ell/ell.c

1339 lines
35 KiB
C
Raw Normal View History

2017-05-06 14:12:46 +02:00
/*
* ell.c: an experimental little language
*
* Copyright (c) 2017, Přemysl Janouch <p.janouch@gmail.com>
*
* Permission to use, copy, modify, and/or distribute this software for any
* purpose with or without fee is hereby granted, provided that the above
* copyright notice and this permission notice appear in all copies.
*
* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
* SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
* OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
* CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
*
*/
2017-05-06 13:27:02 +02:00
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <ctype.h>
#include <errno.h>
#include <stdarg.h>
#include <assert.h>
#include <time.h>
#include <stdbool.h>
#include <setjmp.h>
2017-05-06 13:27:02 +02:00
#if defined __GNUC__
#define ATTRIBUTE_PRINTF(x, y) __attribute__ ((format (printf, x, y)))
#else // ! __GNUC__
#define ATTRIBUTE_PRINTF(x, y)
#endif // ! __GNUC__
#define N_ELEMENTS(a) (sizeof (a) / sizeof ((a)[0]))
// --- Utilities ---------------------------------------------------------------
static char *format (const char *format, ...) ATTRIBUTE_PRINTF (1, 2);
static char *
vformat (const char *format, va_list ap) {
2017-05-06 13:27:02 +02:00
va_list aq;
va_copy (aq, ap);
int size = vsnprintf (NULL, 0, format, aq);
va_end (aq);
if (size < 0)
return NULL;
2017-05-20 21:03:04 +02:00
char *buf = malloc (size + 1);
if (buf && vsnprintf (buf, size + 1, format, ap) < 0) {
free (buf);
2017-05-06 13:27:02 +02:00
return NULL;
2017-05-20 21:03:04 +02:00
}
return buf;
2017-05-06 13:27:02 +02:00
}
static char *
format (const char *format, ...) {
va_list ap;
va_start (ap, format);
char *result = vformat (format, ap);
2017-05-06 13:27:02 +02:00
va_end (ap);
return result;
}
// --- Generic buffer ----------------------------------------------------------
struct buffer {
char *s; ///< Buffer data
size_t alloc; ///< Number of bytes allocated
size_t len; ///< Number of bytes used
bool memory_failure; ///< Memory allocation failed
};
#define BUFFER_INITIALIZER { NULL, 0, 0, false }
static bool
buffer_append (struct buffer *self, const void *s, size_t n) {
if (self->memory_failure)
return false;
if (!self->s)
self->s = malloc (self->alloc = 8);
while (self->len + n > self->alloc)
self->s = realloc (self->s, self->alloc <<= 1);
if (!self->s) {
self->memory_failure = true;
return false;
}
memcpy (self->s + self->len, s, n);
self->len += n;
return true;
}
inline static bool
buffer_append_c (struct buffer *self, char c) {
return buffer_append (self, &c, 1);
}
// --- Data types --------------------------------------------------------------
enum item_type { ITEM_STRING, ITEM_LIST };
struct item {
enum item_type type; ///< The type of this object
struct item *next; ///< Next item on the list/stack
struct item *head; ///< The head of the list
size_t len; ///< Length of the string (sans '\0')
char value[]; ///< The null-terminated string value
};
const char *
item_type_to_str (enum item_type type) {
switch (type) {
case ITEM_STRING: return "string";
case ITEM_LIST: return "list";
}
abort ();
}
// --- Item management ---------------------------------------------------------
static void item_free_list (struct item *);
static struct item *new_clone_list (const struct item *);
static void
item_free (struct item *item) {
if (item->type == ITEM_LIST)
item_free_list (item->head);
2017-05-06 13:27:02 +02:00
free (item);
}
static void
item_free_list (struct item *item) {
while (item) {
struct item *link = item;
item = item->next;
item_free (link);
}
}
static struct item *
new_clone (const struct item *item) {
2017-05-18 19:45:56 +02:00
size_t size = sizeof *item + 1;
2017-05-06 13:27:02 +02:00
if (item->type == ITEM_STRING)
2017-05-18 19:45:56 +02:00
size += item->len;
2017-05-06 13:27:02 +02:00
struct item *clone = malloc (size);
if (!clone)
return NULL;
memcpy (clone, item, size);
if (item->type == ITEM_LIST && clone->head) {
if (!(clone->head = new_clone_list (clone->head))) {
2017-05-06 13:27:02 +02:00
free (clone);
return NULL;
}
}
clone->next = NULL;
return clone;
}
static struct item *
new_clone_list (const struct item *item) {
struct item *head = NULL;
2017-05-06 13:27:02 +02:00
for (struct item **out = &head; item; item = item->next) {
if (!(*out = new_clone (item))) {
2017-05-06 13:27:02 +02:00
item_free_list (head);
return NULL;
}
out = &(*out)->next;
2017-05-06 13:27:02 +02:00
}
return head;
}
static struct item *
2017-05-20 21:03:04 +02:00
new_string (const char *s, size_t len) {
2017-05-06 13:27:02 +02:00
struct item *item = calloc (1, sizeof *item + len + 1);
if (!item)
return NULL;
item->type = ITEM_STRING;
item->len = len;
memcpy (item->value, s, len);
item->value[len] = '\0';
return item;
}
static struct item *
new_list (struct item *head) {
2017-05-18 19:45:56 +02:00
struct item *item = calloc (1, sizeof *item + 1);
if (!item) {
item_free_list (head);
2017-05-06 13:27:02 +02:00
return NULL;
}
2017-05-06 13:27:02 +02:00
item->type = ITEM_LIST;
item->head = head;
return item;
}
// --- Lexer -------------------------------------------------------------------
2017-05-19 09:19:35 +02:00
enum token { T_ABORT, T_LPAREN, T_RPAREN, T_LBRACKET, T_RBRACKET,
T_LBRACE, T_RBRACE, T_STRING, T_NEWLINE, T_AT };
2017-05-06 13:27:02 +02:00
static const char *token_names[] = {
[T_ABORT] = "end of input",
[T_LPAREN] = "left parenthesis",
[T_RPAREN] = "right parenthesis",
[T_LBRACKET] = "left bracket",
[T_RBRACKET] = "right bracket",
[T_LBRACE] = "left brace",
[T_RBRACE] = "right brace",
[T_STRING] = "string",
[T_NEWLINE] = "newline",
[T_AT] = "at symbol",
};
2017-05-06 13:27:02 +02:00
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
struct lexer {
const char *p; ///< Current position in input
size_t len; ///< How many bytes of input are left
2017-05-17 23:34:58 +02:00
unsigned line, column; ///< Current line and column
struct buffer string; ///< Parsed string value
2017-05-06 13:27:02 +02:00
};
/// Input has to be null-terminated anyway
static void
lexer_init (struct lexer *self, const char *p, size_t len) {
memset (self, 0, sizeof *self);
self->p = p;
self->len = len;
}
static void
lexer_free (struct lexer *self) {
free (self->string.s);
2017-05-06 13:27:02 +02:00
}
static bool lexer_is_ignored (int c) { return strchr (" \t", c); }
2017-05-18 20:45:46 +02:00
static bool lexer_is_word_char (int c) {
2017-05-20 23:00:43 +02:00
return !lexer_is_ignored (c) && !strchr ("()[]{}\n;@#'", c);
2017-05-18 20:45:46 +02:00
}
2017-05-06 13:27:02 +02:00
static int
lexer_advance (struct lexer *self) {
int c = *self->p++;
if (c == '\n') {
self->column = 0;
self->line++;
} else
self->column++;
self->len--;
return c;
}
static bool
lexer_hexa_escape (struct lexer *self, struct buffer *output) {
2017-05-06 13:27:02 +02:00
int i;
unsigned char code = 0;
for (i = 0; self->len && i < 2; i++) {
unsigned char c = tolower (*self->p);
if (c >= '0' && c <= '9')
code = (code << 4) | (c - '0');
else if (c >= 'a' && c <= 'f')
code = (code << 4) | (c - 'a' + 10);
else
break;
lexer_advance (self);
}
if (!i)
return false;
buffer_append_c (output, code);
2017-05-06 13:27:02 +02:00
return true;
}
2017-05-19 09:48:53 +02:00
static const char *
lexer_escape_sequence (struct lexer *self, struct buffer *output) {
2017-05-18 20:45:46 +02:00
if (!self->len)
2017-05-19 09:48:53 +02:00
return "premature end of escape sequence";
2017-05-06 13:27:02 +02:00
unsigned char c = *self->p;
switch (c) {
2017-05-06 13:27:02 +02:00
case '"': break;
case '\\': break;
case 'a': c = '\a'; break;
case 'b': c = '\b'; break;
case 'f': c = '\f'; break;
case 'n': c = '\n'; break;
case 'r': c = '\r'; break;
case 't': c = '\t'; break;
case 'v': c = '\v'; break;
case 'x':
case 'X':
lexer_advance (self);
if (lexer_hexa_escape (self, output))
2017-05-19 09:48:53 +02:00
return NULL;
return "invalid hexadecimal escape";
2017-05-06 13:27:02 +02:00
default:
2017-05-19 09:48:53 +02:00
return "unknown escape sequence";
2017-05-06 13:27:02 +02:00
}
buffer_append_c (output, c);
2017-05-06 13:27:02 +02:00
lexer_advance (self);
2017-05-19 09:48:53 +02:00
return NULL;
2017-05-06 13:27:02 +02:00
}
2017-05-19 09:48:53 +02:00
static const char *
lexer_string (struct lexer *self, struct buffer *output) {
2017-05-06 13:27:02 +02:00
unsigned char c;
2017-05-19 09:48:53 +02:00
const char *e = NULL;
2017-05-06 13:27:02 +02:00
while (self->len) {
if ((c = lexer_advance (self)) == '\'')
2017-05-19 09:48:53 +02:00
return NULL;
2017-05-06 13:27:02 +02:00
if (c != '\\')
buffer_append_c (output, c);
2017-05-19 09:48:53 +02:00
else if ((e = lexer_escape_sequence (self, output)))
return e;
2017-05-06 13:27:02 +02:00
}
2017-05-19 09:48:53 +02:00
return "premature end of string";
2017-05-06 13:27:02 +02:00
}
static enum token
lexer_next (struct lexer *self, const char **e) {
2017-05-06 13:27:02 +02:00
// Skip over any whitespace between tokens
2017-05-18 20:45:46 +02:00
while (self->len && lexer_is_ignored (*self->p))
2017-05-06 13:27:02 +02:00
lexer_advance (self);
if (!self->len)
return T_ABORT;
free (self->string.s);
self->string = (struct buffer) BUFFER_INITIALIZER;
2017-05-06 13:27:02 +02:00
switch (*self->p) {
case '(': lexer_advance (self); return T_LPAREN;
case ')': lexer_advance (self); return T_RPAREN;
case '[': lexer_advance (self); return T_LBRACKET;
case ']': lexer_advance (self); return T_RBRACKET;
case '{': lexer_advance (self); return T_LBRACE;
case '}': lexer_advance (self); return T_RBRACE;
case '\n': lexer_advance (self); return T_NEWLINE;
2017-05-20 23:00:43 +02:00
case ';': lexer_advance (self); return T_NEWLINE;
2017-05-06 13:27:02 +02:00
case '@': lexer_advance (self); return T_AT;
case '#':
// Comments go until newline
while (self->len)
if (lexer_advance (self) == '\n')
return T_NEWLINE;
return T_ABORT;
case '\'':
lexer_advance (self);
2017-05-19 09:48:53 +02:00
if ((*e = lexer_string (self, &self->string)))
2017-05-06 13:27:02 +02:00
return T_ABORT;
return T_STRING;
}
assert (lexer_is_word_char (*self->p));
do
buffer_append_c (&self->string, lexer_advance (self));
2017-05-06 13:27:02 +02:00
while (lexer_is_word_char (*self->p));
return T_STRING;
}
static char *lexer_errorf (struct lexer *self, const char *fmt, ...)
ATTRIBUTE_PRINTF (2, 3);
static char *
lexer_errorf (struct lexer *self, const char *fmt, ...) {
va_list ap;
va_start (ap, fmt);
char *description = vformat (fmt, ap);
va_end (ap);
if (!description)
return NULL;
char *e = format ("near line %u, column %u: %s",
self->line + 1, self->column + 1, description);
free (description);
return e;
}
2017-05-06 13:27:02 +02:00
// --- Parsing -----------------------------------------------------------------
static void
print_string (const char *s) {
putc ('\'', stdout);
for (; *s; s++)
if (*s == '\n') printf ("\\n");
else if (*s == '\\') putc ('\\', stdout);
else putc (*s, stdout);
putc ('\'', stdout);
}
static void
print_tree (struct item *tree, int level) {
// TODO: also re-add syntax sugar
for (struct item *iter = tree; iter; iter = iter->next) {
if (iter != tree)
printf ("%*s", level, "");
if (iter->type == ITEM_STRING) {
print_string (iter->value);
} else if (iter->head->type == ITEM_STRING
&& !strcmp (iter->head->value, "list")) {
printf ("[");
print_tree (iter->head->next, level + 1);
printf ("]");
} else {
printf ("(");
print_tree (iter->head, level + 1);
printf (")");
}
if (iter->next)
printf ("\n");
}
}
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
struct parser {
struct lexer lexer; ///< Tokenizer
char *error; ///< Tokenizer error
enum token token; ///< Current token in the lexer
bool replace_token; ///< Replace the token
bool memory_failure; ///< Memory allocation failed
};
static void
parser_init (struct parser *self, const char *script, size_t len) {
memset (self, 0, sizeof *self);
lexer_init (&self->lexer, script, len);
// As reading in tokens may cause exceptions, we wait for the first peek()
// to replace the initial T_ABORT.
self->replace_token = true;
}
static void
parser_free (struct parser *self) {
lexer_free (&self->lexer);
if (self->error)
free (self->error);
}
static enum token
parser_peek (struct parser *self, jmp_buf out) {
2017-05-19 09:19:35 +02:00
if (self->replace_token) {
const char *e = NULL;
self->token = lexer_next (&self->lexer, &e);
if (e) {
self->memory_failure =
!(self->error = lexer_errorf (&self->lexer, "%s", e));
longjmp (out, 1);
}
if (self->token == T_STRING && self->lexer.string.memory_failure)
longjmp (out, 1);
self->replace_token = false;
}
return self->token;
}
static bool
parser_accept (struct parser *self, enum token token, jmp_buf out) {
return self->replace_token = (parser_peek (self, out) == token);
}
static void
parser_expect (struct parser *self, enum token token, jmp_buf out) {
if (parser_accept (self, token, out))
return;
self->memory_failure = !(self->error = lexer_errorf (&self->lexer,
"unexpected `%s', expected `%s'",
token_names[self->token], token_names[token]));
longjmp (out, 1);
}
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
// We don't need no generator, but a few macros will come in handy.
// From time to time C just doesn't have the right features.
#define PEEK() parser_peek (self, err)
#define ACCEPT(token) parser_accept (self, token, err)
#define EXPECT(token) parser_expect (self, token, err)
#define SKIP_NL() do {} while (ACCEPT (T_NEWLINE))
static struct item *
parser_check (struct parser *self, struct item *item, jmp_buf out) {
if (!item) {
self->memory_failure = true;
longjmp (out, 1);
}
return item;
}
// Beware that this jumps to the "out" buffer directly
#define CHECK(item) parser_check (self, (item), out)
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
2017-05-06 13:27:02 +02:00
static struct item *
parse_prefix_list (struct item *list, const char *name) {
struct item *prefix;
if (!(prefix = new_string (name, strlen (name)))) {
item_free_list (list);
return NULL;
}
prefix->next = list;
return new_list (prefix);
}
2017-05-17 23:34:58 +02:00
static struct item * parse_line (struct parser *self, jmp_buf out);
static struct item *
parse_item (struct parser *self, jmp_buf out) {
jmp_buf err;
struct item *volatile result = NULL, *volatile *tail = &result;
if (setjmp (err)) {
item_free_list (result);
longjmp (out, 1);
}
SKIP_NL ();
if (ACCEPT (T_STRING))
return CHECK (new_string
(self->lexer.string.s, self->lexer.string.len));
if (ACCEPT (T_AT)) {
result = parse_item (self, out);
return CHECK (parse_prefix_list (result, "set"));
}
if (ACCEPT (T_LPAREN)) {
while (!ACCEPT (T_RPAREN)) {
tail = &(*tail = parse_item (self, err))->next;
SKIP_NL ();
2017-05-17 23:34:58 +02:00
}
return CHECK (new_list (result));
2017-05-17 23:34:58 +02:00
}
if (ACCEPT (T_LBRACKET)) {
while (!ACCEPT (T_RBRACKET)) {
tail = &(*tail = parse_item (self, err))->next;
SKIP_NL ();
}
return CHECK (parse_prefix_list (result, "list"));
}
if (ACCEPT (T_LBRACE)) {
while ((*tail = parse_line (self, err)))
tail = &(*tail)->next;
EXPECT (T_RBRACE);
result = CHECK (new_list (result));
return CHECK (parse_prefix_list (result, "quote"));
}
self->memory_failure = !(self->error = lexer_errorf (&self->lexer,
"unexpected `%s', expected a value", token_names[self->token]));
longjmp (out, 1);
}
static struct item *
parse_line (struct parser *self, jmp_buf out) {
jmp_buf err;
struct item *volatile result = NULL, *volatile *tail = &result;
if (setjmp (err)) {
item_free_list (result);
longjmp (out, 1);
}
while (PEEK () != T_RBRACE && PEEK () != T_ABORT) {
if (!ACCEPT (T_NEWLINE)) {
tail = &(*tail = parse_item (self, err))->next;
} else if (result) {
return CHECK (new_list (result));
}
2017-05-17 23:34:58 +02:00
}
if (result)
return CHECK (new_list (result));
return NULL;
2017-05-06 13:27:02 +02:00
}
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
#undef PEEK
#undef ACCEPT
#undef EXPECT
#undef SKIP_NL
#undef CHECK
static struct item *
parser_run (struct parser *self, const char **e) {
jmp_buf err;
struct item *volatile result = NULL, *volatile *tail = &result;
if (setjmp (err)) {
item_free_list (result);
*e = self->error;
if (self->memory_failure || self->lexer.string.memory_failure)
*e = "memory allocation failure";
return NULL;
}
while ((*tail = parse_line (self, err)))
tail = &(*tail)->next;
parser_expect (self, T_ABORT, err);
return result;
}
2017-05-06 13:27:02 +02:00
// --- Runtime -----------------------------------------------------------------
struct context {
struct item *variables; ///< List of variables
2017-05-21 09:47:24 +02:00
struct native_fn *native; ///< Maps strings to C functions
2017-05-06 13:27:02 +02:00
char *error; ///< Error information
bool error_is_fatal; ///< Whether the error can be catched
bool memory_failure; ///< Memory allocation failure
void *user_data; ///< User data
};
2017-05-21 09:47:24 +02:00
typedef bool (*handler_fn) (struct context *, struct item *, struct item **);
struct native_fn {
struct native_fn *next; ///< The next link in the chain
handler_fn handler; ///< Internal C handler, or NULL
char name[]; ///< The name of the function
};
2017-05-06 13:27:02 +02:00
static void
context_init (struct context *ctx) {
2017-05-17 23:22:43 +02:00
memset (ctx, 0, sizeof *ctx);
2017-05-06 13:27:02 +02:00
}
static void
context_free (struct context *ctx) {
2017-05-21 09:47:24 +02:00
struct native_fn *next, *iter;
for (iter = ctx->native; iter; iter = next) {
next = iter->next;
free (iter);
}
item_free_list (ctx->variables);
2017-05-06 13:27:02 +02:00
free (ctx->error);
}
static bool
check (struct context *ctx, struct item *item) {
return !(ctx->memory_failure |= !item);
}
2017-05-18 19:47:10 +02:00
static struct item *
get (struct context *ctx, const char *name) {
for (struct item *iter = ctx->variables; iter; iter = iter->next)
if (!strcmp (iter->head->value, name))
return iter->head->next;
return NULL;
}
static bool
set (struct context *ctx, const char *name, struct item *value) {
struct item **p;
for (p = &ctx->variables; *p; p = &(*p)->next)
if (!strcmp ((*p)->head->value, name)) {
struct item *tmp = *p;
*p = (*p)->next;
item_free (tmp);
2017-05-18 19:47:10 +02:00
break;
}
if (!value)
return true;
struct item *key, *pair;
if (!check (ctx, (key = new_string (name, strlen (name))))
|| !check (ctx, (pair = new_list (key)))) {
item_free_list (value);
2017-05-18 19:47:10 +02:00
return false;
}
key->next = value;
pair->next = ctx->variables;
ctx->variables = pair;
return true;
2017-05-18 19:47:10 +02:00
}
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
2017-05-21 09:47:24 +02:00
static struct native_fn *
native_find (struct context *ctx, const char *name) {
for (struct native_fn *fn = ctx->native; fn; fn = fn->next)
if (!strcmp (fn->name, name))
return fn;
return NULL;
}
static bool
native_register (struct context *ctx, const char *name, handler_fn handler) {
struct native_fn *fn = native_find (ctx, name);
if (!fn) {
if (!(fn = calloc (1, sizeof *fn + strlen (name) + 1)))
return false;
strcpy (fn->name, name);
fn->next = ctx->native;
ctx->native = fn;
}
fn->handler = handler;
return true;
}
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
2017-05-06 13:27:02 +02:00
static bool
set_error (struct context *ctx, const char *format, ...) {
va_list ap;
va_start (ap, format);
free (ctx->error);
if (!(ctx->error = vformat (format, ap)))
2017-05-06 13:27:02 +02:00
ctx->memory_failure = true;
va_end (ap);
2017-05-06 13:27:02 +02:00
return false;
}
2017-05-18 19:47:10 +02:00
static bool
rename_arguments (struct context *ctx, struct item *names) {
size_t i = 0;
for (; names; names = names->next) {
if (names->type != ITEM_STRING)
return set_error (ctx, "argument names must be strings");
2017-05-18 19:47:10 +02:00
char buf[64];
(void) snprintf (buf, sizeof buf, "%zu", i++);
struct item *value = get (ctx, buf);
if (value && !check (ctx, (value = new_clone (value))))
return false;
2017-05-18 20:45:46 +02:00
if (!set (ctx, names->value, value))
return false;
2017-05-18 19:47:10 +02:00
}
return true;
2017-05-18 18:01:57 +02:00
}
2017-05-18 23:08:55 +02:00
static bool execute_statement (struct context *, struct item *, struct item **);
2017-05-18 19:47:10 +02:00
static bool execute (struct context *ctx, struct item *body, struct item **);
2017-05-06 13:27:02 +02:00
2017-05-18 23:08:55 +02:00
static bool
execute_args (struct context *ctx, struct item *args, struct item **res) {
for (; args; args = args->next) {
2017-05-18 23:08:55 +02:00
struct item *evaluated = NULL;
if (!execute_statement (ctx, args, &evaluated))
2017-05-18 23:08:55 +02:00
return false;
2017-05-21 08:27:47 +02:00
// Arguments should not evaporate, default to a nil value
if (!evaluated && !check (ctx, (evaluated = new_list (NULL))))
return false;
item_free_list (evaluated->next);
evaluated->next = NULL;
res = &(*res = evaluated)->next;
2017-05-18 23:08:55 +02:00
}
return true;
}
2017-05-21 00:23:51 +02:00
static bool
set_arg (struct context *ctx, size_t arg, struct item *value) {
char buf[64];
(void) snprintf (buf, sizeof buf, "%zu", arg);
return check (ctx, (value = new_clone (value)))
&& set (ctx, buf, value);
2017-05-21 00:23:51 +02:00
}
// TODO: we should probably maintain arguments in a separate list,
// either that or at least remember the count so that we can reset them
2017-05-18 23:08:55 +02:00
static bool
2017-05-20 21:24:11 +02:00
execute_and_set_args (struct context *ctx, struct item *following) {
2017-05-18 23:08:55 +02:00
struct item *args = NULL;
if (!execute_args (ctx, following, &args)) {
2017-05-18 23:08:55 +02:00
item_free_list (args);
return false;
}
size_t i = 0;
2017-05-21 00:23:51 +02:00
for (struct item *arg = args; arg; arg = arg->next)
if (!set_arg (ctx, i++, arg))
return false;
2017-05-18 23:08:55 +02:00
item_free_list (args);
return true;
}
static bool
execute_native (struct context *ctx,
2017-05-21 09:47:24 +02:00
struct native_fn *fn, struct item *following, struct item **result) {
struct item *args = NULL;
2017-05-21 09:47:24 +02:00
bool ok = execute_args (ctx, following, &args)
&& fn->handler (ctx, args, result);
item_free_list (args);
return ok;
}
2017-05-06 13:27:02 +02:00
static bool
2017-05-18 19:47:10 +02:00
execute_statement
(struct context *ctx, struct item *statement, struct item **result) {
if (statement->type == ITEM_STRING)
return check (ctx, (*result = new_clone (statement)));
2017-05-18 19:47:10 +02:00
2017-05-21 08:27:47 +02:00
// Executing a nil value results in no value. It's not very different from
// calling a block that returns no value--it's for our callers to resolve.
2017-05-18 19:47:10 +02:00
struct item *body;
if (!(body = statement->head))
2017-05-06 13:27:02 +02:00
return true;
2017-05-18 23:08:55 +02:00
struct item *following = body->next;
2017-05-18 19:47:10 +02:00
const char *name = "(anonymous)";
if (body->type == ITEM_STRING) {
name = body->value;
2017-05-18 23:08:55 +02:00
// TODO: these could be just regular handlers, only top priority
// TODO: these should also get a stack trace the normal way
if (!strcmp (name, "quote"))
return !following
|| check (ctx, (*result = new_clone_list (following)));
2017-05-18 19:47:10 +02:00
if (!strcmp (name, "arg"))
2017-05-18 23:08:55 +02:00
return rename_arguments (ctx, following);
body = get (ctx, name);
2017-05-18 19:47:10 +02:00
}
2017-05-21 00:23:51 +02:00
// XXX: not sure whether it makes more sense to evaluate a list
// instead of leaving it as it is -> probably evaluate, since then
// lambdas will be better: { arg _n; print @_n } 'hello\n'
//
// Even though it's practically useless being able to dereference
// function names directly that way: `@list 1 2 3` vs `list 1 2 3`
//
// Maybe something like (choose [@f1 @f2 @f3]) arg1 arg2 arg3
2017-05-18 23:08:55 +02:00
if (!body) {
2017-05-21 09:47:24 +02:00
struct native_fn *fn = native_find (ctx, name);
2017-05-18 23:08:55 +02:00
if (!fn)
return set_error (ctx, "unknown function: %s", name);
if (execute_native (ctx, fn, following, result))
return true;
} else if (body->type == ITEM_STRING) {
// Recursion could be pretty fatal, let's not do that
if (check (ctx, (*result = new_clone (body))))
2017-05-18 23:08:55 +02:00
return true;
} else {
2017-05-20 21:24:11 +02:00
// FIXME: this creates a confusing backtrace for argument evaluation
if (execute_and_set_args (ctx, following)
2017-05-18 23:08:55 +02:00
&& execute (ctx, body->head, result))
return true;
2017-05-18 19:47:10 +02:00
}
2017-05-21 08:28:45 +02:00
item_free_list (*result);
*result = NULL;
2017-05-21 09:38:18 +02:00
// In that case, `error' is NULL and there's nothing else to do anyway.
// Errors starting with an underscore are exceptions and would not work
// with stack traces generated this way.
if (!ctx->memory_failure && ctx->error[0] != '_') {
char *tmp = ctx->error;
set_error (ctx, "%s -> %s", name, tmp);
free (tmp);
}
2017-05-06 13:27:02 +02:00
return false;
}
2017-05-18 18:01:57 +02:00
// Execute a block and return whatever the last statement returned
2017-05-18 19:47:10 +02:00
static bool
execute (struct context *ctx, struct item *body, struct item **result) {
for (; body; body = body->next) {
item_free_list (*result);
*result = NULL;
if (!execute_statement (ctx, body, result))
return false;
2017-05-06 13:27:02 +02:00
}
2017-05-18 19:47:10 +02:00
return true;
2017-05-06 13:27:02 +02:00
}
// --- Runtime library ---------------------------------------------------------
2017-05-18 19:47:10 +02:00
#define defn(name) static bool name \
(struct context *ctx, struct item *args, struct item **result)
2017-05-06 13:27:02 +02:00
2017-05-21 10:03:29 +02:00
static struct item *
2017-05-21 10:33:04 +02:00
new_number (double n) {
char *s;
if (!(s = format ("%f", n)))
2017-05-21 10:03:29 +02:00
return NULL;
2017-05-21 10:33:04 +02:00
char *p = strchr (s, 0);
while (--p > s && *p == '0')
*p = 0;
if (*p == '.')
*p = 0;
2017-05-21 10:03:29 +02:00
struct item *item = new_string (s, strlen (s));
free (s);
return item;
}
2017-05-21 09:38:18 +02:00
#define E_BREAK "_break"
static bool
eat_error (struct context *ctx, const char *name) {
if (!ctx->error || strcmp (ctx->error, name))
return false;
free (ctx->error);
ctx->error = NULL;
return true;
}
2017-05-06 13:27:02 +02:00
static bool
2017-05-20 21:24:11 +02:00
truthy (struct item *item) {
return item
&& ((item->type == ITEM_STRING && item->len != 0) || item->head);
2017-05-06 13:27:02 +02:00
}
2017-05-21 10:39:39 +02:00
static struct item * new_boolean (bool b) { return new_string ("1", b); }
2017-05-21 09:38:18 +02:00
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
2017-05-18 19:47:10 +02:00
defn (fn_set) {
struct item *name = args;
if (!name || name->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
2017-05-18 19:47:10 +02:00
struct item *value;
if ((value = name->next))
return check (ctx, (value = new_clone (value)))
&& set (ctx, name->value, value);
2017-05-18 19:47:10 +02:00
2017-05-20 15:46:52 +02:00
// We return an empty list for a nil value
if (!(value = get (ctx, name->value)))
return check (ctx, (*result = new_list (NULL)));
return check (ctx, (*result = new_clone (value)));
}
defn (fn_list) {
struct item *values = NULL;
if (args && !check (ctx, (values = new_clone_list (args))))
return false;
return check (ctx, (*result = new_list (values)));
2017-05-18 19:47:10 +02:00
}
2017-05-20 21:24:11 +02:00
defn (fn_if) {
struct item *cond, *body, *keyword;
for (cond = args; ; cond = keyword->next) {
if (!cond)
return set_error (ctx, "missing condition");
if (!(body = cond->next))
return set_error (ctx, "missing body");
struct item *res = NULL;
if (!execute_statement (ctx, cond, &res))
return false;
bool match = truthy (res);
item_free_list (res);
if (match)
return execute_statement (ctx, body, result);
if (!(keyword = body->next))
break;
if (keyword->type != ITEM_STRING)
return set_error (ctx, "expected keyword, got list");
if (!strcmp (keyword->value, "else")) {
if (!(body = keyword->next))
return set_error (ctx, "missing body");
return execute_statement (ctx, body, result);
}
if (strcmp (keyword->value, "elif"))
return set_error (ctx, "invalid keyword: %s", keyword->value);
}
return true;
}
defn (fn_for) {
struct item *list = args, *body;
if (!list || list->type != ITEM_LIST)
return set_error (ctx, "first argument must be a list");
if (!(body = list->next) || body->type != ITEM_LIST)
return set_error (ctx, "second argument must be a function");
(void) result;
2017-05-21 00:23:51 +02:00
for (struct item *v = list->head; v; v = v->next) {
2017-05-20 21:24:11 +02:00
struct item *res = NULL;
2017-05-21 00:23:51 +02:00
bool ok = set_arg (ctx, 0, v)
&& execute (ctx, body->head, &res);
2017-05-20 21:24:11 +02:00
item_free_list (res);
2017-05-21 09:38:18 +02:00
if (eat_error (ctx, E_BREAK))
break;
2017-05-20 21:24:11 +02:00
if (!ok)
return false;
}
return true;
}
2017-05-21 09:38:18 +02:00
defn (fn_break) {
(void) args; (void) result; return set_error (ctx, E_BREAK);
}
2017-05-20 21:24:11 +02:00
defn (fn_map) {
struct item *body = args, *values;
if (!body || body->type != ITEM_LIST)
return set_error (ctx, "first argument must be a function");
if (!(values = body->next) || values->type != ITEM_LIST)
return set_error (ctx, "second argument must be a list");
struct item *res = NULL, **out = &res;
for (struct item *v = values->head; v; v = v->next) {
2017-05-21 00:23:51 +02:00
if (!set_arg (ctx, 0, v)
|| !execute (ctx, body->head, out)) {
2017-05-20 21:24:11 +02:00
item_free_list (res);
return false;
}
2017-05-21 00:23:51 +02:00
while (*out)
2017-05-20 21:24:11 +02:00
out = &(*out)->next;
}
return check (ctx, (*result = new_list (res)));
}
defn (fn_filter) {
struct item *body = args, *values;
if (!body || body->type != ITEM_LIST)
return set_error (ctx, "first argument must be a function");
if (!(values = body->next) || values->type != ITEM_LIST)
return set_error (ctx, "second argument must be a list");
struct item *res = NULL, **out = &res;
for (struct item *v = values->head; v; v = v->next) {
2017-05-21 00:23:51 +02:00
struct item *keep = NULL;
if (!set_arg (ctx, 0, v)
|| !execute (ctx, body->head, &keep)) {
2017-05-20 21:24:11 +02:00
item_free_list (res);
return false;
}
2017-05-21 00:23:51 +02:00
bool match = truthy (keep);
item_free_list (keep);
2017-05-20 21:24:11 +02:00
if (!match)
continue;
struct item *copy;
if (!check (ctx, (copy = new_clone (v)))) {
item_free_list (res);
return false;
}
out = &(*out = copy)->next;
}
return check (ctx, (*result = new_list (res)));
}
2017-05-17 23:22:43 +02:00
defn (fn_print) {
2017-05-18 19:47:10 +02:00
(void) result;
for (; args; args = args->next) {
if (args->type != ITEM_STRING)
// TODO: print lists as their parsable representation
return set_error (ctx, "cannot print lists");
if (fwrite (args->value, 1, args->len, stdout) != args->len)
return set_error (ctx, "write failed: %s", strerror (errno));
2017-05-06 13:27:02 +02:00
}
return true;
}
defn (fn_concatenate) {
2017-05-18 19:47:10 +02:00
struct buffer buf = BUFFER_INITIALIZER;
for (; args; args = args->next) {
if (args->type != ITEM_STRING) {
free (buf.s);
return set_error (ctx, "cannot concatenate lists");
}
2017-05-18 19:47:10 +02:00
buffer_append (&buf, args->value, args->len);
}
buffer_append_c (&buf, '\0');
2017-05-20 21:38:08 +02:00
bool ok = !(ctx->memory_failure |= buf.memory_failure)
&& check (ctx, (*result = new_string (buf.s, buf.len - 1)));
2017-05-18 19:47:10 +02:00
free (buf.s);
return ok;
2017-05-06 13:27:02 +02:00
}
2017-05-21 10:03:29 +02:00
defn (fn_system) {
struct item *command = args;
if (!command || command->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
if (command->next)
return set_error (ctx, "cannot deal with multiple arguments");
2017-05-21 10:33:04 +02:00
return check (ctx, (*result = new_number (system (command->value))));
2017-05-21 10:03:29 +02:00
}
2017-05-21 10:24:14 +02:00
defn (fn_plus) {
double res = 0.0;
for (; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
res += strtod (args->value, NULL);
}
2017-05-21 10:33:04 +02:00
return check (ctx, (*result = new_number (res)));
2017-05-21 10:24:14 +02:00
}
defn (fn_minus) {
2017-05-21 10:33:04 +02:00
double res = 0.0;
2017-05-21 10:24:14 +02:00
for (; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
res -= strtod (args->value, NULL);
}
2017-05-21 10:33:04 +02:00
return check (ctx, (*result = new_number (res)));
2017-05-21 10:24:14 +02:00
}
defn (fn_multiply) {
double res = 1.0;
for (; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
res *= strtod (args->value, NULL);
}
2017-05-21 10:33:04 +02:00
return check (ctx, (*result = new_number (res)));
2017-05-21 10:24:14 +02:00
}
defn (fn_divide) {
if (!args || args->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
double res = strtod (args->value, NULL), x;
for (args = args->next; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
if (!(x = strtod (args->value, NULL)))
return set_error (ctx, "division by zero");
res /= x;
}
2017-05-21 10:33:04 +02:00
return check (ctx, (*result = new_number (res)));
2017-05-21 10:24:14 +02:00
}
2017-05-21 10:39:39 +02:00
defn (fn_not) {
if (!args)
return set_error (ctx, "missing argument");
return check (ctx, (*result = new_boolean (!truthy (args))));
}
// TODO: "and" and "or" should be short-circuiting special forms
2017-05-21 10:39:39 +02:00
defn (fn_and) {
bool res = true;
for (; args; args = args->next)
res &= truthy (args);
return check (ctx, (*result = new_boolean (res)));
}
defn (fn_or) {
bool res = false;
for (; args; args = args->next)
res |= truthy (args);
return check (ctx, (*result = new_boolean (res)));
}
2017-05-21 11:26:59 +02:00
defn (fn_eq) {
struct item *etalon = args;
if (!etalon || etalon->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
bool res = true;
for (args = etalon->next; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
if (!(res &= !strcmp (etalon->value, args->value)))
break;
}
return check (ctx, (*result = new_boolean (res)));
}
defn (fn_lt) {
struct item *etalon = args;
if (!etalon || etalon->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
bool res = true;
for (args = etalon->next; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
if (!(res &= strcmp (etalon->value, args->value) < 0))
break;
etalon = args;
}
return check (ctx, (*result = new_boolean (res)));
}
2017-05-21 12:14:43 +02:00
defn (fn_equals) {
struct item *etalon = args;
if (!etalon || etalon->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
bool res = true;
for (args = etalon->next; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
if (!(res &= strtod (etalon->value, NULL)
== strtod (args->value, NULL)))
break;
}
return check (ctx, (*result = new_boolean (res)));
}
defn (fn_less) {
struct item *etalon = args;
if (!etalon || etalon->type != ITEM_STRING)
return set_error (ctx, "first argument must be string");
bool res = true;
for (args = etalon->next; args; args = args->next) {
if (args->type != ITEM_STRING)
return set_error (ctx, "arguments must be strings");
if (!(res &= strtod (etalon->value, NULL) < strtod (args->value, NULL)))
break;
etalon = args;
}
return check (ctx, (*result = new_boolean (res)));
}
2017-05-20 21:24:11 +02:00
// - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
static bool
init_runtime_library (struct context *ctx) {
struct {
const char *name; ///< Name of the function
const char *definition; ///< The defining script
} functions[] = {
// FIXME: this "unless" is probably not going to work
{ "unless", "arg _cond _body; if (not (eval @_cond)) @_body" },
// TODO: we should be able to apply them to all arguments
2017-05-21 12:46:21 +02:00
{ "ne?", "arg _ne1 _ne2; not (eq? @_ne1 @_ne2)" },
{ "ge?", "arg _ge1 _ge2; not (lt? @_ge1 @_ge2)" },
{ "le?", "arg _le1 _le2; ge? @_le2 @_le1" },
{ "gt?", "arg _gt1 _gt2; lt? @_gt2 @_gt1" },
2017-05-21 12:46:21 +02:00
{ "<>", "arg _<>1 _<>2; not (= @_<>1 @_<>2)" },
{ ">=", "arg _>=1 _>=2; not (< @_>=1 @_>=2)" },
2017-05-21 12:14:43 +02:00
{ "<=", "arg _<=1 _<=2; >= @_<=2 @_<=1" },
{ ">", "arg _>1 _>2; < @_>2 @_>1" },
2017-05-20 21:24:11 +02:00
};
2017-05-21 09:47:24 +02:00
bool ok = true;
2017-05-20 21:24:11 +02:00
for (size_t i = 0; i < N_ELEMENTS (functions); i++) {
struct parser parser;
parser_init (&parser,
functions[i].definition, strlen (functions[i].definition));
const char *e = NULL;
struct item *body = parser_run (&parser, &e);
if (e) {
printf ("error parsing internal function `%s': %s\n",
functions[i].name, e);
ok = false;
} else if (!check (ctx, (body = new_list (body)))
|| !set (ctx, functions[i].name, body)) {
ok = false;
2017-05-20 21:24:11 +02:00
} else
body = NULL;
2017-05-20 21:24:11 +02:00
item_free_list (body);
parser_free (&parser);
}
2017-05-21 09:47:24 +02:00
return ok
&& native_register (ctx, "set", fn_set)
&& native_register (ctx, "list", fn_list)
&& native_register (ctx, "if", fn_if)
&& native_register (ctx, "for", fn_for)
&& native_register (ctx, "break", fn_break)
&& native_register (ctx, "map", fn_map)
&& native_register (ctx, "filter", fn_filter)
&& native_register (ctx, "print", fn_print)
2017-05-21 10:03:29 +02:00
&& native_register (ctx, "..", fn_concatenate)
2017-05-21 10:24:14 +02:00
&& native_register (ctx, "system", fn_system)
&& native_register (ctx, "+", fn_plus)
&& native_register (ctx, "-", fn_minus)
&& native_register (ctx, "*", fn_multiply)
2017-05-21 10:39:39 +02:00
&& native_register (ctx, "/", fn_divide)
&& native_register (ctx, "not", fn_not)
&& native_register (ctx, "and", fn_and)
2017-05-21 11:26:59 +02:00
&& native_register (ctx, "or", fn_or)
&& native_register (ctx, "eq?", fn_eq)
2017-05-21 12:14:43 +02:00
&& native_register (ctx, "lt?", fn_lt)
&& native_register (ctx, "=", fn_equals)
&& native_register (ctx, "<", fn_less);
2017-05-20 21:24:11 +02:00
}
2017-05-06 13:27:02 +02:00
// --- Main --------------------------------------------------------------------
2017-05-17 23:34:58 +02:00
int
main (int argc, char *argv[]) {
FILE *fp = stdin;
if (argc > 1 && !(fp = fopen (argv[1], "rb"))) {
fprintf (stderr, "%s: %s\n", argv[1], strerror (errno));
return 1;
}
int c;
struct buffer buf = BUFFER_INITIALIZER;
while ((c = fgetc (fp)) != EOF)
buffer_append_c (&buf, c);
buffer_append_c (&buf, 0);
fclose (fp);
2017-05-17 23:34:58 +02:00
struct parser parser;
parser_init (&parser, buf.s, buf.len - 1);
const char *e = NULL;
struct item *program = parser_run (&parser, &e);
#ifndef NDEBUG
printf ("\x1b[1m%s\x1b[0m\n", buf.s);
print_tree (program, 0);
printf ("\n\n");
#endif
free (buf.s);
if (e) {
printf ("%s: %s\n", "parse error", e);
return 1;
}
parser_free (&parser);
2017-05-06 13:27:02 +02:00
struct context ctx;
context_init (&ctx);
2017-05-21 09:47:24 +02:00
if (!init_runtime_library (&ctx))
2017-05-18 18:01:57 +02:00
printf ("%s\n", "runtime library initialization failed");
2017-05-18 19:47:10 +02:00
struct item *result = NULL;
(void) execute (&ctx, program, &result);
2017-05-18 19:47:10 +02:00
item_free_list (result);
item_free_list (program);
2017-05-06 13:27:02 +02:00
const char *failure = ctx.error;
2017-05-06 13:27:02 +02:00
if (ctx.memory_failure)
failure = "memory allocation failure";
if (failure)
printf ("%s: %s\n", "runtime error", failure);
2017-05-06 13:27:02 +02:00
context_free (&ctx);
return 0;
}