Karel Zak wrote:
The problem with CSV is that it will correctly work with new protocol only. Because old versions of clients are newline sensitive. And CSV can contains newline in by quotation marks defined attributes:
"John", "Smith", "The White House 1600 Pennsylvania Avenue NW Washington, DC 20500", "male", "open source software office"
It is one record.
(Long Live President Smith!)
I have never seen such a beast, and if I had been asked off the cuff would have said that it was probably illegal, except that I know of no standard actually defining the format. Perhaps others (Joshua Drake or Josh Berkus?) have wider experience. I think in any case we should ignore those for now and handle the straightforward case.
I *have* seen monstrosities like fields that do not begin with the quote character but then break into a quote, e.g.:
1,2,a,123"abc""def",6,7,8
It's difficult to say it :-), but your DELIMITER idea is better than my
suggested API. Andrew, go ahead. I thought about some data streaming,
but COPY is probably bad place for it.
Just to be clear, I have no current plans for doing this. I have about 3 pg related things I am working on, in addition to doing paid work so I can eat! If I do it it will not be for quite a while.
There are some wrinkles, though, concerning the interaction of CSV's notion of escaping and COPY's notion of escaping. If someone want to undertake this I can flesh those out in a further email.
Someone was asking the other day about a newbie or student project - this might be a good one (impact relatively isolated, very easy to test) although possibly not as sexy as some might like.
cheers
andrew
---------------------------(end of broadcast)--------------------------- TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/docs/faqs/FAQ.html
