githaven/vendor/github.com/olivere/elastic/v7/termvectors.go
Lunny Xiao 5dbf36f356
Issue search support elasticsearch (#9428)
* Issue search support elasticsearch

* Fix lint

* Add indexer name on app.ini

* add a warnning on SearchIssuesByKeyword

* improve code
2020-02-13 14:06:17 +08:00

521 lines
14 KiB
Go
Vendored

// Copyright 2012-present Oliver Eilhard. All rights reserved.
// Use of this source code is governed by a MIT-license.
// See http://olivere.mit-license.org/license.txt for details.
package elastic
import (
"context"
"fmt"
"net/http"
"net/url"
"strings"
"github.com/olivere/elastic/v7/uritemplates"
)
// TermvectorsService returns information and statistics on terms in the
// fields of a particular document. The document could be stored in the
// index or artificially provided by the user.
//
// See https://www.elastic.co/guide/en/elasticsearch/reference/7.0/docs-termvectors.html
// for documentation.
type TermvectorsService struct {
client *Client
pretty *bool // pretty format the returned JSON response
human *bool // return human readable values for statistics
errorTrace *bool // include the stack trace of returned errors
filterPath []string // list of filters used to reduce the response
headers http.Header // custom request-level HTTP headers
id string
index string
typ string
dfs *bool
doc interface{}
fieldStatistics *bool
fields []string
filter *TermvectorsFilterSettings
perFieldAnalyzer map[string]string
offsets *bool
parent string
payloads *bool
positions *bool
preference string
realtime *bool
routing string
termStatistics *bool
version interface{}
versionType string
bodyJson interface{}
bodyString string
}
// NewTermvectorsService creates a new TermvectorsService.
func NewTermvectorsService(client *Client) *TermvectorsService {
return &TermvectorsService{
client: client,
}
}
// Pretty tells Elasticsearch whether to return a formatted JSON response.
func (s *TermvectorsService) Pretty(pretty bool) *TermvectorsService {
s.pretty = &pretty
return s
}
// Human specifies whether human readable values should be returned in
// the JSON response, e.g. "7.5mb".
func (s *TermvectorsService) Human(human bool) *TermvectorsService {
s.human = &human
return s
}
// ErrorTrace specifies whether to include the stack trace of returned errors.
func (s *TermvectorsService) ErrorTrace(errorTrace bool) *TermvectorsService {
s.errorTrace = &errorTrace
return s
}
// FilterPath specifies a list of filters used to reduce the response.
func (s *TermvectorsService) FilterPath(filterPath ...string) *TermvectorsService {
s.filterPath = filterPath
return s
}
// Header adds a header to the request.
func (s *TermvectorsService) Header(name string, value string) *TermvectorsService {
if s.headers == nil {
s.headers = http.Header{}
}
s.headers.Add(name, value)
return s
}
// Headers specifies the headers of the request.
func (s *TermvectorsService) Headers(headers http.Header) *TermvectorsService {
s.headers = headers
return s
}
// Index in which the document resides.
func (s *TermvectorsService) Index(index string) *TermvectorsService {
s.index = index
return s
}
// Type of the document.
//
// Deprecated: Types are in the process of being removed.
func (s *TermvectorsService) Type(typ string) *TermvectorsService {
s.typ = typ
return s
}
// Id of the document.
func (s *TermvectorsService) Id(id string) *TermvectorsService {
s.id = id
return s
}
// Dfs specifies if distributed frequencies should be returned instead
// shard frequencies.
func (s *TermvectorsService) Dfs(dfs bool) *TermvectorsService {
s.dfs = &dfs
return s
}
// Doc is the document to analyze.
func (s *TermvectorsService) Doc(doc interface{}) *TermvectorsService {
s.doc = doc
return s
}
// FieldStatistics specifies if document count, sum of document frequencies
// and sum of total term frequencies should be returned.
func (s *TermvectorsService) FieldStatistics(fieldStatistics bool) *TermvectorsService {
s.fieldStatistics = &fieldStatistics
return s
}
// Fields a list of fields to return.
func (s *TermvectorsService) Fields(fields ...string) *TermvectorsService {
if s.fields == nil {
s.fields = make([]string, 0)
}
s.fields = append(s.fields, fields...)
return s
}
// Filter adds terms filter settings.
func (s *TermvectorsService) Filter(filter *TermvectorsFilterSettings) *TermvectorsService {
s.filter = filter
return s
}
// PerFieldAnalyzer allows to specify a different analyzer than the one
// at the field.
func (s *TermvectorsService) PerFieldAnalyzer(perFieldAnalyzer map[string]string) *TermvectorsService {
s.perFieldAnalyzer = perFieldAnalyzer
return s
}
// Offsets specifies if term offsets should be returned.
func (s *TermvectorsService) Offsets(offsets bool) *TermvectorsService {
s.offsets = &offsets
return s
}
// Parent id of documents.
func (s *TermvectorsService) Parent(parent string) *TermvectorsService {
s.parent = parent
return s
}
// Payloads specifies if term payloads should be returned.
func (s *TermvectorsService) Payloads(payloads bool) *TermvectorsService {
s.payloads = &payloads
return s
}
// Positions specifies if term positions should be returned.
func (s *TermvectorsService) Positions(positions bool) *TermvectorsService {
s.positions = &positions
return s
}
// Preference specify the node or shard the operation
// should be performed on (default: random).
func (s *TermvectorsService) Preference(preference string) *TermvectorsService {
s.preference = preference
return s
}
// Realtime specifies if request is real-time as opposed to
// near-real-time (default: true).
func (s *TermvectorsService) Realtime(realtime bool) *TermvectorsService {
s.realtime = &realtime
return s
}
// Routing is a specific routing value.
func (s *TermvectorsService) Routing(routing string) *TermvectorsService {
s.routing = routing
return s
}
// TermStatistics specifies if total term frequency and document frequency
// should be returned.
func (s *TermvectorsService) TermStatistics(termStatistics bool) *TermvectorsService {
s.termStatistics = &termStatistics
return s
}
// Version an explicit version number for concurrency control.
func (s *TermvectorsService) Version(version interface{}) *TermvectorsService {
s.version = version
return s
}
// VersionType specifies a version type ("internal", "external", or "external_gte").
func (s *TermvectorsService) VersionType(versionType string) *TermvectorsService {
s.versionType = versionType
return s
}
// BodyJson defines the body parameters. See documentation.
func (s *TermvectorsService) BodyJson(body interface{}) *TermvectorsService {
s.bodyJson = body
return s
}
// BodyString defines the body parameters as a string. See documentation.
func (s *TermvectorsService) BodyString(body string) *TermvectorsService {
s.bodyString = body
return s
}
// buildURL builds the URL for the operation.
func (s *TermvectorsService) buildURL() (string, url.Values, error) {
var pathParam = map[string]string{
"index": s.index,
}
path := "/{index}"
var err error
if s.typ != "" {
pathParam["type"] = s.typ
path += "/{type}"
} else {
path += "/_termvectors"
}
if s.id != "" {
pathParam["id"] = s.id
path += "/{id}"
}
if s.typ != "" {
path += "/_termvectors"
}
path, err = uritemplates.Expand(path, pathParam)
if err != nil {
return "", url.Values{}, err
}
// Add query string parameters
params := url.Values{}
if v := s.pretty; v != nil {
params.Set("pretty", fmt.Sprint(*v))
}
if v := s.human; v != nil {
params.Set("human", fmt.Sprint(*v))
}
if v := s.errorTrace; v != nil {
params.Set("error_trace", fmt.Sprint(*v))
}
if len(s.filterPath) > 0 {
params.Set("filter_path", strings.Join(s.filterPath, ","))
}
if v := s.dfs; v != nil {
params.Set("dfs", fmt.Sprint(*v))
}
if v := s.fieldStatistics; v != nil {
params.Set("field_statistics", fmt.Sprint(*v))
}
if len(s.fields) > 0 {
params.Set("fields", strings.Join(s.fields, ","))
}
if v := s.offsets; v != nil {
params.Set("offsets", fmt.Sprint(*v))
}
if s.parent != "" {
params.Set("parent", s.parent)
}
if v := s.payloads; v != nil {
params.Set("payloads", fmt.Sprint(*v))
}
if v := s.positions; v != nil {
params.Set("positions", fmt.Sprint(*v))
}
if s.preference != "" {
params.Set("preference", s.preference)
}
if v := s.realtime; v != nil {
params.Set("realtime", fmt.Sprint(*v))
}
if s.routing != "" {
params.Set("routing", s.routing)
}
if v := s.termStatistics; v != nil {
params.Set("term_statistics", fmt.Sprint(*v))
}
if s.version != nil {
params.Set("version", fmt.Sprintf("%v", s.version))
}
if s.versionType != "" {
params.Set("version_type", s.versionType)
}
return path, params, nil
}
// Validate checks if the operation is valid.
func (s *TermvectorsService) Validate() error {
var invalid []string
if s.index == "" {
invalid = append(invalid, "Index")
}
if len(invalid) > 0 {
return fmt.Errorf("missing required fields: %v", invalid)
}
return nil
}
// Do executes the operation.
func (s *TermvectorsService) Do(ctx context.Context) (*TermvectorsResponse, error) {
// Check pre-conditions
if err := s.Validate(); err != nil {
return nil, err
}
// Get URL for request
path, params, err := s.buildURL()
if err != nil {
return nil, err
}
// Setup HTTP request body
var body interface{}
if s.bodyJson != nil {
body = s.bodyJson
} else if s.bodyString != "" {
body = s.bodyString
} else {
data := make(map[string]interface{})
if s.doc != nil {
data["doc"] = s.doc
}
if len(s.perFieldAnalyzer) > 0 {
data["per_field_analyzer"] = s.perFieldAnalyzer
}
if s.filter != nil {
src, err := s.filter.Source()
if err != nil {
return nil, err
}
data["filter"] = src
}
if len(data) > 0 {
body = data
}
}
// Get HTTP response
res, err := s.client.PerformRequest(ctx, PerformRequestOptions{
Method: "GET",
Path: path,
Params: params,
Body: body,
Headers: s.headers,
})
if err != nil {
return nil, err
}
// Return operation response
ret := new(TermvectorsResponse)
if err := s.client.decoder.Decode(res.Body, ret); err != nil {
return nil, err
}
return ret, nil
}
// -- Filter settings --
// TermvectorsFilterSettings adds additional filters to a Termsvector request.
// It allows to filter terms based on their tf-idf scores.
// See https://www.elastic.co/guide/en/elasticsearch/reference/7.0/docs-termvectors.html#_terms_filtering
// for more information.
type TermvectorsFilterSettings struct {
maxNumTerms *int64
minTermFreq *int64
maxTermFreq *int64
minDocFreq *int64
maxDocFreq *int64
minWordLength *int64
maxWordLength *int64
}
// NewTermvectorsFilterSettings creates and initializes a new TermvectorsFilterSettings struct.
func NewTermvectorsFilterSettings() *TermvectorsFilterSettings {
return &TermvectorsFilterSettings{}
}
// MaxNumTerms specifies the maximum number of terms the must be returned per field.
func (fs *TermvectorsFilterSettings) MaxNumTerms(value int64) *TermvectorsFilterSettings {
fs.maxNumTerms = &value
return fs
}
// MinTermFreq ignores words with less than this frequency in the source doc.
func (fs *TermvectorsFilterSettings) MinTermFreq(value int64) *TermvectorsFilterSettings {
fs.minTermFreq = &value
return fs
}
// MaxTermFreq ignores words with more than this frequency in the source doc.
func (fs *TermvectorsFilterSettings) MaxTermFreq(value int64) *TermvectorsFilterSettings {
fs.maxTermFreq = &value
return fs
}
// MinDocFreq ignores terms which do not occur in at least this many docs.
func (fs *TermvectorsFilterSettings) MinDocFreq(value int64) *TermvectorsFilterSettings {
fs.minDocFreq = &value
return fs
}
// MaxDocFreq ignores terms which occur in more than this many docs.
func (fs *TermvectorsFilterSettings) MaxDocFreq(value int64) *TermvectorsFilterSettings {
fs.maxDocFreq = &value
return fs
}
// MinWordLength specifies the minimum word length below which words will be ignored.
func (fs *TermvectorsFilterSettings) MinWordLength(value int64) *TermvectorsFilterSettings {
fs.minWordLength = &value
return fs
}
// MaxWordLength specifies the maximum word length above which words will be ignored.
func (fs *TermvectorsFilterSettings) MaxWordLength(value int64) *TermvectorsFilterSettings {
fs.maxWordLength = &value
return fs
}
// Source returns JSON for the query.
func (fs *TermvectorsFilterSettings) Source() (interface{}, error) {
source := make(map[string]interface{})
if fs.maxNumTerms != nil {
source["max_num_terms"] = *fs.maxNumTerms
}
if fs.minTermFreq != nil {
source["min_term_freq"] = *fs.minTermFreq
}
if fs.maxTermFreq != nil {
source["max_term_freq"] = *fs.maxTermFreq
}
if fs.minDocFreq != nil {
source["min_doc_freq"] = *fs.minDocFreq
}
if fs.maxDocFreq != nil {
source["max_doc_freq"] = *fs.maxDocFreq
}
if fs.minWordLength != nil {
source["min_word_length"] = *fs.minWordLength
}
if fs.maxWordLength != nil {
source["max_word_length"] = *fs.maxWordLength
}
return source, nil
}
// -- Response types --
type TokenInfo struct {
StartOffset int64 `json:"start_offset"`
EndOffset int64 `json:"end_offset"`
Position int64 `json:"position"`
Payload string `json:"payload"`
}
type TermsInfo struct {
DocFreq int64 `json:"doc_freq"`
Score float64 `json:"score"`
TermFreq int64 `json:"term_freq"`
Ttf int64 `json:"ttf"`
Tokens []TokenInfo `json:"tokens"`
}
type FieldStatistics struct {
DocCount int64 `json:"doc_count"`
SumDocFreq int64 `json:"sum_doc_freq"`
SumTtf int64 `json:"sum_ttf"`
}
type TermVectorsFieldInfo struct {
FieldStatistics FieldStatistics `json:"field_statistics"`
Terms map[string]TermsInfo `json:"terms"`
}
// TermvectorsResponse is the response of TermvectorsService.Do.
type TermvectorsResponse struct {
Index string `json:"_index"`
Type string `json:"_type"`
Id string `json:"_id,omitempty"`
Version int `json:"_version"`
Found bool `json:"found"`
Took int64 `json:"took"`
TermVectors map[string]TermVectorsFieldInfo `json:"term_vectors"`
}