thesis-lisp/compiler/parse.c

#include "parse.h"
#include "node.h"

#include <string.h>
#include <assert.h>
#include <stdlib.h>
#include <errno.h>
#include <stdio.h>

static int vm_str_parser_peek(struct vm_parser *p) {
    return *(const char *) p->ctx;
}

static int vm_str_parser_pop(struct vm_parser *p) {
    char c = *(const char *) p->ctx;
    if (c) {
        ++p->ctx;
    }
    return c;
}

static inline int is_space(int ch) {
    return ch == ' ' || ch == '\t' || ch == '\n';
}

static inline int is_digit(int ch) {
    return ch >= '0' && ch <= '9';
}

static inline int is_ident0(int ch) {
    return (ch >= 'a' && ch <= 'z') ||
           (ch >= 'A' && ch <= 'Z') ||
           strchr("+-*&^%$@!~<>/=.", ch);
}

static int skip_whitespace(struct vm_parser *in) {
    while (is_space(in->peek(in))) {
        assert(in->pop(in) > 0);
    }
    return 0;
}

static int vm_parse_int(struct vm_parser *in, enum node_type *_type, intmax_t *_value) {
    int ch;
    intmax_t value = 0;
    //char type_str[16];
    //size_t idx;
    enum node_type type;

    ch = in->peek(in);
    if (ch == '0') {
        assert(in->pop(in) == ch);
        ch = in->peek(in);

        if (ch == 'x') {
            assert(in->pop(in) == ch);

            assert(0 && "TODO hex\n");
        }
    } else {
        while (1) {
            ch = in->peek(in);
            if (!ch || (ch >= 'a' && ch <= 'z') || strchr(" \n\t()", ch)) {
                break;
            }
            if (!is_digit(ch)) {
                return -EINVAL;
            }
            assert(in->pop(in) == ch);

            value *= 10;
            value += ch - '0';
        }
    }

    type = N_INTEGER;
    //ch = in->peek(in);
    //idx = 0;
    //if (ch >= 'a' && ch <= 'z') {
    //    while ((ch = in->peek(in)) && ((ch >= 'a' && ch <= 'z') || (ch >= '0' && ch <= '9'))) {
    //        assert(idx < sizeof(type_str) - 1);
    //        assert(in->pop(in) == ch);
    //        type_str[idx++] = ch;
    //    }
    //    type_str[idx] = 0;
    //}

    //if (idx) {
    //    if (!strcmp(type_str, "usize") || !strcmp(type_str, "u")) {
    //        type = VT_USIZE;
    //    } else if (!strcmp(type_str, "i8")) {
    //        type = VT_I8;
    //    } else if (!strcmp(type_str, "u8")) {
    //        type = VT_U8;
    //    } else if (!strcmp(type_str, "i16")) {
    //        type = VT_I16;
    //    } else if (!strcmp(type_str, "u16")) {
    //        type = VT_U16;
    //    } else if (!strcmp(type_str, "i32")) {
    //        type = VT_I32;
    //    } else if (!strcmp(type_str, "u32")) {
    //        type = VT_U32;
    //    } else {
    //        panic("Unknown type suffix: %s\n", type_str);
    //    }
    //} else {
    //    type = VT_I32;
    //}

    *_value = value;
    *_type = type;
    return 0;
}

int vm_parse(struct vm_parser *in, struct node **out) {
    (void) out;
    int res;
    int ch;

    if ((res = skip_whitespace(in)) != 0) {
        return res;
    }

    ch = in->peek(in);
    if (ch == 0) {
        return -1;
    } else if (ch == '(') {
        struct node *head, *tail, *car, *pair;
        assert(in->pop(in) == ch);
        head = NULL;
        while (1) {
            if ((res = skip_whitespace(in)) != 0) {
                return res;
            }
            ch = in->peek(in);
            if (ch == ')') {
                assert(in->pop(in) == ch);
                *out = head;
                break;
            }

            if ((res = vm_parse(in, &car)) != 0) {
                return res;
            }

            pair = cons(car, NULL);
            if (head) {
                tail->n_cons.cdr = pair;
            } else {
                head = pair;
            }
            tail = pair;
        }
        return 0;
    } else if (is_digit(ch)) {
        intmax_t value;
        enum node_type type;
        if ((res = vm_parse_int(in, &type, &value)) != 0) {
            return res;
        }
        (void) type;
        *out = integer(value);
        return 0;
    } else if (ch == '#') {
        assert(in->pop(in) == ch);
        ch = in->peek(in);
        if (ch == '\\') {
            assert(in->pop(in) == ch);
            ch = in->peek(in);
            switch (ch) {
            case 'n':
                *out = integer('\n');
                break;
            default:
                fprintf(stderr, "Unknown escape sequence: #\\%c\n", ch);
                abort();
            }
            assert(in->pop(in) == ch);
        } else if (!ch) {
            return -EINVAL;
        } else {
            assert(in->pop(in) == ch);
            *out = integer(ch);
        }
        return 0;
    } else if (is_ident0(ch)) {
        char buf[24];
        size_t i = 0;

        assert(in->pop(in) == ch);

        while (1) {
            assert(i < sizeof(buf) - 1);
            buf[i++] = ch;

            ch = in->peek(in);
            if (!ch || strchr(" \n\t()", ch)) {
                break;
            }
            assert(in->pop(in) == ch);
        }
        buf[i] = 0;
        *out = ident(buf);
        return 0;
    } else if (ch == '"') {
        char *buf;
        size_t cap, len;
        cap = 32;
        len = 0;
        buf = malloc(cap);
        if (!buf) {
            return -ENOMEM;
        }

        assert(in->pop(in) == ch);
        while (1) {
            ch = in->peek(in);
            if (!ch) {
                return -EINVAL;
            }
            assert(ch != '\\' && "Not implemented");
            assert(in->pop(in) == ch);
            if (ch == '"') {
                break;
            }

            if (len == cap - 1) {
                cap += 32;
                buf = realloc(buf, cap);
                assert(buf);
            }

            buf[len++] = ch;
        }
        buf[len] = 0;
        *out = string_from_owned(buf);
        return 0;
    }

    fprintf(stderr, "Unrecognized character: #%c (%d)\n", ch, ch);
    abort();
}

void vm_str_parser(struct vm_parser *p, const char *expr) {
    p->ctx = (void *) expr;
    p->ch = 0;
    p->peek = vm_str_parser_peek;
    p->pop = vm_str_parser_pop;
}

int vm_parse_str(const char *expr, struct node **out) {
    struct vm_parser p;
    vm_str_parser(&p, expr);
    return vm_parse(&p, out);
}

static int file_peek(struct vm_parser *p) {
    if (p->ch == -1) {
        int ch = fgetc(p->ctx);
        p->ch = ch;
    }
    if (p->ch == -1) {
        return 0;
    }
    return p->ch;
}

static int file_pop(struct vm_parser *p) {
    int ch = file_peek(p);
    if (ch == -1) {
        return 0;
    }
    p->ch = fgetc(p->ctx);
    return ch;
}

struct node *vm_load_file(FILE *fp) {
    struct node *head, *tail, *expr;
    struct vm_parser p;

    p.ch = -1;
    p.ctx = fp;
    p.peek = file_peek;
    p.pop = file_pop;

    head = NULL;
    while (vm_parse(&p, &expr) == 0) {
        expr = cons(expr, NULL);
        if (head) {
            tail->n_cons.cdr = expr;
        } else {
            head = expr;
        }
        tail = expr;
    }

    return head;
}