Markus Neteler wrote:
using the second keyword of each module, I have auto-generated
a topic oriented GRASS manual. I wrote a small shell script
for this.Draft version available at:
http://gis.cri.fmach.it/download/grass_html/topics.htmlTry for example:
- classification
- hydrology
- statisticsThe second keyword list should be carefully revisited (while
not breaking too many translations) to better group the commands.
However, even in the current state it will be useful.Markus
PS: Likely it would be better to rewrite my shell script to
Python. It could then be run on the GRASS web site weekly.
Hi,
it will be good to finally make better use of the keywords!
instead of relying on keyword order, might I suggest to create a
growing table, where every new keyword creates a new bin.
bins with counts >=2 create a page like the ones you've made,
singleton keywords get ignored.
Probably python would be best for that..
instead of lots of pages with 2-3 modules listed on them, perhaps
make into 1-cell tables instead, and put all on the same page, with
a 3-4 column TOC at the top by keyword.
except maybe if >=25 hits in the group, in which case create a separate
page and link to it from the main list.
...
I though it would be neat to make one of those word jumbles with size
proportional to frequency of keyword occurrence, perhaps with live links
from the words within it to the <a name> or <a href> of the keyword's
1-cell table described above. There's a good java app called Wordle for
it, and a number of word cloud packages for R.
here are some quick samples:
# extract keywords:
grep -A1 '^\.SH KEYWORDS$' "$GISBASE/man/man1"/* | \
grep -v '\.SH KEYWORDS$\|\.SH SYNOPSIS\|^--$' | \
cut -f2- -d- | tr ',' '\n' | \
sed -e 's/^ //' -e 's/ /~/' \
> keywords_lists~.txt
cut & paste into http://www.wordle.net/create
click on "random" .. I used the "moss" color scheme.
see attached and some PDF proofs:
http://bambi.otago.ac.nz/hamish/grass/GRASS_Keywordle.zip (~1mb)
Hamish
(attachments)
