better indexing and classifier
This commit is contained in:
parent
38c2a72fe0
commit
98b53d3a33
9 changed files with 168 additions and 32 deletions
30
pkg/index/classify.go
Normal file
30
pkg/index/classify.go
Normal file
|
|
@ -0,0 +1,30 @@
|
|||
package index
|
||||
|
||||
import (
|
||||
"regexp"
|
||||
)
|
||||
|
||||
var classes = []class{
|
||||
{regexp.MustCompile(`(?i)[^e]rechnung|quittung|beleg|invoice|bill|check`), []string{"invoice", "rechnung"}},
|
||||
{regexp.MustCompile(`(?i)steuer|lohn|tax|salary`), []string{"tax", "steuer"}},
|
||||
{regexp.MustCompile(`(?i)vertrag|kündigung|vereinbarung|contract|termination|agreement`), []string{"contract", "vertrag"}},
|
||||
{regexp.MustCompile(`(?i)versicherung|versichert|insurance|insured`), []string{"insurance", "versicherung"}},
|
||||
}
|
||||
|
||||
type class struct {
|
||||
r *regexp.Regexp
|
||||
k []string
|
||||
}
|
||||
|
||||
func classify(text string) (keyword []string) {
|
||||
keyword = []string{"", ""}
|
||||
n := 0
|
||||
for _, v := range classes {
|
||||
num := len(v.r.FindAllString(text, -1))
|
||||
if num > n && num >= 2 {
|
||||
keyword = v.k
|
||||
n = num
|
||||
}
|
||||
}
|
||||
return
|
||||
}
|
||||
|
|
@ -7,16 +7,25 @@ import (
|
|||
"git.giftfish.de/ston1th/docstore/pkg/core"
|
||||
"git.giftfish.de/ston1th/godrop/v2"
|
||||
"github.com/blevesearch/bleve"
|
||||
"github.com/blevesearch/bleve/mapping"
|
||||
"html/template"
|
||||
"os"
|
||||
"strings"
|
||||
"time"
|
||||
)
|
||||
|
||||
const maxSearchResult = 101
|
||||
|
||||
type indexDocument struct {
|
||||
Text string `json:"text"`
|
||||
Date string `json:"date"`
|
||||
Type string `json:"type"`
|
||||
const docType = "document"
|
||||
|
||||
type document struct {
|
||||
Text string `json:"text"`
|
||||
Keyword []string `json:"keyword"`
|
||||
Created time.Time `json:"created"`
|
||||
}
|
||||
|
||||
func (d *document) Type() string {
|
||||
return docType
|
||||
}
|
||||
|
||||
type Index struct {
|
||||
|
|
@ -30,8 +39,10 @@ func NewIndex(path string) (i *Index, err error) {
|
|||
return
|
||||
}
|
||||
if _, err = os.Stat(path); os.IsNotExist(err) {
|
||||
mapping := bleve.NewIndexMapping()
|
||||
bi, err = bleve.New(path, mapping)
|
||||
var m *mapping.IndexMappingImpl
|
||||
m, err = createMapping()
|
||||
//m := bleve.NewIndexMapping()
|
||||
bi, err = bleve.New(path, m)
|
||||
if err != nil {
|
||||
err = errors.New("index: " + err.Error())
|
||||
return
|
||||
|
|
@ -51,7 +62,7 @@ func (i *Index) Close() error {
|
|||
return i.i.Close()
|
||||
}
|
||||
|
||||
func (i *Index) Add(text, docType string) (id string, err error) {
|
||||
func (i *Index) Add(text string) (id string, err error) {
|
||||
id = newID()
|
||||
for {
|
||||
doc, _ := i.i.Document(id)
|
||||
|
|
@ -60,7 +71,8 @@ func (i *Index) Add(text, docType string) (id string, err error) {
|
|||
}
|
||||
id = newID()
|
||||
}
|
||||
err = i.i.Index(id, indexDocument{text, core.Now(), docType})
|
||||
k := classify(text)
|
||||
err = i.i.Index(id, &document{text, k, time.Now()})
|
||||
return
|
||||
}
|
||||
|
||||
|
|
@ -86,12 +98,20 @@ func (i *Index) Search(search string) (results []core.Result, err error) {
|
|||
continue
|
||||
}
|
||||
for fragField, frags := range hit.Fragments {
|
||||
if fragField == "text" || fragField == "type" || fragField == "date" {
|
||||
if fragField == "text" || fragField == "created" {
|
||||
for _, f := range frags {
|
||||
r.HTML += template.HTML(f)
|
||||
}
|
||||
break
|
||||
}
|
||||
if fragField == "keyword" {
|
||||
var k string
|
||||
for _, f := range frags {
|
||||
k += f
|
||||
}
|
||||
r.Keyword = strings.ReplaceAll(strings.ReplaceAll(k, "<mark>", ""), "</mark>", "")
|
||||
break
|
||||
}
|
||||
}
|
||||
results = append(results, r)
|
||||
}
|
||||
|
|
|
|||
47
pkg/index/mapping.go
Normal file
47
pkg/index/mapping.go
Normal file
|
|
@ -0,0 +1,47 @@
|
|||
package index
|
||||
|
||||
import (
|
||||
"github.com/blevesearch/bleve"
|
||||
"github.com/blevesearch/bleve/analysis/analyzer/custom"
|
||||
"github.com/blevesearch/bleve/analysis/token/edgengram"
|
||||
"github.com/blevesearch/bleve/analysis/token/lowercase"
|
||||
"github.com/blevesearch/bleve/analysis/tokenizer/unicode"
|
||||
"github.com/blevesearch/bleve/mapping"
|
||||
)
|
||||
|
||||
func createMapping() (m *mapping.IndexMappingImpl, err error) {
|
||||
m = bleve.NewIndexMapping()
|
||||
|
||||
err = m.AddCustomTokenFilter("bigram_tokenfilter", map[string]interface{}{
|
||||
"type": edgengram.Name,
|
||||
"side": edgengram.FRONT,
|
||||
"min": 3.0,
|
||||
"max": 25.0,
|
||||
})
|
||||
if err != nil {
|
||||
return
|
||||
}
|
||||
err = m.AddCustomAnalyzer("fulltext_ngram", map[string]interface{}{
|
||||
"type": custom.Name,
|
||||
"tokenizer": unicode.Name,
|
||||
"token_filters": []string{
|
||||
lowercase.Name,
|
||||
"bigram_tokenfilter",
|
||||
},
|
||||
})
|
||||
if err != nil {
|
||||
return
|
||||
}
|
||||
|
||||
doc := bleve.NewDocumentMapping()
|
||||
tm := bleve.NewTextFieldMapping()
|
||||
tm.Analyzer = "fulltext_ngram"
|
||||
doc.AddFieldMappingsAt("text", tm)
|
||||
km := bleve.NewTextFieldMapping()
|
||||
doc.AddFieldMappingsAt("keyword", km)
|
||||
cm := bleve.NewDateTimeFieldMapping()
|
||||
doc.AddFieldMappingsAt("created", cm)
|
||||
|
||||
m.AddDocumentMapping(docType, doc)
|
||||
return
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue