pegasus

Pegasus is a highly-modular, durable and scalable crawler for clojure.

Parallelism is achieved with core.async. Durability is achieved with durable-queue and LMDB.

A blog post on how pegasus works: [link]

Usage

Leiningen dependencies:

A few example crawls:

This one crawls 20 docs from my blog (http://blog.shriphani.com).

URLs are extracted using enlive selectors.

(ns pegasus.foo
  (:require [pegasus.core :refer [crawl]]
            [pegasus.dsl :refer :all])
  (:import (java.io StringReader)))

(defn crawl-sp-blog
  []
  (crawl {:seeds ["http://blog.shriphani.com"]
          :user-agent "Pegasus web crawler"
          :corpus-size 20 ;; crawl 20 documents
          :job-dir "/tmp/sp-blog-corpus"})) ;; store all crawl data in /tmp/sp-blog-corpus/

(defn crawl-sp-blog-custom-extractor
  []
  (crawl {:seeds ["http://blog.shriphani.com"]
          :user-agent "Pegasus web crawler"
          :extractor (defextractors
                       (extract :at-selector [:article :header :h2 :a]

                                :follow :href

                                :with-regex #"blog.shriphani.com")
                       
                       (extract :at-selector [:ul.pagination :a]

                                :follow :href
                                
                                :with-regex #"blog.shriphani.com"))
          
          :corpus-size 20 ;; crawl 20 documents
          :job-dir "/tmp/sp-blog-corpus"}))

Say you want more control and want to avoid the DSL, you can use the underlying machinery directly. Here's an example using XPaths to extract links.

(ns your.namespace
  (:require [org.bovinegenius.exploding-fish :as uri]
            [net.cgrand.enlive-html :as html]
            [pegasus.core :refer [crawl]]
            [clj-xpath.core :refer [$x $x:text xml->doc]]))

(deftype XpathExtractor []
  process/PipelineComponentProtocol
  
  (initialize
    [this config]
    config)
  
  (run
    [this obj config]
    (when (= "blog.shriphani.com"
             (-> obj :url uri/host))
      
      (let [url (:url obj)
            resource (try (-> obj
                              :body
                              xml->doc)
                          (catch Exception e nil))
            
            ;; extract the articles
            articles (map
                      :text
                      (try ($x "//item/link" resource)
                           (catch Exception e nil)))]
        
        ;; add extracted links to the supplied object
        (merge obj
               {:extracted articles}))))

  (clean
    [this config]
    nil))

(defn crawl-sp-blog-xpaths
  []
  (crawl {:seeds ["http://blog.shriphani.com/feeds/all.rss.xml"]
          :user-agent "Pegasus web crawler"
          :extractor (->XpathExtractor)
          
          :corpus-size 20 ;; crawl 20 documents
          :job-dir "/tmp/sp-blog-corpus"}))

;; start crawling
(crawl-sp-blog-xpaths)

License

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.

raffy2010/pegasus

pegasus

Usage

License