Fix external schema bug in fast_match (#5324)

- The results of fast_match SQL don't include shema name, so we need
add shema name to extSchemaTable for fast_match
- Remove locationStr which is unused.

gpMgmt/bin/gpload.py

@@ -2081,9 +2081,9 @@ class gpload:
     # names and types
     #
     # This function will return the SQL to run in order to find out whether
-    # such a table exists.
+    # such a table exists. The results of this SQl are table names without schema
     #
-    def get_fast_match_exttable_query(self, locationStr, formatType, formatOpts, limitStr, schemaName, log_errors):
+    def get_fast_match_exttable_query(self, formatType, formatOpts, limitStr, schemaName, log_errors):
 
         sqlFormat = """select relname from pg_class
                     join
@@ -2341,7 +2341,7 @@ class gpload:
                 # process the single quotes in order to successfully find an existing external table to reuse.
                 self.formatOpts = self.formatOpts.replace("E'\\''","'\''")
                 if self.fast_match:
-                    sql = self.get_fast_match_exttable_query(locationStr, formatType, self.formatOpts,
+                    sql = self.get_fast_match_exttable_query(formatType, self.formatOpts,
                         limitStr, self.extSchemaName, self.log_errors)
                 else:
                     sql = self.get_reuse_exttable_query(formatType, self.formatOpts,
@@ -2351,7 +2351,11 @@ class gpload:
                 if len(resultList) > 0:
                     # found an external table to reuse. no need to create one. we're done here.
                     self.extTableName = (resultList[0])[0]
-                    self.extSchemaTable = self.extTableName
+                    # fast match result is only table name, so we need add schema info
+                    if self.fast_match:
+                        self.extSchemaTable = self.get_ext_schematable(quote_unident(self.extSchemaName), self.extTableName)
+                    else:
+                        self.extSchemaTable = self.extTableName
                     self.log(self.INFO, "reusing external table %s" % self.extSchemaTable)
                     return
 

gpMgmt/bin/gpload_test/gpload2/TEST.py

@@ -437,7 +437,7 @@ class GPLoad_FormatOpts_TestCase(unittest.TestCase):
 
     def test_00_gpload_formatOpts_setup(self):
         "0  gpload setup"
-        for num in range(1,33):
+        for num in range(1,34):
            f = open(mkpath('query%d.sql' % num),'w')
            f.write("\! gpload -f "+mkpath('config/config_file')+ " -d reuse_gptest\n"+"\! gpload -f "+mkpath('config/config_file')+ " -d reuse_gptest\n")
            f.close()
@@ -683,7 +683,7 @@ class GPLoad_FormatOpts_TestCase(unittest.TestCase):
 
     def test_30_gpload_reuse_table_update_mode_with_fast_match(self):
         "30  gpload update mode with fast match"
-        #drop_tables()
+        drop_tables()
         copy_data('external_file_04.txt','data_file.txt')
         write_config_file(mode='update',reuse_flag='true',fast_match='true',file='data_file.txt')
         self.doTest(30)
@@ -701,6 +701,13 @@ class GPLoad_FormatOpts_TestCase(unittest.TestCase):
         copy_data('external_file_08.txt','data_file.txt')
         write_config_file(mode='update',reuse_flag='false',fast_match='true',file='data_file.txt')
         self.doTest(32)
+    def test_33_gpload_reuse_table_merge_mode_with_fast_match_and_external_schema(self):
+        "33  gpload update mode with fast match and external schema"
+        file = mkpath('setup.sql')
+        runfile(file)
+        copy_data('external_file_04.txt','data_file.txt')
+        write_config_file(mode='merge',reuse_flag='true',fast_match='true',file='data_file.txt',externalSchema='test')
+        self.doTest(33)
 
 if __name__ == '__main__':
     suite = unittest.TestLoader().loadTestsFromTestCase(GPLoad_FormatOpts_TestCase)

gpMgmt/bin/gpload_test/gpload2/query33.ans

+2018-07-24 06:14:29|INFO|gpload session started 2018-07-24 06:14:29
+2018-07-24 06:14:29|INFO|setting schema 'public' for table 'texttable'
+2018-07-24 06:14:29|INFO|started gpfdist -p 8081 -P 8082 -f "/home/gpadmin/workspace/gpdb/gpMgmt/bin/gpload_test/gpload2/data_file.txt" -t 30
+2018-07-24 06:14:29|INFO|did not find a staging table to reuse. creating staging_gpload_reusable_afbaac0da7ced19791c9ab9c537f41d3
+2018-07-24 06:14:29|INFO|did not find an external table to reuse. creating test.ext_gpload_reusable_d2e95f76_8f08_11e8_8c76_0242ac110002
+2018-07-24 06:14:29|INFO|running time: 0.40 seconds
+2018-07-24 06:14:29|INFO|rows Inserted          = 16
+2018-07-24 06:14:29|INFO|rows Updated           = 0
+2018-07-24 06:14:29|INFO|data formatting errors = 0
+2018-07-24 06:14:29|INFO|gpload succeeded
+2018-07-24 06:14:30|INFO|gpload session started 2018-07-24 06:14:30
+2018-07-24 06:14:30|INFO|setting schema 'public' for table 'texttable'
+2018-07-24 06:14:30|INFO|started gpfdist -p 8081 -P 8082 -f "/home/gpadmin/workspace/gpdb/gpMgmt/bin/gpload_test/gpload2/data_file.txt" -t 30
+2018-07-24 06:14:30|INFO|reusing staging table staging_gpload_reusable_afbaac0da7ced19791c9ab9c537f41d3
+2018-07-24 06:14:30|INFO|reusing external table test.ext_gpload_reusable_d2e95f76_8f08_11e8_8c76_0242ac110002
+2018-07-24 06:14:30|INFO|running time: 0.31 seconds
+2018-07-24 06:14:30|INFO|rows Inserted          = 0
+2018-07-24 06:14:30|INFO|rows Updated           = 16
+2018-07-24 06:14:30|INFO|data formatting errors = 0
+2018-07-24 06:14:30|INFO|gpload succeeded