> I've been using Pandoc as a BBEdit filter to write documentation that will
> (eventually) be LaTeX but is temporarily HTML (in the BBEdit preview
> window).

> Pandoc automatically attaches link-anchors to headers in these documents
> (sections, subsections, H1, H2 etc) giving them formatted id-strings
> "behind the scenes". The format of the link-anchors is indicated in the
> User Guide.

> I wanted to automate the creation of these links in my text (e.g. when I
> create an internal link within the document to another section).

> Here is the hackish shell script that I run as a BBEdit text-filter on the
> selected Header name to transform it in-place to the Pandoc link format.

> I'm sure you can improve on this string of perls.

> #!/bin/sh

> # Text filter: when run on the selected string in BBEdit
> # will convert it to a form used by Pandoc for a header identifier

> perl -pe 's/[^\w\.\- ]//g' | perl -pe 's/^[0-9\. ]*//g'| perl -pe '$_=lc'
> | perl -pe 's/[ \t]/-/g'

> perl -pe 's/[^\w\.\-_ ]//g' | perl -pe 's/^[0-9\. ]*//g'| perl -pe '$_=lc'
> | perl -pe 's/[ \t]/-/g'

> On 2012-11-15 07:50, peter gallagher wrote:
> > Oops... I copied it wrong, omitting the underscore before the space
> > character in the first regex. The perl string *should* be:

> > perl -pe 's/[^\w\.\-_ ]//g' | perl -pe 's/^[0-9\. ]*//g'| perl -pe
> '$_=lc'
> > | perl -pe 's/[ \t]/-/g'

> Why not just make all four substitutions in the same
> perl process? I'm not saying you're doing wrong, just
> wondering why. Also you might want to make sure to
> remove all formatting by first piping trough pandoc
> with plain output, for which reason you'll want to trim
> off leading and trailing whitespace as well.

>      pandoc -w plain | perl -pe's/[^\w\.\-_ ]//g; s/^[0-9\. ]*//g;
> $_=lc; s/[ \t]/-/g'

> Also you might want to prepare for Unicode in your input
> by doing

>      perl -Mopen=:utf8,:std -pe'...

> As it happens, when the subject of heading ids and
> links to them came up here a while ago I wrote a perl
> script which collects all ATX headings[^1] in its input
> and outputs blank-line separated reference link
> definitions with the actual heading texts as link
> identifiers and the properly formatted HTML ids as
> URLs. Optionally it also adds the heading text as a
> title attribute text.

> [^1]: Actually all lines beginning with 1-6 hash marks
>      -- take care to backslash escape any false positives!

> I've added some help text/documentation and attached it.

> I use it like this

>      perl mkd-head-links.pl chap1.mkd chap2.mkd >headlinks.mkd
>      pandoc chap1.mkd chap2.mkd headlinks.mkd

> Just do

> perl mkd-head-links.pl --help|pandoc -o mkd-head-links.pdf

> to see the examples nicely typeset!

> /bpj