#include "language/lexer/macro.h"
#include "language/lexer/scan.h"
#include "language/lexer/token.h"
+#include "libpspp/intern.h"
#include "libpspp/message.h"
#include "gl/xalloc.h"
#include "gettext.h"
#define _(msgid) gettext (msgid)
-static bool
-force_macro_id (struct lexer *lexer)
-{
- return lex_token (lexer) == T_MACRO_ID || lex_force_id (lexer);
-}
-
static bool
match_macro_id (struct lexer *lexer, const char *keyword)
{
struct string_lexer slex;
string_lexer_init (&slex, s.string, s.length, SEG_MODE_INTERACTIVE, true);
struct token another_token = { .type = T_STOP };
- if (!string_lexer_next (&slex, token)
- || string_lexer_next (&slex, &another_token))
+ if (string_lexer_next (&slex, token) != SLR_TOKEN
+ || string_lexer_next (&slex, &another_token) != SLR_END)
{
token_uninit (token);
token_uninit (&another_token);
return true;
}
+static bool
+dup_arg_type (struct lexer *lexer, bool *saw_arg_type)
+{
+ if (*saw_arg_type)
+ {
+ lex_error (lexer, _("Only one of !TOKENS, !CHAREND, !ENCLOSE, or "
+ "!CMDEND is allowed."));
+ return false;
+ }
+ else
+ {
+ *saw_arg_type = true;
+ return true;
+ }
+}
+
int
cmd_define (struct lexer *lexer, struct dataset *ds UNUSED)
{
- if (!force_macro_id (lexer))
- return CMD_FAILURE;
+ /* Parse macro name.
+
+ The macro name is a T_STRING token, even though it's an identifier,
+ because that's the way that the segmenter prevents it from getting
+ macro-expanded. */
+ if (lex_token (lexer) != T_STRING)
+ {
+ lex_error (lexer, _("expecting identifier"));
+ return CMD_FAILURE;
+ }
+ const char *name = lex_tokcstr (lexer);
+ if (!id_is_plausible (name + (name[0] == '!'), false))
+ {
+ lex_error (lexer, _("expecting identifier"));
+ return CMD_FAILURE;
+ }
- /* Parse macro name. */
struct macro *m = xmalloc (sizeof *m);
- *m = (struct macro) {
- .name = ss_xstrdup (lex_tokss (lexer)),
- .file_name = xstrdup_if_nonnull (lex_get_file_name (lexer)),
- .first_line = lex_get_first_line_number (lexer, 0),
- };
+ *m = (struct macro) { .name = xstrdup (name) };
+ struct msg_point macro_start = lex_ofs_start_point (lexer, lex_ofs (lexer));
lex_get (lexer);
if (!lex_force_match (lexer, T_LPAREN))
p->positional = false;
p->name = xasprintf ("!%s", lex_tokcstr (lexer));
lex_get (lexer);
-
- if (!lex_force_match (lexer, T_EQUALS))
- goto error;
}
+ lex_match (lexer, T_EQUALS);
- /* Parse default value. */
- if (match_macro_id (lexer, "!DEFAULT"))
+ bool saw_default = false;
+ bool saw_arg_type = false;
+ for (;;)
{
- if (!lex_force_match (lexer, T_LPAREN))
- goto error;
-
- /* XXX Should this handle balanced inner parentheses? */
- while (!lex_match (lexer, T_RPAREN))
+ if (match_macro_id (lexer, "!DEFAULT"))
{
- if (lex_token (lexer) == T_ENDCMD)
+ if (saw_default)
{
- lex_error_expecting (lexer, ")");
+ lex_error (lexer,
+ _("!DEFAULT is allowed only once per argument."));
goto error;
}
- char *syntax = lex_next_representation (lexer, 0, 0);
- const struct macro_token mt = {
- .token = *lex_next (lexer, 0),
- .syntax = ss_cstr (syntax),
- };
- macro_tokens_add (&p->def, &mt);
- free (syntax);
+ saw_default = true;
+
+ if (!lex_force_match (lexer, T_LPAREN))
+ goto error;
+ /* XXX Should this handle balanced inner parentheses? */
+ while (!lex_match (lexer, T_RPAREN))
+ {
+ if (lex_token (lexer) == T_ENDCMD)
+ {
+ lex_error_expecting (lexer, ")");
+ goto error;
+ }
+ char *syntax = lex_next_representation (lexer, 0, 0);
+ const struct macro_token mt = {
+ .token = *lex_next (lexer, 0),
+ .syntax = ss_cstr (syntax),
+ };
+ macro_tokens_add (&p->def, &mt);
+ free (syntax);
+
+ lex_get (lexer);
+ }
+ }
+ else if (match_macro_id (lexer, "!NOEXPAND"))
+ p->expand_arg = false;
+ else if (match_macro_id (lexer, "!TOKENS"))
+ {
+ if (!dup_arg_type (lexer, &saw_arg_type)
+ || !lex_force_match (lexer, T_LPAREN)
+ || !lex_force_int_range (lexer, "!TOKENS", 1, INT_MAX))
+ goto error;
+ p->arg_type = ARG_N_TOKENS;
+ p->n_tokens = lex_integer (lexer);
lex_get (lexer);
+ if (!lex_force_match (lexer, T_RPAREN))
+ goto error;
}
- }
+ else if (match_macro_id (lexer, "!CHAREND"))
+ {
+ if (!dup_arg_type (lexer, &saw_arg_type))
+ goto error;
- if (match_macro_id (lexer, "!NOEXPAND"))
- p->expand_arg = false;
+ p->arg_type = ARG_CHAREND;
- if (match_macro_id (lexer, "!TOKENS"))
- {
- if (!lex_force_match (lexer, T_LPAREN)
- || !lex_force_int_range (lexer, "!TOKENS", 1, INT_MAX))
- goto error;
- p->arg_type = ARG_N_TOKENS;
- p->n_tokens = lex_integer (lexer);
- lex_get (lexer);
- if (!lex_force_match (lexer, T_RPAREN))
- goto error;
- }
- else if (match_macro_id (lexer, "!CHAREND"))
- {
- p->arg_type = ARG_CHAREND;
- p->charend = (struct token) { .type = T_STOP };
+ if (!lex_force_match (lexer, T_LPAREN)
+ || !parse_quoted_token (lexer, &p->end)
+ || !lex_force_match (lexer, T_RPAREN))
+ goto error;
+ }
+ else if (match_macro_id (lexer, "!ENCLOSE"))
+ {
+ if (!dup_arg_type (lexer, &saw_arg_type))
+ goto error;
- if (!lex_force_match (lexer, T_LPAREN)
- || !parse_quoted_token (lexer, &p->charend)
- || !lex_force_match (lexer, T_RPAREN))
- goto error;
- }
- else if (match_macro_id (lexer, "!ENCLOSE"))
- {
- p->arg_type = ARG_ENCLOSE;
- p->enclose[0] = p->enclose[1] = (struct token) { .type = T_STOP };
-
- if (!lex_force_match (lexer, T_LPAREN)
- || !parse_quoted_token (lexer, &p->enclose[0])
- || !lex_force_match (lexer, T_COMMA)
- || !parse_quoted_token (lexer, &p->enclose[1])
- || !lex_force_match (lexer, T_RPAREN))
- goto error;
+ p->arg_type = ARG_ENCLOSE;
+
+ if (!lex_force_match (lexer, T_LPAREN)
+ || !parse_quoted_token (lexer, &p->start)
+ || !lex_force_match (lexer, T_COMMA)
+ || !parse_quoted_token (lexer, &p->end)
+ || !lex_force_match (lexer, T_RPAREN))
+ goto error;
+ }
+ else if (match_macro_id (lexer, "!CMDEND"))
+ {
+ if (!dup_arg_type (lexer, &saw_arg_type))
+ goto error;
+
+ p->arg_type = ARG_CMDEND;
+ }
+ else
+ break;
}
- else if (match_macro_id (lexer, "!CMDEND"))
- p->arg_type = ARG_CMDEND;
- else
+ if (!saw_arg_type)
{
- lex_error_expecting (lexer, "!TOKENS", "!CHAREND",
- "!ENCLOSE", "!CMDEND");
+ lex_error_expecting (lexer, "!TOKENS", "!CHAREND", "!ENCLOSE",
+ "!CMDEND");
goto error;
}
ds_put_byte (&body, '\n');
lex_get (lexer);
}
- m->last_line = lex_get_last_line_number (lexer, 0);
+
+ struct msg_point macro_end = lex_ofs_end_point (lexer, lex_ofs (lexer) - 1);
+ m->location = xmalloc (sizeof *m->location);
+ *m->location = (struct msg_location) {
+ .file_name = intern_new_if_nonnull (lex_get_file_name (lexer)),
+ .start = { .line = macro_start.line },
+ .end = { .line = macro_end.line },
+ };
macro_tokens_from_string (&m->body, body.ss, lex_get_syntax_mode (lexer));
ds_destroy (&body);