message: Introduce underlining for error message regions.

[pspp] / src / language / control / define.c

diff --git a/src/language/control/define.c b/src/language/control/define.c
index 48dda4c7afb0331afa63113d9cc0fc5e36ae7a08..e1d9622ff4067b882a92a8bc259c1bdf8dac2025 100644 (file)
--- a/src/language/control/define.c
+++ b/src/language/control/define.c
@@ -23,6 +23,7 @@
 #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"
@@ -30,12 +31,6 @@
 #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)
 {
@@ -96,19 +91,26 @@ dup_arg_type (struct lexer *lexer, bool *saw_arg_type)
 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)),
-    .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))
@@ -219,10 +221,9 @@ cmd_define (struct lexer *lexer, struct dataset *ds UNUSED)
                 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;
             }
@@ -232,12 +233,11 @@ cmd_define (struct lexer *lexer, struct dataset *ds UNUSED)
                 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;
             }
@@ -276,7 +276,14 @@ cmd_define (struct lexer *lexer, struct dataset *ds UNUSED)
       ds_put_byte (&body, '\n');
       lex_get (lexer);
     }
-  m->location->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);