1
This commit is contained in:
Alex Yatskov 2015-08-15 17:37:34 +09:00
parent 347ae73fe5
commit 6ae001c044
4 changed files with 5833 additions and 5835 deletions

File diff suppressed because it is too large Load Diff

View File

@ -20,7 +20,7 @@
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
package geo
package main
import (
"encoding/json"
@ -31,22 +31,22 @@ import (
"github.com/kellydunn/golang-geo"
)
type Coord struct {
type geoPos struct {
Latitude float64
Longitude float64
}
type Cache struct {
type geoCache struct {
filename string
data map[string]Coord
data map[string]geoPos
ticker *time.Ticker
coder geo.GoogleGeocoder
}
func NewCache(filename string) (*Cache, error) {
cache := &Cache{
func newGeoCache(filename string) (*geoCache, error) {
cache := &geoCache{
filename: filename,
data: make(map[string]Coord),
data: make(map[string]geoPos),
ticker: time.NewTicker(time.Millisecond * 200),
}
@ -57,7 +57,7 @@ func NewCache(filename string) (*Cache, error) {
return cache, nil
}
func (c *Cache) load() error {
func (c *geoCache) load() error {
file, err := os.Open(c.filename)
if os.IsNotExist(err) {
return nil
@ -70,7 +70,7 @@ func (c *Cache) load() error {
return json.NewDecoder(file).Decode(&c.data)
}
func (c *Cache) Save() error {
func (c *geoCache) save() error {
js, err := json.MarshalIndent(c.data, "", " ")
if err != nil {
return err
@ -79,19 +79,19 @@ func (c *Cache) Save() error {
return ioutil.WriteFile(c.filename, js, 0644)
}
func (c *Cache) Decode(address string) (Coord, error) {
if coord, ok := c.data[address]; ok {
return coord, nil
func (c *geoCache) decode(address string) (geoPos, error) {
if pos, ok := c.data[address]; ok {
return pos, nil
}
<-c.ticker.C
point, err := c.coder.Geocode(address)
if err != nil {
return Coord{}, err
return geoPos{}, err
}
coord := Coord{point.Lat(), point.Lng()}
c.data[address] = coord
return coord, nil
pos := geoPos{point.Lat(), point.Lng()}
c.data[address] = pos
return pos, nil
}

View File

@ -31,8 +31,6 @@ import (
"strings"
"sync"
"github.com/FooSoft/scrape/geo"
"github.com/FooSoft/scrape/web"
"github.com/PuerkitoBio/goquery"
)
@ -91,13 +89,13 @@ func dumpReviews(filename string, in chan tabelogReview) error {
return nil
}
func decodeReviews(in chan tabelogReview, out chan tabelogReview, gc *geo.Cache) {
func decodeReviews(in chan tabelogReview, out chan tabelogReview, gc *geoCache) {
for {
if review, ok := <-in; ok {
coord, err := gc.Decode(review.Address)
pos, err := gc.decode(review.Address)
if err == nil {
review.Latitude = coord.Latitude
review.Longitude = coord.Longitude
review.Latitude = pos.Latitude
review.Longitude = pos.Longitude
out <- review
} else {
log.Printf("failed to decode address for %s (%v)", review.Url, err)
@ -109,10 +107,10 @@ func decodeReviews(in chan tabelogReview, out chan tabelogReview, gc *geo.Cache)
}
}
func scrapeReview(url string, out chan tabelogReview, wc *web.Cache, wg *sync.WaitGroup) {
func scrapeReview(url string, out chan tabelogReview, wc *webCache, wg *sync.WaitGroup) {
defer wg.Done()
doc, err := wc.Load(url)
doc, err := wc.load(url)
if err != nil {
log.Printf("failed to scrape review at %s (%v)", url, err)
return
@ -148,8 +146,8 @@ func scrapeReview(url string, out chan tabelogReview, wc *web.Cache, wg *sync.Wa
out <- review
}
func scrapeIndex(url string, out chan tabelogReview, wc *web.Cache, wg *sync.WaitGroup) {
doc, err := wc.Load(url)
func scrapeIndex(url string, out chan tabelogReview, wc *webCache, wg *sync.WaitGroup) {
doc, err := wc.load(url)
if err != nil {
log.Printf("failed to scrape index at %s (%v)", url, err)
return
@ -178,7 +176,7 @@ func scrapeIndex(url string, out chan tabelogReview, wc *web.Cache, wg *sync.Wai
}
}
func scrapeReviews(url string, out chan tabelogReview, wc *web.Cache) error {
func scrapeReviews(url string, out chan tabelogReview, wc *webCache) error {
var wg sync.WaitGroup
scrapeIndex(url, out, wc, &wg)
wg.Wait()
@ -188,12 +186,12 @@ func scrapeReviews(url string, out chan tabelogReview, wc *web.Cache) error {
}
func scrapeTabelog(url, resultFile, webCacheDir, geoCacheFile string) error {
wc, err := web.NewCache(webCacheDir)
wc, err := newWebCache(webCacheDir)
if err != nil {
return err
}
gc, err := geo.NewCache(geoCacheFile)
gc, err := newGeoCache(geoCacheFile)
if err != nil {
return err
}
@ -205,5 +203,5 @@ func scrapeTabelog(url, resultFile, webCacheDir, geoCacheFile string) error {
scrapeReviews(url, scrapeChan, wc)
dumpReviews(resultFile, decodeChan)
return gc.Save()
return gc.save()
}

View File

@ -20,7 +20,7 @@
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
package web
package main
import (
"bytes"
@ -35,17 +35,17 @@ import (
"github.com/PuerkitoBio/goquery"
)
type Cache struct {
type webCache struct {
directory string
ticker *time.Ticker
}
func NewCache(directory string) (*Cache, error) {
func newWebCache(directory string) (*webCache, error) {
if err := os.MkdirAll(directory, 0755); err != nil {
return nil, err
}
cache := &Cache{
cache := &webCache{
directory: directory,
ticker: time.NewTicker(time.Millisecond * 100),
}
@ -53,13 +53,13 @@ func NewCache(directory string) (*Cache, error) {
return cache, nil
}
func (c *Cache) urlToLocal(url string) string {
func (c *webCache) urlToLocal(url string) string {
hash := md5.New()
hash.Write([]byte(url))
return path.Join(c.directory, fmt.Sprintf("%x.html", hash.Sum(nil)))
}
func (c *Cache) Load(url string) (*goquery.Document, error) {
func (c *webCache) load(url string) (*goquery.Document, error) {
localPath := c.urlToLocal(url)
if file, err := os.Open(localPath); err == nil {