[PATCH 034/188] updating wat scanner

From 428c7173cf6a7d363ffa8e1ba748337a277189d3 Mon Sep 17 00:00:00 2001
From: Dmitry Boyarintsev <skalogryz.lists@gmail.com>
Date: Tue, 19 Nov 2019 23:26:17 -0500

git-svn-id: branches/wasm@46030 -
This commit is contained in:
nickysn 2020-08-03 12:59:19 +00:00
parent a676023c7f
commit f7ae46356b

View File

@ -5,59 +5,180 @@ unit watparser;
interface
uses
SysUtils, Classes, parseutils;
SysUtils, Classes, parseutils, wasmtext;
type
TWatEntity = (weNone, weError,
weIdent, weString, weNumber, weSymbol, weOpenBrace, weCloseBrace);
TWatToken = (weNone, weError,
weIdent,
weString, weNumber, weOpenBrace, weCloseBrace,
weAsmSymbol,
{ TWatParser }
weInstr,
weFunc,
weParam, weResult,
weModule, weMut, weFuncRef,
wei32, wei64,
wef32, wef64,
weType,
weImport, weGlobal, weTable, weMemory, weLocal, weExport,
weElem, weData, weOffset
);
TWatParser = class(TObject)
{ TWatScanner }
TWatScanner = class(TObject)
protected
procedure DoComment(const cmt: string);
function CommentIsSymbol(const cmt: string): Boolean;
public
buf : string;
idx : integer;
buf : string;
idx : integer;
ofs : integer;
entity : TWatEntity;
resText : string;
instrCode : byte;
ofs : integer;
token : TWatToken;
resText : string;
procedure SetSource(const abuf: string);
function Next: Boolean;
end;
const
SymbolChars : TCharSet = ['(',')'];
AlphaNumCharsUnd : TCharSet = AlphaNumChars + ['_','$'];
// see Identifiers of Textual format
IdStart = '$';
IdBody = AlphaNumChars
+ [ '!' ,'#' ,'$' ,'%' ,'&' ,'''' ,'*'
,'+' ,'-' ,'.' ,'/' ,':' ,'<' ,'='
,'>' ,'?' ,'@' ,'\' ,'^' ,'_' ,'`'
,'|' ,'~'];
GrammarChars = AlphaNumChars+['.','_'];
procedure GetGrammar(const txt: string; out entity: TWatToken; out instByte: byte);
const
KEY_MODULE = 'module';
KEY_FUNC = 'func';
KEY_FUNCREF = 'funcref';
KEY_I32 = 'i32';
KEY_I64 = 'i64';
KEY_F32 = 'f32';
KEY_F64 = 'f64';
KEY_PARAM = 'param';
KEY_RESULT = 'result';
KEY_MUT = 'mut';
KEY_TYPE = 'type';
KEY_IMPORT = 'import';
KEY_GLOBAL = 'global';
KEY_TABLE = 'table';
KEY_MEMORY = 'memory';
KEY_LOCAL = 'local';
KEY_EXPORT = 'export';
KEY_ELEM = 'elem';
KEY_DATA = 'data';
KEY_OFFSET = 'offset';
function ScanString(const buf: string; var idx: integer): string;
implementation
{ TWatParser }
procedure GetGrammar(const txt: string; out entity: TWatToken; out instByte: byte);
begin
instByte:=0;
entity:=weError;
if txt='' then Exit;
case txt[1] of
'a':
if txt='anyfunc' then entity:=weFuncRef
else if TextToInst(txt, instByte) then entity:=weInstr;
'd':
if txt=KEY_DATA then entity:=weData
else if TextToInst(txt, instByte) then entity:=weInstr;
'e':
if txt=KEY_EXPORT then entity:=weExport
else if txt=KEY_ELEM then entity:=weElem
else if TextToInst(txt, instByte) then entity:=weInstr;
'i':
if txt=KEY_I32 then entity:=wei32
else if txt=KEY_I64 then entity:=wei64
else if txt=KEY_IMPORT then entity:=weImport
else if TextToInst(txt, instByte) then entity:=weInstr;
'g':
if txt=KEY_GLOBAL then entity:=weGlobal
else if TextToInst(txt, instByte) then entity:=weInstr;
'f':
if txt=KEY_FUNC then entity:=weFunc
else if txt=KEY_FUNCREF then entity:=weFuncRef
else if txt=KEY_F32 then entity:=wef32
else if txt=KEY_F64 then entity:=wef64
else if TextToInst(txt, instByte) then entity:=weInstr;
'l':
if txt=KEY_LOCAL then entity:=weLocal
else if TextToInst(txt, instByte) then entity:=weInstr;
'm':
if txt=KEY_MODULE then entity:=weModule
else if txt = KEY_MUT then entity:=weMut
else if txt = KEY_MEMORY then entity:=weMemory
else if TextToInst(txt, instByte) then entity:=weInstr;
'o':
if txt=KEY_OFFSET then entity:=weOffset
else if TextToInst(txt, instByte) then entity:=weInstr;
'p':
if txt=KEY_PARAM then entity:=weParam
else if TextToInst(txt, instByte) then entity:=weInstr;
'r':
if txt=KEY_RESULT then entity:=weResult
else if TextToInst(txt, instByte) then entity:=weInstr;
't':
if txt=KEY_TYPE then entity:=weType
else if txt=KEY_TABLE then entity:=weTable
else if TextToInst(txt, instByte) then entity:=weInstr;
else
if TextToInst(txt, instByte) then entity:=weInstr;
end;
end;
procedure TWatParser.DoComment(const cmt: string);
{ TWatScanner }
procedure TWatScanner.DoComment(const cmt: string);
begin
end;
function TWatParser.CommentIsSymbol(const cmt: string): Boolean;
function TWatScanner.CommentIsSymbol(const cmt: string): Boolean;
begin
Result := false;
end;
procedure TWatParser.SetSource(const abuf: string);
procedure TWatScanner.SetSource(const abuf: string);
begin
buf:=abuf;
idx:=1;
end;
function TWatParser.Next: Boolean;
function ScanString(const buf: string; var idx: integer): string;
var
j : integer;
begin
if buf[idx]<>'"' then begin
Result:='';
Exit;
end;
j:=idx;
inc(idx);
while (buf[idx]<>'"') and (idx<length(buf)) do begin
if buf[idx]='\' then inc(idx);
inc(idx);
end;
inc(idx);
Result:=Copy(buf, j, idx-j);
end;
function TWatScanner.Next: Boolean;
var
has2chars: Boolean;
cmt : string;
done: boolean;
j: integer;
begin
Result := idx<=length(buf);
if not Result then Exit;
@ -65,10 +186,10 @@ begin
done:=false;
resText:='';
while not done do begin
ScanWhile(buf, idx, WhiteSpaceChars);
ScanWhile(buf, idx, SpaceEolnChars);
Result := idx<=length(buf);
if not Result then Exit;
j:=idx;
ofs:=idx;
has2chars := idx<length(buf);
if has2chars then begin
if (buf[idx]=';') and (buf[idx+1]=';') then begin
@ -80,7 +201,7 @@ begin
cmt := ScanToSubstr(buf, idx, ';)');
if CommentIsSymbol(cmt) then begin
entity:=weSymbol;
token:=weAsmSymbol;
done:=true;
end else
DoComment(cmt);
@ -89,19 +210,26 @@ begin
if not done then begin
done:=true;
if buf[idx] = '(' then begin
entity:=weOpenBrace;
token:=weOpenBrace;
inc(idx);
end else if buf[idx]=')' then begin
entity:=weCloseBrace;
token:=weCloseBrace;
inc(idx);
end else if buf[idx] in AlphabetChars then begin
entity:=weIdent;
resText:=ScanWhile(buf, idx, AlphaNumCharsUnd);
end else if buf[idx]='"' then begin
token:=weString;
resText:=ScanString(buf, idx);
end else if buf[idx] = IdStart then begin
token:=weIdent;
resText:=ScanWhile(buf, idx, IdBody);
end else if buf[idx] in AlphaNumChars then begin
resText:=ScanWhile(buf, idx, GrammarChars);
GetGrammar(resText, token, instrCode);
done:=true;
end else if buf[idx] in NumericChars then begin
entity:=weNumber;
token:=weNumber;
resText:=ScanWhile(buf, idx, NumericChars);
end else begin
entity:=weError;
token:=weError;
inc(idx);
done:=true;
end;
@ -109,7 +237,7 @@ begin
end;
if resText='' then
resText := Copy(buf, j, idx-j);
resText := Copy(buf, ofs, idx-ofs);
end;
end.