buildmimetree.py: make Multipart nodes hashable

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

diff --git a/.config/neomutt/buildmimetree.py b/.config/neomutt/buildmimetree.py
index c16e519db3ab3e137e04a915c16be9e0ac96df61..97ada97a0fe1170686b634d9a25a1e28a89c8d40 100755 (executable)
--- a/.config/neomutt/buildmimetree.py
+++ b/.config/neomutt/buildmimetree.py
@@ -6,8 +6,10 @@
 #
 # Configuration:
 #   neomuttrc (needs to be a single line):
 #
 # Configuration:
 #   neomuttrc (needs to be a single line):

+#     set my_mdwn_extensions="extra,admonition,codehilite,sane_lists,smarty"

 #     macro compose B "\
 #     macro compose B "\

-#       <enter-command> source '$my_confdir/buildmimetree.py setup|'<enter>\
+#       <enter-command> source '$my_confdir/buildmimetree.py \
+#       --tempdir $tempdir --extensions $my_mdwn_extensions|'<enter>\

 #       <enter-command> sourc e \$my_mdwn_postprocess_cmd_file<enter>\
 #     " "Convert message into a modern MIME tree with inline images"
 #
 #       <enter-command> sourc e \$my_mdwn_postprocess_cmd_file<enter>\
 #     " "Convert message into a modern MIME tree with inline images"
 #


@@ -21,8 +23,8 @@
 #   - python3-markdown
 # Optional:
 #   - pytest
 #   - 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
 #
 # Latest version:
 #   https://git.madduck.net/etc/neomutt.git/blob_plain/HEAD:/.config/neomutt/buildmimetree.py


@@ -53,65 +55,66 @@ def parse_cli_args(*args, **kwargs):
         )
     )
     parser.epilog = (
         )
     )
     parser.epilog = (

-        "Copyright © 2022 martin f. krafft <madduck@madduck.net>.\n"
+        "Copyright © 2023 martin f. krafft <madduck@madduck.net>.\n"

         "Released under the MIT licence"
     )
 
         "Released under the MIT licence"
     )
 

-    subp = parser.add_subparsers(help="Sub-command parsers", dest="mode")
-    parser_setup = subp.add_parser("setup", help="Setup phase")
-    parser_massage = subp.add_parser("massage", help="Massaging phase")
-
-    parser_setup.add_argument(
-        "--debug-commands",
-        action="store_true",
-        help="Turn on debug logging of commands generated to stderr",
+    parser.add_argument(
+        "--extensions",
+        type=str,
+        default="",
+        help="Markdown extension to use (comma-separated list)",

     )
 
     )
 

-    parser_setup.add_argument(
-        "--extension",
-        "-x",
-        metavar="EXTENSION",
-        dest="extensions",
-        nargs="?",
-        default=[],
-        action="append",
-        help="Markdown extension to add to the list of extensions use",
-    )
+    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_setup.add_argument(
-        "--send-message",
+    parser.add_argument(
+        "--only-build",

         action="store_true",
         action="store_true",

-        help="Generate command(s) to send the message after processing",
+        help="Only build, don't send the message",

     )
 
     )
 

-    parser_massage.add_argument(
-        "--debug-commands",
-        action="store_true",
-        help="Turn on debug logging of commands generated to stderr",
+    parser.add_argument(
+        "--tempdir",
+        default=None,
+        help="Specify temporary directory to use for attachments",

     )
 
     )
 

-    parser_massage.add_argument(
-        "--debug-walk",
+    parser.add_argument(
+        "--debug-commands",

         action="store_true",
         action="store_true",

-        help="Turn on debugging to stderr of the MIME tree walk",
+        help="Turn on debug logging of commands generated to stderr",

     )
 
     )
 

-    parser_massage.add_argument(
-        "--extensions",
-        metavar="EXTENSIONS",
-        type=str,
-        default="",
-        help="Markdown extension to use (comma-separated list)",
+    subp = parser.add_subparsers(help="Sub-command parsers", dest="mode")
+    massage_p = subp.add_parser(
+        "massage", help="Massaging phase (internal use)"

     )
 
     )
 

-    parser_massage.add_argument(
+    massage_p.add_argument(

         "--write-commands-to",
         "--write-commands-to",

+        "-o",

         metavar="PATH",
         dest="cmdpath",
         metavar="PATH",
         dest="cmdpath",

+        required=True,

         help="Temporary file path to write commands to",
     )
 
         help="Temporary file path to write commands to",
     )
 

-    parser_massage.add_argument(
+    massage_p.add_argument(
+        "--debug-walk",
+        action="store_true",
+        help="Turn on debugging to stderr of the MIME tree walk",
+    )
+
+    massage_p.add_argument(

         "MAILDRAFT",
         nargs="?",
         help="If provided, the script is invoked as editor on the mail draft",
         "MAILDRAFT",
         nargs="?",
         help="If provided, the script is invoked as editor on the mail draft",


@@ -169,11 +172,15 @@ class InlineImageExtension(Extension):
         return self._images
 
 
         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)
     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()
     htmltext = mdwn.convert(text)
 
     images = inline_image_handler.get_images()


@@ -188,6 +195,38 @@ def markdown_with_inline_image_support(text, *, extensions=None):
     return text, htmltext, images
 
 
     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 ] ########################################################
 
 
 # [ PARTS GENERATION ] ########################################################
 
 


@@ -213,6 +252,9 @@ class Multipart(
     def __str__(self):
         return f"<multipart/{self.subtype}> children={len(self.children)}"
 
     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:
 
 def filewriter_fn(path, content, mode="w", **kwargs):
     with open(path, mode, **kwargs) as out_f:


@@ -228,16 +270,17 @@ def collect_inline_images(
 
         mimetype = data.headers["Content-Type"]
         ext = mimetypes.guess_extension(mimetype)
 
         mimetype = data.headers["Content-Type"]
         ext = mimetypes.guess_extension(mimetype)

-        tempfilename = tempfile.mkstemp(prefix="img", suffix=ext, dir=tempdir)[
-            1
-        ]
-        path = pathlib.Path(tempfilename)
+        tempfilename = tempfile.mkstemp(prefix="img", suffix=ext, dir=tempdir)
+        path = pathlib.Path(tempfilename[1])

 
         filewriter_fn(path, data.read(), "w+b")
 
         relparts.append(
             Part(
 
         filewriter_fn(path, data.read(), "w+b")
 
         relparts.append(
             Part(

-                *mimetype.split("/"), path, cid=info.cid, desc=info.desc
+                *mimetype.split("/"),
+                path,
+                cid=info.cid,
+                desc=f"Image: {info.desc}",

             )
         )
 
             )
         )
 


@@ -245,10 +288,22 @@ def collect_inline_images(
 
 
 def convert_markdown_to_html(
 
 
 def convert_markdown_to_html(

-    origtext, draftpath, *, filewriter_fn=filewriter_fn, extensions=None
+    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, htmltext, images = markdown_with_inline_image_support(

-        origtext, extensions=extensions
+        origtext, extensions=extensions, extension_configs=extension_configs

     )
 
     filewriter_fn(draftpath, origtext, encoding="utf-8")
     )
 
     filewriter_fn(draftpath, origtext, encoding="utf-8")


@@ -256,6 +311,8 @@ def convert_markdown_to_html(
         "text", "plain", draftpath, "Plain-text version", orig=True
     )
 
         "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"
     htmlpath = draftpath.with_suffix(".html")
     filewriter_fn(
         htmlpath, htmltext, encoding="utf-8", errors="xmlcharrefreplace"


@@ -266,7 +323,9 @@ def convert_markdown_to_html(
         "alternative", [textpart, htmlpart], "Group of alternative content"
     )
 
         "alternative", [textpart, htmlpart], "Group of alternative content"
     )
 

-    imgparts = collect_inline_images(images, filewriter_fn=filewriter_fn)
+    imgparts = collect_inline_images(
+        images, tempdir=tempdir, filewriter_fn=filewriter_fn
+    )

     if imgparts:
         return Multipart(
             "relative", [altpart] + imgparts, "Group of related content"
     if imgparts:
         return Multipart(
             "relative", [altpart] + imgparts, "Group of related content"


@@ -381,19 +440,18 @@ class MuttCommands:
 
 
 def do_setup(
 
 
 def do_setup(

-    extensions=None, *, out_f=sys.stdout, temppath=None, debug_commands=False
+    *,
+    out_f=sys.stdout,
+    temppath=None,
+    tempdir=None,
+    debug_commands=False,

 ):
 ):

-    extensions = extensions or []

     temppath = temppath or pathlib.Path(
     temppath = temppath or pathlib.Path(

-        tempfile.mkstemp(prefix="muttmdwn-")[1]
+        tempfile.mkstemp(prefix="muttmdwn-", dir=tempdir)[1]

     )
     cmds = MuttCommands(out_f, debug=debug_commands)
 
     )
     cmds = MuttCommands(out_f, debug=debug_commands)
 

-    editor = f"{sys.argv[0]} massage --write-commands-to {temppath}"
-    if extensions:
-        editor = f'{editor} --extensions {",".join(extensions)}'
-    if debug_commands:
-        editor = f'{editor} --debug-commands'
+    editor = f"{' '.join(sys.argv)} massage --write-commands-to {temppath}"

 
     cmds.cmd('set my_editor="$editor"')
     cmds.cmd('set my_edit_headers="$edit_headers"')
 
     cmds.cmd('set my_editor="$editor"')
     cmds.cmd('set my_edit_headers="$edit_headers"')


@@ -410,7 +468,10 @@ def do_massage(
     cmd_f,
     *,
     extensions=None,
     cmd_f,
     *,
     extensions=None,

+    cssfile=None,

     converter=convert_markdown_to_html,
     converter=convert_markdown_to_html,

+    only_build=False,
+    tempdir=None,

     debug_commands=False,
     debug_walk=False,
 ):
     debug_commands=False,
     debug_walk=False,
 ):


@@ -433,7 +494,13 @@ def do_massage(
     cmds.flush()
 
     extensions = extensions.split(",") if extensions else []
     cmds.flush()
 
     extensions = extensions.split(",") if extensions else []

-    tree = converter(draft_f.read(), draftpath, extensions=extensions)
+    tree = converter(
+        draft_f.read(),
+        draftpath,
+        cssfile=cssfile,
+        tempdir=tempdir,
+        extensions=extensions,
+    )

 
     mimetree = MIMETreeDFWalker(debug=debug_walk)
 
 
     mimetree = MIMETreeDFWalker(debug=debug_walk)
 


@@ -443,6 +510,8 @@ def do_massage(
         depth-first, and responsible for telling NeoMutt how to assemble
         the tree.
         """
         depth-first, and responsible for telling NeoMutt how to assemble
         the tree.
         """

+        KILL_LINE = r"\Ca\Ck"
+

         if isinstance(item, Part):
             # We've hit a leaf-node, i.e. an alternative or a related part
             # with actual content.
         if isinstance(item, Part):
             # We've hit a leaf-node, i.e. an alternative or a related part
             # with actual content.


@@ -463,7 +532,7 @@ def do_massage(
             # information, then we might just as well update the NeoMutt
             # tree now:
             if item.cid:
             # information, then we might just as well update the NeoMutt
             # tree now:
             if item.cid:

-                cmds.push(f"<edit-content-id>\\Ca\\Ck{item.cid}<enter>")
+                cmds.push(f"<edit-content-id>{KILL_LINE}{item.cid}<enter>")

 
         elif isinstance(item, Multipart):
             # This node has children, but we already visited them (see
 
         elif isinstance(item, Multipart):
             # This node has children, but we already visited them (see


@@ -472,7 +541,7 @@ def do_massage(
             # appropriate grouping:
             if item.subtype == "alternative":
                 cmds.push("<group-alternatives>")
             # appropriate grouping:
             if item.subtype == "alternative":
                 cmds.push("<group-alternatives>")

-            elif item.subtype == "relative":
+            elif item.subtype in ("relative", "related"):

                 cmds.push("<group-related>")
             elif item.subtype == "multilingual":
                 cmds.push("<group-multilingual>")
                 cmds.push("<group-related>")
             elif item.subtype == "multilingual":
                 cmds.push("<group-multilingual>")


@@ -483,7 +552,7 @@ def do_massage(
 
         # If the item has a description, we might just as well add it
         if item.desc:
 
         # If the item has a description, we might just as well add it
         if item.desc:

-            cmds.push(f"<edit-description>\\Ca\\Ck{item.desc}<enter>")
+            cmds.push(f"<edit-description>{KILL_LINE}{item.desc}<enter>")

 
         # Finally, if we're at non-root level, tag the new container,
         # as it might itself be part of a container, to be processed
 
         # Finally, if we're at non-root level, tag the new container,
         # as it might itself be part of a container, to be processed


@@ -498,6 +567,9 @@ def do_massage(
     # function
     mimetree.walk(tree, visitor_fn=visitor_fn)
 
     # function
     mimetree.walk(tree, visitor_fn=visitor_fn)
 

+    if not only_build:
+        cmds.push("<send-message>")
+

     # Finally, cleanup. Since we're responsible for removing the temporary
     # file, how's this for a little hack?
     try:
     # Finally, cleanup. Since we're responsible for removing the temporary
     # file, how's this for a little hack?
     try:


@@ -514,11 +586,11 @@ def do_massage(
 if __name__ == "__main__":
     args = parse_cli_args()
 
 if __name__ == "__main__":
     args = parse_cli_args()
 

-    if args.mode == "setup":
-        if args.send_message:
-            raise NotImplementedError()
-
-        do_setup(args.extensions, debug_commands=args.debug_commands)
+    if args.mode is None:
+        do_setup(
+            tempdir=args.tempdir,
+            debug_commands=args.debug_commands,
+        )

 
     elif args.mode == "massage":
         with open(args.MAILDRAFT, "r") as draft_f, open(
 
     elif args.mode == "massage":
         with open(args.MAILDRAFT, "r") as draft_f, open(


@@ -529,6 +601,9 @@ if __name__ == "__main__":
                 pathlib.Path(args.MAILDRAFT),
                 cmd_f,
                 extensions=args.extensions,
                 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,
                 debug_walk=args.debug_walk,
             )
                 debug_commands=args.debug_commands,
                 debug_walk=args.debug_walk,
             )


@@ -664,33 +739,12 @@ try:
             mimetree.walk(basic_mime_tree)
             assert len(items) == 5
 
             mimetree.walk(basic_mime_tree)
             assert len(items) == 5
 

-        def test_do_setup_no_extensions(self, const1, capsys):
-            "Assert basics about the setup command output"
-            do_setup(temppath=const1, out_f=sys.stdout)
-            captout = capsys.readouterr()
-            lines = captout.out.splitlines()
-            assert lines[2].endswith(f'{const1}"')
-            assert lines[4].endswith(const1)
-            assert "first-entry" in lines[-1]
-            assert "edit-file" in lines[-1]
-
-        def test_do_setup_extensions(self, const1, const2, capsys):
-            "Assert that extensions are passed to editor"
-            do_setup(
-                temppath=const1, extensions=[const2, const1], out_f=sys.stdout
-            )
-            captout = capsys.readouterr()
-            lines = captout.out.splitlines()
-            # assert comma-separated list of extensions passed
-            assert lines[2].endswith(f'{const2},{const1}"')
-            assert lines[4].endswith(const1)
-

         @pytest.fixture
         def string_io(self, const1, text=None):
             return StringIO(text or const1)
 
         def test_do_massage_basic(self, const1, string_io, capsys):
         @pytest.fixture
         def string_io(self, const1, text=None):
             return StringIO(text or const1)
 
         def test_do_massage_basic(self, const1, string_io, capsys):

-            def converter(drafttext, draftpath, extensions):
+            def converter(drafttext, draftpath, cssfile, extensions, tempdir):

                 return Part("text", "plain", draftpath, orig=True)
 
             do_massage(
                 return Part("text", "plain", draftpath, orig=True)
 
             do_massage(


@@ -706,6 +760,7 @@ try:
             assert '="$my_edit_headers"' in lines.pop(0)
             assert "unset my_editor" == lines.pop(0)
             assert "unset my_edit_headers" == lines.pop(0)
             assert '="$my_edit_headers"' in lines.pop(0)
             assert "unset my_editor" == lines.pop(0)
             assert "unset my_edit_headers" == lines.pop(0)

+            assert "send-message" in lines.pop(0)

             assert "update-encoding" in lines.pop(0)
             assert "source 'rm -f " in lines.pop(0)
             assert "unset my_mdwn_postprocess_cmd_file" == lines.pop(0)
             assert "update-encoding" in lines.pop(0)
             assert "source 'rm -f " in lines.pop(0)
             assert "unset my_mdwn_postprocess_cmd_file" == lines.pop(0)


@@ -713,7 +768,7 @@ try:
         def test_do_massage_fulltree(
             self, string_io, const1, basic_mime_tree, capsys
         ):
         def test_do_massage_fulltree(
             self, string_io, const1, basic_mime_tree, capsys
         ):

-            def converter(drafttext, draftpath, extensions):
+            def converter(drafttext, draftpath, cssfile, extensions, tempdir):

                 return basic_mime_tree
 
             do_massage(
                 return basic_mime_tree
 
             do_massage(


@@ -725,6 +780,7 @@ try:
 
             captured = capsys.readouterr()
             lines = captured.out.splitlines()[4:]
 
             captured = capsys.readouterr()
             lines = captured.out.splitlines()[4:]

+            assert "send-message" in lines.pop(0)

             assert "Related" in lines.pop(0)
             assert "group-related" in lines.pop(0)
             assert "tag-entry" in lines.pop(0)
             assert "Related" in lines.pop(0)
             assert "group-related" in lines.pop(0)
             assert "tag-entry" in lines.pop(0)


@@ -861,27 +917,26 @@ try:
                 assert k == f"file://{imgpath}"
                 break
 
                 assert k == f"file://{imgpath}"
                 break
 

-        def test_markdown_inline_image_processor_base64(self):
-            img = (
+        @pytest.fixture
+        def test_png(self):
+            return (

                 "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAE"
                 "AAAABCAAAAAA6fptVAAAACklEQVQI12P4DwABAQEAG7buVgAA"
             )
                 "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAE"
                 "AAAABCAAAAAA6fptVAAAACklEQVQI12P4DwABAQEAG7buVgAA"
             )

-            text = f"![1px white inlined]({img})"
+
+        def test_markdown_inline_image_processor_base64(self, test_png):
+            text = f"![1px white inlined]({test_png})"

             text, html, images = markdown_with_inline_image_support(text)
 
             assert 'src="cid:' in html
             assert "](cid:" in text
             assert len(images) == 1
             text, html, images = markdown_with_inline_image_support(text)
 
             assert 'src="cid:' in html
             assert "](cid:" in text
             assert len(images) == 1

-            assert img in images
+            assert test_png in images

 
         def test_converter_tree_inline_image_base64(
 
         def test_converter_tree_inline_image_base64(

-            self, const1, fake_filewriter
+            self, test_png, const1, fake_filewriter

         ):
         ):

-            img = (
-                "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAE"
-                "AAAABCAAAAAA6fptVAAAACklEQVQI12P4DwABAQEAG7buVgAA"
-            )
-            text = f"![inline base64 image]({img})"
+            text = f"![inline base64 image]({test_png})"

             path = pathlib.Path(const1)
             tree = convert_markdown_to_html(
                 text, path, filewriter_fn=fake_filewriter
             path = pathlib.Path(const1)
             tree = convert_markdown_to_html(
                 text, path, filewriter_fn=fake_filewriter


@@ -891,7 +946,40 @@ try:
             assert tree.children[1].subtype == "png"
             written = fake_filewriter.pop()
             assert tree.children[1].path == written[0]
             assert tree.children[1].subtype == "png"
             written = fake_filewriter.pop()
             assert tree.children[1].path == written[0]

-            assert written[1] == request.urlopen(img).read()
+            assert written[1] == request.urlopen(test_png).read()
+
+        def test_inline_image_collection(
+            self, test_png, const1, const2, fake_filewriter
+        ):
+            test_images = {test_png: InlineImageInfo(cid=const1, desc=const2)}
+            relparts = collect_inline_images(
+                test_images, filewriter_fn=fake_filewriter
+            )
+
+            written = fake_filewriter.pop()
+            assert b"PNG" in written[1]
+
+            assert relparts[0].subtype == "png"
+            assert relparts[0].path == written[0]
+            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
 
 except ImportError:
     pass