Quoth hiro:
> > A lot of the time my web browsing consists of (or rather is
> > replaced by) a script invoking pandoc and mupdf from newsbeuter,
> > which is pleasant.
>
> Can you give a demonstration of this workflow and how it looks?
Sure. Newsbeuter has RSS feeds for most of the websites I regularly
care about. When I want to read an article from the article list
there, I press ,. which is a macro defined in the newsbeuter config
file like so:
macro . pipe-to readnicely ; toggle-article-read ; prev
This pipes the text (not the html) straight to a script called
readnicely, which is this:
#!/bin/sh
test $# -ne 0 && echo "usage: pipe text in, to get a latex pdf displayed" &&
exit 1
cd /tmp
t=`mktemp`
pandoc -s -f markdown -t latex > $t.tex
pdflatex $t.tex >/dev/null
mupdf -r 120 $t.pdf 2>/dev/null
rm ${t}*
Which turns the plain text into latex, compiles it, and opens the
result in mupdf. /tmp is a ramdisk on my computer, so this all takes
around half a second.
The same script with '-f html' to pandoc could undoubtedly be used
for other tasks to make webpages readable.
Ideally I'd like a fuse based or similar RSS reader, as this would
be nicer and more flexible than the ncurses newsbeuter, but I don't
know of such a thing (I haven't really looked, though).
This obviously only works for full text feeds, but almost all the
sites I care about provide these, and the couple that didn't
switched to them when I asked about it.
Another plus (as with most full text rss options) is that this
requires no internet connection, and there's no analytics or other
tracking beyond initial feedburner tracking.
There are no pictures, but I never miss them.
