
+    'KW_ORDER',
+    'KW_INGEST',
+    'KW_COMPACT',
+    'KW_FLUSH',
+    'KW_SUBMIT',
+    'KW_ASC',
+    'KW_DISTINCT',
+    'KW_FETCH',
+    'KW_PROP',
+    'KW_BALANCE',
+    'KW_STOP',
+    'KW_LIMIT',
+    'KW_OFFSET',
+    'KW_GROUP',
+    'KW_IS',
+    'KW_NULL',
+    'KW_FORCE',
+    'KW_RECOVER'
+]
+
+
+def get_unreserved_keyword(file_path):
+    parser_file = open(file_path)
+    flag = 0
+    unreserved_key_words = []
+    for line in parser_file.readlines():
+        if line.strip() == 'unreserved_keyword':
+            flag = 1
+            continue
+        if flag == 1:
+            if line.strip() == ';':
+                break
+            unreserved_key_words.append(re.sub('\\s+[:|]\\s+(\\w+)\\s+.*', '\\1', line).strip())
+            continue
+
+    parser_file.close()
+    return unreserved_key_words
+
+
+if __name__ == '__main__':
+    cmd = 'git diff --diff-filter=ACMRTUXB HEAD -p ' + SCANNER_FILE_PATH + '|grep "^+"|grep -v "^+++"|grep "KW_"'
+    content = os.popen(cmd)
+    keywords=[]
+    for line in  content.readlines():
+        keyword = re.sub('.*(KW_\\w+)\s*;.*','\\1',line.strip())
+        keywords.append(keyword)
+
+    if len(keywords) == 0:
+        exit(0)
+    unreserved_key_words = get_unreserved_keyword(PASER_FILE_PATH)
+    new_key_words = [word for word in keywords if word not in reserved_key_words]
+    if len(new_key_words) == 0:
+        exit(0)
+    result = [word for word in new_key_words if word not in unreserved_key_words]
+    if len(result) == 0:
+        exit(0)
+    print('Keywords \"{}\" in src/parser/scanner.lex are not in the unreserved keyword list.'.format(result))
+    print('Please add them to the unreserved keyword in the src/parser/parser.yy.')
+    exit(1)

.linters/cpp/hooks/pre-commit.sh

@@ -6,6 +6,15 @@
 # attached with Common Clause Condition 1.0, found in the LICENSES directory.
 
 CPPLINT=`dirname $0`/../../.linters/cpp/cpplint.py
+CHECKKEYWORD=`dirname $0`/../../.linters/cpp/checkKeyword.py
+
+echo "Performing checkout keyword..."
+python $CHECKKEYWORD
+
+if [ $? -ne 0 ]; then
+    echo "Checkout keyword failed"
+    exit 1
+fi
 
 if [ $# -eq 0 ];then
     # Since cpplint.py could only apply on our working tree,

src/graph/test/SchemaTest.cpp

@@ -1044,6 +1044,22 @@ TEST_F(SchemaTest, TestTagAndEdge) {
         code = client->execute(query, resp);
         ASSERT_EQ(cpp2::ErrorCode::SUCCEEDED, code);
     }
+    // Test tagName is reserved keyword
+    {
+        cpp2::ExecutionResponse resp;
+        std::string query = "USE my_space; CREATE TAG `tag` (`edge` string)";
+        auto code = client->execute(query, resp);
+        ASSERT_EQ(cpp2::ErrorCode::SUCCEEDED, code);
+    }
+    {
+        cpp2::ExecutionResponse resp;
+        std::string query = "DESCRIBE TAG `tag`";
+        client->execute(query, resp);
+        std::vector<uniform_tuple_t<std::string, 2>> expected{
+                {"edge", "string"},
+        };
+        ASSERT_TRUE(verifyResult(resp, expected));
+    }
     {
         cpp2::ExecutionResponse resp;
         std::string query = "SHOW SPACES";

src/parser/parser.yy

@@ -110,7 +110,7 @@ static constexpr size_t MAX_ABS_INTEGER = 9223372036854775808ULL;
 %token KW_IF KW_NOT KW_EXISTS KW_WITH
 %token KW_COUNT KW_COUNT_DISTINCT KW_SUM KW_AVG KW_MAX KW_MIN KW_STD KW_BIT_AND KW_BIT_OR KW_BIT_XOR
 %token KW_BY KW_DOWNLOAD KW_HDFS KW_UUID KW_CONFIGS KW_FORCE KW_STATUS
-%token KW_VARIABLES KW_GET KW_DECLARE KW_GRAPH KW_META KW_STORAGE
+%token KW_GET KW_DECLARE KW_GRAPH KW_META KW_STORAGE
 %token KW_TTL KW_TTL_DURATION KW_TTL_COL KW_DATA KW_STOP
 %token KW_FETCH KW_PROP KW_UPDATE KW_UPSERT KW_WHEN
 %token KW_ORDER KW_ASC KW_LIMIT KW_OFFSET KW_GROUP
@@ -281,23 +281,33 @@ unreserved_keyword
      | KW_DATA               { $$ = new std::string("data"); }
      | KW_LEADER             { $$ = new std::string("leader"); }
      | KW_UUID               { $$ = new std::string("uuid"); }
-     | KW_VARIABLES          { $$ = new std::string("variables"); }
      | KW_JOB                { $$ = new std::string("job"); }
      | KW_JOBS               { $$ = new std::string("jobs"); }
-     | KW_SUBMIT             { $$ = new std::string("submit"); }
-     | KW_RECOVER            { $$ = new std::string("recover"); }
-     | KW_FLUSH              { $$ = new std::string("flush"); }
-     | KW_COMPACT            { $$ = new std::string("compact"); }
      | KW_BIDIRECT           { $$ = new std::string("bidirect"); }
      | KW_OFFLINE            { $$ = new std::string("offline"); }
      | KW_FORCE              { $$ = new std::string("force"); }
      | KW_STATUS             { $$ = new std::string("status"); }
-     | KW_REBUILD            { $$ = new std::string("rebuild"); }
      | KW_PART               { $$ = new std::string("part"); }
      | KW_PARTS              { $$ = new std::string("parts"); }
      | KW_DEFAULT            { $$ = new std::string("default"); }
      | KW_CONFIGS            { $$ = new std::string("configs"); }
      | KW_ACCOUNT            { $$ = new std::string("account"); }
+     | KW_HDFS               { $$ = new std::string("hdfs"); }
+     | KW_PARTITION_NUM      { $$ = new std::string("partition_num"); }
+     | KW_REPLICA_FACTOR     { $$ = new std::string("replica_factor"); }
+     | KW_CHARSET            { $$ = new std::string("charset"); }
+     | KW_COLLATE            { $$ = new std::string("collate"); }
+     | KW_COLLATION          { $$ = new std::string("collation"); }
+     | KW_TTL_DURATION       { $$ = new std::string("ttl_duration"); }
+     | KW_TTL_COL            { $$ = new std::string("ttl_col"); }
+     | KW_SNAPSHOT           { $$ = new std::string("snapshot"); }
+     | KW_SNAPSHOTS          { $$ = new std::string("snapshots"); }
+     | KW_GRAPH              { $$ = new std::string("graph"); }
+     | KW_META               { $$ = new std::string("meta"); }
+     | KW_STORAGE            { $$ = new std::string("storage"); }
+     | KW_ALL                { $$ = new std::string("all"); }
+     | KW_SHORTEST           { $$ = new std::string("shortest"); }
+     | KW_COUNT_DISTINCT     { $$ = new std::string("count_distinct"); }
      ;
 
 agg_function