Merge branch 'mistune-2.0.2' into 'main'

Port to Mistune 2.0.2

See merge request mx-moment/moment!16

requirements.txt

@@ -5,7 +5,7 @@ cairosvg          >= 2.4.2,  < 3
 filetype          >= 1.0.7,  < 2
 html_sanitizer    >= 1.9.1,  < 2
 lxml              >= 4.5.1,  < 5
-mistune           >= 0.8.4,  < 0.9
+mistune           >= 2.0.0,  < 3.0
 pymediainfo       >= 4.2.1,  < 5
 plyer             >= 1.4.3,  < 2
 sortedcontainers  >= 2.2.2,  < 3

src/backend/html_markdown.py

@@ -18,84 +18,32 @@ from lxml.html import HtmlElement, etree  # nosec
 from .color import SVGColor
 
 
-class MarkdownInlineGrammar(mistune.InlineGrammar):
-    """Markdown inline elements syntax modifications for the Mistune parser.
+def parse_colour(inline, m, state):
+    colour = m.group(1)
+    text = m.group(2)
+    return "colour", colour, text
 
-    - Disable underscores for bold/italics (e.g. `__bold__`)
 
-    - Add syntax for coloring text: `<color>(text)`,
-      e.g. `<red>(Lorem ipsum)` or `<#000040>(sit dolor amet...)`
-    """
+def render_html_colour(colour, text):
+    return f'<span data-mx-color="{colour}">{text}</span>'
 
-    escape          = re.compile(r"^\\([\\`*{}\[\]()#+\-.!_<>~|])")  # Add <
-    emphasis        = re.compile(r"^\*((?:\*\*|[^\*])+?)\*(?!\*)")
-    double_emphasis = re.compile(r"^\*{2}([\s\S]+?)\*{2}(?!\*)")
 
+def plugin_matrix(md):
     # test string: r"<b>(x) <r>(x) \<a>b>(x) <a\>b>(x) <b>(\(z) <c>(foo\)xyz)"
-    color = re.compile(
-        r"^<(.+?)>"          # capture the color in `<color>`
+    colour = (
+        r"^<(.+?)>"          # capture the colour in `<colour>`
         r"\((.+?)"           # capture text in `(text`
         r"(?<!\\)(?:\\\\)*"  # ignore the next `)` if it's \escaped
-        r"\)",               # finish on a `)`
+        r"\)"                # finish on a `)`
     )
 
+    # Mark colour as high priority as otherwise e.g. <red>(hi) matches the
+    # inline_html rule instead of the colour rule.
+    md.inline.rules.insert(1, "colour")
+    md.inline.register_rule("colour", colour, parse_colour)
 
-class MarkdownBlockGrammar(mistune.BlockGrammar):
-    """Markdown block elements syntax modifications for the Mistune parser.
-
-    - Require a space after # for titles
-    """
-
-    heading = re.compile(r"^ *(#{1,6}) +([^\n]+?) *#* *(?:\n+|$)")
-
-
-class MarkdownInlineLexer(mistune.InlineLexer):
-    """Apply the changes from `MarkdownInlineGrammar` for Mistune."""
-
-    grammar_class = MarkdownInlineGrammar
-
-    default_rules = [
-        "escape", "color", "autolink", "url",  # Add color
-        "footnote", "link", "reflink", "nolink",
-        "double_emphasis", "emphasis", "code",
-        "linebreak", "strikethrough", "text",
-    ]
-    inline_html_rules = [
-        "escape", "color", "autolink", "url", "link", "reflink",  # Add color
-        "nolink", "double_emphasis", "emphasis", "code",
-        "linebreak", "strikethrough", "text",
-    ]
-
-
-    def output_double_emphasis(self, m):
-        return self.renderer.double_emphasis(self.output(m.group(1)))
-
-
-    def output_emphasis(self, m):
-        return self.renderer.emphasis(self.output(m.group(1)))
-
-
-    def output_color(self, m):
-        color = m.group(1)
-        text  = self.output(m.group(2))
-        return self.renderer.color(color, text)
-
-
-class MarkdownBlockLexer(mistune.BlockLexer):
-    """Apply the changes from `MarkdownBlockGrammar` for Mistune."""
-
-    grammar_class = MarkdownBlockGrammar
-
-
-class MarkdownRenderer(mistune.Renderer):
-    def color(self, color: str, text: str) -> str:
-        """Render given text with a color using `<span data-mx-color=#hex>`."""
-
-        # This may be a SVG color name, try to get a #hex code from it:
-        with suppress(KeyError):
-            color = SVGColor[color.lower().replace(" ", "")].value.hex
-
-        return f'<span data-mx-color="{color}">{text}</span>'
+    if md.renderer.NAME == "html":
+        md.renderer.register("colour", render_html_colour)
 
 
 class HTMLProcessor:
@@ -201,19 +149,13 @@ class HTMLProcessor:
 
         # hard_wrap: convert all \n to <br> without required two spaces
         # escape: escape HTML characters in the input string, e.g. tags
-        self._markdown_to_html = mistune.Markdown(
+        self._markdown_to_html = mistune.create_markdown(
             hard_wrap = True,
             escape    = True,
-            inline    = MarkdownInlineLexer,
-            block     = MarkdownBlockLexer,
-            renderer  = MarkdownRenderer(),
+            renderer  = "html",
+            plugins   = ['strikethrough', plugin_matrix],
         )
 
-        self._markdown_to_html.block.default_rules = [
-            rule for rule in self._markdown_to_html.block.default_rules
-            if rule != "block_quote"
-        ]
-
 
     def mentions_in_html(self, html: str) -> List[Tuple[str, str]]:
         """Return list of (text, href) tuples for all mention links in html."""