diagnostics: give m4 precise locations

Currently we pass only the columns based on the screen-width, which is
important for the carets.  But we don't pass the bytes-based columns,
which is important for the colors.  Pass both.

* src/muscle-tab.c (muscle_boundary_grow): Also pass the byte-based column.
* src/location.c (location_caret): Clarify.
(boundary_set_from_string): Adjust to the new format.
* tests/diagnostics.at (Tabulations and multibyte characters from M4): New.

3 files changed, 33 insertions, 20 deletions

diff --git a/src/location.c b/src/location.c
index e2299c2e..2e8b62b8 100644
--- a/src/location.c
+++ b/src/location.c
@@ -238,8 +238,8 @@ location_caret (location loc, const char *style, FILE *out)
     int c = getc (caret_info.source);
     if (c != EOF)
       {
-        /* Quote the file (at most first line in the case of multiline
-           location).  Indent by a single column.  */
+        /* Quote the file (at most the first line in the case of
+           multiline locations).  */
         fprintf (out, "%5d | ", loc.start.line);
         bool single_line = loc.start.line == loc.end.line;
         /* Consider that single point location (with equal boundaries)
@@ -269,9 +269,9 @@ location_caret (location loc, const char *style, FILE *out)
           putc ('^', out);
           /* Underlining a multiline location ends with the first
              line.  */
-          int len = loc.start.line != loc.end.line
-            ? ftell (caret_info.source) - caret_info.offset
-            : loc.end.column;
+          int len = single_line
+            ? loc.end.column
+            : ftell (caret_info.source) - caret_info.offset;
           for (int i = loc.start.column + 1; i < len; ++i)
             putc ('~', out);
           end_use_class (style, out);
@@ -289,17 +289,29 @@ location_empty (location loc)
 }
 
 void
-boundary_set_from_string (boundary *bound, char *loc_str)
+boundary_set_from_string (boundary *bound, char *str)
 {
-  /* Must search in reverse since the file name field may
-   * contain '.' or ':'.  */
-  char *delim = strrchr (loc_str, '.');
-  aver (delim);
-  *delim = '\0';
-  bound->byte = bound->column = atoi (delim+1);
-  delim = strrchr (loc_str, ':');
-  aver (delim);
-  *delim = '\0';
-  bound->line = atoi (delim+1);
-  bound->file = uniqstr_new (loc_str);
+  /* Must search in reverse since the file name field may contain '.'
+     or ':'.  */
+  char *at = strrchr (str, '@');
+  if (at)
+    {
+      *at = '\0';
+      bound->byte = atoi (at+1);
+    }
+  {
+    char *dot = strrchr (str, '.');
+    aver (dot);
+    *dot = '\0';
+    bound->column = atoi (dot+1);
+    if (!at)
+      bound->byte = bound->column;
+  }
+  {
+    char *colon = strrchr (str, ':');
+    aver (colon);
+    *colon = '\0';
+    bound->line = atoi (colon+1);
+  }
+  bound->file = uniqstr_new (str);
 }
diff --git a/src/location.h b/src/location.h
index 7e473824..2539f27f 100644
--- a/src/location.h
+++ b/src/location.h
@@ -135,7 +135,8 @@ location_cmp (location a, location b)
 /* Whether this is the empty location.  */
 bool location_empty (location loc);
 
-/* LOC_STR must be formatted as 'file:line.column', it will be modified.  */
-void boundary_set_from_string (boundary *bound, char *loc_str);
+/* STR must be formatted as 'file:line.column@byte' or 'file:line.column',
+   it will be modified.  */
+void boundary_set_from_string (boundary *bound, char *str);
 
 #endif /* ! defined LOCATION_H_ */
diff --git a/src/muscle-tab.c b/src/muscle-tab.c
index cd4beeb2..fbb80fc2 100644
--- a/src/muscle-tab.c
+++ b/src/muscle-tab.c
@@ -275,7 +275,7 @@ muscle_boundary_grow (char const *key, boundary bound)
 {
   obstack_sgrow  (&muscle_obstack, "[[");
   obstack_escape (&muscle_obstack, bound.file);
-  obstack_printf (&muscle_obstack, ":%d.%d]]", bound.line, bound.column);
+  obstack_printf (&muscle_obstack, ":%d.%d@@%d]]", bound.line, bound.column, bound.byte);
   char const *extension = obstack_finish0 (&muscle_obstack);
   muscle_grow (key, extension, "", "");
   obstack_free (&muscle_obstack, extension);