Support validating anchors containing emojis (#67)

.gitignore

@@ -43,3 +43,6 @@ test/integration/site
 
 # PyCharm
 .idea
+
+# Project specfic
+tests/integration/site

htmlproofer/plugin.py

@@ -33,6 +33,13 @@
 ATTRLIST_ANCHOR_PATTERN = re.compile(r'\{.*?\#([^\s\}]*).*?\}')
 ATTRLIST_PATTERN = re.compile(r'\{.*?\}')
 
+# Example emojis:
+#   :banana:
+#   :smiley_cat:
+#   :octicons-apps-16:
+#   :material-star:
+EMOJI_PATTERN = re.compile(r'\:[a-z0-9_-]+\:')
+
 urllib3.disable_warnings()
 
 
@@ -225,7 +232,7 @@ def contains_anchor(markdown: str, anchor: str) -> bool:
                 # # Heading {: #testanchor .testclass }
                 # # Heading {.testclass #testanchor}
                 # # Heading {.testclass}
-                # these can override the headings anchor id, or alternativly just provide additional class etc.
+                # these can override the headings anchor id, or alternatively just provide additional class etc.
                 attr_list_anchor_match = ATTRLIST_ANCHOR_PATTERN.match(heading)
                 if attr_list_anchor_match is not None:
                     attr_list_anchor = heading_match.groups()[1]
@@ -239,6 +246,10 @@ def contains_anchor(markdown: str, anchor: str) -> bool:
                 # But these images are not included in the generated anchor, so remove them.
                 heading = re.sub(IMAGE_PATTERN, '', heading)
 
+                # Headings are allowed to have emojis in them under certain Mkdocs themes.
+                # https://squidfunk.github.io/mkdocs-material/setup/extensions/python-markdown-extensions/#emoji
+                heading = re.sub(EMOJI_PATTERN, '', heading)
+
                 anchor_slug = slugify(heading, '-')
                 if anchor == anchor_slug:
                     return True

tests/integration/docs/nested/page1.md

@@ -21,8 +21,9 @@ or to a nested page
 [Invalid Anchor](./page2.md#BAD_ANCHOR).
 
 But allows valid anchors such as
-[Main Page](../index.md#mkdocs-htmlproofer-plugin) and
-[Table of Contents](../index.md#table-of-contents).
+[Main Page](../index.md#mkdocs-htmlproofer-plugin),
+[Table of Contents](../index.md#table-of-contents), and
+[Emoji Anchor](./page2.md#title-with-emojis).
 
 ## Image Link absolute/relative
 

tests/integration/docs/nested/page2.md

@@ -1 +1,3 @@
 # Second Nested Test Page
+
+## :smile_cat: Title with Emojis :material-star: :octicons-apps-16: