Quote URLs

7 files changed, 51 insertions, 13 deletions

diff --git a/README.creole b/README.creole
index b6c2dbb..658e3c4 100644
--- a/README.creole
+++ b/README.creole
@@ -221,9 +221,9 @@ Note: In this case you must install **docutils**! See above.
 
 = history =
 
-* [[https://github.com/jedie/python-creole/compare/v1.5.0.rc2...master|*dev*]]
+* [[https://github.com/jedie/python-creole/compare/v1.5.0.rc3...master|*dev*]]
 ** TBC
-* [[https://github.com/jedie/python-creole/compare/v1.4.10...v1.5.0.rc2|v1.5.0.rc2 - 2022-08-20]]
+* [[https://github.com/jedie/python-creole/compare/v1.4.10...v1.5.0.rc3|v1.5.0.rc3 - 2022-08-20]]
 ** NEW: html2markdown
 ** creole2html bugfixes:
 *** replace wrong {{{<tt>}}} with {{{<code>}}}
diff --git a/README.md b/README.md
index 81c75ed..c4bfc77 100644
--- a/README.md
+++ b/README.md
@@ -226,9 +226,9 @@ Note: In this case you must install **docutils**! See above.
 # history
 
 
-* [*dev*](https://github.com/jedie/python-creole/compare/v1.5.0.rc2...master)
+* [*dev*](https://github.com/jedie/python-creole/compare/v1.5.0.rc3...master)
   * TBC
-* [v1.5.0.rc2 - 2022-08-20](https://github.com/jedie/python-creole/compare/v1.4.10...v1.5.0.rc2)
+* [v1.5.0.rc3 - 2022-08-20](https://github.com/jedie/python-creole/compare/v1.4.10...v1.5.0.rc3)
   * NEW: html2markdown
   * creole2html bugfixes:
     * replace wrong `<tt>` with `<code>`
@@ -438,4 +438,4 @@ first source code was written 27.11.2008: [Forum thread (de)](http://www.python-
 
 ------------
 
-``Note: this file is generated from README.creole 2022-08-21 00:49:59 with "python-creole"``
\ No newline at end of file
+``Note: this file is generated from README.creole 2022-08-23 07:53:38 with "python-creole"``
\ No newline at end of file
diff --git a/README.rst b/README.rst
index 5e6ded5..3991f0c 100644
--- a/README.rst
+++ b/README.rst
@@ -291,11 +291,11 @@ Note: In this case you must install **docutils**! See above.
 history
 =======
 
-* `*dev* <https://github.com/jedie/python-creole/compare/v1.5.0.rc2...master>`_ 
+* `*dev* <https://github.com/jedie/python-creole/compare/v1.5.0.rc3...master>`_ 
 
     * TBC
 
-* `v1.5.0.rc2 - 2022-08-20 <https://github.com/jedie/python-creole/compare/v1.4.10...v1.5.0.rc2>`_ 
+* `v1.5.0.rc3 - 2022-08-20 <https://github.com/jedie/python-creole/compare/v1.4.10...v1.5.0.rc3>`_ 
 
     * NEW: html2markdown
 
@@ -708,4 +708,4 @@ donation
 
 ------------
 
-``Note: this file is generated from README.creole 2022-08-21 00:49:59 with "python-creole"``
\ No newline at end of file
+``Note: this file is generated from README.creole 2022-08-23 07:53:38 with "python-creole"``
\ No newline at end of file
diff --git a/creole/__init__.py b/creole/__init__.py
index f3b056a..5d394f7 100644
--- a/creole/__init__.py
+++ b/creole/__init__.py
@@ -24,7 +24,7 @@ from creole.parser.creol2html_parser import CreoleParser
 from creole.parser.html_parser import HtmlParser
 
 
-__version__ = "1.5.0.rc2"
+__version__ = "1.5.0.rc3"
 __api__ = "1.0"  # Creole 1.0 spec - http://wikicreole.org/
 
 VERSION_STRING = __version__  # remove in future
diff --git a/creole/emitter/html2markdown_emitter.py b/creole/emitter/html2markdown_emitter.py
index 848add0..5e6140a 100644
--- a/creole/emitter/html2markdown_emitter.py
+++ b/creole/emitter/html2markdown_emitter.py
@@ -8,7 +8,7 @@
     :license: GNU GPL v3 or above, see LICENSE for more details.
 
 """
-
+from urllib.parse import ParseResult, quote, urlparse, urlunparse
 
 from creole.parser.html_parser import HtmlParser
 from creole.shared.base_emitter import BaseEmitter
@@ -16,6 +16,19 @@ from creole.shared.document_tree import DocNode
 from creole.shared.markup_table import MarkupTable
 
 
+def quote_link(uri):
+    """
+    >>> quote_link('http://foo.tld/a image with spaces.png')
+    'http://foo.tld/a%20image%20with%20spaces.png'
+
+    >>> quote_link('https://foo.tld/a image.png?bar=1#anchor')
+    'https://foo.tld/a%20image.png?bar=1#anchor'
+    """
+    scheme, netloc, url, params, query, fragment = urlparse(uri)
+    url = quote(url)
+    return urlunparse(ParseResult(scheme, netloc, url, params, query, fragment))
+
+
 class MarkdownEmitter(BaseEmitter):
     """
     Build from a document_tree (html2creole.parser.HtmlParser instance) a
@@ -133,7 +146,9 @@ class MarkdownEmitter(BaseEmitter):
 
     def a_emit(self, node: DocNode):
         link_text = self.emit_children(node)
-        url = node.attrs['href']
+
+        url = quote_link(node.attrs['href'])
+
         title = node.attrs.get('title')
         if title:
             return f'[{link_text}]({url} "{title}")'
@@ -141,7 +156,7 @@ class MarkdownEmitter(BaseEmitter):
             return f'[{link_text}]({url})'
 
     def img_emit(self, node: DocNode):
-        src = node.attrs['src']
+        src = quote_link(node.attrs['src'])
 
         title = node.attrs.get('title')
         alt = node.attrs.get('alt', '')
diff --git a/creole/tests/test_html2markdown.py b/creole/tests/test_html2markdown.py
index 2651339..90d4daf 100644
--- a/creole/tests/test_html2markdown.py
+++ b/creole/tests/test_html2markdown.py
@@ -163,3 +163,26 @@ class MarkdownTests(BaseCreoleTest):
             ),
             # debug=True,
         )
+    def test_links_with_spaces(self):
+        self.assert_html2markdown(
+            markdown_string=cleandoc(
+                '''
+                [one](/foo%20bar.png)
+
+                [/somewhere/foo bar.exe](https://somewhere/foo%20bar.exe?bar=1#anchor "Foo Bar")
+
+                ![Alt text](https://foo.tld/a%20image.png?bar=1#anchor)
+                '''
+            ),
+            html_string=cleandoc(
+                '''
+                <p><a href="/foo bar.png">one</a></p>
+                <p>
+                    <a href="https://somewhere/foo bar.exe?bar=1#anchor" title="Foo Bar">
+                    /somewhere/foo bar.exe
+                    </a>
+                </p>
+                <p><img alt="Alt text" src="https://foo.tld/a image.png?bar=1#anchor" /></p>
+                '''
+            ),
+        )
diff --git a/pyproject.toml b/pyproject.toml
index 1764142..166a480 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "python-creole"
-version = "1.5.0.rc2"
+version = "1.5.0.rc3"
 description = "python-creole is an open-source (GPL) markup converter in pure Python for: creole2html, html2creole, html2ReSt, html2textile"
 
 # Will be generated from README.creole with: "poetry run update_rst_readme"