blekko.rkt

#lang racket

(require net/url)

(require (planet dherman/json:4:0))


; Blekko API implementation

(struct blekko (queries api-key))

(define *BLEKKO-BASE-URL* "http://blekko.com/ws/")

(define (issue-queries a-blekko gather-results-file)
  
    (for ((q (blekko-queries a-blekko)))
      (display q)
      (newline)
      (append-to-file 'url (issue-query q (blekko-api-key a-blekko) 0 '()) gather-results-file)))


; collect results from all pages
(define (issue-query q auth page-no acc-results)
  (let* ((cur-pg-results (with-handlers 
			     ([exn:fail? (lambda (x) (make-hash))])
			   (read-json (get-pure-port 
				     (struct-copy 
				      url
				      (string->url *BLEKKO-BASE-URL*)
				      (query (list
					      (cons 'q (string-join 
							(list q 
							      " /json /ps=100") ""))
					      (cons 'auth auth)
					      (cons 'p (number->string page-no))))))))))
    (if (member 'RESULT (hash-keys cur-pg-results))
	(begin
	  (sleep 1)
	  (display "next page")
	  (newline)
;	  (display (map (lambda (r) (hash-ref r 'url)) (hash-ref cur-pg-results 'RESULT)))
;	  (newline)
	  (issue-query q auth (+ page-no 1) (append acc-results (hash-ref cur-pg-results 'RESULT))))
	acc-results)))

; given a field in the blekko results obj, we dump the associated data obj
(define (dump-to-file result-field results filename)
  (display-lines-to-file (map (lambda (x) (hash-ref x result-field)) results) filename #:mode 'text))

(define (append-to-file result-field results filename)
  (display-lines-to-file (map (lambda (x) (hash-ref x result-field)) results) filename #:mode 'text #:exists 'append))
  
(provide (all-defined-out))