Unicode file names and visually identical spellings
Unicode file names can contain visually identical text encoded as different character sequences, so normalization and filesystem comparison rules affect path matching.
A title such as Café.mkv looks straightforward on screen. Its accented letter can be represented by one Unicode code point, or by a base letter followed by a combining accent. The text can look identical while the stored sequences differ.
That matters when a path moves between tools or systems that compare, preserve or normalize names differently.
One visible word, two sequences
Composed: C a f é . m k v
U+0043 U+0061 U+0066 U+00E9 ...
Decomposed: C a f e ◌́ . m k v
U+0043 U+0061 U+0066 U+0065 U+0301 ...
Unicode normalization defines standard ways to transform canonically equivalent sequences. NFC commonly produces the composed form where a composition exists; NFD produces canonical decompositions. In this example, both names normalize to the same NFC string.
| Comparison | Result for these two names |
|---|---|
| Exact code-point sequence | Different |
| NFC-normalized sequence | Equal |
| Original appearance in a typical font | Often indistinguishable |
| Upper/lowercase comparison | A separate rule from normalization |
Normalization is not a universal test for all visually similar text. A Latin letter and a similar-looking letter from another script need not be canonically equivalent. It also does not remove every restriction on reserved characters or pathname lengths.
The file system chooses its own naming behavior
Windows naming documentation describes Unicode storage for long names in its modern file systems. Linux ext4 directory entries store names as byte strings. Other file systems or network services may use particular normalization and comparison rules.
It is therefore incorrect to assume that every file system converts all names to NFC, or that every Linux directory contains valid UTF-8 names. The application, transfer tool and destination can each influence the resulting representation.
An Emby library may contain a correctly displayed title while an external script fails to open the visually matching path. The displayed media title can also come from metadata rather than from the filename, so it is not a reliable source for reconstructing that filename.
Inspect the actual directory entry
When an apparently matching name fails, compare the name enumerated from the directory with the requested name, including code points or encoded bytes where appropriate. Preserve the exact path during that investigation.
Changing names can affect sidecars, playlists and existing path references. A name comparison should establish the difference first. Normalization explains this particular mismatch; it does not imply that every accented filename needs renaming.