org-tag-eldoc-wikipedia.el

   1 ;;; org-tag-eldoc-wikipedia.el --- Web Scraping on Wikipedia -*- lexical-binding: t; -*-
   2 ;; -*- coding: utf-8 -*-
   3
   4 ;; Copyright (C) 2024-2025 Christopher M. Miles, all rights reserved.
   5
   6 ;;; Commentary:
   7
   8 ;; search API: https://wikipedia.org/w/index.php?search=[query]
   9 ;; tag URL: https://en.wikipedia.org/wiki/Computer_science
  10
  11 ;;; Code:
  12 \f
  13 (require 'request)
  14 (require 'dom)
  15 (require 'elquery nil t) ; optionally use elquery.el to replace dom.el.
  16 (require 'org-tag-eldoc-common)
  17
  18
  19 (defun org-tag-eldoc-wikipedia--request (tag)
  20   "Send HTTP GET request to get TAG explanation data."
  21   (let ((request-backend 'url-retrieve) ; use `url-retrieve' backend for proxy.
  22         (url-proxy-services org-tag-eldoc-request-proxy)
  23         (url (format "https://en.wikipedia.org/wiki/%s" (capitalize tag)))
  24         (summary))
  25     (cond
  26      ((featurep 'dom)
  27       (request url
  28         :type "GET"
  29         :parser (lambda () (libxml-parse-html-region (point-min) (point-max))) ; convert HTML -> Elisp alist structure
  30         :success (cl-function
  31                   (lambda (&key data &allow-other-keys)
  32                     ;; DEBUG: (setq request-result data)
  33                     ;; for `libxml' parser
  34                     (setq org-tag-eldoc--explanation
  35                           (org-tag-eldoc-common--format-explanation
  36                            (dom-texts (dom-by-class data "mw-content-container"))))))
  37         :error (cl-function
  38                 (lambda (&rest args &key error-thrown &allow-other-keys)
  39                   (error "request error!")))
  40         :status-code '((404 . (lambda (&rest _) (message (format "Endpoint %s does not exist." url))))
  41                        (500 . (lambda (&rest _) (message (format "Error from %s." url)))))))
  42      ((featurep 'elquery)
  43       (request url
  44         :type "GET"
  45         :parser (lambda () (elquery-read-buffer (current-buffer))) ; convert HTML -> elquery object structure
  46         :success (cl-function
  47                   (lambda (&key data &allow-other-keys)
  48                     ;; DEBUG: (setq request-result data)
  49                     ;; for `elquery' parser
  50                     (setq org-tag-eldoc--explanation
  51                           (org-tag-eldoc-common--format-explanation
  52                            (elquery-text (car (elquery-$ ".mw-content-container" data)))))))
  53         :error (cl-function
  54                 (lambda (&rest args &key error-thrown &allow-other-keys)
  55                   (error "[org-tag-eldoc] request error!")
  56                   "request error" ; return identity string for `org-tag-eldoc-tag-explanation''s `seq-some' -> `pcase'.
  57                   ))
  58         :status-code '((404 . (lambda (&rest _) (message (format "Endpoint %s does not exist." url))))
  59                        (500 . (lambda (&rest _) (message (format "Error from %s." url))))))))))
  60
  61 ;;; TEST: https://en.wikipedia.org/wiki/Computer_science
  62 ;; (org-tag-eldoc-wikipedia--request "computer_science")
  63
  64 (defun org-tag-eldoc-wikipedia-query (tag)
  65   "Query TAG on Wikipedia then return a cons cell of tag and explanation."
  66   (org-tag-eldoc-database-query tag)
  67   (if (stringp org-tag-eldoc--explanation)
  68       org-tag-eldoc--explanation
  69     (org-tag-eldoc-wikipedia--request tag)
  70     (sit-for 1.0)
  71     (org-tag-eldoc-database-save tag org-tag-eldoc--explanation)))
  72
  73 \f
  74
  75 (provide 'org-tag-eldoc-wikipedia)
  76
  77 ;;; org-tag-eldoc-wikipedia.el ends here