buildmimetree.py: handle image URLs starting with cid:

[etc/neomutt.git] / .config / neomutt / buildmimetree.py

diff --git a/.config/neomutt/buildmimetree.py b/.config/neomutt/buildmimetree.py
index c277e6344477f5fa025eebf36176c22dfeb26046..e2a23a741bbee012d1fd42c6eecf3fbc5fcef7e6 100755 (executable)
--- a/.config/neomutt/buildmimetree.py
+++ b/.config/neomutt/buildmimetree.py
@@ -23,8 +23,8 @@
 #   - python3-markdown
 # Optional:
 #   - pytest
-#   - Pynliner
-#   - Pygments, if installed, then syntax highlighting is enabled
+#   - Pynliner, provides --css-file and thus inline styling of HTML output
+#   - Pygments, then syntax highlighting for fenced code is enabled
 #
 # Latest version:
 #   https://git.madduck.net/etc/neomutt.git/blob_plain/HEAD:/.config/neomutt/buildmimetree.py
@@ -66,6 +66,16 @@ def parse_cli_args(*args, **kwargs):
         help="Markdown extension to use (comma-separated list)",
     )
 
+    if _PYNLINER:
+        parser.add_argument(
+            "--css-file",
+            type=str,
+            default="",
+            help="CSS file to merge with the final HTML",
+        )
+    else:
+        parser.set_defaults(css_file=None)
+
     parser.add_argument(
         "--only-build",
         action="store_true",
@@ -91,8 +101,10 @@ def parse_cli_args(*args, **kwargs):
 
     massage_p.add_argument(
         "--write-commands-to",
+        "-o",
         metavar="PATH",
         dest="cmdpath",
+        required=True,
         help="Temporary file path to write commands to",
     )
 
@@ -160,11 +172,15 @@ class InlineImageExtension(Extension):
         return self._images
 
 
-def markdown_with_inline_image_support(text, *, extensions=None):
+def markdown_with_inline_image_support(
+    text, *, extensions=None, extension_configs=None
+):
     inline_image_handler = InlineImageExtension()
     extensions = extensions or []
     extensions.append(inline_image_handler)
-    mdwn = markdown.Markdown(extensions=extensions)
+    mdwn = markdown.Markdown(
+        extensions=extensions, extension_configs=extension_configs
+    )
     htmltext = mdwn.convert(text)
 
     images = inline_image_handler.get_images()
@@ -179,6 +195,38 @@ def markdown_with_inline_image_support(text, *, extensions=None):
     return text, htmltext, images
 
 
+# [ CSS STYLING ] #############################################################
+
+try:
+    import pynliner
+
+    _PYNLINER = True
+
+except ImportError:
+    _PYNLINER = False
+
+try:
+    from pygments.formatters import get_formatter_by_name
+
+    _CODEHILITE_CLASS = "codehilite"
+
+    _PYGMENTS_CSS = get_formatter_by_name(
+        "html", style="default"
+    ).get_style_defs(f".{_CODEHILITE_CLASS}")
+
+except ImportError:
+    _PYGMENTS_CSS = None
+
+
+def apply_styling(html, css):
+    return (
+        pynliner.Pynliner()
+        .from_string(html)
+        .with_cssString("\n".join(s for s in [_PYGMENTS_CSS, css] if s))
+        .run()
+    )
+
+
 # [ PARTS GENERATION ] ########################################################
 
 
@@ -204,6 +252,9 @@ class Multipart(
     def __str__(self):
         return f"<multipart/{self.subtype}> children={len(self.children)}"
 
+    def __hash__(self):
+        return hash(str(self.subtype) + "".join(str(self.children)))
+
 
 def filewriter_fn(path, content, mode="w", **kwargs):
     with open(path, mode, **kwargs) as out_f:
@@ -215,6 +266,9 @@ def collect_inline_images(
 ):
     relparts = []
     for path, info in images.items():
+        if path.startswith("cid:"):
+            continue
+
         data = request.urlopen(path)
 
         mimetype = data.headers["Content-Type"]
@@ -240,12 +294,19 @@ def convert_markdown_to_html(
     origtext,
     draftpath,
     *,
+    cssfile=None,
     filewriter_fn=filewriter_fn,
     tempdir=None,
     extensions=None,
+    extension_configs=None,
 ):
+    # TODO extension_configs need to be handled differently
+    extension_configs = extension_configs or {}
+    extension_configs.setdefault("pymdownx.highlight", {})
+    extension_configs["pymdownx.highlight"]["css_class"] = _CODEHILITE_CLASS
+
     origtext, htmltext, images = markdown_with_inline_image_support(
-        origtext, extensions=extensions
+        origtext, extensions=extensions, extension_configs=extension_configs
     )
 
     filewriter_fn(draftpath, origtext, encoding="utf-8")
@@ -253,6 +314,8 @@ def convert_markdown_to_html(
         "text", "plain", draftpath, "Plain-text version", orig=True
     )
 
+    htmltext = apply_styling(htmltext, cssfile)
+
     htmlpath = draftpath.with_suffix(".html")
     filewriter_fn(
         htmlpath, htmltext, encoding="utf-8", errors="xmlcharrefreplace"
@@ -408,6 +471,7 @@ def do_massage(
     cmd_f,
     *,
     extensions=None,
+    cssfile=None,
     converter=convert_markdown_to_html,
     only_build=False,
     tempdir=None,
@@ -540,6 +604,7 @@ if __name__ == "__main__":
                 pathlib.Path(args.MAILDRAFT),
                 cmd_f,
                 extensions=args.extensions,
+                cssfile=args.css_file,
                 only_build=args.only_build,
                 tempdir=args.tempdir,
                 debug_commands=args.debug_commands,
@@ -682,7 +747,7 @@ try:
             return StringIO(text or const1)
 
         def test_do_massage_basic(self, const1, string_io, capsys):
-            def converter(drafttext, draftpath, extensions, tempdir):
+            def converter(drafttext, draftpath, cssfile, extensions, tempdir):
                 return Part("text", "plain", draftpath, orig=True)
 
             do_massage(
@@ -706,7 +771,7 @@ try:
         def test_do_massage_fulltree(
             self, string_io, const1, basic_mime_tree, capsys
         ):
-            def converter(drafttext, draftpath, extensions, tempdir):
+            def converter(drafttext, draftpath, cssfile, extensions, tempdir):
                 return basic_mime_tree
 
             do_massage(
@@ -886,6 +951,23 @@ try:
             assert tree.children[1].path == written[0]
             assert written[1] == request.urlopen(test_png).read()
 
+        def test_converter_tree_inline_image_cid(
+            self, const1, fake_filewriter
+        ):
+            text = f"![inline base64 image](cid:{const1})"
+            path = pathlib.Path(const1)
+            tree = convert_markdown_to_html(
+                text,
+                path,
+                filewriter_fn=fake_filewriter,
+                related_to_html_only=False,
+            )
+            assert len(tree.children) == 2
+            assert tree.children[0].cid != const1
+            assert tree.children[0].type != "image"
+            assert tree.children[1].cid != const1
+            assert tree.children[1].type != "image"
+
         def test_inline_image_collection(
             self, test_png, const1, const2, fake_filewriter
         ):
@@ -902,5 +984,22 @@ try:
             assert relparts[0].cid == const1
             assert relparts[0].desc.endswith(const2)
 
+        def test_apply_stylesheet(self):
+            if _PYNLINER:
+                html = "<p>Hello, world!</p>"
+                css = "p { color:red }"
+                out = apply_styling(html, css)
+                assert 'p style="color' in out
+
+        def test_apply_stylesheet_pygments(self):
+            if _PYGMENTS_CSS:
+                html = (
+                    f'<div class="{_CODEHILITE_CLASS}">'
+                    "<pre>def foo():\n    return</pre></div>"
+                )
+                out = apply_styling(html, _PYGMENTS_CSS)
+                assert f'{_CODEHILITE_CLASS}" style="' in out
+
+
 except ImportError:
     pass