From f0751ec5dd938e2b8c93bcc9030b26a6f62e4fce Mon Sep 17 00:00:00 2001
From: Yohan Boniface <yohanboniface@free.fr>
Date: Wed, 29 Jan 2025 20:23:43 +0100
Subject: [PATCH] fix: correctly parse http link including another http link in
 the path

fix #2457
---
 umap/static/umap/js/modules/utils.js | 17 +++++++----------
 umap/static/umap/unittests/utils.js  | 14 ++++++++++++++
 2 files changed, 21 insertions(+), 10 deletions(-)

diff --git a/umap/static/umap/js/modules/utils.js b/umap/static/umap/js/modules/utils.js
index 3f9fdf0e..228d94c1 100644
--- a/umap/static/umap/js/modules/utils.js
+++ b/umap/static/umap/js/modules/utils.js
@@ -159,16 +159,13 @@ export function toHTML(r, options) {
   r = r.replace(/\*(.*?)\*/g, '<em>$1</em>')
 
   // links
-  r = r.replace(/(\[\[http)/g, '[[h_t_t_p') // Escape for avoiding clash between [[http://xxx]] and http://xxx
-  r = r.replace(/({{http)/g, '{{h_t_t_p')
-  r = r.replace(/(=http)/g, '=h_t_t_p') // http://xxx as query string, see https://github.com/umap-project/umap/issues/607
-  r = r.replace(/(https?:[^ \<)\n]*)/g, `<a target="_${target}" href="$1">$1</a>`)
   r = r.replace(
-    /\[\[(h_t_t_ps?:[^\]|]*?)\]\]/g,
-    `<a target="_${target}" href="$1">$1</a>`
+    /(^|\s|>|\()(https?:[^ \<)\n]*)/g,
+    `$1<a target="_${target}" href="$2">$2</a>`
   )
+  r = r.replace(/\[\[(https?:[^\]|]*?)\]\]/g, `<a target="_${target}" href="$1">$1</a>`)
   r = r.replace(
-    /\[\[(h_t_t_ps?:[^|]*?)\|(.*?)\]\]/g,
+    /\[\[(https?:[^|]*?)\|(.*?)\]\]/g,
     `<a target="_${target}" href="$1">$2</a>`
   )
   r = r.replace(/\[\[([^\]|]*?)\]\]/g, `<a target="_${target}" href="$1">$1</a>`)
@@ -176,15 +173,15 @@ export function toHTML(r, options) {
 
   // iframe
   r = r.replace(
-    /{{{(h_t_t_ps?[^|{]*)}}}/g,
+    /{{{(https?[^|{]*)}}}/g,
     '<div><iframe frameborder="0" src="$1" width="100%" height="300px"></iframe></div>'
   )
   r = r.replace(
-    /{{{(h_t_t_ps?[^|{]*)\|(\d*)(px)?}}}/g,
+    /{{{(https?[^|{]*)\|(\d*)(px)?}}}/g,
     '<div><iframe frameborder="0" src="$1" width="100%" height="$2px"></iframe></div>'
   )
   r = r.replace(
-    /{{{(h_t_t_ps?[^|{]*)\|(\d*)(px)?\*(\d*)(px)?}}}/g,
+    /{{{(https?[^|{]*)\|(\d*)(px)?\*(\d*)(px)?}}}/g,
     '<div><iframe frameborder="0" src="$1" width="$4px" height="$2px"></iframe></div>'
   )
 
diff --git a/umap/static/umap/unittests/utils.js b/umap/static/umap/unittests/utils.js
index 8d097f07..fb06900b 100644
--- a/umap/static/umap/unittests/utils.js
+++ b/umap/static/umap/unittests/utils.js
@@ -54,6 +54,13 @@ describe('Utils', () => {
       )
     })
 
+    it('should handle links with another url in path', () => {
+      assert.equal(
+        Utils.toHTML('A simple https://osm.org/https://anotherurl.com link'),
+        'A simple <a href="https://osm.org/https://anotherurl.com" target="_blank">https://osm.org/https://anotherurl.com</a> link'
+      )
+    })
+
     it('should handle simple link inside parenthesis', () => {
       assert.equal(
         Utils.toHTML('A simple link (http://osm.org)'),
@@ -172,6 +179,13 @@ describe('Utils', () => {
       )
     })
 
+    it('http link with http link as in path', () => {
+      assert.equal(
+        Utils.toHTML('A phrase with a [[https://iframeurl.com/https://another.com]].'),
+        'A phrase with a <a href="https://iframeurl.com/https://another.com" target="_blank">https://iframeurl.com/https://another.com</a>.'
+      )
+    })
+
     it('simple bullet points', () => {
       assert.equal(
         Utils.toHTML('* First point\n* Second point\n* Last point'),