Limit regex match to one line, accept identicals

This commit makes two small tweaks, each likely to improve the behavior
of the library.

The first is to prevent any newlines in a regex match.  They're invalid
in a regex and would indicate an overbroad matched region.

The second is to accept a diff if the matched regex is identical to the
text it should 'match'.  Doing otherwise would be creating issues where
there were none, since the point is to match the old and the new.

Author: mnemnion

src/ohsnap.zig

@@ -87,7 +87,7 @@ pub fn snapfmt(ohsnap: OhSnap, location: SourceLocation, text: []const u8) Snap
 }
 
 // Regex for detecting embedded regexen
-const ignore_regex_string = "<\\^.+?\\$>";
+const ignore_regex_string = "<\\^[^\n]+?\\$>";
 const regex_finder = mvzr.compile(ignore_regex_string).?;
 
 pub const Snap = struct {
@@ -216,12 +216,17 @@ pub const Snap = struct {
             const snap_end = found.end;
             const got_start = diffz.diffIndex(diffs.*, snap_start);
             const got_end = diffz.diffIndex(diffs.*, snap_end);
+            // Check if these are identical (use/mention distinction!)
+            if (std.mem.eql(u8, found.slice, got[got_start..got_end])) {
+                // That's fine then
+                continue :regex_while;
+            }
             // Trim the angle brackets off the regex.
             const exclude_regex = found.slice[1 .. found.slice.len - 1];
             const maybe_matcher = UserRegex.compile(exclude_regex);
             if (maybe_matcher == null) {
-                std.debug.print("issue with mvzr or regex, hard to say.\n", .{});
-                break :regex_while;
+                std.debug.print("Issue with mvzr or regex, hard to say. Regex string: {s}\n", .{exclude_regex});
+                continue :regex_while;
             }
             const matcher = maybe_matcher.?;
             const maybe_match = matcher.match(got[got_start..got_end]);