#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)
{
{
if (*saw_arg_type)
{
- lex_error (lexer, _("Only one of !TOKENS, !CHAREND, !ENCLOSE, or "
- "!CMDEND is allowed."));
+ lex_next_error (lexer, -1, -1,
+ _("Only one of !TOKENS, !CHAREND, !ENCLOSE, or "
+ "!CMDEND is allowed."));
return false;
}
else
}
}
+static bool
+parse_macro_body (struct lexer *lexer, struct macro_tokens *mts)
+{
+ *mts = (struct macro_tokens) { .n = 0 };
+ struct string body = DS_EMPTY_INITIALIZER;
+ struct msg_point start = lex_ofs_start_point (lexer, lex_ofs (lexer));
+ while (!match_macro_id (lexer, "!ENDDEFINE"))
+ {
+ if (lex_token (lexer) != T_STRING)
+ {
+ lex_error (lexer,
+ _("Syntax error expecting macro body or !ENDDEFINE."));
+ ds_destroy (&body);
+ return false;
+ }
+
+ ds_put_substring (&body, lex_tokss (lexer));
+ ds_put_byte (&body, '\n');
+ lex_get (lexer);
+ }
+
+ struct segmenter segmenter = segmenter_init (lex_get_syntax_mode (lexer),
+ true);
+ struct substring p = body.ss;
+ bool ok = true;
+ while (p.length > 0)
+ {
+ enum segment_type type;
+ int seg_len = segmenter_push (&segmenter, p.string,
+ p.length, true, &type);
+ assert (seg_len >= 0);
+
+ struct macro_token mt = {
+ .token = { .type = T_STOP },
+ .syntax = ss_head (p, seg_len),
+ };
+ enum tokenize_result result
+ = token_from_segment (type, mt.syntax, &mt.token);
+ ss_advance (&p, seg_len);
+
+ switch (result)
+ {
+ case TOKENIZE_EMPTY:
+ break;
+
+ case TOKENIZE_TOKEN:
+ macro_tokens_add (mts, &mt);
+ break;
+
+ case TOKENIZE_ERROR:
+ {
+ size_t start_offset = mt.syntax.string - body.ss.string;
+ size_t end_offset = start_offset + (mt.syntax.length ? mt.syntax.length - 1 : 0);
+
+ const struct msg_location loc = {
+ .file_name = intern_new_if_nonnull (lex_get_file_name (lexer)),
+ .start = msg_point_advance (start, ss_buffer (body.ss.string, start_offset)),
+ .end = msg_point_advance (start, ss_buffer (body.ss.string, end_offset)),
+ .src = CONST_CAST (struct lex_source *, lex_source (lexer)),
+ };
+ msg_at (SE, &loc, "%s", mt.token.string.string);
+ intern_unref (loc.file_name);
+
+ ok = false;
+ }
+ break;
+ }
+
+ token_uninit (&mt.token);
+ }
+ ds_destroy (&body);
+ return ok;
+}
+
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, _("Syntax error expecting identifier."));
+ return CMD_FAILURE;
+ }
+ const char *name = lex_tokcstr (lexer);
+ if (!id_is_plausible (name + (name[0] == '!')))
+ {
+ lex_error (lexer, _("Syntax error expecting identifier."));
+ return CMD_FAILURE;
+ }
- /* Parse macro name. */
struct macro *m = xmalloc (sizeof *m);
- *m = (struct macro) {
- .name = ss_xstrdup (lex_tokss (lexer)),
- .location = xmalloc (sizeof *m->location),
- };
- *m->location = (struct msg_location) {
- .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))
goto error;
size_t allocated_params = 0;
+ int keyword_ofs = 0;
while (!lex_match (lexer, T_RPAREN))
{
if (m->n_params >= allocated_params)
{
if (param_index > 0 && !m->params[param_index - 1].positional)
{
- lex_error (lexer, _("Positional parameters must precede "
- "keyword parameters."));
+ lex_next_error (lexer, -1, -1,
+ _("Positional parameters must precede "
+ "keyword parameters."));
+ lex_ofs_msg (lexer, SN, keyword_ofs, keyword_ofs,
+ _("Here is a previous keyword parameter."));
goto error;
}
}
else
{
+ if (keyword_ofs == 0)
+ keyword_ofs = lex_ofs (lexer);
if (lex_token (lexer) == T_MACRO_ID)
{
lex_error (lexer, _("Keyword macro parameter must be named in "
{
if (saw_default)
{
- lex_error (lexer,
- _("!DEFAULT is allowed only once per argument."));
+ lex_next_error (
+ lexer, -1, -1,
+ _("!DEFAULT is allowed only once per argument."));
goto error;
}
saw_default = true;
goto error;
p->arg_type = ARG_CHAREND;
- p->charend = (struct token) { .type = T_STOP };
if (!lex_force_match (lexer, T_LPAREN)
- || !parse_quoted_token (lexer, &p->charend)
+ || !parse_quoted_token (lexer, &p->end)
|| !lex_force_match (lexer, T_RPAREN))
goto error;
}
goto error;
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])
+ || !parse_quoted_token (lexer, &p->start)
|| !lex_force_match (lexer, T_COMMA)
- || !parse_quoted_token (lexer, &p->enclose[1])
+ || !parse_quoted_token (lexer, &p->end)
|| !lex_force_match (lexer, T_RPAREN))
goto error;
}
goto error;
}
- struct string body = DS_EMPTY_INITIALIZER;
- while (!match_macro_id (lexer, "!ENDDEFINE"))
- {
- if (lex_token (lexer) != T_STRING)
- {
- lex_error (lexer, _("Expecting macro body or !ENDDEFINE"));
- ds_destroy (&body);
- goto error;
- }
-
- ds_put_substring (&body, lex_tokss (lexer));
- ds_put_byte (&body, '\n');
- lex_get (lexer);
- }
- m->location->last_line = lex_get_last_line_number (lexer, 0);
+ if (!parse_macro_body (lexer, &m->body))
+ goto error;
- macro_tokens_from_string (&m->body, body.ss, lex_get_syntax_mode (lexer));
- ds_destroy (&body);
+ 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 },
+ };
lex_define_macro (lexer, m);
projects / pspp / blobdiff