Merge remote-tracking branch 'refs/remotes/origin/pr/973'

CHANGELOG.md

@@ -4,6 +4,8 @@
 
 - [Issue #1068](https://github.com/influxdb/influxdb/issues/1068). Print
   the processor chain when the query is started
+- [Issue #973](https://github.com/influxdb/influxdb/issues/973). Support joining
+  using a regex or list of time series
 
 ### Bugfixes
 

coordinator/coordinator.go

@@ -242,6 +242,9 @@ func (self *Coordinator) getShardsAndProcessor(querySpec *parser.QuerySpec, writ
 			shardIds[i] = s.Id()
 		}
 		writer, err = engine.NewQueryEngine(writer, q, shardIds)
+		if err != nil {
+			log.Info("Coordinator processor chain: %s", engine.ProcessorChain(writer))
+		}
 		return shards, writer, err
 	}
 

engine/common_merge_engine.go

@@ -1,6 +1,9 @@
 package engine
 
-import "github.com/influxdb/influxdb/protocol"
+import (
+	"code.google.com/p/log4go"
+	"github.com/influxdb/influxdb/protocol"
+)
 
 type CommonMergeEngine struct {
 	merger  *Merger
@@ -39,6 +42,7 @@ func (cme *CommonMergeEngine) Close() error {
 }
 
 func (cme *CommonMergeEngine) Yield(s *protocol.Series) (bool, error) {
+	log4go.Fine("CommonMergeEngine.Yield(): %s", s)
 	stream := cme.streams[s.GetShardId()]
 	stream.Yield(s)
 	return cme.merger.Update()

engine/join_engine.go

@@ -1,31 +1,44 @@
 package engine
 
 import (
+	"code.google.com/p/log4go"
 	"github.com/influxdb/influxdb/parser"
 	"github.com/influxdb/influxdb/protocol"
 )
 
+// TODO: Explain how JoinEngine work
 type JoinEngine struct {
-	query                    *parser.SelectQuery
-	next                     Processor
-	table1, table2           string
-	name                     string // the output table name
-	lastPoint1, lastPoint2   *protocol.Point
-	lastFields1, lastFields2 []string
+	query       *parser.SelectQuery
+	next        Processor
+	name        string // the output table name
+	tableIdx    map[string]int
+	tablesState []joinEngineState
+	pts         int
 }
 
+// Create and return a new JoinEngine given the shards that will be
+// processed and the query
 func NewJoinEngine(shards []uint32, query *parser.SelectQuery, next Processor) Processor {
-	table1 := query.GetFromClause().Names[0].GetAlias()
-	table2 := query.GetFromClause().Names[1].GetAlias()
-	name := table1 + "_join_" + table2
+	tableNames := query.GetFromClause().Names
+	name := query.GetFromClause().GetString()
+	log4go.Debug("NewJoinEngine: shards=%v, query=%s, next=%s, tableNames=%v, name=%s",
+		shards, query.GetQueryString(), next.Name(), tableNames, name)
 
 	joinEngine := &JoinEngine{
-		next:   next,
-		name:   name,
-		table1: table1,
-		table2: table2,
-		query:  query,
+		next:        next,
+		name:        name,
+		tablesState: make([]joinEngineState, len(tableNames)),
+		tableIdx:    make(map[string]int, len(tableNames)),
+		query:       query,
+		pts:         0,
 	}
+
+	for i, tn := range tableNames {
+		alias := tn.GetAlias()
+		joinEngine.tablesState[i] = joinEngineState{}
+		joinEngine.tableIdx[alias] = i
+	}
+
 	mergeEngine := NewCommonMergeEngine(shards, false, query.Ascending, joinEngine)
 	return mergeEngine
 }
@@ -39,42 +52,49 @@ func (je *JoinEngine) Close() error {
 }
 
 func (je *JoinEngine) Yield(s *protocol.Series) (bool, error) {
-	if *s.Name == je.table1 {
-		je.lastPoint1 = s.Points[len(s.Points)-1]
-		if je.lastFields1 == nil {
-			for _, f := range s.Fields {
-				je.lastFields1 = append(je.lastFields1, je.table1+"."+f)
-			}
+	log4go.Fine("JoinEngine.Yield(): %s", s)
+	idx := je.tableIdx[s.GetName()]
+	state := &je.tablesState[idx]
+	// If the state for this table didn't contain a point already,
+	// increment the number of tables ready to emit a point by
+	// incrementing `pts`
+	if state.lastPoint == nil {
+		je.pts++
+	}
+	state.lastPoint = s.Points[len(s.Points)-1]
+	// update the fields for this table. the fields shouldn't change
+	// after the first point, so we only need to set them once
+	if state.lastFields == nil {
+		for _, f := range s.Fields {
+			state.lastFields = append(state.lastFields, s.GetName()+"."+f)
 		}
 	}
 
-	if *s.Name == je.table2 {
-		je.lastPoint2 = s.Points[len(s.Points)-1]
-		if je.lastFields2 == nil {
-			for _, f := range s.Fields {
-				je.lastFields2 = append(je.lastFields2, je.table2+"."+f)
-			}
-		}
-	}
-
-	if je.lastPoint1 == nil || je.lastPoint2 == nil {
+	log4go.Fine("JoinEngine: pts = %d", je.pts)
+	// if the number of tables ready to emit a point isn't equal to the
+	// total number of tables being joined, then return
+	if je.pts != len(je.tablesState) {
 		return true, nil
 	}
 
+	// we arbitrarily use the timestamp of the first table's point as
+	// the timestamp of the resulting point. may be we should use the
+	// smalles (or largest) timestamp.
+	ts := je.tablesState[0].lastPoint.Timestamp
 	newSeries := &protocol.Series{
 		Name:   &je.name,
-		Fields: append(je.lastFields1, je.lastFields2...),
+		Fields: je.fields(),
 		Points: []*protocol.Point{
 			{
-				Values:    append(je.lastPoint1.Values, je.lastPoint2.Values...),
-				Timestamp: je.lastPoint2.Timestamp,
+				Timestamp: ts,
+				Values:    je.values(),
 			},
 		},
 	}
 
-	je.lastPoint1 = nil
-	je.lastPoint2 = nil
-
+	// filter the point. the user may have a where clause with the join,
+	// e.g. `select * from join(foo1, foo2) where foo1.val > 10`. we
+	// can't evaluate the where clause until after join happens
 	filteredSeries, err := Filter(je.query, newSeries)
 	if err != nil {
 		return false, err
@@ -89,3 +109,33 @@ func (je *JoinEngine) Yield(s *protocol.Series) (bool, error) {
 func (self *JoinEngine) Next() Processor {
 	return self.next
 }
+
+// private
+
+type joinEngineState struct {
+	lastFields []string
+	lastPoint  *protocol.Point
+}
+
+// Returns the field names from all tables appended together
+func (je *JoinEngine) fields() []string {
+	fs := []string{}
+	for _, s := range je.tablesState {
+		fs = append(fs, s.lastFields...)
+	}
+	return fs
+}
+
+// Returns the field values from all tables appended together
+func (je *JoinEngine) values() []*protocol.FieldValue {
+	vs := make([]*protocol.FieldValue, 0)
+	for i := range je.tablesState {
+		// Take the address of the slice element, since we set lastPoint
+		// to nil
+		s := &je.tablesState[i]
+		vs = append(vs, s.lastPoint.Values...)
+		s.lastPoint = nil
+	}
+	je.pts = 0
+	return vs
+}

integration/data_test.go

@@ -549,6 +549,41 @@ func (self *DataTestSuite) TestWhereAndArithmetic(c *C) {
 	c.Assert(maps[0]["expr0"], Equals, 2.0)
 }
 
+// issue #524
+func (self *DataTestSuite) TestJoinRegex(c *C) {
+	t1 := time.Now().Truncate(time.Hour).Add(-4 * time.Hour)
+	t2 := t1.Add(time.Hour)
+	data := fmt.Sprintf(`[
+{
+  "name":"foo1",
+  "columns":["time", "val"],
+  "points":[[%d, 1],[%d, 2]]
+},
+{
+  "name":"foo2",
+  "columns":["time", "val"],
+  "points":[[%d, 3],[%d, 4]]
+
+},
+{
+  "name":"foo3",
+  "columns":["time", "val"],
+  "points":[[%d, 5],[%d, 6]]
+
+}]`, t1.Unix(), t2.Unix(), t1.Unix(), t2.Unix(), t1.Unix(), t2.Unix())
+	self.client.WriteJsonData(data, c, "s")
+	serieses := self.client.RunQuery("select * from join(/foo\\d+/)", c, "m")
+	c.Assert(serieses, HasLen, 1)
+	maps := ToMap(serieses[0])
+	c.Assert(maps, HasLen, 2)
+	c.Assert(maps[0]["foo1.val"], Equals, 2.0)
+	c.Assert(maps[0]["foo2.val"], Equals, 4.0)
+	c.Assert(maps[0]["foo3.val"], Equals, 6.0)
+	c.Assert(maps[1]["foo1.val"], Equals, 1.0)
+	c.Assert(maps[1]["foo2.val"], Equals, 3.0)
+	c.Assert(maps[1]["foo3.val"], Equals, 5.0)
+}
+
 // issue #524
 func (self *DataTestSuite) TestJoinAndArithmetic(c *C) {
 	t1 := time.Now().Truncate(time.Hour).Add(-4 * time.Hour)

parser/from_clause.go

@@ -15,8 +15,9 @@ type FromClauseType int
 const (
 	FromClauseArray      FromClauseType = C.FROM_ARRAY
 	FromClauseMerge      FromClauseType = C.FROM_MERGE
-	FromClauseInnerJoin  FromClauseType = C.FROM_INNER_JOIN
+	FromClauseInnerJoin  FromClauseType = C.FROM_JOIN
 	FromClauseMergeRegex FromClauseType = C.FROM_MERGE_REGEX
+	FromClauseJoinRegex  FromClauseType = C.FROM_JOIN_REGEX
 )
 
 func (self *TableName) GetAlias() string {

parser/parser.go

@@ -473,7 +473,7 @@ func GetFromClause(fromClause *C.from_clause) (*FromClause, error) {
 	var regex *regexp.Regexp
 
 	switch t {
-	case FromClauseMergeRegex:
+	case FromClauseMergeRegex, FromClauseJoinRegex:
 		val, err := GetValue(fromClause.regex_value)
 		if err != nil {
 			return nil, err

parser/query.yacc

@@ -450,9 +450,22 @@ FROM_CLAUSE:
           $$->names->elems[1] = malloc(sizeof(table_name));
           $$->names->elems[1]->name = $6;
           $$->names->elems[1]->alias = $7;
-          $$->from_clause_type = FROM_INNER_JOIN;
+          $$->from_clause_type = FROM_JOIN;
+        }
+        |
+        FROM JOIN '(' SIMPLE_TABLE_VALUES ')'
+        {
+          $$ = calloc(1, sizeof(from_clause));
+          $$->names = $4;
+          $$->from_clause_type = FROM_JOIN;
+        }
+        |
+        FROM JOIN '(' REGEX_VALUE ')'
+        {
+          $$ = calloc(1, sizeof(from_clause));
+          $$->from_clause_type = FROM_JOIN_REGEX;
+          $$->regex_value = $4;
         }
-
 
 WHERE_CLAUSE:
         WHERE CONDITION

parser/query_types.h

@@ -72,8 +72,9 @@ typedef struct {
   enum {
     FROM_ARRAY,
     FROM_MERGE,
-    FROM_INNER_JOIN,
-    FROM_MERGE_REGEX
+    FROM_JOIN,
+    FROM_MERGE_REGEX,
+    FROM_JOIN_REGEX,
   } from_clause_type;
   // in case of merge or join, it's guaranteed that the names array
   // will have two table names only and they aren't regex.

parser/rewrite.go

@@ -12,13 +12,18 @@ type RegexMatcher func(r *regexp.Regexp) []string
 // the query will be rewritten to
 //     select * from merge(foobar, foobaz)
 func RewriteMergeQuery(query *SelectQuery, rm RegexMatcher) {
-	if query.FromClause.Type != FromClauseMergeRegex {
+	resultFromClauseType := FromClauseMerge
+	switch query.FromClause.Type {
+	case FromClauseMergeRegex:
+	case FromClauseJoinRegex:
+		resultFromClauseType = FromClauseInnerJoin
+	default:
 		return
 	}
 
 	series := rm(query.FromClause.Regex)
 	f := query.FromClause
-	f.Type = FromClauseMerge
+	f.Type = resultFromClauseType
 	f.Regex = nil
 	for _, s := range series {
 		f.Names = append(f.Names, &TableName{