123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141 |
- package main
- import (
- "fmt"
- "jy/admin/track"
- "jy/clear"
- "jy/extract"
- . "jy/mongodbutil"
- "log"
- "os"
- "qfw/util"
- "regexp"
- "strconv"
- "strings"
- "testing"
- "time"
- )
- func Test_han(t *testing.T) {
- str := `[\u4e00-\u9fa5]` //"[\u4e00-\u9fa5]"
- //var rg = regexp.MustCompile(`[\u4e00-\u9fa5]`)会出错
- if strings.Contains(str, "\\u") {
- pattern, _ := strconv.Unquote(`"` + str + `"`)
- log.Println(pattern)
- }
- var rg = regexp.MustCompile(str)
- fmt.Println(rg.MatchString(str))
- os.Exit(0)
- }
- func Test_task(t *testing.T) {
- Mgo = MgoFactory(1, 3, 120, "192.168.3.207:27092", "extract_kf")
- //extract.StartExtractTaskId("5b8f804025e29a290415aee1")5c528686698414055c47b115
- //extract.StartExtractTestTask("5e103206234ddc34b406c5d1", "5df59ee3e9d1f601e46fc3f9", "1", "mxs_v1", "mxs_v1")
- extract.StartExtractTestTask("5cdd3025698414032c8322b1", "5df50776e9d1f601e4964179", "1", "mxs_v1", "mxs_v2")
- //extract.StartExtractTestTask("5c3d75c96984142998eb00e1", "5c2a3d28a5cb26b9b76144dd", "100", "mxs_v3", "mxs_v3")
- time.Sleep(5 * time.Second)
- }
- func Test_extractcity(t *testing.T) {
- Mgo = MgoFactory(1, 3, 120, "192.168.3.207:27082", "extract_kf")
- extract.InitDFA2()
- //查询采购单位信息
- extract.FindBuyer()
- }
- func Test_reg(t *testing.T) {
- reg1, _ := regexp.Compile("((([((]\\d{3,4}[))])?(\\d{6,12}([×―—-\\-]+\\d{3,4})?|\\d{3,5}[×―—-\\-]+[\u3000\u2003\u00a0\\s]*\\d{6,12}([×―—-\\-]+\\d{3,})?|(\\d{2}[×―—-\\-])+\\d{8}[×―—-\\-](\\d{3}[、])+)(转\\d{3,4})?[或/、,,;;\u3000\u2003\u00a0\\s]*)+(\\d{3,})?)")
- log.Println("---", reg1.FindAllString("05939-5365001(兰陵县芦柞镇人民政府)", -1))
- reg2, _ := regexp.Compile("^\\d*[×―—-\\-]*[\u3000\u2003\u00a0\\s]*\\d*$")
- log.Println("---", reg2.MatchString("张女士/"))
- filterK := regexp.MustCompile("[((\\[【].*?[))\\]】]|<[^>].+?>|[①②③¥·;;‘“'’”,*<>((\\[【、))/\\]】??,。.\".\\s\u3000\u2003\u00a0]+|^[一二三四五六七八九十0-91234567890]+")
- log.Println(filterK.FindString("二)采购项目联系人(代理机构)"))
- }
- func Test_reg1(t *testing.T) {
- context := `sss<input name="AgentCode" size="30" maxsize="50" value="91370800688271668P" class="textbox">
- dfdf<input type="hidden" name="AgentCode" size="30" maxsize="50" value="tttt" class="textbox"></input>`
- reg := regexp.MustCompile(`<\s*input.*value=['"](.[^'"]+).+>`)
- tmp := reg.ReplaceAllString(context, "$1")
- log.Println(tmp)
- }
- func Test_paths(t *testing.T) {
- Mgo = MgoFactory(1, 3, 120, "192.168.3.207:27082", "extract_v3")
- tracks := track.GetTrackPath("5b8dd276a5cb26b9b7faaa7c", "projectname", "rack_v3", "result_v3")
- for code, v := range tracks {
- if tmp, ok := v.([]map[string]interface{}); ok {
- for k, v := range tmp {
- if k == 0 {
- log.Println(v)
- } else {
- log.Println(code, v["code"], v["value"])
- }
- }
- }
- break
- }
- }
- func Test_clear(t *testing.T) {
- value := "法拉(盛(客{)户)端副科级沙发俩括号的"
- log.Println("pre---", value)
- startChars := []string{"[((]", "[\\[【]", "[{{]", "[<《]", "〔"}
- endChars := []string{"[))]", "[\\]】]", "[}}]", "[>》]", "〕"}
- for k, v := range startChars {
- sReg := regexp.MustCompile(v)
- eReg := regexp.MustCompile(endChars[k])
- sIndex := sReg.FindAllStringIndex(value, -1)
- eIndex := eReg.FindAllStringIndex(value, -1)
- sCount := len(sIndex)
- eCount := len(eIndex)
- if sCount == eCount {
- continue
- }
- log.Println("value1---", value, sCount, eCount)
- //清理前面
- if sCount > eCount {
- value = value[sIndex[eCount][1]:]
- }
- log.Println("value2---", value)
- //清理后面
- if sCount < eCount {
- value = value[:eIndex[sCount][0]]
- }
- log.Println("value3---", value)
- }
- log.Println("value4---", value)
- //交叉出现情况处理
- sReplReg := regexp.MustCompile("[((\\[【{{〔<《][^))\\]】}}〕>》]*$")
- eReplReg := regexp.MustCompile("^[^((\\[【{{〔<《]*[))\\]】}}〕>》]")
- if sReplReg.MatchString(value) || eReplReg.MatchString(value) {
- value = sReplReg.ReplaceAllString(value, "")
- value = eReplReg.ReplaceAllString(value, "")
- //value = fmt.Sprint(childCutNotPrs([]interface{}{value, data[1]}, count+1)[0])
- }
- log.Println("result---", value)
- }
- func Test_buyer(t *testing.T) {
- Mgo = MgoFactory(1, 3, 120, "192.168.3.207:27092", "extract_kf")
- demo, _ := Mgo.Find("demo_data", nil, `{"_id:1"}`, `{"buyer":1,"title":1}`, false, -1, -1)
- result, _ := Mgo.Find("mxs_buyer", nil, `{"_id:1"}`, `{"buyer":1}`, false, -1, -1)
- for _, d := range *demo {
- id1 := util.BsonIdToSId(d["_id"])
- buyer1 := util.ObjToString(d["buyer"])
- //title := util.ObjToString(d["title"])
- for _, r := range *result {
- id2 := util.BsonIdToSId(r["_id"])
- buyer2 := util.ObjToString(r["buyer"])
- if id1 == id2 {
- if buyer1 != buyer2 {
- util.Debug(id1, buyer1, buyer2)
- }
- break
- }
- }
- }
- }
- func Test_util1(t *testing.T) {
- data := clear.CutSymbol([]interface{}{"----------123123", "-----123123"})
- fmt.Println(data)
- }
|