rss.lisp 4.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118
  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 (puri:uri url)))
  9. (puri:render-uri
  10. (if (null (puri:uri-scheme uri))
  11. (puri:uri (format nil "http://~A" url))
  12. uri)
  13. nil)))
  14. (defun get-by-tag (node tag)
  15. (nreverse (plump::get-elements-by-tag-name node tag)))
  16. (defun url-parse (url)
  17. (multiple-value-bind (body-stream status headers uri stream)
  18. (drakma:http-request (http-default url)
  19. :force-binary t
  20. :want-stream t
  21. :decode-content t)
  22. (declare (ignore status headers stream))
  23. (values
  24. (plump:parse (flex:make-flexi-stream body-stream :external-format :utf-8))
  25. uri)))
  26. (defun find-rss-links (url)
  27. (handler-case
  28. (multiple-value-bind (root uri) (url-parse url)
  29. (loop for link in (get-by-tag root "link")
  30. when (string= "application/rss+xml" (plump:attribute link "type"))
  31. collect (list (plump:attribute link "title")
  32. (puri:render-uri
  33. (puri:merge-uris
  34. (puri:uri (plump:attribute link "href"))
  35. uri) nil))))
  36. (error (e) (log:error e))))
  37. (defun build-feed (url)
  38. (let ((root (url-parse url)))
  39. (alexandria:when-let (rss (car (get-by-tag root "rss")))
  40. (make-feed :url url :title (child-text rss "title")))))
  41. (defun adjust-period (feed coeff)
  42. "Adjust the period of feed based on whenever there were new items. With clamping"
  43. (let* ((p (feed-period feed))
  44. (diff (round (* p *rss-change-rate*))))
  45. (setf (feed-period feed)
  46. (min *rss-max-period*
  47. (max *rss-min-period*
  48. (- p (* coeff diff)))))))
  49. (defun need-fetch-p (feed)
  50. (or (null (feed-next-fetch feed))
  51. (local-time:timestamp> (local-time:now) (feed-next-fetch feed))))
  52. (defun fetch-new-items (feed)
  53. (let ((items
  54. (loop for item in (parse-rss (feed-url feed))
  55. unless (member (feed-item-guid item) (feed-seen-guids feed) :test #'equal)
  56. do (pushnew (feed-item-guid item) (feed-seen-guids feed) :test #'equal)
  57. and collect item)))
  58. (adjust-period feed (if (consp items) (length items) -1))
  59. (setf (feed-next-fetch feed)
  60. (local-time:timestamp+ (local-time:now)
  61. (feed-period feed)
  62. :sec))
  63. items))
  64. (defun trim-nil (text)
  65. (when text
  66. (let ((text (string-trim " " text)))
  67. (unless (zerop (length text))
  68. text))))
  69. (defun text-with-cdata (node)
  70. "Compiles all text nodes within the nesting-node into one string."
  71. (with-output-to-string (stream)
  72. (labels ((r (node)
  73. (loop for child across (plump:children node)
  74. do (typecase child
  75. (plump:text-node (write-string (plump:text child) stream))
  76. (plump:cdata (write-string (plump:text child) stream))
  77. (plump:nesting-node (r child))))))
  78. (r node))))
  79. (defun child-text (node tag)
  80. (alexandria:when-let (child (car (get-by-tag node tag)))
  81. (trim-nil (text-with-cdata child))))
  82. (defun clean-text (text)
  83. (when text (trim-nil (plump:text (plump:parse text)))))
  84. (defun parse-rss (url)
  85. (let ((plump:*tag-dispatchers* plump:*xml-tags*))
  86. (loop for item in (get-by-tag (url-parse url) "item")
  87. collect (make-feed-item :guid (child-text item "guid")
  88. :link (child-text item "link")
  89. :title (clean-text (child-text item "title"))
  90. :description (clean-text (child-text item "description"))
  91. :published (local-time:universal-to-timestamp
  92. (date-time-parser:parse-date-time (child-text item "pubDate")))))))
  93. (defun format-feed-item (feed item)
  94. (format nil "~A~@[ @ ~A~]~%~A~%~A"
  95. (feed-title feed)
  96. (alexandria:when-let (ts (feed-item-published item))
  97. (local-time:format-timestring
  98. nil ts
  99. :format '((:year 2) "-" (:month 2) "-" (:day 2)
  100. " " (:hour 2) ":" (:min 2))))
  101. (feed-item-title item)
  102. (feed-item-link item)))