org-tag-eldoc-baidu-baike.el

   1 ;;; org-tag-eldoc-baidu-baike.el --- Web Scraping on Baidu Baike -*- 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: http://baike.baidu.com/search/none?word=[query]
   9 ;; tag URL: https://baike.baidu.com/item/[tag]
  10
  11 ;;; Code:
  12 \f
  13 (require 'url)
  14 (require 'url-http) ; for `url-http-end-of-headers'
  15 (require 'request)
  16 (require 'dom)
  17 (require 'elquery nil t) ; optionally use elquery.el to replace dom.el.
  18 (require 'org-tag-eldoc-common)
  19
  20
  21 (defun org-tag-eldoc-baidu-baike--request (tag)
  22   "Send HTTP GET request to get TAG explanation data."
  23   (let ((request-backend 'url-retrieve) ; use `url-retrieve' backend for proxy.
  24         (request-message-level -1)
  25         (url-proxy-services org-tag-eldoc-request-proxy)
  26         (url (url-encode-url ; encode Chinese query text into %X URL.
  27               (format "https://baike.baidu.com/item/%s" (capitalize tag)))))
  28     (request url
  29       :type "GET"
  30       :parser (lambda ()
  31                 (cond
  32                  ((fboundp 'libxml-parse-html-region) ; convert HTML -> Elisp alist structure
  33                   (libxml-parse-html-region (point-min) (point-max)))
  34                  ((featurep 'elquery) ; convert HTML -> elquery object structure
  35                   (elquery-read-buffer (current-buffer)))))
  36       :success (cl-function
  37                 (lambda (&key data &allow-other-keys)
  38                   ;; DEBUG: (setq request-result data)
  39                   (cond
  40                    ((featurep 'dom)
  41                     ;; for `libxml' parser
  42                     (setq org-tag-eldoc--explanation
  43                           (org-tag-eldoc-common--format-explanation
  44                            (dom-texts (dom-by-class data "lemmaSummary_wv32m\\ J-summary")))))
  45                    ((featurep 'elquery)
  46                     ;; for `elquery' parser
  47                     (setq org-tag-eldoc--explanation
  48                           (elquery-text (car (elquery-$ "#J-lemma-main-wrapper > div.contentWrapper_QxCiu > div > div.mainContent_E1dlD > div > div.lemmaSummary_wv32m.J-summary" data))))))))
  49       :error (cl-function
  50               (lambda (&rest args &key error-thrown &allow-other-keys)
  51                 ;; (message "[org-tag-eldoc] (百度百科) request error %s!" error-thrown)
  52                 nil))
  53       :status-code '((404 . (lambda (&rest _) nil))
  54                      (500 . (lambda (&rest _) nil))))
  55
  56     ;; (with-current-buffer (url-retrieve-synchronously url)
  57     ;;   (let ((dom (progn
  58     ;;                (goto-char url-http-end-of-headers)
  59     ;;                (libxml-parse-html-region (point) (point-max)))))
  60     ;;     ;; DEBUG: (setq request-result dom)
  61     ;;     (org-tag-eldoc-common--format-explanation
  62     ;;      (cond
  63     ;;       ((featurep 'dom)
  64     ;;        (dom-texts
  65     ;;         (dom-search (dom-by-class dom "mw-content-ltr\\ mw-parser-output")
  66     ;;                     (lambda (node) (and (eq (cl-first node) 'p) (null (cl-second node)))))))
  67     ;;       ((featurep 'elquery)
  68     ;;        (elquery-text (car (elquery-$ "#mw-content-text > div.mw-content-ltr.mw-parser-output > p" data))))))))
  69     ))
  70
  71 ;;; TEST: https://baike.baidu.com/item/Linux
  72 ;; (org-tag-eldoc-baidu-baike--request "Linux")
  73
  74 ;;; TEST
  75 ;; (let ((url-proxy-services org-tag-eldoc-request-proxy))
  76 ;;   (with-current-buffer
  77 ;;       (url-retrieve-synchronously "https://baike.baidu.com/item/Linux")
  78 ;;     (let ((dom (progn
  79 ;;                  (goto-char url-http-end-of-headers)
  80 ;;                  (libxml-parse-html-region (point) (point-max)))))
  81 ;;       (setq request-result dom)
  82 ;;       (setq org-tag-eldoc--explanation
  83 ;;             (dom-texts (dom-by-class dom "lemmaSummary_wv32m\\ J-summary"))))))
  84
  85 ;;; TEST
  86 ;; (dom-texts (dom-by-class request-result "lemmaSummary_wv32m\\ J-summary"))
  87
  88
  89 (defun org-tag-eldoc-baidu-baike-query (tag)
  90   "Query TAG on Baidu Baike then return a cons cell of tag and explanation."
  91   (if (stringp org-tag-eldoc--explanation)
  92       org-tag-eldoc--explanation
  93     (org-tag-eldoc-baidu-baike--request tag)
  94     (sit-for 1.0)
  95     (org-tag-eldoc-database-save tag org-tag-eldoc--explanation)))
  96
  97 \f
  98
  99 (provide 'org-tag-eldoc-baidu-baike)
 100
 101 ;;; org-tag-eldoc-baidu-baike.el ends here