Merge pull request #105 from macbre/not_used_tables-add-context

not_used_tables: report rows count, table size and its schema

Author: macbre

README.md

@@ -161,9 +161,19 @@ not_used_indices → table affected: 0002_not_used_indices
   - not_used_index: KEY test_id_idx (test, id)
 
 ------------------------------------------------------------
-not_used_tables → table affected: 0000_the_table
+not_used_tables → table affected: 0020_big_table
 
-✗ "0000_the_table" table was not used by provided queries
+✗ "0020_big_table" table was not used by provided queries
+
+  - schema: CREATE TABLE `0020_big_table` (
+      `id` int(9) NOT NULL AUTO_INCREMENT,
+      `val` int(9) NOT NULL,
+      `text` char(5) NOT NULL,
+      PRIMARY KEY (`id`),
+      KEY `text_idx` (`text`)
+    ) ENGINE=InnoDB AUTO_INCREMENT=100001 DEFAULT CHARSET=utf8
+  - table_size_mb: 5.03125
+  - rows_estimated: 100405
 
 ------------------------------------------------------------
 insert_ignore → table affected: 0070_insert_ignore

indexdigest/linters/linter_0006_not_used_columns_and_tables.py

@@ -3,7 +3,7 @@
 """
 import logging
 
-from collections import defaultdict
+from collections import defaultdict, OrderedDict
 from sql_metadata import get_query_columns, get_query_tables
 
 from indexdigest.database import IndexDigestQueryError
@@ -63,8 +63,17 @@ def check_not_used_tables(database, queries):
 
     # generate reports
     for table in not_used_tables:
+        metadata = database.get_table_metadata(table)
+        context = OrderedDict()
+
+        context['schema'] = database.get_table_schema(table)
+        context['table_size_mb'] = \
+            1. * (metadata['data_size'] + metadata['index_size']) / 1024 / 1024
+        context['rows_estimated'] = database.get_table_rows_estimate(table)
+
         yield LinterEntry(linter_type='not_used_tables', table_name=table,
-                          message='"{}" table was not used by provided queries'.format(table))
+                          message='"{}" table was not used by provided queries'.format(table),
+                          context=context)
 
 
 def check_not_used_columns(database, queries):

indexdigest/test/linters/test_0006_not_used_columns_and_tables.py

@@ -32,6 +32,12 @@ def test_not_used_tables(self):
         self.assertEqual(str(reports[0]), '0006_not_used_tables: "0006_not_used_tables" table was not used by provided queries')
         self.assertEqual(reports[0].table_name, '0006_not_used_tables')
 
+        assert str(reports[0].context['schema']).startswith('CREATE TABLE `0006_not_used_tables` (\n')
+
+        # these are estimates, can't assert a certain value
+        assert reports[0].context['table_size_mb'] > 0.0001
+        assert reports[0].context['rows_estimated'] > 0
+
     def test_get_used_tables_from_queries(self):
         queries = [
             'SELECT /* a comment */ foo FROM `0006_not_used_columns` WHERE id = 1;',

sql/0006-not-used-columns-and-tables.sql

@@ -22,4 +22,7 @@ CREATE TABLE `0006_not_used_tables` (
 	PRIMARY KEY (`id`)
 );
 
-INSERT INTO 0006_not_used_tables VALUES(1, 'foo');
+INSERT INTO 0006_not_used_tables VALUES
+    (1, 'foo'),
+    (2, 'foo'),
+    (3, 'foo');