commit 4e30b242e98097476132b2b00420109c36fd3544
parent e0745f0500fa255357bb08de421fc3fcf6016334
Author: citbl <citbl@citbl.org>
Date: Sun, 31 May 2026 18:24:10 +1000
wip lexer
Diffstat:
6 files changed, 151 insertions(+), 37 deletions(-)
diff --git a/makefile b/makefile
@@ -1,10 +1,10 @@
.SILENT:
-default: fmt
+
+default:
mkdir -p ./bin/
gnatmake -o bin/nightshade src/nightshade.adb
rm -rf *.o *.ali
- ./bin/nightshade test.txt
+ ./bin/nightshade test.mty
fmt:
gnatformat src/*.adb src/*.ads
-
diff --git a/src/lexer.adb b/src/lexer.adb
@@ -6,35 +6,138 @@ package body Lexer is
function Init (File_Name : String; File_Contents : String) return Lexer is
L : Lexer :=
- (Source => To_Unbounded_String (File_Contents),
- File_Name => To_Unbounded_String (File_Name),
- Tokens => Token_Vectors.Empty_Vector);
+ (Source => File_Contents,
+ Src_Len => File_Contents'Length,
+ File_Name => File_Name,
+ FN_Len => File_Name'Length,
+ Tokens => Token_Vectors.Empty_Vector,
+ Pos => 1,
+ Line => 1,
+ Col => 1);
begin
return L;
end Init;
- procedure Lex (L : in out Lexer) is
- SRC : constant String := To_String (L.Source);
- I : Positive := SRC'First;
- I2 : Positive := 1;
- Line : Positive := 1;
- Col : Positive := 1;
- C : Character;
- begin
- Put_Line (Natural'Image (Length (L.Source)));
-
- while I <= SRC'Last loop
- if I < SRC'Last then
- I2 := I + 1;
- else
- I2 := I;
+ function Peek (L : in Lexer) return Character is
+ begin
+ return L.Source (L.Pos);
+ end;
+
+ function Peek2 (L : in Lexer) return Character is
+ C : Character;
+ begin
+ C := L.Source (L.Pos);
+ if L.Pos < L.Source'Length then
+ C := L.Source (L.Pos + 1);
+ end if;
+ return C;
+ end;
+
+ -- nudge
+
+ procedure Nudge (L : in out Lexer) is
+ C : Character;
+ begin
+ C := Peek (L);
+ if Is_Line_Terminator (C) then
+ L.Line := L.Line + 1;
+ L.Col := 1;
+ else
+ L.Col := L.Col + 1;
+ end if;
+ L.Pos := L.Pos + 1;
+ end;
+
+ -- skip spaces
+
+ procedure Skip_Spaces_Comments (L : in out Lexer) is
+ C : Character;
+ begin
+ while true loop
+ while true loop
+ C := Peek (L);
+ if Is_Space (C) or Is_Line_Terminator (C) then
+ Nudge (L);
+ else
+ exit;
+ end if;
+ end loop;
+ if Peek (L) = '/' and Peek2 (L) = '/' then
+ while not Is_Line_Terminator (Peek (L)) loop
+ Nudge (L);
+ end loop;
+ goto Continue;
end if;
- C := SRC (I);
+ exit;
+ <<Continue>>
+ null;
+ end loop;
+ end Skip_Spaces_Comments;
+
+ -- next token
- I := I + 1;
+ function Next_Token (L : in out Lexer) return Token is
+ T : Token;
+ TK : Token_Kind;
+ begin
+ Skip_Spaces_Comments (L);
+ Put (Peek (L));
+ if L.Pos < L.Source'Length then
+ TK := Unknown;
+ else
+ TK := EOF;
+ end if;
+ T :=
+ (Kind => TK,
+ Lexeme => To_Unbounded_String (""),
+ Line => 1,
+ Col => 1);
+ Nudge (L);
+ return T;
+ end Next_Token;
+
+ -- add token
+
+ procedure Add_Token (L : in out Lexer; T : Token) is
+ begin
+ L.Tokens.Append (T);
+ end Add_Token;
+
+ procedure Lex (L : in out Lexer) is
+ Tok : Token;
+ begin
+
+ while true loop
+ Tok := Next_Token (L);
+ Add_Token (L, Tok);
+ exit when Tok.Kind = EOF;
end loop;
- end;
+ -- Put_Line (Natural'Image (Length (L.Source)));
+
+ -- while I <= SRC'Last loop
+ -- if I < SRC'Last then
+ -- I2 := I + 1;
+ -- else
+ -- I2 := I;
+ -- end if;
+
+ -- C := SRC (I);
+
+ -- if Is_Space (C) or Is_Line_Terminator (C) then
+ -- if C = ASCII.LF then
+ -- Put_Line ("found return at line:" & I'Image);
+ -- end if;
+ -- end if;
+ -- if Is_Letter (C) then
+
+ -- null;
+ -- end if;
+
+ -- I := I + 1;
+ -- end loop;
+
+ end Lex;
end Lexer;
diff --git a/src/lexer.ads b/src/lexer.ads
@@ -2,12 +2,17 @@ with Ada.Containers.Vectors;
with Ada.Strings.Unbounded; use Ada.Strings.Unbounded;
package Lexer is
-
- type Token_Type is
- (Ident, Float_Literal, Int_Literal, Str_Literal, Bool_Literal, Unknown);
+ type Token_Kind is
+ (Ident,
+ Float_Literal,
+ Int_Literal,
+ Str_Literal,
+ Bool_Literal,
+ Unknown,
+ EOF);
type Token is record
- Kind : Token_Type;
+ Kind : Token_Kind;
Lexeme : Unbounded_String;
Line, Col : Positive;
end record;
@@ -15,10 +20,13 @@ package Lexer is
package Token_Vectors is new
Ada.Containers.Vectors (Index_Type => Natural, Element_Type => Token);
- type Lexer is record
- Source : Unbounded_String;
- File_Name : Unbounded_String;
+ type Lexer(Src_Len: Positive; FN_Len: Positive) is record
+ Source : String(1 .. Src_Len);
+ File_Name : String(1 .. FN_Len);
Tokens : Token_Vectors.Vector;
+ Pos : Positive;
+ Line : Positive;
+ Col : Positive;
end record;
function Init (File_Name : String; File_Contents : String) return Lexer;
diff --git a/src/nightshade.adb b/src/nightshade.adb
@@ -18,6 +18,7 @@ begin
L : Lexer.Lexer := Lexer.Init (File_Name, File_Contents);
begin
Lexer.Lex (L);
+ Put_Line ("");
Put_Line ("done");
end;
end Nightshade;
diff --git a/test.mty b/test.mty
@@ -0,0 +1,8 @@
+use io
+
+// this is a test program
+
+fx main() ::
+ print("hello")
+end
+
diff --git a/test.txt b/test.txt
@@ -1,6 +0,0 @@
-This is some text to be injected.
-
-Hello world. I am sam.
-
-I am not sam, but for the sake of this example, I am Andre.
-
