Prevent PyHTML from rendering in quirks mode

COMP1010UNSW · Apr 27, 2024 · d2027e9 · d2027e9
1 parent 1b2e076
commit d2027e9
Show file tree

Hide file tree

Showing 12 changed files with 137 additions and 64 deletions.
diff --git a/README.md b/README.md
@@ -32,6 +32,7 @@ with improved documentation and type safety.
 ...     ),
 ... )
 >>> print(str(my_website))
+<!DOCTYPE html>
 <html>
   <head>
     <title>
@@ -159,6 +160,7 @@ instantiated element.
 >>> p.br
 <class 'pyhtml.__tags.generated.br'>
 >>> print(str(p.html(p.body(p.br))))
+<!DOCTYPE html>
 <html>
   <body>
     <br/>

diff --git a/meta/generate_tag_defs.py b/meta/generate_tag_defs.py
@@ -17,6 +17,11 @@ def _escape_children(self) -> bool:
         return False
 """
 
+GET_PRE_CONTENT = """
+    def _get_tag_pre_content(self) -> Optional[str]:
+        return {}
+"""
+
 
 def get_template_class(name: str):
     try:
@@ -96,6 +101,13 @@ def generate_tag_class(output: TextIO, tag: TagInfo):
     if not tag.escape_children:
         print(NO_ESCAPE_CHILDREN, file=output)
 
+    # Add pre-content declaration if needed
+    if tag.pre_content is not None:
+        print(
+            GET_PRE_CONTENT.replace("{}", f"'{tag.pre_content}'"),
+            file=output,
+        )
+
     # And a nice trailing newline to make flake8 happy
     print(file=output)
 

diff --git a/meta/scrape_tags.py b/meta/scrape_tags.py
@@ -136,6 +136,11 @@ class TagsYmlItem(TypedDict):
     escape_children: NotRequired[bool]
     """Whether to escape the contents of the tag (default True)"""
 
+    pre_content: NotRequired[str]
+    """
+    Pre-content for the element (eg `<!DOCTYPE html>`)
+    """
+
 
 TagsYaml = dict[str, TagsYmlItem]
 """Type alias for type of tags.yml file"""
@@ -204,6 +209,11 @@ class TagInfo:
     List of attributes and their documentation.
     """
 
+    pre_content: Optional[str]
+    """
+    Pre-content for the element (eg `<!DOCTYPE html>`)
+    """
+
 
 def fetch_mdn():
     """
@@ -449,6 +459,16 @@ def get_tag_escape_children(tags: TagsYaml, tag_name: str) -> bool:
     return tag.get('escape_children', True)
 
 
+def get_tag_pre_content(tags: TagsYaml, tag_name: str) -> Optional[str]:
+    """
+    Return pre-content for the tag
+    """
+    if tag_name not in tags:
+        return None
+    tag = tags[tag_name]
+    return tag.get('pre_content', None)
+
+
 def make_mdn_link(tag: str) -> str:
     """Generate an MDN docs link for the given tag"""
     return f"{MDN_ELEMENT_PAGE}/{tag}"
@@ -475,6 +495,7 @@ def elements_to_element_structs(
             mdn_link=make_mdn_link(name),
             escape_children=get_tag_escape_children(tag_attrs, name),
             attributes=attr_entries_to_object(tag_attrs, name),
+            pre_content=get_tag_pre_content(tag_attrs, name)
         ))
 
     return output

diff --git a/meta/tags.yml b/meta/tags.yml
@@ -1,6 +1,13 @@
 %YAML 1.2
 ---
 
+html:
+  pre_content: "<!DOCTYPE html>"
+  attributes:
+    lang:
+      doc: Language used by the document
+      type: Optional[str]
+
 # Rename <del> tag as it is a reserved keyword in Python
 del:
   rename: del_

diff --git a/pyhtml/__tag_base.py b/pyhtml/__tag_base.py
@@ -3,7 +3,7 @@
 
 Tag base class, including rendering logic
 """
-from typing import TypeVar
+from typing import Optional, TypeVar
 from . import __util as util
 from .__types import ChildrenType, AttributeType
 
@@ -74,6 +74,15 @@ def _get_default_attributes(
         """
         return {}
 
+    def _get_tag_pre_content(self) -> Optional[str]:
+        """
+        Return "pre-content" for the tag.
+
+        This is used by the `<html>` tag to add a `<!DOCTYPE html>` before the
+        tag.
+        """
+        return None
+
     def _escape_children(self) -> bool:
         """
         Returns whether the contents of the element should be escaped, or
@@ -97,6 +106,11 @@ def _render(self) -> list[str]:
 
         # Tag and attributes
         opening = f"<{self._get_tag_name()}"
+
+        # Add pre-content
+        if (pre := self._get_tag_pre_content()) is not None:
+            opening = f"{pre}\n{opening}"
+
         if len(attributes):
             opening += f" {util.render_tag_attributes(attributes)}>"
         else:

diff --git a/pyhtml/__tags/generated.py b/pyhtml/__tags/generated.py
@@ -15,48 +15,52 @@ class html(Tag):
     """
     Represents the root (top-level element) of an HTML document, so it is also referred to as the _root element_. All other elements must be descendants of this element.
 
-    
+    * `lang`: Language used by the document
 
     [View full documentation](https://developer.mozilla.org/en-US/docs/Web/HTML/Element/html)
     """
     def __init__(
         self,
         *children: ChildrenType,
-
+        lang: Optional[str] = None,
         **attributes: AttributeType,
     ) -> None:
         """
         Represents the root (top-level element) of an HTML document, so it is also referred to as the _root element_. All other elements must be descendants of this element.
 
-        
+        * `lang`: Language used by the document
 
         [View full documentation](https://developer.mozilla.org/en-US/docs/Web/HTML/Element/html)
         """
         attributes |= {
-
+            'lang': lang,
         }
         super().__init__(*children, **attributes)
 
     def __call__(  # type: ignore
         self,
         *children: ChildrenType,
-
+        lang: Optional[str] = None,
         **attributes: AttributeType,
     ):
         """
         Represents the root (top-level element) of an HTML document, so it is also referred to as the _root element_. All other elements must be descendants of this element.
 
-        
+        * `lang`: Language used by the document
 
         [View full documentation](https://developer.mozilla.org/en-US/docs/Web/HTML/Element/html)
         """
         attributes |= {
-
+            'lang': lang,
         }
         return super().__call__(*children, **attributes)
 
     def _get_default_attributes(self, given: dict[str, AttributeType]) -> dict[str, AttributeType]:
-        return {}
+        return {'lang': None}
+
+
+    def _get_tag_pre_content(self) -> Optional[str]:
+        return '<!DOCTYPE html>'
 
 
 class base(SelfClosingTag):