Baslinux mailing list

[baslinux AT lists.ibiblio.org]

> sqlite 3
> Also imports csv files.
I can't figure out how this works.

I created database.csv which I wanted to make into database.db
csv has comma rather than | as separator

sqlite3 database.db
> .separator ',' 
> .import database.csv database

(instructions are .import FILE TABLE)

TABLE appears NOT to refer to the name of my database file.

> dbf convertor can convert dbf to csv, a much shorter format.
> I am hoping any garbage will get removed from my file by converting dbf to
> csv and thence to sqlite db format.

I found documentation for the later dbf (0.9.0) but as dbf.sgml, for which 
an hour's online hunting did not find me a convertor to HMTL.  You 
apparently need to have the sgml creating package to get docbook-to-manand 
then use man2html (which I do have).  Anyway, I read (between the many 
large tags) about the --trim command, recommended for making much smaller 
sql files. r - l - or b (b is both left and right).  I think it removes 
stray spaces.

dbf --trim b  --sql file.sql file.dbf

(The --trim b must precede the --sql.)

My 437K dbf file, without trimming, produced 548K sql > 1MB db
Trimming produced 347K sql and 152K db.  Which makes sqlite's format much 
more efficient than dbf or even csv (294K because of all the " ") even 
despite a lot of garbage.

There is still lots of garbage in the file (created by my buggy dbf 
database program) but about 640K less of it.  Fragments of records are 
found in the header section. Long stretches of ^@ are found at random 
intervals.

Also, sqlite chopped off the entire first page (25 lines or so) of records 
from the sql file when reading it in.  (It did not do this to the small 
non-buggy sample file, just my big actual buggy one).

I can enter this infomratoin back manually while referring to the dbf file,
once I figure out how.

Sindi