Charset Detection

Character encoding detection, charset detection, or code page detection is the process of heuristically guessing the character encoding of a series of bytes that represent text. The technique is recognised to be unreliable and is only used when specific metadata, such as a HTTP Content-Type: header is either not available, or is assumed to be untrustworthy.

This algorithm usually involves statistical analysis of byte patterns, like frequency distribution of trigraphs of various languages encoded in each code page that will be detected; such statistical analysis can also be used to perform language detection. This process is not foolproof because it depends on statistical data; for example, some versions of the Windows operating system would mis-detect the phrase "Bush hid the facts" in ASCII as Chinese UTF-16LE.

One of the few cases where charset detection works reliably is detecting UTF-8. This is due to the large percentage of invalid byte sequences in UTF-8, so that text in any other encoding that uses bytes with the high bit set is extremely unlikely to pass a UTF-8 validity test. Unfortunately badly written charset detection routines do not run the reliable UTF-8 test first, and may decide that UTF-8 is some other encoding.

Charset detection is particularly unreliable in Europe, in an environment of mixed ISO-8859 encodings. These are closely related eight-bit encodings that share an overlap in their lower half with ASCII. There is no technical way to tell these encodings apart and recognising them relies on identifying language features, such as letter frequencies or spellings.

Due to the unreliability of heuristic detection, it is better to properly label datasets with the correct encoding. HTML documents served across the web by HTTP should have their encoding stated out-of-band using the Content-Type: header.

An isolated HTML document, such as one being edited as a file on disc, may embed such a header within the file:

If the document is Unicode, then some UTF encodings explicitly label the document with an embedded initial byte order mark.