diff options
| -rw-r--r-- | modules/elfeed-config.el | 30 | ||||
| -rw-r--r-- | tests/test-elfeed-config--decode-html-entities.el | 31 |
2 files changed, 43 insertions, 18 deletions
diff --git a/modules/elfeed-config.el b/modules/elfeed-config.el index ad7bda83a..7712f48db 100644 --- a/modules/elfeed-config.el +++ b/modules/elfeed-config.el @@ -126,23 +126,13 @@ Returns the stream URL or nil on failure." (cmd-args (append '("yt-dlp" "-q" "-g") format-args (list url))) - ;; DEBUG: Log the command - (_ (cj/log-silently "DEBUG: Extracting with command: %s" - (mapconcat #'shell-quote-argument cmd-args " "))) (output (with-temp-buffer (let ((exit-code (apply #'call-process (car cmd-args) nil t nil (cdr cmd-args)))) (if (zerop exit-code) (string-trim (buffer-string)) - (progn - ;; DEBUG: Log failure - (cj/log-silently "DEBUG: yt-dlp failed with exit code %d" exit-code) - (cj/log-silently "DEBUG: Error output: %s" (buffer-string)) - nil)))))) - ;; DEBUG: Log the result - (cj/log-silently "DEBUG: Extracted URL: %s" - (if output (truncate-string-to-width output 100) "nil")) + nil))))) (when (and output (string-match-p "^https?://" output)) output))) @@ -223,6 +213,15 @@ Note: Function name kept for backwards compatibility." "Seconds to wait for a synchronous YouTube page fetch before giving up. Without a timeout a hung request would block Emacs indefinitely.") +(defun cj/--decode-html-entities (text) + "Decode the common HTML entities in TEXT. +Handles & < > " ' and ' -- the entities YouTube's +og:title meta tag emits. Decoded left-to-right, & first." + (let ((entities '(("&" . "&") ("<" . "<") (">" . ">") + (""" . "\"") ("'" . "'") ("'" . "'")))) + (dolist (pair entities text) + (setq text (replace-regexp-in-string (car pair) (cdr pair) text))))) + (defun cj/youtube-to-elfeed-feed-format (url type) "Convert YouTube URL to elfeed-feeds format. @@ -274,13 +273,8 @@ TYPE should be either \='channel or \='playlist." (goto-char (point-min)) (when (re-search-forward "<meta property=\"og:title\" content=\"\\([^\"]+\\)\"" nil t) (setq title (match-string 1)) - ;; Simple HTML entity decoding - (setq title (replace-regexp-in-string "&" "&" title)) - (setq title (replace-regexp-in-string "<" "<" title)) - (setq title (replace-regexp-in-string ">" ">" title)) - (setq title (replace-regexp-in-string """ "\"" title)) - (setq title (replace-regexp-in-string "'" "'" title)) - (setq title (replace-regexp-in-string "'" "'" title)))))) + ;; Decode HTML entities in the extracted title + (setq title (cj/--decode-html-entities title)))))) ;; Always kill the temporary URL buffer, even when extraction failed -- ;; the old code only killed it when an ID was found, leaking it otherwise. (when (buffer-live-p buffer) diff --git a/tests/test-elfeed-config--decode-html-entities.el b/tests/test-elfeed-config--decode-html-entities.el new file mode 100644 index 000000000..a3fba3c49 --- /dev/null +++ b/tests/test-elfeed-config--decode-html-entities.el @@ -0,0 +1,31 @@ +;;; test-elfeed-config--decode-html-entities.el --- Tests for cj/--decode-html-entities -*- lexical-binding: t; -*- + +;;; Commentary: +;; cj/--decode-html-entities replaces the six inline replace-regexp-in-string +;; calls that cj/youtube-to-elfeed-feed-format used to hand-decode an og:title. + +;;; Code: + +(require 'ert) + +(add-to-list 'load-path (expand-file-name "modules" user-emacs-directory)) +(require 'elfeed-config) + +(ert-deftest test-elfeed-decode-html-entities-all () + "Normal: every supported entity is decoded." + (should (equal (cj/--decode-html-entities + "a & b <c> "d" 'e'") + "a & b <c> \"d\" 'e'"))) + +(ert-deftest test-elfeed-decode-html-entities-no-entities () + "Boundary: text without entities is unchanged." + (should (equal (cj/--decode-html-entities "plain title") "plain title")) + (should (equal (cj/--decode-html-entities "") ""))) + +(ert-deftest test-elfeed-decode-html-entities-amp-first () + "Boundary: & is decoded before the others (no double-decoding chains)." + (should (equal (cj/--decode-html-entities "Tom & Jerry <3") + "Tom & Jerry <3"))) + +(provide 'test-elfeed-config--decode-html-entities) +;;; test-elfeed-config--decode-html-entities.el ends here |