Merge pull request #2105 from influxdb/fix_tag_neq

Fix tag NEQ
influxdata · Mar 31, 2015 · 3eb0b04 · 3eb0b04
2 parents ef409d6 + a7d9047
commit 3eb0b04
Show file tree

Hide file tree

Showing 3 changed files with 80 additions and 4 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -5,6 +5,8 @@
 - [#2100](https://github.com/influxdb/influxdb/pull/2100): Synchronize access to shard index.
 - [#2131](https://github.com/influxdb/influxdb/pull/2131): Optimize marshalTags().
 - [#2130](https://github.com/influxdb/influxdb/pull/2130): Make fewer calls to marshalTags().
+- [#2105](https://github.com/influxdb/influxdb/pull/2105): Support != for tag values. Fix issue #2097, thanks to @smonkewitz for bug report.
+- [#2105](https://github.com/influxdb/influxdb/pull/2105): Support !~ tags values.
 
 ## v0.9.0-rc17 [2015-03-29]
 

diff --git a/cmd/influxd/server_integration_test.go b/cmd/influxd/server_integration_test.go
@@ -592,16 +592,73 @@ func runTestsData(t *testing.T, testName string, nodes Cluster, database, retent
 		{
 			reset: true,
 			name:  "WHERE tags SELECT single field (EQ tag value1)",
-			write: `{"database" : "%DB%", "retentionPolicy" : "%RP%", "points": [{"name": "cpu", "timestamp": "2015-02-28T01:03:36.703820946Z", "tags": {"host": "server01"}, "fields": {"value": 100}},
-			                                                                     {"name": "cpu", "timestamp": "2010-02-28T01:03:37.703820946Z", "tags": {"host": "server02"}, "fields": {"value": 200}}]}`,
+			write: `{"database" : "%DB%", "retentionPolicy" : "%RP%", "points": [{"name": "cpu", "timestamp": "2015-02-28T01:03:36.703820946Z", "tags": {"host": "server01", "region": "us-west"}, "fields": {"value": 100}},
+			                                                                     {"name": "cpu", "timestamp": "2010-02-28T01:03:37.703820946Z", "tags": {"host": "server02"}, "fields": {"value": 200}},
+			                                                                     {"name": "cpu", "timestamp": "2012-02-28T01:03:38.703820946Z", "tags": {"host": "server03"}, "fields": {"value": 300}}]}`,
 			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host = 'server01'`,
 			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2015-02-28T01:03:36.703820946Z",100]]}]}]}`,
 		},
+		{
+			name:     "WHERE tags SELECT single field (2 EQ tags)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host = 'server01' AND region = 'us-west'`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2015-02-28T01:03:36.703820946Z",100]]}]}]}`,
+		},
+		{
+			name:     "WHERE tags SELECT single field (1 EQ and 1 NEQ tag)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host = 'server01' AND region != 'us-west'`,
+			expected: `{"results":[{}]}`,
+		},
 		{
 			name:     "WHERE tags SELECT single field (EQ tag value2)",
 			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host = 'server02'`,
 			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2010-02-28T01:03:37.703820946Z",200]]}]}]}`,
 		},
+		{
+			name:     "WHERE tags SELECT single field (NEQ tag value1)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host != 'server01'`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2010-02-28T01:03:37.703820946Z",200],["2012-02-28T01:03:38.703820946Z",300]]}]}]}`,
+		},
+		{
+			name:     "WHERE tags SELECT single field (NEQ tag value1 AND NEQ tag value2)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host != 'server01' AND host != 'server02'`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2012-02-28T01:03:38.703820946Z",300]]}]}]}`,
+		},
+		{
+			name:     "WHERE tags SELECT single field (NEQ tag value1 OR NEQ tag value2)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host != 'server01' OR host != 'server02'`, // Yes, this is always true, but that's the point.
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2010-02-28T01:03:37.703820946Z",200],["2012-02-28T01:03:38.703820946Z",300],["2015-02-28T01:03:36.703820946Z",100]]}]}]}`,
+		},
+		{
+			name:     "WHERE tags SELECT single field (NEQ tag value1 AND NEQ tag value2 AND NEQ tag value3)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host != 'server01' AND host != 'server02' AND host != 'server03'`,
+			expected: `{"results":[{}]}`,
+		},
+		{
+			reset: true,
+			name:  "WHERE tags SELECT single field (NEQ tag value1, point without any tags)",
+			write: `{"database" : "%DB%", "retentionPolicy" : "%RP%", "points": [{"name": "cpu", "timestamp": "2015-02-28T01:03:36.703820946Z", "tags": {"host": "server01"}, "fields": {"value": 100}},
+                                                                                 {"name": "cpu", "timestamp": "2012-02-28T01:03:38.703820946Z", "fields": {"value": 200}}]}`,
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host != 'server01'`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2012-02-28T01:03:38.703820946Z",200]]}]}]}`,
+		},
+		{
+			reset: true,
+			name:  "WHERE tags SELECT single field (regex tag no match)",
+			write: `{"database" : "%DB%", "retentionPolicy" : "%RP%", "points": [{"name": "cpu", "timestamp": "2015-02-28T01:03:36.703820946Z", "tags": {"host": "server01"}, "fields": {"value": 100}},
+                                                                                 {"name": "cpu", "timestamp": "2012-02-28T01:03:38.703820946Z", "fields": {"value": 200}}]}`,
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host !~ /server01/`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2012-02-28T01:03:38.703820946Z",200]]}]}]}`,
+		},
+		{
+			name:     "WHERE tags SELECT single field (regex tag match)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host =~ /server01/`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2015-02-28T01:03:36.703820946Z",100]]}]}]}`,
+		},
+		{
+			name:     "WHERE tags SELECT single field (regex tag match)",
+			query:    `SELECT value FROM "%DB%"."%RP%".cpu WHERE host !~ /server[23]/`,
+			expected: `{"results":[{"series":[{"name":"cpu","columns":["time","value"],"values":[["2012-02-28T01:03:38.703820946Z",200],["2015-02-28T01:03:36.703820946Z",100]]}]}]}`,
+		},
 
 		// WHERE fields queries
 		{

diff --git a/database.go b/database.go
@@ -376,19 +376,36 @@ func (m *Measurement) idsForExpr(n *influxql.BinaryExpr) (seriesIDs, bool, influ
 		return nil, true, nil
 	}
 
-	// if we're looking for series with a specific tag value
+	// if we're looking for series with specific tag values
 	if str, ok := value.(*influxql.StringLiteral); ok {
-		return tagVals[str.Val], true, nil
+		var ids seriesIDs
+
+		if n.Op == influxql.EQ {
+			// return series that have a tag of specific value.
+			ids = tagVals[str.Val]
+		} else if n.Op == influxql.NEQ {
+			ids = m.seriesIDs.reject(tagVals[str.Val])
+		}
+		return ids, true, nil
 	}
 
 	// if we're looking for series with tag values that match a regex
 	if re, ok := value.(*influxql.RegexLiteral); ok {
 		var ids seriesIDs
+
+		// The operation is a NEQREGEX, code must start by assuming all match, even
+		// series without any tags.
+		if n.Op == influxql.NEQREGEX {
+			ids = m.seriesIDs
+		}
+
 		for k := range tagVals {
 			match := re.Val.MatchString(k)
 
 			if (match && n.Op == influxql.EQREGEX) || (!match && n.Op == influxql.NEQREGEX) {
 				ids = ids.union(tagVals[k])
+			} else if match && n.Op == influxql.NEQREGEX {
+				ids = ids.reject(tagVals[k])
 			}
 		}
 		return ids, true, nil