Added filter for query

cmd/csearch/csearch.go

@@ -11,11 +11,11 @@ import (
 	"os"
 	"runtime/pprof"
 
-	"github.com/google/codesearch/index"
+	"github.com/Evengining/codesearch/index"
 	"github.com/google/codesearch/regexp"
 )
 
-var usageMessage = `usage: csearch [-c] [-f fileregexp] [-h] [-i] [-l] [-n] regexp
+var usageMessage = `usage: csearch [-c] [-f fileregexp] [-h] [-i] [-l] [-n] regexp [filter for regex]
 
 Csearch behaves like grep over all indexed files, searching for regexp,
 an RE2 (nearly PCRE) regular expression.
@@ -66,7 +66,7 @@ func Main() {
 	flag.Parse()
 	args := flag.Args()
 
-	if len(args) != 1 {
+	if len(args) != 2 {
 		usage()
 	}
 
@@ -96,7 +96,16 @@ func Main() {
 			log.Fatal(err)
 		}
 	}
+	filter := "(?m)" + args[1]
+	if *iFlag {
+		filter = "(?i)" + filter
+	}
+	filterRe, err := regexp.Compile(filter)
+	if err != nil {
+		log.Fatal(err)
+	}
 	q := index.RegexpQuery(re.Syntax)
+	filterQuery := index.RegexpQuery(filterRe.Syntax)
 	if *verboseFlag {
 		log.Printf("query: %s\n", q)
 	}
@@ -105,9 +114,9 @@ func Main() {
 	ix.Verbose = *verboseFlag
 	var post []uint32
 	if *bruteFlag {
-		post = ix.PostingQuery(&index.Query{Op: index.QAll})
+		post = ix.PostingQuery(&index.Query{Op: index.QAll}, filterQuery)
 	} else {
-		post = ix.PostingQuery(q)
+		post = ix.PostingQuery(q, filterQuery)
 	}
 	if *verboseFlag {
 		log.Printf("post query identified %d possible files\n", len(post))

index/read.go

@@ -330,11 +330,25 @@ func (ix *Index) postingOr(list []uint32, trigram uint32, restrict []uint32) []u
 	return x
 }
 
-func (ix *Index) PostingQuery(q *Query) []uint32 {
-	return ix.postingQuery(q, nil)
+func (ix *Index) PostingQuery(q *Query, filter *Query) []uint32 {
+	return ix.postingQuery(q, filter, nil)
 }
 
-func (ix *Index) postingQuery(q *Query, restrict []uint32) (ret []uint32) {
+func (ix *Index) filterPostingQuery(q *Query, list []uint32) (ret []uint32) {
+	var finds = ix.postingQuery(q, nil, nil)
+	var findsIds = make([]uint32, 0, len(finds))
+	for _, id := range finds {
+		for _, listId := range list {
+			if id != listId {
+				continue
+			}
+			findsIds = append(findsIds, id)
+		}
+	}
+	return findsIds
+}
+
+func (ix *Index) postingQuery(q *Query, filter *Query, restrict []uint32) (ret []uint32) {
 	var list []uint32
 	switch q.Op {
 	case QNone:
@@ -347,7 +361,9 @@ func (ix *Index) postingQuery(q *Query, restrict []uint32) (ret []uint32) {
 		for i := range list {
 			list[i] = uint32(i)
 		}
-		return list
+		if filter != nil {
+			list = ix.filterPostingQuery(filter, list)
+		}
 	case QAnd:
 		for _, t := range q.Trigram {
 			tri := uint32(t[0])<<16 | uint32(t[1])<<8 | uint32(t[2])
@@ -364,11 +380,14 @@ func (ix *Index) postingQuery(q *Query, restrict []uint32) (ret []uint32) {
 			if list == nil {
 				list = restrict
 			}
-			list = ix.postingQuery(sub, list)
+			list = ix.postingQuery(sub, nil, list)
 			if len(list) == 0 {
 				return nil
 			}
 		}
+		if filter != nil {
+			list = ix.filterPostingQuery(filter, list)
+		}
 	case QOr:
 		for _, t := range q.Trigram {
 			tri := uint32(t[0])<<16 | uint32(t[1])<<8 | uint32(t[2])
@@ -379,9 +398,12 @@ func (ix *Index) postingQuery(q *Query, restrict []uint32) (ret []uint32) {
 			}
 		}
 		for _, sub := range q.Sub {
-			list1 := ix.postingQuery(sub, restrict)
+			list1 := ix.postingQuery(sub, nil, restrict)
 			list = mergeOr(list, list1)
 		}
+		if filter != nil {
+			list = ix.filterPostingQuery(filter, list)
+		}
 	}
 	return list
 }