Merge pull request #82 from AStenbaek/integer-literal-frontend

Integer literal frontend

Author: aslanix

compiler/src/Lexer.x

@@ -16,12 +16,15 @@ import Control.Monad.Except
 import Data.Char (isSpace, toLower)
 import Data.List (dropWhileEnd)
 import Data.Char ( chr )
-import Numeric ( readDec )
+import Numeric ( readDec, readBin, readOct, readHex )
 import Control.Monad (when)
 }
 
 %wrapper "monadUserState"
 
+$bindigit = [01]
+$octdigit = 0-7
+$hexdigit = [0-9A-Fa-f]
 $digit = 0-9
 $alpha = [a-zA-Z]
 $alpha_ = [$alpha \_]
@@ -30,7 +33,10 @@ $graphic    = $printable # $white
 @sym        = $alpha_ [$alpha $digit \_ \']*
 @string     = \" ($printable # \")* \"
 @label      = \`\{ ($printable # \})*  \}\`
-
+@declit     = $digit[\_$digit]*
+@binlit     = 0[bB]$bindigit[\_$bindigit]*
+@octlit     = 0[oO]$octdigit[\_$octdigit]*
+@hexlit     = 0[xX]$hexdigit[\_$hexdigit]*
 
 tokens:-
 -- Whitespace insensitive
@@ -109,7 +115,12 @@ tokens:-
 <state_dclabel> "#root-integrity"       { mkL TokenDCRootInteg }
 <state_dclabel> "#null-confidentiality" { mkL TokenDCNullConf }
 <state_dclabel> "#null-integrity"       { mkL TokenDCNullInteg }
-<0>   $digit+                        { mkLs (\s -> TokenNum (read s)) }
+-- Integer literal parsing inspired by https://github.com/ocaml/ocaml/blob/trunk/parsing/lexer.mll
+<0>   @declit                        { mkLs (\s -> TokenNum (read (filter (/='_') s))) }
+<0>   @binlit                        { mkLs (\s -> TokenNum (fst (head (readBin (filter (/='_') (drop 2 s)))))) }
+<0>   @octlit                        { mkLs (\s -> TokenNum (fst (head (readOct (filter (/='_') (drop 2 s)))))) }
+<0>   @hexlit                        { mkLs (\s -> TokenNum (fst (head (readHex (filter (/='_') (drop 2 s)))))) }
+<0>   (@declit|@binlit|@octlit|@hexlit)@sym { \(_, _, _, s) _ -> lexerError ("Invalid literal " ++ s) }
 <0>   [\<][\<]                       { mkL TokenBinShiftLeft }
 <0>   [\>][\>]                       { mkL TokenBinShiftRight }
 <0>   [\~][\>][\>]                   { mkL TokenBinZeroShiftRight }
@@ -386,9 +397,6 @@ lexerError msg =
   where
     trim = reverse . dropWhile (== ' ') . reverse . dropWhile (== ' ')
 
-
-
-
 -- we use a custom version of monadScan so that we have full
 -- control over the error reporting; this one is based on 
 -- the built-in alexMonadScan

tests/cmp/int-lit1.golden

@@ -0,0 +1,3 @@
+Parse Error:
+Invalid literal 0x_A
+ at 1:5 before end of line

tests/cmp/int-lit1.trp

@@ -0,0 +1 @@
+0x_A

tests/cmp/int-lit2.golden

@@ -0,0 +1,3 @@
+Parse Error:
+Invalid literal 1_x_0
+ at 1:6 before end of line

tests/cmp/int-lit2.trp

@@ -0,0 +1 @@
+1_x_0

tests/cmp/int-lit3.golden

@@ -0,0 +1,3 @@
+Parse Error:
+Invalid literal 0b12
+ at 1:5 before end of line

tests/cmp/int-lit3.trp

@@ -0,0 +1 @@
+0b12

tests/cmp/int-lit4.golden

@@ -0,0 +1,3 @@
+Parse Error:
+Invalid literal 0o8
+ at 1:4 before end of line

tests/cmp/int-lit4.trp

@@ -0,0 +1 @@
+0o8

tests/rt/pos/core/int-lit1.golden

@@ -0,0 +1,2 @@
+2025-08-24T19:08:18.468Z [RTM] info: Skipping network creation. Observe that all external IO operations will yield a runtime error.
+>>> Main thread finished with value: 1000000@{}%{}