MaximeBarniaudy
/
minproto
forkeado de piumarta/minproto
1
0
Fork 0
Código Incidencias 0 Pull Requests 0 Lanzamientos 0 Wiki Actividad
Explorar el Código

Add emitByteCode functions to support using the peg VM, add support for @ actions, automatically generate the minproto.grammar file from minproto.leg and use that in the parser generator. Fix issue with backslashes being stupid.

master
MaximeBarniaudy hace 10 meses
padre
c3125a863f
commit
168631e246
Se han modificado 6 ficheros con 1051 adiciones y 111 borrados
Dividir vista
Opciones de diferencias
Mostrar estadísticas Descargar archivo de parche Descargar archivo de diferencias
  1. +4
    -1
      Makefile
  2. +682
    -79
      grammar_parser.meta
  3. +316
    -0
      minproto.grammar
  4. +27
    -18
      minproto.leg
  5. +6
    -5
      rawgrammar.leg
  6. +16
    -8
      rawminproto.leg

+ 4
- 1
Makefile Ver fichero

@ -15,7 +15,7 @@ MAIN = minproto
all : $(MAIN)
% : %.c
% : %.c %.grammar
$(CC) $(GFLAGS) $(CFLAGS) $(LDFLAGS) -o $@ $< $(LDLIBS)
%-opt : %.c
@ -27,6 +27,9 @@ all : $(MAIN)
%.c : %.leg
leg -o $@ $<
%.grammar : %.leg
( sed "/%{/,/%}/d" $< | sed "/%%/,\$$d" ) > $@
tests : .FORCE
$(MAKE) clean all GFLAGS="-Wno-unused -g -DTYPECODES=1 -DELOPT=1"
-./$(MAIN) test.txt > test.out 2>&1

+ 682
- 79
grammar_parser.meta
La diferencia del archivo ha sido suprimido porque es demasiado grande
Ver fichero


+ 316
- 0
minproto.grammar Ver fichero

@ -0,0 +1,316 @@
# minproto.leg -- minimal prototype langauge for semantic experiments
#
# last edited: 2024-07-04 10:07:00 by piumarta on zora
start = - ( s:stmt { global yysval = s }
| !. { global yysval = 0 }
| < (!EOL .)* > { syntaxError(yytext) }
)
stmt = WHILE LPAREN c:expr RPAREN s:stmt { $$ = newWhile(c, s) }
| IF LPAREN c:expr RPAREN s:stmt
( ELSE t:stmt { $$ = newIf(c, s, t ) }
| { $$ = newIf(c, s, nil) }
)
| CONT EOS { $$ = newContinue() }
| BREAK e:expr EOS { $$ = newBreak(e) }
| BREAK EOS { $$ = newBreak(nil) }
| RETURN e:expr EOS { $$ = newReturn(e) }
| RETURN EOS { $$ = newReturn(nil) }
| FOR LPAREN i:id IN e:expr RPAREN
s:stmt { $$ = newForIn(i, e, s) }
| FOR LPAREN i:id FROM a:expr
TO b:expr RPAREN s:stmt { $$ = newForFromTo(i, a, b, s) }
| FOR LPAREN i:expr SEMI c:expr SEMI
u:expr RPAREN s:stmt { $$ = newFor(i, c, u, s) }
| TRY t:stmt
( CATCH LPAREN i:id RPAREN c:stmt { $$ = newTryCatch(t, i, c) }
| ENSURE e:stmt { $$ = newTryEnsure(t, e) }
)
| RAISE e:expr EOS { $$ = newRaise(e) }
| LOCAL i:id p:params b:block { $$ = newSetLocal (i, newLambda(p, b, nil, i)) }
| GLOBAL i:id p:params b:block { $$ = newSetGlobal(i, newLambda(p, b, nil, i)) }
| i:id p:params b:block { $$ = newSetVar (i, newLambda(p, b, nil, i)) }
| v:proto DOT i:id p:params b:block { $$ = newSetProp(v, i, newLambda(p, b, v, i)) }
| b:block { $$ = newBlock(b) }
| e:expr EOS { $$ = e }
proto = v:var ( DOT j:id !LPAREN { v = newGetProp(v, j) }
)* { $$ = v }
EOS = SEMI+ | &RBRACE | &ELSE | &CATCH
expr = LOCAL i:id ASSIGN e:expr { $$ = newSetLocal (i, e) }
| GLOBAL i:id ASSIGN e:expr { $$ = newSetGlobal(i, e) }
| i:id ASSIGN e:expr { $$ = newSetVar (i, e) }
| l:logor ( ASSIGN r:expr { l = assign(l, r) }
| PLUSEQ r:expr { l = newBinop(opPreAdd, lvalue(l), r) }
| MINUSEQ r:expr { l = newBinop(opPreSub, lvalue(l), r) }
| STAREQ r:expr { l = newBinop(opPreMul, lvalue(l), r) }
| SLASHEQ r:expr { l = newBinop(opPreDiv, lvalue(l), r) }
| PCENTEQ r:expr { l = newBinop(opPreMod, lvalue(l), r) }
| SHLEQ r:expr { l = newBinop(opPreShl, lvalue(l), r) }
| SHREQ r:expr { l = newBinop(opPreShr, lvalue(l), r) }
| ANDEQ r:expr { l = newBinop(opPreAnd, lvalue(l), r) }
| XOREQ r:expr { l = newBinop(opPreXor, lvalue(l), r) }
| OREQ r:expr { l = newBinop(opPreOr, lvalue(l), r) }
)? { $$ = l }
logor = l:logand ( BARBAR r:logand { l = newBinop(opLogOr, l, r) }
)* { $$ = l }
logand = l:bitor ( ANDAND r:bitor { l = newBinop(opLogAnd, l, r) }
)* { $$ = l }
bitor = l:bitxor ( OR r:bitxor { l = newBinop(opBitOr, l, r) }
)* { $$ = l }
bitxor = l:bitand ( XOR r:bitand { l = newBinop(opBitXor, l, r) }
)* { $$ = l }
bitand = l:eq ( AND r:eq { l = newBinop(opBitAnd, l, r) }
)* { $$ = l }
eq = l:ineq ( EQ r:ineq { l = newBinop(opEq, l, r) }
| NOTEQ r:ineq { l = newBinop(opNotEq, l, r) }
)* { $$ = l }
ineq = l:shift ( LESS r:shift { l = newBinop(opLess, l, r) }
| LESSEQ r:shift { l = newBinop(opLessEq, l, r) }
| GRTREQ r:shift { l = newBinop(opGrtrEq, l, r) }
| GRTR r:shift { l = newBinop(opGrtr, l, r) }
)* { $$ = l }
shift = l:sum ( SHL r:sum { l = newBinop(opShl, l, r) }
| SHR r:sum { l = newBinop(opShr, l, r) }
)* { $$ = l }
sum = l:prod ( PLUS r:prod { l = newBinop(opAdd, l, r) }
| MINUS r:prod { l = newBinop(opSub, l, r) }
)* { $$ = l }
prod = l:prefix ( STAR r:prefix { l = newBinop(opMul, l, r) }
| SLASH r:prefix { l = newBinop(opDiv, l, r) }
| PCENT r:prefix { l = newBinop(opMod, l, r) }
) * { $$ = l }
prefix = PPLUS p:prefix { $$ = newBinop(opPreAdd, lvalue(p), newInteger(1)) }
| MMINUS p:prefix { $$ = newBinop(opPreSub, lvalue(p), newInteger(1)) }
| PLING p:prefix { $$ = newUnyop(opNot, p) }
| MINUS p:prefix { $$ = newUnyop(opNeg, p) }
| TILDE p:prefix { $$ = newUnyop(opCom, p) }
| BQUOTE s:stmt { $$ = newUnyop(opQuasiquote, s) }
| COMMAT e:expr { $$ = newUnyop(opUnquote, e) }
| postfix
postfix = SUPER DOT i:id a:args { $$ = newSuper(i, a) }
| p:primary
( LBRAK
( COLON ( RBRAK { p = newGetSlice(p, nil, nil) }
| e:xexpr RBRAK { p = newGetSlice(p, nil, e) }
)
| s:xexpr ( COLON ( RBRAK { p = newGetSlice(p, s, nil) }
| e:xexpr RBRAK { p = newGetSlice(p, s, e) }
)
| RBRAK { p = newGetArray(p, s) }
)
)
| DOT i:id ( a:args !LBRACE { p = newInvoke(p, i, a) }
| { p = newGetProp(p, i) }
)
| a:args !LBRACE { p = newApply(p, a) }
)*
( PPLUS { p = newBinop(opPostAdd, lvalue(p), newInteger( 1)) }
| MMINUS { p = newBinop(opPostAdd, lvalue(p), newInteger(-1)) }
)? { $$ = p }
args = LPAREN a:mkobj
( RPAREN
| ( k:id COLON e:xexpr { Object_put(a, k, e) }
| e:xexpr { Object_push(a, e) }
)
( COMMA ( k:id COLON e:xexpr { Object_put(a, k, e) }
| e:xexpr { Object_push(a, e) }
) )* RPAREN ) { $$ = a }
params = LPAREN p:mkobj
( RPAREN
| i:id ( COLON e:expr { Object_put(p, i, e) }
| { Object_push(p, i) }
)
( COMMA i:id ( COLON e:expr { Object_put(p, i, e) }
| { Object_push(p, i) }
)
)* RPAREN ) { $$ = p }
mkobj = { $$ = (global new)(pObject) }
primary = nil | number | string | symbol | var | lambda | subexpr | literal # | regex
lambda = p:params b:block { $$ = newLambda(p, b, nil, nil) }
subexpr = LPAREN e:expr RPAREN { $$ = e }
| b:block { $$ = newBlock(b) }
literal = LBRAK o:mkobj
( RBRAK
| ( ( i:id COLON e:expr { Object_put(o, i, e) }
| e:expr { Object_push(o, e) }
) ( COMMA ( i:id COLON e:expr { Object_put(o, i, e) }
| e:expr { Object_push(o, e) }
) )* )? RBRAK ) { $$ = newLiteral(o) }
block = LBRACE b:mkobj
( e:stmt { Object_push(b, e) }
)* ( RBRACE { $$ = b }
| error @{ expected("statement or \x7D", yytext) }
)
nil = NIL { $$ = nil }
number = "-" n:unsign { $$ = neg(n) }
| "+" n:number { $$ = n }
| n:unsign { $$ = n }
unsign = < DIGIT* '.' DIGIT+ EXP? > - { $$ = newFloat(strtod(yytext, 0)) }
| "0" [bB] < BIGIT+ > - { $$ = newInteger(strtol(yytext, 0, 2)) }
| "0" [xX] < HIGIT+ > - { $$ = newInteger(strtol(yytext, 0, 16)) }
| "0" < OIGIT* > - { $$ = newInteger(strtol(yytext, 0, 8)) }
| < DIGIT+ > - { $$ = newInteger(strtol(yytext, 0, 10)) }
| "'" < char > "'" - { $$ = newInteger(_get(newStringUnescaped(yytext), String,value)[0]) }
string = '"' < ( !'"' char )* > '"' - { $$ = newStringUnescaped(yytext) }
char = '\\' [abefnrtv'"\[\]\\]
| '\\' [0-3][0-7][0-7]
| '\\' [xX] HIGIT*
| '\\' [0-7][0-7]?
| !'\\' .
# char = "\\" ( ["'\\abfnrtv]
# | [xX] HIGIT*
# | [0-7][0-7]?[0-7]?
# )
# | .
symbol = HASH i:id { $$ = i }
var = LOCAL i:id { $$ = newGetLocal (i) }
| GLOBAL i:id { $$ = newGetGlobal(i) }
| i:id { $$ = newGetVar (i) }
id = < LETTER ALNUM* > - { $$ = intern(yytext) }
# regex = SLASH a:alts SLASH { $$ = a }
# alts = s:seq ( OR t:seq { s = Alt_append(t) }
# )* { $$ = s }
# seq = p:pre ( q:pre { s = Seq_append(t) }
# )* { $$ = s }
# elt = action | pre
# action = b:block { $$ = newAction(b) }
# pre = PLING p:pre { $$ = newNot(p) }
# | AND p:pre { $$ = newAnd(p) }
# | post
# post = a:atom ( STAR { a = newMany(a) }
# | PLUS { a = newMore(a) }
# | QUERY { a = newMore(a) }
# )? { $$ = a }
# atom = DOT { $$ = newDot() }
# | "[" ( !"]" "\\"? . )* "]" - { $$ = newClass(yytext) }
# | '"' xxxxxx
# class = LBRAK
BIGIT = [0-1]
OIGIT = [0-7]
DIGIT = [0-9]
HIGIT = [0-9A-Fa-f]
LETTER = [A-Za-z_$?]
ALNUM = LETTER | DIGIT
SIGN = [-+]
EXP = [eE] SIGN DIGIT+
- = SPACE*
SPACE = [ \t] | EOL | SLC | MLC
EOL = [\n\r] { ++lineno }
SLC = "//" (!EOL .)*
MLC = "/*" ( MLC | !"*/" (EOL | .))* "*/" -
NIL = "nil" !ALNUM -
WHILE = "while" !ALNUM -
IF = "if" !ALNUM -
ELSE = "else" !ALNUM -
FOR = "for" !ALNUM -
IN = "in" !ALNUM -
FROM = "from" !ALNUM -
TO = "to" !ALNUM -
CONT = "continue" !ALNUM -
BREAK = "break" !ALNUM -
RETURN = "return" !ALNUM -
TRY = "try" !ALNUM -
CATCH = "catch" !ALNUM -
ENSURE = "ensure" !ALNUM -
RAISE = "raise" !ALNUM -
GLOBAL = "global" !ALNUM -
LOCAL = "local" !ALNUM -
SUPER = "super" !ALNUM -
BQUOTE = "`" -
COMMAT = "@" -
HASH = "#" -
SEMI = ";" -
ASSIGN = "=" ![=] -
COMMA = "," -
COLON = ":" ![:] -
LPAREN = "(" -
RPAREN = ")" -
LBRAK = "[" -
RBRAK = "]" -
LBRACE = "{" -
RBRACE = "}" -
BARBAR = "||" ![=] -
ANDAND = "&&" ![=] -
OR = "|" ![|=] -
OREQ = "|=" -
XOR = "^" ![=] -
XOREQ = "^=" -
AND = "&" ![&=] -
ANDEQ = "&=" -
EQ = "==" -
NOTEQ = "!=" -
LESS = "<" ![<=] -
LESSEQ = "<=" -
GRTREQ = ">=" -
GRTR = ">" ![=] -
SHL = "<<" ![=] -
SHLEQ = "<<=" -
SHR = ">>" ![=] -
SHREQ = ">>=" -
PLUS = "+" ![+=] -
PLUSEQ = "+=" -
PPLUS = "++" -
MINUS = "-" ![-=] -
MINUSEQ = "-=" -
MMINUS = "--" -
STAR = "*" ![=] -
STAREQ = "*=" -
SLASH = "/" ![/=] -
SLASHEQ = "/=" -
PCENT = "%" ![=] -
PCENTEQ = "%=" -
DOT = "." ![.] -
PLING = "!" ![=] -
TILDE = "~" -
error = - < (!EOL .)* >
xexpr = expr | error @{ expected("expression", yytext) }

+ 27
- 18
minproto.leg Ver fichero

@ -3394,10 +3394,12 @@ void expected(char *what, char *where)
fatal("syntax error: %s expected near: %s", what, where);
}
#define global
%}
start = - ( s:stmt { yysval = s }
| !. { yysval = 0 }
start = - ( s:stmt { global yysval = s }
| !. { global yysval = 0 }
| < (!EOL .)* > { syntaxError(yytext) }
)
@ -3537,7 +3539,7 @@ params = LPAREN p:mkobj
)
)* RPAREN ) { $$ = p }
mkobj = { $$ = new(pObject) }
mkobj = { $$ = (global new)(pObject) }
primary = nil | number | string | symbol | var | lambda | subexpr | literal # | regex
@ -3575,11 +3577,16 @@ unsign = < DIGIT* '.' DIGIT+ EXP? > - { $$ = newFloat(strtod(yytext, 0
string = '"' < ( !'"' char )* > '"' - { $$ = newStringUnescaped(yytext) }
char = "\\" ( ["'\\abfnrtv]
| [xX] HIGIT*
| [0-7][0-7]?[0-7]?
)
| .
char = '\\' [abefnrtv'"\[\]\\]
| '\\' [0-3][0-7][0-7]
| '\\' [xX] HIGIT*
| '\\' [0-7][0-7]?
| !'\\' .
# char = "\\" ( ["'\\abfnrtv]
# | [xX] HIGIT*
# | [0-7][0-7]?[0-7]?
# )
# | .
symbol = HASH i:id { $$ = i }
@ -3695,7 +3702,6 @@ SLASHEQ = "/=" -
PCENT = "%" ![=] -
PCENTEQ = "%=" -
DOT = "." ![.] -
DOTDOT = ".." -
PLING = "!" ![=] -
TILDE = "~" -
@ -3705,6 +3711,7 @@ xexpr = expr | error @{ expected("expression", yytext) }
%%
;
#undef global
#if PROFILE
@ -4389,7 +4396,8 @@ oop prim_eval(oop func, oop self, oop args, oop env)
oop *indexed = _get(args, Object,indexed);
oop result = nil;
if (nil != Object_getLocal(args, sym_env)) {
//if (nil != Object_getLocal(args, sym_env)) {
if (Object_find(args, sym_env) >= 0) {
env = Object_getLocal(args, sym_env);
}
@ -4908,7 +4916,7 @@ typedef struct vmState
oop variables;
} vmState;
#define VM_STATE_INITIALISER { nil, nil }
#define VM_STATE_INITIALISER { nil, new(pObject) }
void vmEnter(vmState *state, oop obj, char *yytext, int yyleng)
{
@ -4923,7 +4931,7 @@ void vmSet(vmState *state, oop obj, char *yytext, int yyleng)
void vmAction(vmState *state, oop obj, char *yytext, int yyleng)
{
oop text = yyleng ? newStringLen(yytext, yyleng) : nil;
oop text = yyleng >= 0 ? newStringLen(yytext, yyleng) : nil;
Object_put(state->variables, sym_yytext, text);
Object_put(state->variables, sym_yyleng, newInteger(yyleng));
applyThunkIn(obj, state->variables);
@ -4933,6 +4941,7 @@ void vmLeave(vmState *state, oop obj, char *yytext, int yyleng)
{
state->result = Object_getLocal(state->variables, sym_$$);
state->variables = _getDelegate(state->variables);
Object_put(state->variables, sym_$$, state->result);
}
void vmDisassemble(vmInsn *code, int pc)
@ -5130,6 +5139,8 @@ int vmRun(oop grammar0, oop symbol, char *text, int start, int length)
C##stack[--C##sp]; \
})
saveAction(vmEnter, nil, 0, 0);
for (;;) {
if (opt_d) vmDisassemble(frame.code, frame.pc);
vmInsn *i = frame.code + frame.pc++;
@ -5278,8 +5289,6 @@ int vmRun(oop grammar0, oop symbol, char *text, int start, int length)
break;
}
saveAction(vmLeave, nil, 0, 0);
#undef pop
#undef drop
#undef push
@ -5457,16 +5466,16 @@ int main(int argc, char **argv)
#define stringify(x) #x
#define declareOp(NAME, OP) _set(intern(stringify(__op##NAME)), Symbol,value, newInteger(op##NAME));
#define declareOp(NAME, OP) _set(intern(stringify(op##NAME)), Symbol,value, newInteger(op##NAME));
doBinops(declareOp)
#undef declareOp
#define declareOp(NAME, OP) _set(intern(stringify(__##NAME)), Symbol,value, newInteger(NAME));
#undef stringify
#define declareOp(NAME, OP) _set(intern(#NAME), Symbol,value, newInteger(NAME));
doUnyops(declareOp)
#undef declareOp
#undef stringify
#if TYPECODES
# define defineEvaluator(NAME) \

+ 6
- 5
rawgrammar.leg Ver fichero

@ -9,11 +9,12 @@ expression = s:sequence !BAR { s; }
( BAR s2:sequence { s1.push(s2); }
) * { s1; }
sequence = p:prefix { p = Sequence.new().push(p); }
sequence = p:prefix ( q:prefix { p = Sequence.new().push(p).push(q); }
( q:prefix { p.push(q); }
) * { p; }
) * ) ? { p; }
prefix = AND a:action { ParseTimeAction.new(action: a); }
| AT a:action { ExecuteAction.new(action: a); }
|
( AND s:suffix { And.new(expression: s); }
| NOT s:suffix { Not.new(expression: s); }
@ -33,8 +34,7 @@ primary = i1:identifier COLON i2:ruleCall !ASSIGN { Assignment.new(name:
| c:class { c; }
| DOT { Dot.new(); }
| a:action { a; }
| BEGIN { Begin.new(); }
| END { End.new(); }
| BEGIN e:expression END { Capture.new(expression: e); }
identifier = < [-a-zA-Z_][-a-zA-Z_0-9]* > - { intern(yytext); }
@ -53,7 +53,7 @@ char = '\\' [abefnrtv'"\[\]\\]
| '\\' [0-7][0-7]?
| !'\\' .
action = m:metaLanguage::block - { Action.new(parseTree: m); }
action = m:metaLanguage::block - { Action.new(parseTree: Block.new(body: m)); }
- = ( space | comment )*
space = ' ' | '\t' | end-of-line
@ -76,3 +76,4 @@ PLUS = "+" ![+=] -
STAR = "*" ![=] -
DOT = "." ![.] -
COLON = ":" ![:] -
AT = "@" -

+ 16
- 8
rawminproto.leg Ver fichero

@ -152,7 +152,7 @@ literal = LBRAK o:mkobj
block = LBRACE b:mkobj
( e:stmt { b.push(e) }
)* RBRACE { Block.new(body: b) }
)* RBRACE { b }
nil = NIL { nil }
@ -167,13 +167,21 @@ unsign = < DIGIT* '.' DIGIT+ EXP? > - { yytext.asFloat() }
| < DIGIT+ > - { yytext.asInteger() }
| "'" < char > "'" - { ord(yytext.unescaped()) }
string = '"' < ( !'"' char )* > '"' - { yytext.unescaped() }
string = '"' < ( !'"' char )* > '"' - { yytext }
char = "\\" ( ["'\\abfnrtv]
| [xX] HIGIT*
| [0-7][0-7]?[0-7]?
)
| .
# Version originale, qui ne parse pas deux antislash pour une raison obscure
# char = "\\" ( ["'\\abfnrtv]
# | [xX] HIGIT*
# | [0-7][0-7]?[0-7]?
# )
# | .
# Version de rawgrammar.leg, qui fonctionne sans problème
char = '\\' [abefnrtv'"\[\]\\]
| '\\' [0-3][0-7][0-7]
| '\\' [0-7][0-7]?
| '\\' [xX] HIGIT+
| !'\\' .
symbol = HASH i:id { i }
@ -187,7 +195,7 @@ BIGIT = [0-1]
OIGIT = [0-7]
DIGIT = [0-9]
HIGIT = [0-9A-Fa-f]
LETTER = [A-Za-z_]
LETTER = [A-Za-z_$?]
ALNUM = LETTER | DIGIT
SIGN = [-+]
EXP = [eE] SIGN DIGIT+

Escribir Vista previa
Cargando…
Cancelar
Guardar