http://code.secondbit.org/pqarrays/ en-us pqarrays Changelog http://code.secondbit.org/pqarrays/rev/9a415db0346a http://code.secondbit.org/pqarrays/rev/9a415db0346a changeset 9a415db0346a branch default bookmark tag tip user Paddy <paddy@impractical.co> description Fix whitespace lexing/parsing.

Our consumeWhitespace method had a bug where it wouldn't emit the whitespace
token, meaning we weren't actually skipping the whitespace. Oops.

While in there, I removed an outdated TODO (we already supplied a default
state).

I updated the lexing of unquoted strings to take into account the rules about
spaces and unquoted strings; basically, spaces before or after are ignored as
whitespace, spaces in the middle ocunt, and empty strings are not allowed.

I removed an extra case when detecting what to do when lexing an unquoted
string; we already consumed all the whitespace, so the next character shouldn't
be whitespace, so no need to test for it.

We need to consume whitespace before we start lexing the separator character.

I updated the token debugging to be a bit more useful, by defining a String()
method on the token type itself, so it'll expose both the type and the value.
This makes unexpected errors easier to deal with, and is used in all the errors
raised by the parser now.

I added a bunch of whitespace tests for lexing and parsing. files lexer.go
lexer_test.go
parser.go
parser_test.go
]]> Paddy <paddy@impractical.co> Fri, 26 Feb 2016 01:23:50 -0800 http://code.secondbit.org/pqarrays/rev/ce9c92fc81ab http://code.secondbit.org/pqarrays/rev/ce9c92fc81ab changeset ce9c92fc81ab branch bookmark tag user Paddy <paddy@impractical.co> description Fix bug parsing empty arrays, make golint and go vet happy.

Add comments to make golint happy. Also, because comments are a good thing to
have.

Turn += 1 and -= 1 into ++ and --, respectively, so golint will be happy.

Fix an improperly formated errorf, where a rune was being treated as a string.
Thanks, go vet!

Fix whitespace parsing, returning the parse functions again instead of just
skipping the one character. Now if we have more than one whitespace character in
a row, they'll all be skipped.

Add a parseStringOrNullOrEnd parse function that will be called after the
tokenArrayStart character, to fix a bug where empty arrays were expecting a
string or null and getting the array end character. This is only valid after
tokenArrayStart, however; in other places where parseSeparatorOrDelim is used,
it wouldn't be appropriate.

Add a parser test for an empty array. files array.go
lexer.go
parser.go
parser_test.go
]]> Paddy <paddy@impractical.co> Thu, 25 Feb 2016 23:52:05 -0800 http://code.secondbit.org/pqarrays/rev/bfe2a4af6bdf http://code.secondbit.org/pqarrays/rev/bfe2a4af6bdf changeset bfe2a4af6bdf branch bookmark tag user Paddy <paddy@secondbit.org> description First pass implementation.

Use a lexer to generate tokens out of the Array type responses that PostgreSQL
will send.

Write a parser for string[] array types.

Create a StringArray type that fulfills the driver.Valuer and sql.Scanner
interfaces using the parser and lexer. files .hgignore
array.go
lexer.go
lexer_test.go
parser.go
parser_test.go
]]> Paddy <paddy@secondbit.org> Sun, 19 Apr 2015 23:47:36 -0400