Migrating from anyascii

translit's transliterate() replaces anyascii for Unicode-to-ASCII conversion.

Quick migration

# Before
from anyascii import anyascii
result = anyascii("café")

# After
from translit import transliterate
result = transliterate("café")

Or use the compatibility alias:

from translit import unidecode as anyascii
result = anyascii("café")

API comparison

anyascii translit Notes
anyascii(s) transliterate(s)
— transliterate(s, lang="de") New: language profiles
— transliterate(s, errors="ignore") New: error modes
— transliterate(s, errors="preserve") New: preserve unmapped

Behavioral differences

Transliteration approach

anyascii and translit both provide Unicode → ASCII transliteration, but they use different lookup tables. The core Latin-script mappings are very similar, but edge cases may differ. A detailed character-level comparison across all 83 supported languages shows:

  • 49,089 codepoints across all Unicode blocks tested comprehensively (no sampling)
  • 48,415 mapped by translit vs 48,761 by anyascii — anyascii has broader coverage of some extended script blocks, while translit provides language-aware romanization with 83 language profiles and 1,136 characters only translit maps
  • Most differences are systematic: CJK pinyin casing, Korean romanization, and language-specific national standards
# Common cases — identical
anyascii("café")    # => "cafe"
transliterate("café")  # => "cafe"

# CJK — may differ in romanization style
anyascii("北京")       # romanization varies
transliterate("北京")  # => "bei jing"

Language awareness

anyascii has no language parameter. translit provides 83 language-specific profiles:

from translit import transliterate

# anyascii can't do this
transliterate("München", lang="de")  # => "Muenchen"
transliterate("Malmö", lang="sv")    # => "Malmoe"

Error handling

anyascii silently drops characters with no mapping. translit gives you control:

from translit import transliterate

transliterate("♠", errors="replace", replace_with="?")  # => "?"
transliterate("♠", errors="ignore")                      # => ""
transliterate("♠", errors="preserve")                    # => "♠"

New features in translit

Beyond basic transliteration, translit also provides:

  • slugify() — URL slug generation
  • sanitize_filename() — OS-safe filenames
  • normalize_confusables() — homoglyph normalization
  • TextPipeline — composable text processing
  • strip_accents(), fold_case() — granular text cleaning

All in a single package with one consistent API.