Beginnings of a prototype scraper in racket for s-bahn-hamburg.de

s-bahn-hamburg.scm

#!/usr/bin/env racket
#lang scheme/base
;; downloading and compiling those requires will take ages, but only
;; for the first time!
(require srfi/1
         srfi/13
         net/url
         (planet bzlib/http:1:0)
         (planet neil/htmlprag:1:6)
         ;; (except-in (planet lizorkin/ssax:2:0/ssax)
         ;;            ;; conflicts with srfi-1:
         ;;            fold-right
         ;;            fold
         ;;            filter
         ;;            cons*)
         (planet lizorkin/sxml:2:1/sxml))

(define http_proxy (regexp-match #rx"^([Hh][Tt][Tt][Pp])(?:://)?([^:]+)(?::([0-9]+))?/?$" (getenv "http_proxy")))
(when http_proxy (current-proxy-servers (let ((p (cdr http_proxy)))
                                          `((,(first p)
                                             ,(second p)
                                             ,(string->number (third p)))))))

(define web-url "http://www.s-bahn-hamburg.de/s_hamburg/view/index.shtml")

(define get-sxml (compose html->sxml http-get))

(define s-bahn-sxml (get-sxml web-url))

;; see http://www.metapaper.net/lisovsky/query/examples/xpath/ to get
;; what's happening here
(define verkehrsmeldungen
  ((sxpath '(// *text*)) ((txpath "//div[contains(@class, 'info-box context info')]")
                                s-bahn-sxml)))

(display (format "~a" verkehrsmeldungen))