Extend images regex to also match the appendix after file extension

[Thaodan]

Some images have an appendix after the file extension making the builtin regex fail. An example would be something like https://example.com/foo.svg?size=800x600.

[wasamasa]

The original code tries to match several URLs in a buffer, with this change that no longer works:

(let ((haystack "foo http://xxx.jpg <https://bar.jpg?>http://baz.jpg")
      (needle "\\(https?://[^ ]*?\\.\\(?:png\\|jpg\\|jpeg\\|svg\\|gif\\).*\\)"))
  (with-temp-buffer
    (insert haystack)
    (goto-char (point-min))
    (while (re-search-forward needle nil t)
      (message "%S" (match-string-no-properties 1)))))
;; "http://xxx.jpg <https://bar.jpg?>http://baz.jpg"

Generally, .* patterns should be avoided whenever possible for this reason because they're only terminated by a newline. Instead, it should be delimited by URL delimiters (whitespace, double quote, angular brackets). The existing regex could be rewritten using rx if you prefer.

[Thaodan]

The original code tries to match several URLs in a buffer, with this change that no longer works:

(let ((haystack "foo http://xxx.jpg <https://bar.jpg?>http://baz.jpg")
      (needle "\\(https?://[^ ]*?\\.\\(?:png\\|jpg\\|jpeg\\|svg\\|gif\\).*\\)"))
  (with-temp-buffer
    (insert haystack)
    (goto-char (point-min))
    (while (re-search-forward needle nil t)
      (message "%S" (match-string-no-properties 1)))))
;; "http://xxx.jpg <https://bar.jpg?>http://baz.jpg"

Generally, .* patterns should be avoided whenever possible for this reason because they're only terminated by a newline.

That's what I was wondering to but I thought it only works per line so how can white space by an issue.

Instead, it should be delimited by URL delimiters (whitespace, double quote, angular brackets). The existing regex could be rewritten using rx if you prefer.

That sounds like it make sense.

[Thaodan]

Instead, it should be delimited by URL delimiters (whitespace, double quote, angular brackets). The existing regex could be rewritten using rx if you prefer. What is rx in this case?

[wasamasa]

S-expression notation for regular expressions. Try M-x find-library RET rx RET for more information.