summaryrefslogtreecommitdiffstats
path: root/src/Parser.hs
diff options
context:
space:
mode:
authorJan Tuomi <jans.tuomi@gmail.com>2022-02-19 17:06:18 +0200
committerJan Tuomi <jans.tuomi@gmail.com>2022-02-19 17:06:18 +0200
commit2bd7673f23bf95a97323c632aeaff35df992c927 (patch)
tree0ca1b84695125b126fcf61f52898ad7789476035 /src/Parser.hs
parente8240a95aed965aa2ba82286283a47ef3afc59ca (diff)
Add more docsHEADmain
Diffstat (limited to 'src/Parser.hs')
-rw-r--r--src/Parser.hs23
1 files changed, 19 insertions, 4 deletions
diff --git a/src/Parser.hs b/src/Parser.hs
index 679f318..508be0f 100644
--- a/src/Parser.hs
+++ b/src/Parser.hs
@@ -10,6 +10,8 @@ import qualified Data.Text as T
import LTypes
import Utils
+-- | Convert a string into an 'LWord'.
+parseWord :: String -> LWord
parseWord rawStr
| isIntStr rawStr = LInteger (read rawStr)
| isFloatStr rawStr = LFloat $ read rawStr
@@ -18,9 +20,13 @@ parseWord rawStr
| "$" `isPrefixOf` rawStr = LLabel $ tail rawStr
| otherwise = LSymbol rawStr
+-- | Remove line suffixes starting with the comment marker '--'.
+removeComments :: String -> String
removeComments =
lines .> map (T.pack .> T.splitOn (T.pack "--") .> head .> T.unpack) .> unlines
+-- | Preprocess the source by inserting "string literal references" in place of
+-- string literals (e.g. "hello world") in order to remove significant whitespace from the source.
processStringLiterals :: Maybe String -> Map String [LWord] -> String -> ExceptT LException IO (String, Map String [LWord])
processStringLiterals currentM refMap [] = case currentM of
Just _ -> throwError $ LException "nonterminated string literal"
@@ -46,8 +52,14 @@ processStringLiterals currentM refMap (c : source) = case c of
-- proceed normally
processStringLiterals Nothing refMap source $> fmap (B.first (c :))
-internalNVar n = LLabel $ fmt "$__%%" [show n]
+-- | Format a new internal variable 'LLabel' for use in string literal desugaring.
+produceInternalNVar :: Int -> LWord
+produceInternalNVar n = LLabel $ fmt "$__%%" [show n]
+-- | Desugar a "string phrase" (i.e. phrase containing only chars).
+-- If the string contains the substring '%%', construct a sequence of words that consumes another string
+-- phrase from the stack when evaluated, inserting that string where the '%%' substring was located. Can be used
+-- with N instances of the '%%' substring (consumes N strings from stack).
processSLInterpolations :: LWord -> ExceptT LException IO [LWord]
processSLInterpolations strP@(LPhrase lChars)
| "%%" `isInfixOf` chars = pure [LPhrase interpolated, LSymbol "unphrase"]
@@ -58,18 +70,21 @@ processSLInterpolations strP@(LPhrase lChars)
separatedByMarker = T.pack chars $> T.splitOn (T.pack "%%") .> map (T.unpack .> map LChar)
labelIndices = [0 .. length separatedByMarker - 1 - 1]
-- store n - 1 variables from stack
- varStores = labelIndices $> map (\n -> [internalNVar n, LSymbol "!"]) .> reverse
+ varStores = labelIndices $> map (\n -> [produceInternalNVar n, LSymbol "!"]) .> reverse
part1 = concat varStores
-- interleave the n substrings and n - 1 variable reads
- symbolLookups = labelIndices $> map (\n -> [internalNVar n, LSymbol "@", LSymbol "unphrase"])
+ symbolLookups = labelIndices $> map (\n -> [produceInternalNVar n, LSymbol "@", LSymbol "unphrase"])
part2 = LSymbol "'[" : concat (mix separatedByMarker symbolLookups) ++ [LSymbol "']", LSymbol "phrase"]
-- forget the temp variables used
- varForgets = labelIndices $> map (\n -> [internalNVar n, LSymbol "forget"])
+ varForgets = labelIndices $> map (\n -> [produceInternalNVar n, LSymbol "forget"])
part3 = concat varForgets
-- combine everything into one phrase
interpolated = part1 ++ part2 ++ part3
processSLInterpolations p = throwError $ LException $ fmt "non-phrase in processSLInterpolations: %%" [show p]
+-- | Parse source code string to a list of 'LWord's and a string literal refmap (see 'LTypes.LState' for explanation).
+-- Throws exception on parsing error.
+parseSource :: String -> ExceptT LException IO ([LWord], Map String [LWord])
parseSource source = do
let woComments = source $> removeComments
(woStringLiterals, stringLiteralRefMap) <- processStringLiterals Nothing M.empty woComments