Converting filter to custom writer

Converting filter to custom writer

[BPJ]

[-- Attachment #1: Type: text/plain, Size: 2044 bytes --]

I have a not-so-old "pseudo-writer" implemented as a filter which mostly
injects  tons of raw markup into the document tree so that the output of
the plain writer looks like my target format, returning a table of filters
at the end like this

``````lua
return {
  {
    Meta = get_config
  },
  {
    SmallCaps = SmallCaps,
    Str = Str
  },
  note_filter,
  main_filter
}
``````

where `get_config` just extracts some info from the metadata and populates
a table declared at the top of the script which the functions in the other
filters refer to.

Am I right that I in principle could convert this into a new style custom
writer by removing the Meta filter, replacing `return {<filters>}` above
with `local filters = {<filters>}` and adding the following at the bottom
of the script?

``````lua
function Writer (doc, opts)
  get_config(doc.meta)
  for _,filter in ipairs(filters) do
    doc = doc:walk(filter)
  end
  doc.blocks = blocks
  return pandoc.write(doc, 'plain', opts)
``````

(This got me thinking: what if `walk` accepted one more argument which will
be passed as an extra argument to the functions in the filters, which could
be used to pass config and/or maintain state? It somehow feels more
intuitive than defining a table outside the metadata handler which all
handlers close around.)

(BTW the main reason that the `note_filter` is separate is that it needs to
be topdown because it counts and collects the notes, while `main_filter`
needs to be bottomup because it sometimes relies on the content of elements
already having been processed. I was very happy when I realized that this
would work!)

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/CADAJKhAyQY_deNGV4SN61jodEM4sLUBSa4zpdxRdXg8%2BLLMzTA%40mail.gmail.com.

[-- Attachment #2: Type: text/html, Size: 3159 bytes --]

Re: Converting filter to custom writer

[BPJ]

[-- Attachment #1: Type: text/plain, Size: 2283 bytes --]

I saw no answer to this one. Hopefully it just got lost.
Apologies to anyone who already got it.

Den fre 24 juni 2022 16:09BPJ <> skrev:

> I have a not-so-old "pseudo-writer" implemented as a filter which mostly
> injects  tons of raw markup into the document tree so that the output of
> the plain writer looks like my target format, returning a table of filters
> at the end like this
>
> ``````lua
> return {
>   {
>     Meta = get_config
>   },
>   {
>     SmallCaps = SmallCaps,
>     Str = Str
>   },
>   note_filter,
>   main_filter
> }
> ``````
>
> where `get_config` just extracts some info from the metadata and populates
> a table declared at the top of the script which the functions in the other
> filters refer to.
>
> Am I right that I in principle could convert this into a new style custom
> writer by removing the Meta filter, replacing `return {<filters>}` above
> with `local filters = {<filters>}` and adding the following at the bottom
> of the script?
>
> ``````lua
> function Writer (doc, opts)
>   get_config(doc.meta)
>   for _,filter in ipairs(filters) do
>     doc = doc:walk(filter)
>   end
>   doc.blocks = blocks
>   return pandoc.write(doc, 'plain', opts)
> ``````
>
> (This got me thinking: what if `walk` accepted one more argument which
> will be passed as an extra argument to the functions in the filters, which
> could be used to pass config and/or maintain state? It somehow feels more
> intuitive than defining a table outside the metadata handler which all
> handlers close around.)
>
> (BTW the main reason that the `note_filter` is separate is that it needs
> to be topdown because it counts and collects the notes, while `main_filter`
> needs to be bottomup because it sometimes relies on the content of elements
> already having been processed. I was very happy when I realized that this
> would work!)
>
>

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/CADAJKhBPpu1%3DFnxP0byMDWGEBF-7XAV7mVc%2BTzDSPX2t2GXN%3Dg%40mail.gmail.com.

[-- Attachment #2: Type: text/html, Size: 3573 bytes --]

Re: Converting filter to custom writer

[Albert Krewinkel]

BPJ <melroch-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> writes:

> I saw no answer to this one.

Thanks for the reminder :)

> Den fre 24 juni 2022 16:09BPJ <> skrev:
>
> I have a not-so-old "pseudo-writer" implemented as a filter which
> mostly injects  tons of raw markup into the document tree so that the
> output of the plain writer looks like my target format, returning a
> table of filters at the end like this
>
> [...]
>
> Am I right that I in principle could convert this into a new style
> custom writer by removing the Meta filter, replacing `return
> {<filters>}` above with `local filters = {<filters>}` and adding the
> following at the bottom of the script?
>
> ``````lua
> function Writer (doc, opts)
>   get_config(doc.meta)
>   for _,filter in ipairs(filters) do
>     doc = doc:walk(filter)
>   end
>   doc.blocks = blocks
>   return pandoc.write(doc, 'plain', opts)
> ``````

Yes, I believe that would work. I'm actually starting to use custom
writers in a similar way. The only downside is that you can't use the
default extension mechanism (like `plain-multiline_tables`) on the
command line to tune the writer, but we're working on that.

> (This got me thinking: what if `walk` accepted one more argument which
> will be passed as an extra argument to the functions in the filters,
> which could be used to pass config and/or maintain state? It somehow
> feels more intuitive than defining a table outside the metadata handler
> which all handlers close around.)

Good idea. Although I have to admit that I'm not really enthusiastic
about it due to many small, slightly annoying changes that implementing
this feature would require in the Haskell code.

> (BTW the main reason that the `note_filter` is separate is that it
> needs to be topdown because it counts and collects the notes, while
> `main_filter` needs to be bottomup because it sometimes relies on the
> content of elements already having been processed. I was very happy
> when I realized that this would work!)

😊

-- 
Albert Krewinkel
GPG: 8eed e3e2 e8c5 6f18 81fe  e836 388d c0b2 1f63 1124

-- 
You received this message because you are subscribed to the Google Groups "pandoc-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pandoc-discuss+unsubscribe-/JYPxA39Uh5TLH3MbocFF+G/Ez6ZCGd0@public.gmane.org
To view this discussion on the web visit https://groups.google.com/d/msgid/pandoc-discuss/87edz7bc3h.fsf%40zeitkraut.de.