From e3756891780a5a9bef4fcbc2169893df1b68bcac Mon Sep 17 00:00:00 2001
From: Alex Vandiver <github@chmrr.net>
Date: Tue, 16 Jan 2024 17:09:32 -0500
Subject: [PATCH] Improve support for null atttibute values (#406)

* img: Support null "src" attributes, rather than fail an assert.
* img: Support null "width" and "height" attributes, rather than assert.
* links: Support null "title" attribute with --reference-links.
---
 html2text/__init__.py             | 12 ++++--------
 test/empty-img-src.html           |  1 +
 test/empty-img-src.md             |  2 ++
 test/images_with_size.html        |  4 ++++
 test/images_with_size.md          |  5 ++++-
 test/no_inline_links_example.html |  1 +
 test/no_inline_links_example.md   |  4 +++-
 7 files changed, 19 insertions(+), 10 deletions(-)
 create mode 100644 test/empty-img-src.html
 create mode 100644 test/empty-img-src.md

diff --git a/html2text/__init__.py b/html2text/__init__.py
index ae4e154..9054388 100644
--- a/html2text/__init__.py
+++ b/html2text/__init__.py
@@ -537,8 +537,7 @@ def link_url(self: HTML2Text, link: str, title: str = "") -> None:
                             self.o("][" + str(a_props.count) + "]")
 
         if tag == "img" and start and not self.ignore_images:
-            if "src" in attrs:
-                assert attrs["src"] is not None
+            if "src" in attrs and attrs["src"] is not None:
                 if not self.images_to_alt:
                     attrs["href"] = attrs["src"]
                 alt = attrs.get("alt") or self.default_image_alt
@@ -549,11 +548,9 @@ def link_url(self: HTML2Text, link: str, title: str = "") -> None:
                     self.images_with_size and ("width" in attrs or "height" in attrs)
                 ):
                     self.o("<img src='" + attrs["src"] + "' ")
-                    if "width" in attrs:
-                        assert attrs["width"] is not None
+                    if "width" in attrs and attrs["width"] is not None:
                         self.o("width='" + attrs["width"] + "' ")
-                    if "height" in attrs:
-                        assert attrs["height"] is not None
+                    if "height" in attrs and attrs["height"] is not None:
                         self.o("height='" + attrs["height"] + "' ")
                     if alt:
                         self.o("alt='" + alt + "' ")
@@ -826,8 +823,7 @@ def o(
                             + "]: "
                             + urlparse.urljoin(self.baseurl, link.attrs["href"])
                         )
-                        if "title" in link.attrs:
-                            assert link.attrs["title"] is not None
+                        if "title" in link.attrs and link.attrs["title"] is not None:
                             self.out(" (" + link.attrs["title"] + ")")
                         self.out("\n")
                     else:
diff --git a/test/empty-img-src.html b/test/empty-img-src.html
new file mode 100644
index 0000000..865f94a
--- /dev/null
+++ b/test/empty-img-src.html
@@ -0,0 +1 @@
+<img alt src width="1234" height="auto">
diff --git a/test/empty-img-src.md b/test/empty-img-src.md
new file mode 100644
index 0000000..139597f
--- /dev/null
+++ b/test/empty-img-src.md
@@ -0,0 +1,2 @@
+
+
diff --git a/test/images_with_size.html b/test/images_with_size.html
index fcda9b6..3fccec1 100644
--- a/test/images_with_size.html
+++ b/test/images_with_size.html
@@ -8,3 +8,7 @@
 <img src='image_with_width_and_height.jpg' width='300' height='300' id='ignored-id' />
 <img src='image_with_width_and_height.jpg' id='ignored-id' />
 <img id='ignored-id' />
+
+<img src='image_with_no_width_value.jpg' width height='123' id='ignored-id' />
+<img src='image_with_no_height_value.jpg' width='123' height id='ignored-id' />
+<img src='image_with_no_dimention_values.jpg' width height id='ignored-id' />
diff --git a/test/images_with_size.md b/test/images_with_size.md
index c0548a9..7678ffa 100644
--- a/test/images_with_size.md
+++ b/test/images_with_size.md
@@ -3,4 +3,7 @@ src='image_with_width.jpg' width='300' alt='An image with a width attr' />
 <img src='image_with_width.jpg' height='300' alt='An image with a height attr'
 /> <img src='image_with_width_and_height.jpg' width='300' height='300' alt='An
 image with width and height' /> <img src='image_with_width_and_height.jpg'
-width='300' height='300' /> ![](image_with_width_and_height.jpg)
+width='300' height='300' /> ![](image_with_width_and_height.jpg) <img
+src='image_with_no_width_value.jpg' height='123' /> <img
+src='image_with_no_height_value.jpg' width='123' /> <img
+src='image_with_no_dimention_values.jpg' />
diff --git a/test/no_inline_links_example.html b/test/no_inline_links_example.html
index 5e4c45c..e7d428f 100644
--- a/test/no_inline_links_example.html
+++ b/test/no_inline_links_example.html
@@ -7,3 +7,4 @@
 <a href="http://example.com" title="abc">
 link text
 </a></a></a>
+<a href='http://example.com' title>Empty link title</a>
diff --git a/test/no_inline_links_example.md b/test/no_inline_links_example.md
index c000b59..bd11d62 100644
--- a/test/no_inline_links_example.md
+++ b/test/no_inline_links_example.md
@@ -1,8 +1,10 @@
 [Googler][1] No href No href but title available [ Example][2] [ [ [ link text
-][3]][3]][3]
+][3]][3]][3] [Empty link title][4]
 
    [1]: http://google.com
 
    [2]: http://example.com (Example title)
 
    [3]: http://example.com (abc)
+
+   [4]: http://example.com