1
0

rss.lisp 4.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113
  1. (in-package #:chatikbot)
  2. (defstruct feed url title seen-guids next-fetch (period 300))
  3. (defstruct feed-item guid link title description published)
  4. (defparameter *rss-min-period* 60 "Min rss refresh period in seconds")
  5. (defparameter *rss-max-period* 1800 "Max rss refresh period in seconds")
  6. (defparameter *rss-change-rate* 0.1 "Refresh period adjustment rata")
  7. (defun http-default (url)
  8. (let ((uri (quri:uri url)))
  9. (quri:render-uri
  10. (if (null (quri:uri-scheme uri))
  11. (quri:uri (format nil "http://~A" url))
  12. uri))))
  13. (defun get-by-tag (node tag)
  14. (nreverse (plump:get-elements-by-tag-name node tag)))
  15. (defun url-parse (url)
  16. (plump:parse (flexi-streams:octets-to-string
  17. (dex:get (http-default url) :force-binary t) :external-format :utf-8)))
  18. (defun find-rss-url (url)
  19. (ignore-errors
  20. (multiple-value-bind (body status headers uri stream)
  21. (dex:get (http-default url) :force-binary t)
  22. (declare (ignore status stream))
  23. (let ((root (plump:parse (flexi-streams:octets-to-string body :external-format :utf-8)))
  24. (content-type (gethash "content-type" headers)))
  25. (cond
  26. ((alexandria:starts-with-subseq "text/html" content-type)
  27. (loop for link in (get-by-tag root "link")
  28. when (string= "application/rss+xml" (plump:attribute link "type"))
  29. do (return (quri:render-uri
  30. (quri:merge-uris (quri:uri (plump:attribute link "href"))
  31. uri)))))
  32. ((string= "rss" (plump:tag-name (plump:first-element root)))
  33. (quri:render-uri uri)))))))
  34. (defun build-feed (url)
  35. (let ((root (url-parse url)))
  36. (make-feed :url url :title (child-text root "title"))))
  37. (defun adjust-period (feed coeff)
  38. "Adjust the period of feed based on whenever there were new items. With clamping"
  39. (let* ((p (feed-period feed))
  40. (diff (round (* p *rss-change-rate*))))
  41. (setf (feed-period feed)
  42. (min *rss-max-period*
  43. (max *rss-min-period*
  44. (- p (* coeff diff)))))))
  45. (defun need-fetch-p (feed)
  46. (or (null (feed-next-fetch feed))
  47. (local-time:timestamp> (local-time:now) (feed-next-fetch feed))))
  48. (defun fetch-new-items (feed)
  49. (let ((items
  50. (loop for item in (parse-rss (feed-url feed))
  51. unless (member (feed-item-guid item) (feed-seen-guids feed) :test #'equal)
  52. do (pushnew (feed-item-guid item) (feed-seen-guids feed) :test #'equal)
  53. and collect item)))
  54. (adjust-period feed (if (consp items) (length items) -1))
  55. (setf (feed-next-fetch feed)
  56. (local-time:timestamp+ (local-time:now)
  57. (feed-period feed)
  58. :sec))
  59. items))
  60. (defun trim-nil (text)
  61. (when text
  62. (let ((text (string-trim " " text)))
  63. (unless (zerop (length text))
  64. text))))
  65. (defun text-with-cdata (node)
  66. "Compiles all text nodes within the nesting-node into one string."
  67. (with-output-to-string (stream)
  68. (labels ((r (node)
  69. (loop for child across (plump:children node)
  70. do (typecase child
  71. (plump:text-node (write-string (plump:text child) stream))
  72. (plump:cdata (write-string (plump:text child) stream))
  73. (plump:nesting-node (r child))))))
  74. (r node))))
  75. (defun child-text (node tag)
  76. (alexandria:when-let (child (car (get-by-tag node tag)))
  77. (trim-nil (text-with-cdata child))))
  78. (defun clean-text (text)
  79. (when text (trim-nil (plump:text (plump:parse text)))))
  80. (defun parse-rss (url)
  81. (let ((plump:*tag-dispatchers* plump:*xml-tags*))
  82. (loop for item in (get-by-tag (url-parse url) "item")
  83. collect (make-feed-item :guid (child-text item "guid")
  84. :link (child-text item "link")
  85. :title (clean-text (child-text item "title"))
  86. :description (clean-text (child-text item "description"))
  87. :published (local-time:universal-to-timestamp
  88. (date-time-parser:parse-date-time (child-text item "pubDate")))))))
  89. (defun format-feed-item (item)
  90. (format nil "~A~@[ @ ~A~]~@[~%~%~A~]~%~%~A"
  91. (feed-item-title item)
  92. (alexandria:when-let (ts (feed-item-published item))
  93. (local-time:format-timestring
  94. nil ts
  95. :format '((:year 2) "-" (:month 2) "-" (:day 2) " " (:hour 2) ":" (:min 2))))
  96. (feed-item-description item)
  97. (feed-item-link item)))