Re: [PATCH] org-protocol: decode "+" in query part as space (v2)

[Maxim Nikulin <manikulin@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1254 bytes --]

I have realized that only a half of new apostrophes in doc strings were 
properly escaped, so I am attaching updated patch. I still consider the 
change as a minor improvement.

On 06/04/2021 21:47, Maxim Nikulin wrote:
> 
> When I tried org-protocol for the first time, I was quite surprised that 
> query parameters generated using URLSearchParams JavaScript class are 
> decoded incorrectly. "+" characters representing space are passed as is. 
> I hope, I have found a proper place in the code to handle such case.
> 
> It does not affect previously recommended bookmarklets with abundant 
> calls of encodeURIComponent that escapes spaces as "%20". I believe, new 
> option is more readable:
> 
>      javascript:location.href='org-protocol://capture?' +
>          new URLSearchParams({
>              template: 'x',
>              url: location.href,
>              title: document.title,
>              body: window.getSelection()});
> 
> I guess, with old org-protocol syntax only pure percent encoding was 
> necessary due to each parameter was represented as path component. "+" 
> is allowed only in query part. Such variant was just missed when new 
> query-like syntax was introduced.


[-- Attachment #2: 0001-org-protocol.el-decode-in-query-part-as-space.patch --]
[-- Type: text/x-patch, Size: 7902 bytes --]

From 9bacd0761acd14fb5809042d96977bfd3088bee3 Mon Sep 17 00:00:00 2001
From: Max Nikulin <manikulin@gmail.com>
Date: Tue, 6 Apr 2021 21:30:06 +0700
Subject: [PATCH] org-protocol.el: decode "+" in query part as space

* lisp/org-protocol.el (org-protocol-convert-query-to-plist):
Replace "+" chars by spaces before passing parameter string
to decoder.  Allow making org-protocol URIs with help of URLSearchParams
JavaScript class.
* lisp/org-protocol.el doc/org-manual.org etc/ORG-NEWS: Add examples
demonstrating new opportunity for browser bookmarklets.

Make parsing of URI parameters a bit closer to URL standard
https://url.spec.whatwg.org/#urlencoded-parsing
---
 doc/org-manual.org   | 22 ++++++++++++++++++++
 etc/ORG-NEWS         | 11 ++++++++++
 lisp/org-protocol.el | 48 ++++++++++++++++++++++++++++++++++++++------
 3 files changed, 75 insertions(+), 6 deletions(-)

diff --git a/doc/org-manual.org b/doc/org-manual.org
index b547a8a5f..d91aa7a50 100644
--- a/doc/org-manual.org
+++ b/doc/org-manual.org
@@ -19596,11 +19596,20 @@ slashes, and probably quote those for the shell.
 To use this feature from a browser, add a bookmark with an arbitrary
 name, e.g., =Org: store-link= and enter this as /Location/:
 
+#+begin_example
+javascript:location.href='org-protocol://store-link?' +
+      new URLSearchParams({url:location.href, title:document.title});
+#+end_example
+
+Title is an optional parameter.  Another expression was recommended earlier:
+
 #+begin_example
 javascript:location.href='org-protocol://store-link?url='+
       encodeURIComponent(location.href);
 #+end_example
 
+The latter form is compatible with older Org versions from 9.0 to 9.4.
+
 *** The ~capture~ protocol
 :PROPERTIES:
 :DESCRIPTION: Fill a buffer with external information.
@@ -19616,6 +19625,15 @@ using acapture template.
 To use this feature, add a bookmark with an arbitrary name, e.g.,
 =Org: capture=, and enter this as =Location=:
 
+#+begin_example
+javascript:location.href='org-protocol://capture?' +
+      new URLSearchParams({
+            template: 'x', url: window.location.href,
+            title: document.title, body: window.getSelection()});
+#+end_example
+
+You might have seen another expression:
+
 #+begin_example
 javascript:location.href='org-protocol://capture?template=x'+
       '&url='+encodeURIComponent(window.location.href)+
@@ -19623,6 +19641,10 @@ javascript:location.href='org-protocol://capture?template=x'+
       '&body='+encodeURIComponent(window.getSelection());
 #+end_example
 
+It is a bit more cluttered than the former one, but it is compatible with
+previous Org versions 9.0-9.4. In these versions encoding of space as "+"
+character was not supported by URI decoder.
+
 #+vindex: org-protocol-default-template-key
 The capture template to be used can be specified in the bookmark (like
 =X= above).  If unspecified, the template key is set in the variable
diff --git a/etc/ORG-NEWS b/etc/ORG-NEWS
index 83dca5c03..724adc486 100644
--- a/etc/ORG-NEWS
+++ b/etc/ORG-NEWS
@@ -269,6 +269,17 @@ back to the filename as before.
 
 The function does not allow for a third optional parameter anymore.
 
+*** Make org-protocol compatible with =URLSearchParams= JavaScript class
+
+Decoder of query part of org-protocol URI recognizes "+" as an encoded
+space characters now, so it is possible to avoid call to =encodeURIComponent=
+for each parameter and use more readable expression in bookmarklet:
+
+#+begin_example
+'org-protocol://store-link?' + new URLSearchParams({
+      url: location.href, title: document.title})
+#+end_example
+
 * Version 9.4
 ** Incompatible changes
 *** Possibly broken internal file links: please check and fix
diff --git a/lisp/org-protocol.el b/lisp/org-protocol.el
index 30f5bc040..a09d7fbed 100644
--- a/lisp/org-protocol.el
+++ b/lisp/org-protocol.el
@@ -94,6 +94,15 @@
 ;; You may use the same bookmark URL for all those standard handlers and just
 ;; adjust the sub-protocol used:
 ;;
+;;     javascript:location.href='org-protocol://sub-protocol?'+
+;;           new URLSearchParams({
+;;                 url: location.href,
+;;                 title: document.title,
+;;                 body: window.getSelection()})
+;;
+;; Alternatively use the following expression that encodes space as \"%20\"
+;; instead of \"+\", so it is compatible with Org versions from 9.0 to 9.4:
+;;
 ;;     location.href='org-protocol://sub-protocol?url='+
 ;;           encodeURIComponent(location.href)+'&title='+
 ;;           encodeURIComponent(document.title)+'&body='+
@@ -103,6 +112,11 @@
 ;; char that, if present, triggers the use of a special template.
 ;; Example:
 ;;
+;;     location.href='org-protocol://capture?'+
+;;           new URLSearchParams({template:'x', /* ... */})
+;;
+;; or
+;;
 ;;     location.href='org-protocol://capture?template=x'+ ...
 ;;
 ;;  uses template ?x.
@@ -426,7 +440,12 @@ Parameters: url, title (optional), body (optional)
 Old-style links such as org-protocol://store-link://URL/TITLE are
 also recognized.
 
-The location for a browser's bookmark has to look like this:
+The location for a browser's bookmark may look like this:
+
+  javascript:location.href = \\='org-protocol://store-link?\\=' +
+       new URLSearchParams({url:location.href, title:document.title});
+
+or to keep compatibility with Org versions from 9.0 to 9.4 it may be:
 
   javascript:location.href = \\
       \\='org-protocol://store-link?url=\\=' + \\
@@ -435,7 +454,9 @@ The location for a browser's bookmark has to look like this:
 
 Don't use `escape()'!  Use `encodeURIComponent()' instead.  The
 title of the page could contain slashes and the location
-definitely will.
+definitely will.  Org 9.4 and earlier could not decode \"+\"
+to space, that is why less readable latter expression may be necessary
+for backward compatibility.
 
 The sub-protocol used to reach this function is set in
 `org-protocol-protocol-alist'.
@@ -463,6 +484,14 @@ The sub-protocol used to reach this function is set in
 This function detects an URL, title and optional text, separated
 by `/'.  The location for a browser's bookmark looks like this:
 
+  javascript:location.href = \\='org-protocol://capture?\\=' +
+        new URLSearchParams({
+              url: location.href,
+              title: document.title,
+              body: window.getSelection()})
+
+or to keep compatibility with Org versions from 9.0 to 9.4:
+
   javascript:location.href = \\='org-protocol://capture?url=\\='+ \\
         encodeURIComponent(location.href) + \\='&title=\\=' + \\
         encodeURIComponent(document.title) + \\='&body=\\=' + \\
@@ -518,10 +547,11 @@ Now template ?b will be used."
 (defun org-protocol-convert-query-to-plist (query)
   "Convert QUERY key=value pairs in the URL to a property list."
   (when query
-    (apply 'append (mapcar (lambda (x)
-			     (let ((c (split-string x "=")))
-			       (list (intern (concat ":" (car c))) (cadr c))))
-			   (split-string query "&")))))
+    (let ((plus-decoded (replace-regexp-in-string "\\+" " " query t t)))
+      (apply 'append (mapcar (lambda (x)
+			       (let ((c (split-string x "=")))
+				 (list (intern (concat ":" (car c))) (cadr c))))
+			     (split-string plus-decoded "&"))))))
 
 (defun org-protocol-open-source (fname)
   "Process an org-protocol://open-source?url= style URL with FNAME.
@@ -531,6 +561,12 @@ in `org-protocol-project-alist'.
 
 The location for a browser's bookmark should look like this:
 
+  javascript:location.href = \\='org-protocol://open-source?\\=' +
+        new URLSearchParams({url: location.href})
+
+or if you prefer to keep compatibility with older Org versions (9.0 to 9.4),
+consider the following expression:
+
   javascript:location.href = \\='org-protocol://open-source?url=\\=' + \\
         encodeURIComponent(location.href)"
   ;; As we enter this function for a match on our protocol, the return value
-- 
2.25.1