Bump the parsers #95
Bump the parsers #95
Conversation
| block = symbol Block *> children (makeTerm'' <$> location <*> manyTerm expression) | ||
|
|
||
| block' :: Assignment Term | ||
| block' = symbol Block *> children (makeTerm <$> location <*> manyTerm expression) |
There was a problem hiding this comment.
I defined these to incur as few changes to the fixtures as possibly; basically I only changed the fixtures where the assignments themselves were outright wrong.
| withItem = symbol WithItem *> children (flip Statement.Let <$> term expression <*> term (expression <|> emptyTerm)) | ||
| <|> flip Statement.Let <$> term expression <*> emptyTerm | ||
| make (val, name) = makeTerm1 . Statement.Let name val | ||
| withItem = symbol WithItem *> children ((,) <$> term expression <*> term (expression <|> emptyTerm)) |
There was a problem hiding this comment.
This had been incorrectly building a sequence of separate Lets instead of a right-nested chain of them.
There was a problem hiding this comment.
children (flip (foldr make) — that's the Good Shit. that's the Shit I Like. Hell yeah.
There was a problem hiding this comment.
Sometimes parsing is Good, Actually.
| tryStatement :: Assignment Term | ||
| tryStatement = makeTerm <$> symbol TryStatement <*> children (Statement.Try <$> term expression <*> manyTerm (expression <|> elseClause)) | ||
| where elseClause = makeTerm <$> symbol ElseClause <*> children (Statement.Else <$> emptyTerm <*> expressions) | ||
| tryStatement = makeTerm <$> symbol TryStatement <*> children (Statement.Try <$> term block <*> manyTerm (expression <|> elseClause)) |
There was a problem hiding this comment.
try statements with multiple statements in the body were shoving all but the first into the list of catches. Whoops!
| , (inject .) . Expression.Minus <$ (symbol AnonMinus <|> symbol AnonMinus' <|> symbol AnonMinus'') | ||
| , (inject .) . Expression.Times <$ (symbol AnonStar <|> symbol AnonStar') | ||
| , (inject .) . Expression.Power <$ symbol AnonStarStar | ||
| , (inject .) . Expression.DividedBy <$ (symbol AnonSlash <|> symbol AnonSlash' <|> symbol AnonSlash'') |
There was a problem hiding this comment.
Pretty trivial; just removing some symbols that don’t exist any more.
| result <- if compatible then | ||
| liftIO $ runParser parser blobSource | ||
| else | ||
| Failed <$ trace "tree-sitter: incompatible versions" |
There was a problem hiding this comment.
Eventually we should do something smarter than this (because we still report this as a parser timeout exception), but for now at least it’s distinguishable by the trace.
| actual `shouldBe` expected | ||
| actual <- fmap verbatim <$> parseFilePath session path | ||
| case actual of | ||
| Left err -> throw err |
There was a problem hiding this comment.
This lets us see the actual exception instead of just ExitFailure or w/e.
| {+(Catch | ||
| {+(Identifier)+} | ||
| {+(Statements)+})+} | ||
| {-(Identifier)-} |
There was a problem hiding this comment.
What are you doing here (Identifier)? You belong in the body!
| {-(Assignment | ||
| {-(Identifier)-} | ||
| {-(Boolean)-})-})-})-})) | ||
| {+(Statements |
There was a problem hiding this comment.
Just un-nested a single layer of [].
| , Language.JSON.Assignment | ||
| , Language.JSON.PrettyPrint | ||
| , Language.MiniRuby.Assignment | ||
| , Language.MiniPython.Assignment |
There was a problem hiding this comment.
These were pretty much just dead weight. 🔥 with much 🔥ing.
| withItem = symbol WithItem *> children (flip Statement.Let <$> term expression <*> term (expression <|> emptyTerm)) | ||
| <|> flip Statement.Let <$> term expression <*> emptyTerm | ||
| make (val, name) = makeTerm1 . Statement.Let name val | ||
| withItem = symbol WithItem *> children ((,) <$> term expression <*> term (expression <|> emptyTerm)) |
There was a problem hiding this comment.
children (flip (foldr make) — that's the Good Shit. that's the Shit I Like. Hell yeah.
This PR bumps the parsers to
master, except fortree-sitter-python, which is on thenode-fieldsbranch, andtree-sitter-typescriptwhich is reset back to what we had it at before tree-sitter/haskell-tree-sitter#86 accidentally changed it. That means we’re now using a recenthaskell-tree-sitterand can work at migrating it & the language packages to hackage!