*Entered: in emacs on gemini pda* |
*Date: 20220606*
## Recutils - text based database
Gnu recutils has been discussed before by tomasino. I
have had a TODO item for years at this point to check
it out, and see if I had a use case for it.
## What to do with it?
Fast forward to just now... I am on a plane to the
Pacific Northwest reading the infotex on recutils and
I am struck that you can trivially perform sed or awk
like things with recutils. You can add fields on the
fly, and assign values to those fields.
So, you could do something like:
```
$ cd /var/gopher/Phlog
$ ls *.post | \
sed 's/^.*$/printf "Date: &\\nTitle: \
$(head -1 & | sed "s@^# @@g;" | \
cut -c-50)\\n\\n"/g;' | \
sh | sed 's/\.post//g' > temp
```
Note: before someone rolls their eyes at piping to
sh, as recently happened to me online -- this is
my data. I do not have any phlog entries where the
first line is ";rm -rf *", so settle down.
Then you can query by year, for example and pipe your
recordset into recfmt, which is a utility that is
typically used for form letters, and have it spit out
dynamic gophermap.
```
$ cat temp | recset -f Server -a "localhost" | \
recset -f Port -a 70 | recsel -e "Date ~ '2022'" | \
recfmt -f /var/gopher/tmplt | sort -r
```
Note: the actual file contains true tab characters
where the \t appears above.
This was the result of screwing around on the command
line. I think if I use this to generate gophermaps for
my phlog or other portions of this gopherhole, I would
not ls the entire directory and dyanamically create the
entire recordset each time. It makes more sense to
have my publishing script append a new record for the
new phlog entry after checking if it exists already to
no create duplicates. I would want to pipe this to a
plaintext gophermap file to avoid slowing down the
server.
## Conclusion
Gnu recutils can be used to store and format data in
whatever manner you desire. If this had dawned upon me
sooner I would be using it in many places.
