Off-by-one in extractHtmlAttr can miss attributes at end of tag

[calin-marian]

Problem

extractHtmlAttr in formatter.zig uses pos + attr_name.len < tag_inner.len as the loop bound. This is off by one — if an attribute name starts at the last valid position, the loop terminates before checking it. For example, a tag like <a href="url"> where href ends exactly at the boundary would fail to match.

Additionally, whitespace checks use manual character comparisons (!= ' ' and != '\t' and != '\n') instead of the standard library's std.ascii.isWhitespace, missing characters like \r.

Fix

• Changed loop bound to <=
• Replaced manual whitespace checks with std.ascii.isWhitespace

References

Fixed in PR #7 (commit 1e12dd1).