X-Git-Url: https://git.madduck.net/etc/neomutt.git/blobdiff_plain/3b6231cac26f319a513946af7a6004e36225999e..d1089abf149a07558a58450115a3e98440fbca14:/.config/neomutt/buildmimetree.py diff --git a/.config/neomutt/buildmimetree.py b/.config/neomutt/buildmimetree.py index e2a23a7..0052632 100755 --- a/.config/neomutt/buildmimetree.py +++ b/.config/neomutt/buildmimetree.py @@ -69,13 +69,34 @@ def parse_cli_args(*args, **kwargs): if _PYNLINER: parser.add_argument( "--css-file", - type=str, - default="", + type=pathlib.Path, help="CSS file to merge with the final HTML", ) else: parser.set_defaults(css_file=None) + parser.add_argument( + "--related-to-html-only", + action="store_true", + help="Make related content be sibling to HTML parts only", + ) + + def positive_integer(value): + try: + if int(value) > 0: + return int(value) + + except ValueError: + pass + + raise ValueError(f"Must be a positive integer") + + parser.add_argument( + "--max-number-other-attachments", + type=positive_integer, + help="Make related content be sibling to HTML parts only", + ) + parser.add_argument( "--only-build", action="store_true", @@ -84,7 +105,7 @@ def parse_cli_args(*args, **kwargs): parser.add_argument( "--tempdir", - default=None, + type=pathlib.Path, help="Specify temporary directory to use for attachments", ) @@ -104,6 +125,7 @@ def parse_cli_args(*args, **kwargs): "-o", metavar="PATH", dest="cmdpath", + type=pathlib.Path, required=True, help="Temporary file path to write commands to", ) @@ -117,6 +139,7 @@ def parse_cli_args(*args, **kwargs): massage_p.add_argument( "MAILDRAFT", nargs="?", + type=pathlib.Path, help="If provided, the script is invoked as editor on the mail draft", ) @@ -294,6 +317,7 @@ def convert_markdown_to_html( origtext, draftpath, *, + related_to_html_only=False, cssfile=None, filewriter_fn=filewriter_fn, tempdir=None, @@ -322,73 +346,105 @@ def convert_markdown_to_html( ) htmlpart = Part("text", "html", htmlpath, "HTML version") - altpart = Multipart( - "alternative", [textpart, htmlpart], "Group of alternative content" - ) - imgparts = collect_inline_images( images, tempdir=tempdir, filewriter_fn=filewriter_fn ) - if imgparts: + + if related_to_html_only: + # If there are inline image part, they will be contained within a + # multipart/related part along with the HTML part only + if imgparts: + # replace htmlpart with a multipart/related container of the HTML + # parts and the images + htmlpart = Multipart( + "relative", [htmlpart] + imgparts, "Group of related content" + ) + return Multipart( - "relative", [altpart] + imgparts, "Group of related content" + "alternative", [textpart, htmlpart], "Group of alternative content" ) + else: - return altpart + # If there are inline image part, they will be siblings to the + # multipart/alternative tree within a multipart/related part + altpart = Multipart( + "alternative", [textpart, htmlpart], "Group of alternative content" + ) + if imgparts: + return Multipart( + "relative", [altpart] + imgparts, "Group of related content" + ) + else: + return altpart class MIMETreeDFWalker: def __init__(self, *, visitor_fn=None, debug=False): - self._visitor_fn = visitor_fn + self._visitor_fn = visitor_fn or self._echovisit self._debug = debug + def _echovisit(self, node, ancestry, debugprint): + debugprint(f"node={node} ancestry={ancestry}") + def walk(self, root, *, visitor_fn=None): """ Recursive function to implement a depth-dirst walk of the MIME-tree rooted at `root`. """ - if isinstance(root, list): - root = Multipart("mixed", children=root) + if len(root) > 1: + root = Multipart("mixed", children=root) + else: + root = root[0] self._walk( root, - stack=[], + ancestry=[], + descendents=[], visitor_fn=visitor_fn or self._visitor_fn, ) - def _walk(self, node, *, stack, visitor_fn): + def _walk(self, node, *, ancestry, descendents, visitor_fn): # Let's start by enumerating the parts at the current level. At the - # root level, stack will be the empty list, and we expect a multipart/* - # container at this level. Later, e.g. within a mutlipart/alternative - # container, the subtree will just be the alternative parts, while the - # top of the stack will be the multipart/alternative container, which - # we will process after the following loop. - - lead = f"{'| '*len(stack)}|-" + # root level, ancestry will be the empty list, and we expect a + # multipart/* container at this level. Later, e.g. within a + # mutlipart/alternative container, the subtree will just be the + # alternative parts, while the top of the ancestry will be the + # multipart/alternative container, which we will process after the + # following loop. + + lead = f"{'│ '*len(ancestry)}" if isinstance(node, Multipart): self.debugprint( - f"{lead}{node} parents={[s.subtype for s in stack]}" + f"{lead}├{node} ancestry={[s.subtype for s in ancestry]}" ) - # Depth-first, so push the current container onto the stack, - # then descend … - stack.append(node) - self.debugprint("| " * (len(stack) + 1)) + # Depth-first, so push the current container onto the ancestry + # stack, then descend … + ancestry.append(node) + self.debugprint(lead + "│ " * 2) for child in node.children: self._walk( child, - stack=stack, + ancestry=ancestry, + descendents=descendents, visitor_fn=visitor_fn, ) - self.debugprint("| " * len(stack)) - assert stack.pop() == node + assert ancestry.pop() == node + sibling_descendents = descendents + descendents.extend(node.children) else: - self.debugprint(f"{lead}{node}") + self.debugprint(f"{lead}├{node}") + sibling_descendents = descendents + + if False and ancestry: + self.debugprint(lead[:-1] + " │") if visitor_fn: - visitor_fn(node, stack, debugprint=self.debugprint) + visitor_fn( + node, ancestry, sibling_descendents, debugprint=self.debugprint + ) def debugprint(self, s, **kwargs): if self._debug: @@ -473,7 +529,9 @@ def do_massage( extensions=None, cssfile=None, converter=convert_markdown_to_html, + related_to_html_only=True, only_build=False, + max_other_attachments=20, tempdir=None, debug_commands=False, debug_walk=False, @@ -501,13 +559,16 @@ def do_massage( draft_f.read(), draftpath, cssfile=cssfile, + related_to_html_only=related_to_html_only, tempdir=tempdir, extensions=extensions, ) mimetree = MIMETreeDFWalker(debug=debug_walk) - def visitor_fn(item, stack, *, debugprint=None): + state = dict(pos=1, tags={}, parts=1) + + def visitor_fn(item, ancestry, descendents, *, debugprint=None): """ Visitor function called for every node (part) of the MIME tree, depth-first, and responsible for telling NeoMutt how to assemble @@ -524,44 +585,101 @@ def do_massage( # The original source already exists in the NeoMutt tree, but # the underlying file may have been modified, so we need to # update the encoding, but that's it: + cmds.push("") cmds.push("") + + # We really just need to be able to assume that at this point, + # NeoMutt is at position 1, and that we've processed only this + # part so far. Nevermind about actual attachments, we can + # safely ignore those as they stay at the end. + assert state["pos"] == 1 + assert state["parts"] == 1 else: # … whereas all other parts need to be added, and they're all # considered to be temporary and inline: cmds.push(f"{item.path}") cmds.push("") + # This added a part at the end of the list of parts, and that's + # just how many parts we've seen so far, so it's position in + # the NeoMutt compose list is the count of parts + state["parts"] += 1 + state["pos"] = state["parts"] + # If the item (including the original) comes with additional # information, then we might just as well update the NeoMutt # tree now: if item.cid: cmds.push(f"{KILL_LINE}{item.cid}") + # Now for the biggest hack in this script, which is to handle + # attachments, such as PDFs, that aren't related or alternatives. + # The problem is that when we add an inline image, it always gets + # appended to the list, i.e. inserted *after* other attachments. + # Since we don't know the number of attachments, we also cannot + # infer the postition of the new attachment. Therefore, we bubble + # it all the way to the top, only to then move it down again: + if state["pos"] > 1: # skip for the first part + for i in range(max_other_attachments): + # could use any number here, but has to be larger than the + # number of possible attachments. The performance + # difference of using a high number is negligible. + # Bubble up the new part + cmds.push(f"") + + # As we push the part to the right position in the list (i.e. + # the last of the subset of attachments this script added), we + # must handle the situation that subtrees are skipped by + # NeoMutt. Hence, the actual number of positions to move down + # is decremented by the number of descendents so far + # encountered. + for i in range(1, state["pos"] - len(descendents)): + cmds.push(f"") + elif isinstance(item, Multipart): # This node has children, but we already visited them (see - # above), and so they have been tagged in NeoMutt's compose - # window. Now it's just a matter of telling NeoMutt to do the - # appropriate grouping: + # above). The tags dictionary of State should contain a list of + # their positions in the NeoMutt compose window, so iterate those + # and tag the parts there: + n_tags = len(state["tags"][item]) + for tag in state["tags"][item]: + cmds.push(f"{tag}") + if item.subtype == "alternative": cmds.push("") elif item.subtype in ("relative", "related"): cmds.push("") elif item.subtype == "multilingual": cmds.push("") + else: + raise NotImplementedError( + f"Handling of multipart/{item.subtype} is not implemented" + ) + + state["pos"] -= n_tags - 1 + state["parts"] += 1 else: # We should never get here - assert not "is valid part" + raise RuntimeError(f"Type {type(item)} is unexpected: {item}") # If the item has a description, we might just as well add it if item.desc: cmds.push(f"{KILL_LINE}{item.desc}") - # Finally, if we're at non-root level, tag the new container, - # as it might itself be part of a container, to be processed - # one level up: - if stack: - cmds.push("") + if ancestry: + # If there's an ancestry, record the current (assumed) position in + # the NeoMutt compose window as needed-to-tag by our direct parent + # (i.e. the last item of the ancestry) + state["tags"].setdefault(ancestry[-1], []).append(state["pos"]) + + lead = "│ " * (len(ancestry) + 1) + "* " + debugprint( + f"{lead}ancestry={[a.subtype for a in ancestry]}\n" + f"{lead}descendents={[d.subtype for d in descendents]}\n" + f"{lead}children_positions={state['tags'][ancestry[-1]]}\n" + f"{lead}pos={state['pos']}, parts={state['parts']}" + ) # ----------------- # End of visitor_fn @@ -601,10 +719,12 @@ if __name__ == "__main__": ) as cmd_f: do_massage( draft_f, - pathlib.Path(args.MAILDRAFT), + args.MAILDRAFT, cmd_f, extensions=args.extensions, cssfile=args.css_file, + related_to_html_only=args.related_to_html_only, + max_other_attachments=args.max_number_other_attachments, only_build=args.only_build, tempdir=args.tempdir, debug_commands=args.debug_commands, @@ -674,7 +794,7 @@ try: assert lines[5] in lines_out[7] @pytest.fixture - def basic_mime_tree(self): + def mime_tree_related_to_alternative(self): return Multipart( "relative", children=[ @@ -699,47 +819,89 @@ try: desc="Related", ) - def test_MIMETreeDFWalker_depth_first_walk(self, basic_mime_tree): + @pytest.fixture + def mime_tree_related_to_html(self): + return Multipart( + "alternative", + children=[ + Part( + "text", + "plain", + "part.txt", + desc="Plain", + orig=True, + ), + Multipart( + "relative", + children=[ + Part("text", "html", "part.html", desc="HTML"), + Part( + "text", + "png", + "logo.png", + cid="logo.png", + desc="Logo", + ), + ], + desc="Related", + ), + ], + desc="Alternative", + ) + + def test_MIMETreeDFWalker_depth_first_walk( + self, mime_tree_related_to_alternative + ): mimetree = MIMETreeDFWalker() items = [] - def visitor_fn(item, stack, debugprint): - items.append((item, len(stack))) + def visitor_fn(item, ancestry, descendents, debugprint): + items.append((item, len(ancestry), len(descendents))) - mimetree.walk(basic_mime_tree, visitor_fn=visitor_fn) + mimetree.walk( + mime_tree_related_to_alternative, visitor_fn=visitor_fn + ) assert len(items) == 5 assert items[0][0].subtype == "plain" assert items[0][1] == 2 + assert items[0][2] == 0 assert items[1][0].subtype == "html" assert items[1][1] == 2 + assert items[1][2] == 0 assert items[2][0].subtype == "alternative" assert items[2][1] == 1 + assert items[2][2] == 2 assert items[3][0].subtype == "png" assert items[3][1] == 1 + assert items[3][2] == 2 assert items[4][0].subtype == "relative" assert items[4][1] == 0 + assert items[4][2] == 4 - def test_MIMETreeDFWalker_list_to_mixed(self, basic_mime_tree): + def test_MIMETreeDFWalker_list_to_mixed(self, const1): mimetree = MIMETreeDFWalker() items = [] - def visitor_fn(item, stack, debugprint): + def visitor_fn(item, ancestry, descendents, debugprint): items.append(item) - mimetree.walk([basic_mime_tree], visitor_fn=visitor_fn) + p = Part("text", "plain", const1) + mimetree.walk([p], visitor_fn=visitor_fn) + assert items[-1].subtype == "plain" + mimetree.walk([p, p], visitor_fn=visitor_fn) assert items[-1].subtype == "mixed" def test_MIMETreeDFWalker_visitor_in_constructor( - self, basic_mime_tree + self, mime_tree_related_to_alternative ): items = [] - def visitor_fn(item, stack, debugprint): + def visitor_fn(item, ancestry, descendents, debugprint): items.append(item) mimetree = MIMETreeDFWalker(visitor_fn=visitor_fn) - mimetree.walk(basic_mime_tree) + mimetree.walk(mime_tree_related_to_alternative) assert len(items) == 5 @pytest.fixture @@ -747,7 +909,14 @@ try: return StringIO(text or const1) def test_do_massage_basic(self, const1, string_io, capsys): - def converter(drafttext, draftpath, cssfile, extensions, tempdir): + def converter( + drafttext, + draftpath, + cssfile, + related_to_html_only, + extensions, + tempdir, + ): return Part("text", "plain", draftpath, orig=True) do_massage( @@ -765,43 +934,60 @@ try: assert "unset my_edit_headers" == lines.pop(0) assert "send-message" in lines.pop(0) assert "update-encoding" in lines.pop(0) + assert "first-entry" in lines.pop(0) assert "source 'rm -f " in lines.pop(0) assert "unset my_mdwn_postprocess_cmd_file" == lines.pop(0) def test_do_massage_fulltree( - self, string_io, const1, basic_mime_tree, capsys + self, string_io, const1, mime_tree_related_to_alternative, capsys ): - def converter(drafttext, draftpath, cssfile, extensions, tempdir): - return basic_mime_tree - + def converter( + drafttext, + draftpath, + cssfile, + related_to_html_only, + extensions, + tempdir, + ): + return mime_tree_related_to_alternative + + max_attachments = 5 do_massage( draft_f=string_io, draftpath=const1, cmd_f=sys.stdout, + max_other_attachments=max_attachments, converter=converter, ) 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 "Logo" in lines.pop(0) - assert "content-id" in lines.pop(0) - assert "toggle-unlink" in lines.pop(0) - assert "logo.png" in lines.pop(0) - assert "tag-entry" in lines.pop(0) - assert "Alternative" in lines.pop(0) - assert "group-alternatives" in lines.pop(0) - assert "tag-entry" in lines.pop(0) - assert "HTML" in lines.pop(0) - assert "toggle-unlink" in lines.pop(0) - assert "part.html" in lines.pop(0) - assert "tag-entry" in lines.pop(0) - assert "Plain" in lines.pop(0) - assert "update-encoding" in lines.pop(0) - assert len(lines) == 2 + lines = captured.out.splitlines()[4:-2] + assert "first-entry" in lines.pop() + assert "update-encoding" in lines.pop() + assert "Plain" in lines.pop() + assert "part.html" in lines.pop() + assert "toggle-unlink" in lines.pop() + for i in range(max_attachments): + assert "move-up" in lines.pop() + assert "move-down" in lines.pop() + assert "HTML" in lines.pop() + assert "jump>1" in lines.pop() + assert "jump>2" in lines.pop() + assert "group-alternatives" in lines.pop() + assert "Alternative" in lines.pop() + assert "logo.png" in lines.pop() + assert "toggle-unlink" in lines.pop() + assert "content-id" in lines.pop() + for i in range(max_attachments): + assert "move-up" in lines.pop() + assert "move-down" in lines.pop() + assert "Logo" in lines.pop() + assert "jump>1" in lines.pop() + assert "jump>4" in lines.pop() + assert "group-related" in lines.pop() + assert "Related" in lines.pop() + assert "send-message" in lines.pop() + assert len(lines) == 0 @pytest.fixture def fake_filewriter(self): @@ -821,9 +1007,7 @@ try: def markdown_non_converter(self, const1, const2): return lambda s, text: f"{const1}{text}{const2}" - def test_converter_tree_basic( - self, const1, const2, fake_filewriter, markdown_non_converter - ): + def test_converter_tree_basic(self, const1, const2, fake_filewriter): path = pathlib.Path(const2) tree = convert_markdown_to_html( const1, path, filewriter_fn=fake_filewriter @@ -942,15 +1126,36 @@ try: text = f"![inline base64 image]({test_png})" path = pathlib.Path(const1) tree = convert_markdown_to_html( - text, path, filewriter_fn=fake_filewriter + text, + path, + filewriter_fn=fake_filewriter, + related_to_html_only=False, ) - assert tree.subtype == "relative" + assert tree.children[0].subtype == "alternative" assert tree.children[1].subtype == "png" written = fake_filewriter.pop() assert tree.children[1].path == written[0] assert written[1] == request.urlopen(test_png).read() + def test_converter_tree_inline_image_base64_related_to_html( + self, test_png, const1, fake_filewriter + ): + text = f"![inline base64 image]({test_png})" + path = pathlib.Path(const1) + tree = convert_markdown_to_html( + text, + path, + filewriter_fn=fake_filewriter, + related_to_html_only=True, + ) + assert tree.subtype == "alternative" + assert tree.children[1].subtype == "relative" + assert tree.children[1].children[1].subtype == "png" + written = fake_filewriter.pop() + assert tree.children[1].children[1].path == written[0] + assert written[1] == request.urlopen(test_png).read() + def test_converter_tree_inline_image_cid( self, const1, fake_filewriter ): @@ -1000,6 +1205,143 @@ try: out = apply_styling(html, _PYGMENTS_CSS) assert f'{_CODEHILITE_CLASS}" style="' in out + def test_mime_tree_relative_within_alternative( + self, string_io, const1, capsys, mime_tree_related_to_html + ): + def converter( + drafttext, + draftpath, + cssfile, + related_to_html_only, + extensions, + tempdir, + ): + return mime_tree_related_to_html + + do_massage( + draft_f=string_io, + draftpath=const1, + cmd_f=sys.stdout, + converter=converter, + ) + + captured = capsys.readouterr() + lines = captured.out.splitlines()[4:-2] + assert "first-entry" in lines.pop() + assert "update-encoding" in lines.pop() + assert "Plain" in lines.pop() + assert "part.html" in lines.pop() + assert "toggle-unlink" in lines.pop() + assert "move-up" in lines.pop() + while True: + top = lines.pop() + if "move-up" not in top: + break + assert "move-down" in top + assert "HTML" in lines.pop() + assert "logo.png" in lines.pop() + assert "toggle-unlink" in lines.pop() + assert "content-id" in lines.pop() + assert "move-up" in lines.pop() + while True: + top = lines.pop() + if "move-up" not in top: + break + assert "move-down" in top + assert "move-down" in lines.pop() + assert "Logo" in lines.pop() + assert "jump>2" in lines.pop() + assert "jump>3" in lines.pop() + assert "group-related" in lines.pop() + assert "Related" in lines.pop() + assert "jump>1" in lines.pop() + assert "jump>2" in lines.pop() + assert "group-alternative" in lines.pop() + assert "Alternative" in lines.pop() + assert "send-message" in lines.pop() + assert len(lines) == 0 + + def test_mime_tree_nested_trees_does_not_break_positioning( + self, string_io, const1, capsys + ): + def converter( + drafttext, + draftpath, + cssfile, + related_to_html_only, + extensions, + tempdir, + ): + return Multipart( + "relative", + children=[ + Multipart( + "alternative", + children=[ + Part( + "text", + "plain", + "part.txt", + desc="Plain", + orig=True, + ), + Multipart( + "alternative", + children=[ + Part( + "text", + "plain", + "part.txt", + desc="Nested plain", + ), + Part( + "text", + "html", + "part.html", + desc="Nested HTML", + ), + ], + desc="Nested alternative", + ), + ], + desc="Alternative", + ), + Part( + "text", + "png", + "logo.png", + cid="logo.png", + desc="Logo", + ), + ], + desc="Related", + ) + + do_massage( + draft_f=string_io, + draftpath=const1, + cmd_f=sys.stdout, + converter=converter, + ) + + captured = capsys.readouterr() + lines = captured.out.splitlines() + while not "logo.png" in lines.pop(): + pass + lines.pop() + assert "content-id" in lines.pop() + assert "move-up" in lines.pop() + while True: + top = lines.pop() + if "move-up" not in top: + break + assert "move-down" in top + # Due to the nested trees, the number of descendents of the sibling + # actually needs to be considered, not just the nieces. So to move + # from position 1 to position 6, it only needs one + # because that jumps over the entire sibling tree. Thus what + # follows next must not be another + assert "Logo" in lines.pop() except ImportError: pass