unstable: initial tags

This commit is contained in:
ston1th 2019-08-20 23:57:43 +02:00
commit b3b62af3a4
28 changed files with 503 additions and 91 deletions

View file

@ -1,32 +0,0 @@
// Copyright (C) 2019 Marius Schellenberger
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
}

View file

@ -19,7 +19,7 @@ const docType = "document"
type document struct {
Text string `json:"text"`
Keyword []string `json:"keyword"`
Tags []string `json:"tags"`
Created time.Time `json:"created"`
}
@ -61,7 +61,7 @@ func (i *Index) Close() error {
return i.i.Close()
}
func (i *Index) Add(text string) (id string, err error) {
func (i *Index) Add(text string, tags []string) (id string, err error) {
id = newID()
for {
doc, _ := i.i.Document(id)
@ -70,8 +70,7 @@ func (i *Index) Add(text string) (id string, err error) {
}
id = newID()
}
k := classify(text)
err = i.i.Index(id, &document{text, k, time.Now()})
err = i.i.Index(id, &document{text, tags, time.Now()})
return
}
@ -97,12 +96,12 @@ func (i *Index) Search(search string) (results core.Results, err error) {
continue
}
for fragField, frags := range hit.Fragments {
if fragField == "keyword" {
var k string
if fragField == "tags" {
var t string
for _, f := range frags {
k += f
t += f
}
r.Keyword = strings.ReplaceAll(strings.ReplaceAll(k, "<mark>", ""), "</mark>", "")
r.Tags = strings.ReplaceAll(strings.ReplaceAll(t, "<mark>", ""), "</mark>", "")
}
if fragField == "text" || fragField == "created" {
for _, f := range frags {