larpcc/lex.c

#include "parse.h"
#include <assert.h>
#include <stddef.h>
#include <stdio.h>
#include <string.h>

void source_char_iter_init(SourceCharIter* r, const char* text)
{
    *r = (SourceCharIter) {
        .text = text,
        .i = 0,
        .line = 1,
        .col = 1,
    };
}

Char source_iter_next(SourceCharIter* r)
{
    char ch = r->text[r->i];
    int line = r->line;
    int col = r->col;

    if (r->text[r->i] != '\0') {
        if (ch == '\n') {
            r->line += 1;
            r->col = 1;
        } else {
            r->col += 1;
        }
        r->i += 1;
    }

    return (Char) { ch, line, col };
}

void lsl_iter_init(LslCharIter* r, SourceCharIter* source_char_iter)
{
    Char curr = source_iter_next(source_char_iter);

    *r = (LslCharIter) {
        .source_char_iter = source_char_iter,
        .curr = curr,
    };
}

Char lsl_iter_next(LslCharIter* r)
{
    Char next = source_iter_next(r->source_char_iter);

    if (r->curr.ch == '\\' && next.ch == '\n') {
        r->curr = source_iter_next(r->source_char_iter);
        return lsl_iter_next(r);
    }

    Char lsl_char = { r->curr.ch, r->curr.line, r->curr.col };

    r->curr = next;
    return lsl_char;
}

void lexer_init(Lexer* r, LslCharIter* lsl_char_iter)
{
    Char curr = lsl_iter_next(lsl_char_iter);

    *r = (Lexer) {
        .lsl_char_iter = lsl_char_iter,
        .curr = curr,
        .lex_whitespace = false,
        .lex_headername = false,
    };
}

static Tok tok(TokTy ty, const char* text, int line, int col)
{
    return (Tok) { ty, strdup(text), line, col };
}

static bool done(Lexer* l)
{
    return l->curr.ch == '\0';
}

constexpr size_t max_token_text_len = 64;

static void step(Lexer* l, char text[], size_t* text_len)
{
    assert(*text_len < max_token_text_len);
    text[*text_len] = l->curr.ch;
    *text_len += 1;

    l->curr = lsl_iter_next(l->lsl_char_iter);
}

Tok lexer_next(Lexer* l)
{
    int line = l->curr.line;
    int col = l->curr.col;

    char text[max_token_text_len] = "";
    size_t text_len = 0;

    if (done(l)) {
        return tok(Tok_Eof, text, line, col);
    }

    bool matched = false;
    while (strchr(" \t\r", l->curr.ch)) {
        matched = true;
        step(l, text, &text_len);
    }
    if (matched) {
        if (l->lex_whitespace) {
            return tok(Tok_Whitespace, text, line, col);
        }
        return lexer_next(l);
    }

    while ((l->curr.ch >= 'a' && l->curr.ch <= 'z')
        || (l->curr.ch >= 'A' && l->curr.ch <= 'Z')
        || (matched && l->curr.ch >= '0' && l->curr.ch <= '9')
        || strchr("_", l->curr.ch)) {
        matched = true;
        step(l, text, &text_len);
    }
    if (matched) {
        return tok(Tok_Ident, text, line, col);
    }

    while ((l->curr.ch >= '1' && l->curr.ch <= '9')
        || (matched && l->curr.ch == '0')) {
        matched = true;
        step(l, text, &text_len);
    }
    if (matched) {
        return tok(Tok_Int, text, line, col);
    }

    if (l->lex_headername && (l->curr.ch == '"' || l->curr.ch == '<')) {
        char ter = l->curr.ch == '<' ? '>' : '"';
        while (l->curr.ch != '\0' && l->curr.ch != '\n' && l->curr.ch != ter) {
            step(l, text, &text_len);
        }
        if (l->curr.ch != ter) {
            fprintf(stderr,
                "error: malformed header name, expected '%c' on line %d\n",
                ter,
                l->curr.line);
            return lexer_next(l);
        }
        step(l, text, &text_len);
        return tok(Tok_HeaderName, text, line, col);
    }

    if (l->curr.ch == '"') {
        step(l, text, &text_len);
        while (l->curr.ch != '"') {
            if (l->curr.ch == '\\') {
                step(l, text, &text_len);
                if (done(l))
                    break;
            }
            step(l, text, &text_len);
        }
        if (done(l) || l->curr.ch != '"') {
            fprintf(stderr,
                "error: malformed string literal on line %d\n",
                l->curr.line);
            return lexer_next(l);
        }
        step(l, text, &text_len);
        return tok(Tok_Str, text, line, col);
    }

    if (l->curr.ch == '\'') {
        step(l, text, &text_len);
        if (l->curr.ch == '\\') {
            step(l, text, &text_len);
        }
        if (!done(l)) {
            step(l, text, &text_len);
        }
        if (done(l) && l->curr.ch != '\'') {
            fprintf(stderr,
                "error: malformed character literal on line %d\n",
                l->curr.line);
            return lexer_next(l);
        }
        step(l, text, &text_len);
        return tok(Tok_Char, text, line, col);
    }

    if (strchr("\n;(){}#", l->curr.ch)) {
        TokTy ty = (TokTy)l->curr.ch;
        step(l, text, &text_len);
        return tok(ty, text, line, col);
    }
    fprintf(stderr,
        "error: illegal character '%c' on line %d\n",
        l->curr.ch,
        l->curr.line);
    step(l, text, &text_len);
    return lexer_next(l);
}