-
|
I'm using Pandoc to convert HTML to plain text in order to render HTML emails in aerc. Their recommended command line is: pandoc -f html -t plainThis works well for simple emails, though many senders use deeply-nested tables for layout, and each table is rendered as an unnecessary border around the content. One of the more extreme examplesThis is a series of Can I get Pandoc to render the table layout but without the borders? I can see there are extensions for different types of table rendering, though I've tried various combinations of enabling/disabling those on the I've also had a quick scan through the variables and writer options but can't see anything obvious to help here. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 6 replies
-
|
The table extensions you have linked are for pandoc's own Markdown flavour, and disabling them shouldn't have any effect when converting from Regarding your issue, I'm not aware of any command line options that might help you here, however one thing you could do is write a Lua filter which 'promotes' the table contents to top-level blocks. If you provide an example HTML file someone may be able to help. |
Beta Was this translation helpful? Give feedback.
-
|
Toy example: Real-world example (producing the "extreme" output above): 📎 email.html |
Beta Was this translation helpful? Give feedback.
-
|
You could try using a crude Lua filter like this: detable.lua function Table(el)
local contents = {}
for _,x in ipairs(el.bodies) do
for _,y in ipairs(x.body) do
for _,z in ipairs(y.cells) do
contents = contents .. z.contents
end
end
end
return contents
end |
Beta Was this translation helpful? Give feedback.
-
|
Neat -- the filter pipeline looks like the best option here, if the table rendering "style" isn't directly controllable. This filter does have the unfortunate side-effect of losing the layout of side-by-side items, but that's a small price to pay to reduce all the border noise from table misuse, and if anything probably improves the flow when reading in a narrow terminal. Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
I wrote a filter as well, you can find it here. It is a lot less aggressive than John's filter and simply replaces tables that only contain another table with their contents, thereby preserving layout. Consequently, if one of your emails actually contains a legitimate table, it will still render properly. The filter also flattens Divs, because your email example contained a case where a Table contained a Div (which in turn contained a Table), thereby circumventing my flattening logic. |
Beta Was this translation helpful? Give feedback.
-
|
Aha, this works pretty well -- we end up with a bit of cruft from "empty" table cells (e.g. rows just used as background-colour lines), but this keeps the layout intact and does remove most of the border spam to make it readable again. 🙂 Looking at the Lua API, E: Added to the start of if pandoc.utils.stringify(tbl):gsub("%s+", "") == "" then
return {}
end |
Beta Was this translation helpful? Give feedback.
-
|
Nice! A few unsolicited tips for your tinkering. For debugging, it is often useful to see what the document's AST representation looks like, since that is what is being manipulated by filters. You can do that by outputting to the |
Beta Was this translation helpful? Give feedback.
I wrote a filter as well, you can find it here. It is a lot less aggressive than John's filter and simply replaces tables that only contain another table with their contents, thereby preserving layout. Consequently, if one of your emails actually contains a legitimate table, it will still render properly. The filter also flattens Divs, because your email example contained a case where a Table contained a Div (which in turn contained a Table), thereby circumventing my flattening logic.