gibheer
/
bleve
0
0
Fork 0
Code

multi-term searches check DisjunctionMaxClauseCount earlier 

regexp, fuzzy and numeric range searchers now check to see if
they will be exceeding a configured DisjunctionMaxClauseCount
and stop work earlier, this does a better job of avoiding
situations which consume all available memory for an operation
they cannot complete
This commit is contained in:
Marty Schoch 2016-04-18 10:06:34 -04:00
parent c7ae842b33
commit 53f7eb2891
4 changed files with 83 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
search/searchers/search_disjunction.go
View File

@ -35,9 +35,20 @@ type DisjunctionSearcher struct {
min float64
}
func tooManyClauses(count int) bool {
if DisjunctionMaxClauseCount != 0 && count > DisjunctionMaxClauseCount {
return true
}
return false
}
func tooManyClausesErr() error {
return fmt.Errorf("TooManyClauses[maxClauseCount is set to %d]", DisjunctionMaxClauseCount)
}
func NewDisjunctionSearcher(indexReader index.IndexReader, qsearchers []search.Searcher, min float64, explain bool) (*DisjunctionSearcher, error) {
if DisjunctionMaxClauseCount != 0 && len(qsearchers) > DisjunctionMaxClauseCount {
return nil, fmt.Errorf("TooManyClauses[maxClauseCount is set to %d]", DisjunctionMaxClauseCount)
if tooManyClauses(len(qsearchers)) {
return nil, tooManyClausesErr()
}
// build the downstream searchers
searchers := make(OrderedSearcherList, len(qsearchers))

58
search/searchers/search_fuzzy.go
View File

@ -32,36 +32,13 @@ func NewFuzzySearcher(indexReader index.IndexReader, term string, prefix, fuzzin
}
}
// find the terms with this prefix
var fieldDict index.FieldDict
var err error
if len(prefixTerm) > 0 {
fieldDict, err = indexReader.FieldDictPrefix(field, []byte(prefixTerm))
} else {
fieldDict, err = indexReader.FieldDict(field)
}
// enumerate terms and check levenshtein distance
candidateTerms := make([]string, 0)
tfd, err := fieldDict.Next()
for err == nil && tfd != nil {
ld, exceeded := search.LevenshteinDistanceMax(&term, &tfd.Term, fuzziness)
if !exceeded && ld <= fuzziness {
candidateTerms = append(candidateTerms, tfd.Term)
}
tfd, err = fieldDict.Next()
}
if err != nil {
return nil, err
}
err = fieldDict.Close()
candidateTerms, err := findFuzzyCandidateTerms(indexReader, &term, fuzziness, field, prefixTerm)
if err != nil {
return nil, err
}
// enumerate all the terms in the range
qsearchers := make([]search.Searcher, 0, 25)
qsearchers := make([]search.Searcher, 0, len(candidateTerms))
for _, cterm := range candidateTerms {
qsearcher, err := NewTermSearcher(indexReader, cterm, field, boost, explain)
@ -87,6 +64,37 @@ func NewFuzzySearcher(indexReader index.IndexReader, term string, prefix, fuzzin
searcher: searcher,
}, nil
}
func findFuzzyCandidateTerms(indexReader index.IndexReader, term *string, fuzziness int, field, prefixTerm string) (rv []string, err error) {
rv = make([]string, 0)
var fieldDict index.FieldDict
if len(prefixTerm) > 0 {
fieldDict, err = indexReader.FieldDictPrefix(field, []byte(prefixTerm))
} else {
fieldDict, err = indexReader.FieldDict(field)
}
defer func() {
if cerr := fieldDict.Close(); cerr != nil && err == nil {
err = cerr
}
}()
// enumerate terms and check levenshtein distance
tfd, err := fieldDict.Next()
for err == nil && tfd != nil {
ld, exceeded := search.LevenshteinDistanceMax(term, &tfd.Term, fuzziness)
if !exceeded && ld <= fuzziness {
rv = append(rv, tfd.Term)
if tooManyClauses(len(rv)) {
return rv, tooManyClausesErr()
}
}
tfd, err = fieldDict.Next()
}
return rv, err
}
func (s *FuzzySearcher) Count() uint64 {
return s.searcher.Count()
}

3
search/searchers/search_numeric_range.go
View File

@ -57,6 +57,9 @@ func NewNumericRangeSearcher(indexReader index.IndexReader, min *float64, max *f
// FIXME hard-coded precision, should match field declaration
termRanges := splitInt64Range(minInt64, maxInt64, 4)
terms := termRanges.Enumerate()
if tooManyClauses(len(terms)) {
return nil, tooManyClausesErr()
}
// enumerate all the terms in the range
qsearchers := make([]search.Searcher, len(terms))
for i, term := range terms {

57
search/searchers/search_regexp.go
View File

@ -27,39 +27,20 @@ type RegexpSearcher struct {
func NewRegexpSearcher(indexReader index.IndexReader, pattern *regexp.Regexp, field string, boost float64, explain bool) (*RegexpSearcher, error) {
prefixTerm, complete := pattern.LiteralPrefix()
candidateTerms := make([]string, 0)
var candidateTerms []string
if complete {
// there is no pattern
candidateTerms = append(candidateTerms, prefixTerm)
candidateTerms = []string{prefixTerm}
} else {
var fieldDict index.FieldDict
var err error
if len(prefixTerm) > 0 {
fieldDict, err = indexReader.FieldDictPrefix(field, []byte(prefixTerm))
} else {
fieldDict, err = indexReader.FieldDict(field)
}
// enumerate the terms and check against regexp
tfd, err := fieldDict.Next()
for err == nil && tfd != nil {
if pattern.MatchString(tfd.Term) {
candidateTerms = append(candidateTerms, tfd.Term)
}
tfd, err = fieldDict.Next()
}
if err != nil {
return nil, err
}
err = fieldDict.Close()
candidateTerms, err = findRegexpCandidateTerms(indexReader, pattern, field, prefixTerm)
if err != nil {
return nil, err
}
}
// enumerate all the terms in the range
qsearchers := make([]search.Searcher, 0, 25)
qsearchers := make([]search.Searcher, 0, len(candidateTerms))
for _, cterm := range candidateTerms {
qsearcher, err := NewTermSearcher(indexReader, cterm, field, boost, explain)
@ -83,6 +64,36 @@ func NewRegexpSearcher(indexReader index.IndexReader, pattern *regexp.Regexp, fi
searcher: searcher,
}, nil
}
func findRegexpCandidateTerms(indexReader index.IndexReader, pattern *regexp.Regexp, field, prefixTerm string) (rv []string, err error) {
rv = make([]string, 0)
var fieldDict index.FieldDict
if len(prefixTerm) > 0 {
fieldDict, err = indexReader.FieldDictPrefix(field, []byte(prefixTerm))
} else {
fieldDict, err = indexReader.FieldDict(field)
}
defer func() {
if cerr := fieldDict.Close(); cerr != nil && err == nil {
err = cerr
}
}()
// enumerate the terms and check against regexp
tfd, err := fieldDict.Next()
for err == nil && tfd != nil {
if pattern.MatchString(tfd.Term) {
rv = append(rv, tfd.Term)
if tooManyClauses(len(rv)) {
return rv, tooManyClausesErr()
}
}
tfd, err = fieldDict.Next()
}
return rv, err
}
func (s *RegexpSearcher) Count() uint64 {
return s.searcher.Count()
}