move collection and extraction of articles into their own file; add custom response cache

This commit is contained in:
2025-01-04 16:53:46 +01:00
parent f31c1d7793
commit d63ee8dcf2
14 changed files with 519 additions and 105 deletions

View File

@@ -0,0 +1,7 @@
package collectors
import "crowsnest/internal/model/sqlite"
type Collector struct {
Responses *sqlite.ResponseModel
}

View File

@@ -0,0 +1,110 @@
package collectors
import (
//"crowsnest/internal/model"
//"regexp"
//"time"
//"strings"
"fmt"
"time"
"github.com/gocolly/colly/v2"
)
func (c *Collector) Spiegel() {
collycollector := colly.NewCollector(
colly.AllowedDomains("www.spiegel.de", "spiegel.de"),
colly.CacheDir("./persistence/spiegel_cache"),
colly.MaxDepth(3),
)
// cascade
collycollector.OnHTML("a[href]", func(e *colly.HTMLElement) {
e.Request.Visit(e.Attr("href"))
})
// cache
collycollector.OnScraped(func(r *colly.Response) {
c.Responses.Insert(r.Request.URL.String(), string(r.Body))
})
// go through archive
startDate := time.Date(2000, time.January, 1, 0, 0, 0, 0, time.UTC)
currentDate := time.Now()
for date := startDate; date.Before(currentDate) || date.Equal(currentDate); date = date.AddDate(0, 0, 1) {
urlDate := date.Format("02.01.2006")
url := fmt.Sprintf("https://www.spiegel.de/nachrichtenarchiv/artikel-%s.html", urlDate)
collycollector.Visit(url)
}
//// create entry if not behind paywall
//paywall_false_pattern := regexp.MustCompile("\"paywall\":{\"attributes\":{\"is_active\":false")
//collycollector.OnResponse(func(r *colly.Response) {
// if paywall_false_pattern.Match(r.Body) {
// url := r.Request.URL.String()
// (*results)[url] = &model.Article{
// SourceUrl: url,
// FetchDate: time.Now(),
// Content: "",
// }
// }
//})
//// check for article type
//collycollector.OnHTML("meta[property='og:type']", func(e *colly.HTMLElement) {
// if e.Attr("content") != "article" {
// delete(*results, e.Request.URL.String())
// }
//})
//// add title
//collycollector.OnHTML("meta[property='og:title']", func(e *colly.HTMLElement) {
// if val, ok := (*results)[e.Request.URL.String()]; ok {
// val.Title = e.Attr("content")
// }
//})
//// prepend description to content of article
//collycollector.OnHTML("meta[name='description']", func(e *colly.HTMLElement) {
// if val, ok := (*results)[e.Request.URL.String()]; ok {
// val.Content = e.Attr("content") + val.Content
// }
//})
//// add publishing date
//collycollector.OnHTML("meta[name='date']", func(e *colly.HTMLElement) {
// if val, ok := (*results)[e.Request.URL.String()]; ok {
// t, err := time.Parse("2006-01-02T15:04:05-07:00", e.Attr("content"))
// if err != nil {
// panic(err)
// }
// val.PublishDate = t
// }
//})
//// add author
//collycollector.OnHTML("meta[name='author']", func(e *colly.HTMLElement) {
// if val, ok := (*results)[e.Request.URL.String()]; ok {
// val.Author = e.Attr("content")
// }
//})
//// add content
//collycollector.OnHTML("main[id='Inhalt'] div > p", func(e *colly.HTMLElement) {
// if val, ok := (*results)[e.Request.URL.String()]; ok {
// cont := val.Content
// pattern := regexp.MustCompile("\\s+")
// cont = string(pattern.ReplaceAll([]byte(cont), []byte(" ")))
// cont = strings.ReplaceAll(cont, "»", "\"")
// cont = strings.ReplaceAll(cont, "«", "\"")
// val.Content = cont + " " + e.Text
// }
//})
}

View File

@@ -0,0 +1,39 @@
package collectors
import (
"fmt"
"time"
"github.com/gocolly/colly/v2"
)
// Gets every page of the archive of zeit.de and stores the responses into the
// database.
func (c *Collector) Zeit() {
collycollector := colly.NewCollector(
colly.AllowedDomains("www.zeit.de", "zeit.de"),
colly.CacheDir("./persistence/zeit_cache"),
colly.MaxDepth(2),
)
// cascade
collycollector.OnHTML("a[href]", func(e *colly.HTMLElement) {
e.Request.Visit(e.Attr("href"))
})
// cache
collycollector.OnScraped(func(r *colly.Response) {
c.Responses.Insert(r.Request.URL.String(), string(r.Body))
})
// go through archive
startDate := time.Date(1946, time.January, 1, 0, 0, 0, 0, time.UTC)
currentDate := time.Now()
for date := startDate; date.Before(currentDate) || date.Equal(currentDate); date = date.AddDate(0, 0, 7) {
year, week := date.ISOWeek()
url := fmt.Sprintf("https://www.zeit.de/%04d/%02d/index", year, week)
collycollector.Visit(url)
}
}

View File

@@ -0,0 +1,8 @@
package extractors
import "crowsnest/internal/model/sqlite"
type Extractor struct {
Responses *sqlite.ResponseModel
Articles *sqlite.ArticleModel
}

View File

@@ -0,0 +1,113 @@
package extractors
import (
"crowsnest/internal/model"
"log"
"regexp"
"strings"
"time"
"github.com/PuerkitoBio/goquery"
)
func (extractor *Extractor) Spiegel() error {
// get urls to process
urls, err := extractor.Responses.UnprocessedUrls()
if err != nil { return err }
paywall_false_pattern := regexp.MustCompile("\"paywall\":{\"attributes\":{\"is_active\":false")
url_pattern := regexp.MustCompile("^https://(www\\.)?spiegel.de.*")
whitespace := regexp.MustCompile("\\s+")
var exists bool
var pagetype, title, content, datestr, author string
var tag *goquery.Selection
var date time.Time
for _, url := range urls {
// check url url pattern
if !url_pattern.Match([]byte(url)) { continue }
// get response
res, err := extractor.Responses.GetByUrl(url)
if err != nil {
log.Println("failed to process url", url, "with", err)
continue
}
// check for paywall
if !paywall_false_pattern.Match([]byte(res.Content)) {
extractor.Responses.Processed(url)
continue
}
// construct goquery doc
doc, err := goquery.NewDocumentFromReader(strings.NewReader(res.Content))
if err != nil {
log.Println("failed to process url", url, "with", err)
continue
}
// check for article type
tag = doc.Find("meta[property='og:type']")
pagetype, exists = tag.Attr("content")
if !exists || pagetype != "article" { extractor.Responses.Processed(url); continue; }
// get title
tag = doc.Find("meta[property='og:title']")
title, exists = tag.Attr("content")
if !exists { extractor.Responses.Processed(url); continue; }
// prepend description to content of article
tag = doc.Find("meta[name='description']")
content, exists = tag.Attr("content")
content += " "
if !exists { extractor.Responses.Processed(url); continue; }
// get publishing date
tag = doc.Find("meta[name='date']")
datestr, exists = tag.Attr("content")
if !exists { extractor.Responses.Processed(url); continue; }
date, err = time.Parse("2006-01-02T15:04:05-07:00", datestr)
if err != nil { extractor.Responses.Processed(url); continue; }
// get author
tag = doc.Find("meta[name='author']")
author, exists = tag.Attr("content")
if !exists { extractor.Responses.Processed(url); continue; }
// get content
tag = doc.Find("main[id='Inhalt'] div > p")
tag.Each(func(index int, p *goquery.Selection) {
content += " " + p.Text()
})
// clean up content string
content = string(whitespace.ReplaceAll([]byte(content), []byte(" ")))
content = strings.ReplaceAll(content, "»", "\"")
content = strings.ReplaceAll(content, "«", "\"")
// insert new article
article := model.Article{
SourceUrl: url,
PublishDate: date,
FetchDate: res.FetchDate,
Title: title,
Content: content,
Author: author,
}
err = extractor.Articles.Insert(&article)
if err != nil {
log.Println("failed to insert", article)
} else {
extractor.Responses.Processed(url)
log.Println("found article at", url)
}
}
return nil
}

View File

@@ -1,114 +1,35 @@
package main
import (
"crowsnest/internal/model"
"crowsnest/cmd/crawler/collectors"
"crowsnest/cmd/crawler/extractors"
"crowsnest/internal/model/sqlite"
"database/sql"
"fmt"
"log"
"regexp"
"strings"
"time"
"github.com/gocolly/colly/v2"
_ "github.com/mattn/go-sqlite3"
)
func spiegelCollector(results *map[string]*model.Article) *colly.Collector {
c := colly.NewCollector(
colly.AllowedDomains("www.spiegel.de", "spiegel.de"),
colly.CacheDir("./persistence/spiegel_cache"),
colly.MaxDepth(5),
)
// create entry if not behind paywall
paywall_false_pattern := regexp.MustCompile("\"paywall\":{\"attributes\":{\"is_active\":false")
c.OnResponse(func(r *colly.Response) {
if paywall_false_pattern.Match(r.Body) {
url := r.Request.URL.String()
(*results)[url] = &model.Article{
SourceUrl: url,
FetchDate: time.Now(),
Content: "",
}
}
})
// check for article type
c.OnHTML("meta[property='og:type']", func(e *colly.HTMLElement) {
if e.Attr("content") != "article" {
}
})
// add title
c.OnHTML("meta[property='og:title']", func(e *colly.HTMLElement) {
if val, ok := (*results)[e.Request.URL.String()]; ok {
val.Title = e.Attr("content")
}
})
// prepend description to content of article
c.OnHTML("meta[name='description']", func(e *colly.HTMLElement) {
if val, ok := (*results)[e.Request.URL.String()]; ok {
val.Content = e.Attr("content") + val.Content
}
})
// add publishing date
c.OnHTML("meta[name='date']", func(e *colly.HTMLElement) {
if val, ok := (*results)[e.Request.URL.String()]; ok {
t, err := time.Parse("2006-01-02T15:04:05-07:00", e.Attr("content"))
if err != nil {
panic(err)
}
val.PublishDate = t
}
})
// add author
c.OnHTML("meta[name='author']", func(e *colly.HTMLElement) {
if val, ok := (*results)[e.Request.URL.String()]; ok {
val.Author = e.Attr("content")
}
})
// add content
c.OnHTML("main[id='Inhalt'] div > p", func(e *colly.HTMLElement) {
if val, ok := (*results)[e.Request.URL.String()]; ok {
cont := val.Content
pattern := regexp.MustCompile("\\s+")
cont = string(pattern.ReplaceAll([]byte(cont), []byte(" ")))
cont = strings.ReplaceAll(cont, "»", "\"")
cont = strings.ReplaceAll(cont, "«", "\"")
val.Content = cont + " " + e.Text
}
})
// cascade
c.OnHTML("a[href]", func(e *colly.HTMLElement) {
e.Request.Visit(e.Attr("href"))
})
return c
}
func main() {
res := make(map[string]*model.Article)
c := spiegelCollector(&res)
c.Visit("https://www.spiegel.de/")
db, err := sql.Open("sqlite3", "./persistence/app.db")
if err != nil { log.Fatal(err) }
db_articles := &sqlite.ArticleModel{ DB: db }
counter := 0
for _, val := range res {
counter++
db_articles.Insert(val)
// open database
db, err := sql.Open("sqlite3", "./persistence/app.db")
if err != nil {
log.Fatal(err)
}
fmt.Println(counter)
// collect websites
_ = collectors.Collector{
Responses: &sqlite.ResponseModel{DB: db},
}
//coll.Spiegel()
//coll.Zeit()
// extract articles from websites
extr := extractors.Extractor{
Responses: &sqlite.ResponseModel{DB: db},
Articles: &sqlite.ArticleModel{DB: db},
}
extr.Spiegel()
}