shithub: lpa

ref: dcb761aebd2815fdf2d04c0b05724292de9dc98b
dir: /scan.c/

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

#include "dat.h"
#include "fns.h"

struct {
	Rune spelling;
	int id;
} primspecs[] = {
	L'+', PrimPlus,
	L'-', PrimMinus,
};

Token *
newtok(TokenList *tokens, int tag)
{
	Token *new;

	tokens->count++;
	tokens->tokens = allocextra(tokens, sizeof(Token) * tokens->count);
	new = tokens->tokens + (tokens->count-1);
	new->tag = tag;

	return new;
}

TokenList *
scan(char *buf, char **errp)
{
	Rune r;
	int n;
	TokenList *tokens = alloc(DataTokenList);
	Token *tok;
	char *cp = buf;

	while(*cp){
		n = chartorune(&r, cp);
		int new = -1;
		switch(r){
		case L'(': new = TokLparen; break;
		case L')': new = TokRparen; break;
		case L'[': new = TokLbrack; break;
		case L']': new = TokRbrack; break;
		case L'{': new = TokLbrace; break;
		case L'}': new = TokRbrace; break;
		case L'\n': new = TokNewline; break;
		case L'⋄': new = TokDiamond; break;
		case L'∇': new = TokDel; break;
		case L'←': new = TokLarrow; break;
		case L';': new = TokSemi; break;
		}
		if(new != -1){
			newtok(tokens, new);
			goto next;
		}
		for(int i = 0; i < nelem(primspecs); i++){
			if(r == primspecs[i].spelling){
				tok = newtok(tokens, TokPrimitive);
				tok->prim = primspecs[i].id;
				tok->nameclass = NameclassFunc;
				goto next;
			}
		}
		if(isspacerune(r))
			goto next;
		if(isdigitrune(r)){
			char *rest;
			vlong num = strtoll(cp, &rest, 10);
			n = rest - cp;
			tok = newtok(tokens, TokNumber);
			tok->num = num;
			goto next;
		}
		if(isalpharune(r)){
			char *start = cp;
			do{
				cp += n;
				n = chartorune(&r, cp);
			}while(isalpharune(r) || isdigitrune(r));
			tok = newtok(tokens, TokName);
			usize size = cp - start;
			tok->name = malloc(size + 1);
			memcpy(tok->name, start, size);
			tok->name[size] = 0;
			continue;
		}
		*errp = "scan error";
		return nil;
next:
		cp += n;
	}
	newtok(tokens, TokEnd);
	if(tokens){
		print("token tags: ");
		for(uvlong i = 0; i < tokens->count; i++)
			print("%d ", tokens->tokens[i].tag);
		print("\n");
	}
	return tokens;
}