alex/restaurant-search
1
Fork 0
Code Releases Activity

Updating scraper

Browse Source
This commit is contained in:
Alex Yatskov 2015-08-12 20:29:45 +09:00
parent 0d6994a764
commit 32862a2797
2 changed files with 64 additions and 52 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
scrape.go
View File

@ -25,7 +25,7 @@ package main
import "log"
func main() {
urlFormat := "http://tabelog.com/en/rstLst/{{.Page}}/?lat=35.465808055555996&lon=139.61964361111&zoom=16&RdoCosTp=2&LstCos=0&LstCosT=11&LstSitu=0&LstRev=0&LstReserve=0&ChkParking=0&LstSmoking=0"
urlFormat := "http://tabelog.com/en/kanagawa/rstLst/{{.Page}}/"
tabelogReviews := "tabelog.json"
if err := scrapeTabelog(tabelogReviews, urlFormat); err != nil {

114
tabelog.go
View File

@ -29,6 +29,7 @@ import (
"log"
"strconv"
"strings"
"sync"
"text/template"
"github.com/PuerkitoBio/goquery"
@ -51,53 +52,6 @@ type tabelogReview struct {
Url string
}
func scrapeReview(url string, out chan tabelogReview) {
doc, err := goquery.NewDocument(url)
if err != nil {
log.Print(err)
return
}
var r tabelogReview
r.Url = url
r.Name = doc.Find("body > article > header > div.rd-header.l-container > div > div.rd-header__headline > h2 > a").Text()
r.Address = strings.TrimSpace(doc.Find("#anchor-rd-detail > section > table > tbody > tr > td > p.rd-detail-info__rst-address > span").Text())
if r.Dishes, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(2)").Text(), 8); err != nil {
return
}
if r.Service, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(4)").Text(), 8); err != nil {
return
}
if r.Atmosphere, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(6)").Text(), 8); err != nil {
return
}
if r.Cost, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(8)").Text(), 8); err != nil {
return
}
if r.Drinks, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(10)").Text(), 8); err != nil {
return
}
out <- r
}
func scrapeIndex(url string, out chan tabelogReview) error {
doc, err := goquery.NewDocument(url)
if err != nil {
return err
}
doc.Find("#js-map-search-result-list > li > div.list-rst__header > p > a").Each(func(index int, sel *goquery.Selection) {
if href, ok := sel.Attr("href"); ok {
go scrapeReview(href, out)
}
})
return nil
}
func dumpReviews(filename string, in chan tabelogReview, out chan error) {
var reviews []tabelogReview
for {
@ -118,6 +72,62 @@ func dumpReviews(filename string, in chan tabelogReview, out chan error) {
out <- ioutil.WriteFile(filename, js, 0644)
}
func scrapeReview(url string, out chan tabelogReview, wg *sync.WaitGroup) {
defer wg.Done()
doc, err := goquery.NewDocument(url)
if err != nil {
log.Print(err)
return
}
addresses := doc.Find("p.rd-detail-info__rst-address")
if addresses.Length() != 2 {
return
}
var review tabelogReview
review.Url = url
review.Name = doc.Find("div.rd-header__headline > h2 > a").Text()
review.Address = strings.TrimSpace(addresses.First().Text())
if review.Dishes, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(2)").Text(), 8); err != nil {
return
}
if review.Service, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(4)").Text(), 8); err != nil {
return
}
if review.Atmosphere, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(6)").Text(), 8); err != nil {
return
}
if review.Cost, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(8)").Text(), 8); err != nil {
return
}
if review.Drinks, err = strconv.ParseFloat(doc.Find("#js-rating-detail > dd:nth-child(10)").Text(), 8); err != nil {
return
}
out <- review
}
func scrapeIndex(url string, out chan tabelogReview, wg *sync.WaitGroup) {
defer wg.Done()
doc, err := goquery.NewDocument(url)
if err != nil {
log.Print(err)
return
}
doc.Find("div.list-rst__header > p > a").Each(func(index int, sel *goquery.Selection) {
if href, ok := sel.Attr("href"); ok {
wg.Add(1)
go scrapeReview(href, out, wg)
}
})
}
func scrapeTabelog(filename, url string) error {
out := make(chan tabelogReview)
in := make(chan error)
@ -126,17 +136,19 @@ func scrapeTabelog(filename, url string) error {
t := template.New("tabelog")
t.Parse(url)
for i := 1; i <= 2; i++ {
var wg sync.WaitGroup
for i := 1; i <= 1; i++ {
var url bytes.Buffer
if err := t.Execute(&url, tabelogParams{i}); err != nil {
log.Fatal(err)
}
if err := scrapeIndex(string(url.Bytes()), out); err != nil {
return err
}
wg.Add(1)
go scrapeIndex(string(url.Bytes()), out, &wg)
}
wg.Wait()
close(out)
return <-in
}