path: root/lexer.c

#include <u.h>
#include <libc.h>
#include <bio.h>

#include "apl9.h"

enum {
	InputTypeBio,
	InputTypeString,
};

typedef struct InputStream InputStream;
struct InputStream
{
	int tag;
	int offset;
	Rune last;
	Biobuf *bio;
	Rune *string;
};

int inputEOF(InputStream *);
Rune getrune(InputStream *);
void ungetrune(InputStream *);

Statement *lexline(InputStream *, int);

Statement *
lexlinebio(Biobuf *bio, int toplevel)
{
	InputStream in;
	in.tag = InputTypeBio;
	in.offset = 0;
	in.last = 0;
	in.bio = bio;
	in.string = nil;
	return lexline(&in, toplevel);
}

Statement *
lexlinestr(Rune *str, int toplevel)
{
	InputStream in;
	in.tag = InputTypeString;
	in.offset = 0;
	in.last = 0;
	in.bio = nil;
	in.string = str;
	return lexline(&in, toplevel);	
}

Statement *
lexline(InputStream *input, int toplevel)
{
	Statement *stmt = emalloc(sizeof(Statement));
	stmt->ntoks = 0;
	stmt->toks = mallocz(sizeof(Datum) * MAX_LINE_TOKENS, 1);
	stmt->guard = nil;
	stmt->next = nil;

	Rune peek = getrune(input);
	while(!inputEOF(input) && (peek != '\n' || toplevel == 0)){
		Rune *p;
		if(isspacerune(peek) && peek != '\n'){
			peek = getrune(input);
			continue;
		}else if(runestrchr(L"←⋄\n⍝⍬", peek)){
			switch(peek){
			case L'←': stmt->toks[stmt->ntoks].tag = ArrowTag; break;
			case L'\n':
			case L'⋄':
				if(stmt->ntoks == 0)
					stmt = lexline(input, toplevel);
				else
					stmt->next = lexline(input, toplevel);
				goto end;
			case L'⍝': goto end;
			case L'⍬':
				stmt->toks[stmt->ntoks].tag = ArrayTag;
				stmt->toks[stmt->ntoks].array = allocarray(AtypeInt, 1, 0);
				stmt->toks[stmt->ntoks].array->shape[0] = 0;
				break;
			}
		}else if(!toplevel && peek == ':'){
			Rune buf[MAX_LINE_LENGTH];
			Rune *p = buf;
			while((peek = getrune(input)) != L'⋄' && peek != '\n' && !inputEOF(input))
				*p++ = peek;
			*p = 0;
			ungetrune(input);
			stmt->guard = lexlinestr(buf, toplevel);
			stmt->ntoks--;
		}else if(peek == '{'){
			int unclosed = 1;
			int oplevel = 0; /* 1 = monadic operator, 2 = dyadic operator */
			Rune buf[MAX_LINE_LENGTH];
			Rune *p = buf;
			while(((peek = getrune(input)) != '}' || unclosed > 1) && !inputEOF(input)){
				if(unclosed == 1 && peek == L'⍶' && oplevel == 0)
					oplevel = 1;
				else if(unclosed && peek == L'⍹')
					oplevel = 2;
				if(peek == '{')
					unclosed++;
				else if(peek == '}')
					unclosed--;
				*p++ = peek;
			}
			if(peek != '}')
				goto syntax_error;
			*p = 0;
			if(oplevel == 0){
				stmt->toks[stmt->ntoks].tag = FunctionTag;
				stmt->toks[stmt->ntoks].func.type = FunctypeDfn;
				stmt->toks[stmt->ntoks].func.dfn = runestrdup(buf);
			}else{
				stmt->toks[stmt->ntoks].tag = oplevel == 1 ? MonadicOpTag : DyadicOpTag;
				stmt->toks[stmt->ntoks].operator.type = OperatortypeDop;
				stmt->toks[stmt->ntoks].operator.dyadic = oplevel == 2;
				stmt->toks[stmt->ntoks].operator.dop = runestrdup(buf);
			}
		}else if(peek == '('){
			int unclosed = 1;
			Rune buf[MAX_LINE_LENGTH];
			Rune *p = buf;
			while(((peek = getrune(input)) != ')' || unclosed > 1) && !inputEOF(input) && peek != '\n'){
				if(peek == '(')
					unclosed++;
				else if(peek == ')')
					unclosed--;
				*p++ = peek;
			}
			if(peek != ')')
				goto syntax_error;
			*p = 0;
			stmt->toks[stmt->ntoks].tag = LParTag;
			stmt->toks[stmt->ntoks].stmt = *lexlinestr(buf, toplevel);
			stmt->ntoks++;
			stmt->toks[stmt->ntoks].tag = RParTag;
		}else if(p = runestrchr(primfuncnames, peek)){
			stmt->toks[stmt->ntoks].tag = FunctionTag;
			stmt->toks[stmt->ntoks].func.type = FunctypePrim;
			stmt->toks[stmt->ntoks].func.code = p-primfuncnames;
		}else if(p = runestrchr(primmonopnames, peek)){
			stmt->toks[stmt->ntoks].tag = MonadicOpTag;
			stmt->toks[stmt->ntoks].operator.type = OperatortypePrim;
			stmt->toks[stmt->ntoks].operator.dyadic = 0;
			stmt->toks[stmt->ntoks].operator.code = p-primmonopnames;
		}else if(p = runestrchr(primdyadopnames, peek)){
			stmt->toks[stmt->ntoks].tag = DyadicOpTag;
			stmt->toks[stmt->ntoks].operator.type = OperatortypePrim;
			stmt->toks[stmt->ntoks].operator.dyadic = 1;
			stmt->toks[stmt->ntoks].operator.code = p-primdyadopnames;
		}else if(p = runestrchr(primhybridnames, peek)){
			stmt->toks[stmt->ntoks].tag = HybridTag;
			stmt->toks[stmt->ntoks].hybrid = p-primhybridnames;
		}else if(isdigitrune(peek) || peek == L'¯'){
			char buf[64];
			char *p = buf;
			int floating = 0;
			if(peek == L'¯'){
				*p++ = '-';
				peek = getrune(input);
			}
get_digits:
			while(isdigitrune(peek)){
				p += runetochar(p, &peek);
				peek = getrune(input);
			}
			if(!floating && peek == '.'){
				p += runetochar(p, &peek);
				peek = getrune(input);
				floating = 1;
				goto get_digits;
			}
			*p = 0;
			ungetrune(input);
			stmt->toks[stmt->ntoks].tag = ArrayTag;
			stmt->toks[stmt->ntoks].array = floating ? mkscalarfloat(atof(buf)) : mkscalarint(atoll(buf));
		}else if(runestrchr(L"⍺⍵⍶⍹", peek)){
			Rune name[2] = {peek, 0};
			stmt->toks[stmt->ntoks].tag = NameTag;
			stmt->toks[stmt->ntoks].symbol = getsym(name, 1);
		}else if(isalpharune(peek)){
			Rune buf[64];
			Rune *p = buf;
			while(isalpharune(peek) || isdigitrune(peek)){
				*p++ = peek;
				peek = getrune(input);
			}
			*p = 0;
			ungetrune(input);
			stmt->toks[stmt->ntoks].tag = NameTag;
			stmt->toks[stmt->ntoks].symbol = getsym(buf, 0);
		}else if(runestrchr(L"⎕⍞", peek)){
			/* quad names */
			Rune buf[64];
			Rune *p = buf;
			*p++ = peek;
			peek = getrune(input);
			while(isalpharune(peek)){
				*p++ = toupperrune(peek);
				peek = getrune(input);
			}
			*p = 0;
			ungetrune(input);
			int valid = 0;
			for(int i = 0; quadnames[i].name != nil && !valid; i++){
				if(runestrcmp(buf, quadnames[i].name) != 0)
					continue;
				valid = 1;
				stmt->toks[stmt->ntoks] = quadnamedatum(quadnames[i]);
			}
			if(!valid)
				goto syntax_error;
		}else if(peek == '\''){
			Rune buf[1024]; /* stupid limit on literal string lengths */
			Rune *b = buf;
			int done = 0;
			peek = getrune(input);
			while(!done && !inputEOF(input)){
				if(peek == '\''){
					peek = getrune(input);
					if(peek != '\''){
						*b = 0;
						done = 1;
						ungetrune(input);
					}else{
						*b++ = '\'';
						peek = getrune(input);
					}
				}else{
					*b++ = peek;
					peek = getrune(input);
				}
			}
			if(!done)
				goto syntax_error;
			stmt->toks[stmt->ntoks].tag = ArrayTag;
			if(runestrlen(buf) == 1)
				stmt->toks[stmt->ntoks].array = mkscalarrune(buf[0]);
			else
				stmt->toks[stmt->ntoks].array = mkrunearray(buf);
		}else{
			Rune *err;
syntax_error:
			err = runesmprint("Can't lex");
			free(stmt->toks);
			free(stmt);
			throwerror(err, ESyntax);
		}
		/* print("Got token: %S\n", ppdatum(stmt->toks[stmt->ntoks])); */
		stmt->ntoks++;
		peek = getrune(input);
	}
end:
	stmt->toks = realloc(stmt->toks, sizeof(Datum) * stmt->ntoks);
	return stmt;
}

int
inputEOF(InputStream *i)
{
	int eof;
	if(i->tag == InputTypeBio)
		eof = i->last == Beof;
	else
		eof = i->last == 0;
	/*if(eof) print("EOF\n");*/
	return eof;
}

Rune
getrune(InputStream *i)
{
	Rune r;
	if(i->tag == InputTypeBio)
		r = Bgetrune(i->bio);
	else{
		if(i->string[i->offset] == 0)
			r = 0;
		else
			r = i->string[i->offset++];
	}
	/* print("Get rune: '%C' (%d)\n", r, r); */
	i->last = r;
	return r;
}

void
ungetrune(InputStream *i)
{
	/* print("Unget rune: '%C' (%d)\n", i->last, i->last); */
	if(inputEOF(i))
		return;

	if(i->tag == InputTypeBio)
		Bungetrune(i->bio);
	else
		i->offset--;
}