UnicodeDB

This library aims to bring the unicode database to Nim. Main goal is having O(1) access for every API and be lightweight in size.

Note: this library doesn't provide Unicode Common Locale Data (UCLD / CLDR data)

Install

nimble install unicodedb

Compatibility

Nim +1.0.0

Usage

Properties

import unicode
import unicodedb/properties

assert Rune('A'.ord).unicodeCategory() == ctgLu  # 'L'etter, 'u'ppercase
assert Rune('A'.ord).unicodeCategory() in ctgLm+ctgLo+ctgLu+ctgLl+ctgLt
assert Rune('A'.ord).unicodeCategory() in ctgL

echo Rune(0x0660).bidirectional() # 'A'rabic, 'N'umber
# "AN"

echo Rune(0x860).combining()
# 0

echo nfcQcNo in Rune(0x0374).quickCheck()
# true

docs

Names

import unicode
import unicodedb/names

echo lookupStrict("LEFT CURLY BRACKET")  # '{'
# Rune(0x007B)

echo "/".runeAt(0).name()
# "SOLIDUS"

docs

Compositions

import unicode
import unicodedb/compositions

echo composition(Rune(108), Rune(803))
# Rune(7735)

docs

Decompositions

import unicode
import unicodedb/decompositions

echo Rune(0x0F9D).decomposition()
# @[Rune(0x0F9C), Rune(0x0FB7)]

docs

Types

import unicode
import unicodedb/types

assert utmDecimal in Rune(0x0030).unicodeTypes()
assert utmDigit in Rune(0x00B2).unicodeTypes()
assert utmNumeric in Rune(0x2CFD).unicodeTypes()
assert utmLowercase in Rune(0x1E69).unicodeTypes()
assert utmUppercase in Rune(0x0041).unicodeTypes()
assert utmCased in Rune(0x0041).unicodeTypes()
assert utmWhiteSpace in Rune(0x0009).unicodeTypes()
assert utmWord in Rune(0x1E69).unicodeTypes()

const alphaNumeric = utmLowercase + utmUppercase + utmNumeric
assert alphaNumeric in Rune(0x2CFD).unicodeTypes()
assert alphaNumeric in Rune(0x1E69).unicodeTypes()
assert alphaNumeric in Rune(0x0041).unicodeTypes()

docs

Widths

import unicode
import unicodedb/widths

assert "🕺".runeAt(0).unicodeWidth() == uwdtWide

docs

Scripts

import unicode
import unicodedb/scripts

assert "諸".runeAt(0).unicodeScript() == sptHan

docs

Casing

import sequtils
import unicode
import unicodedb/casing

assert toSeq("Ⓗ".runeAt(0).lowerCase) == @["ⓗ".runeAt(0)]
assert toSeq("İ".runeAt(0).lowerCase) == @[0x0069.Rune, 0x0307.Rune]

assert toSeq("ⓗ".runeAt(0).upperCase) == @["Ⓗ".runeAt(0)]
assert toSeq("ffi".runeAt(0).upperCase) == @['F'.ord.Rune, 'F'.ord.Rune, 'I'.ord.Rune]

assert toSeq("ß".runeAt(0).titleCase) == @['S'.ord.Rune, 's'.ord.Rune]

assert toSeq("ᾈ".runeAt(0).caseFold) == @["ἀ".runeAt(0), "ι".runeAt(0)]

docs

Segmentation

import unicode
import unicodedb/segmentation

assert 0x000B.Rune.wordBreakProp == sgwNewline

docs

Related libraries

• nim-unicodeplus
• nim-graphemes
• nim-segmentation
• nim-normalize

Storage

Storage is based on multi-stage tables and minimal perfect hashing data-structures.

Sizes

These are the current collections sizes:

• properties is 40KB. Used by properties(1), category(1), bidirectional(1), combining(1) and quickCheck(1)
• compositions is 12KB. Used by: composition(1)
• decompositions is 89KB. Used by decomposition(1) and canonicalDecomposition(1)
• names is 578KB. Used by name(1) and lookupStrict(1)
• names (lookup) is 241KB. Used by lookupStrict(1)

Missing APIs

New APIs will be added from time to time. If you need something that's missing, please open an issue or PR (please, do mention the use-case).

Upgrading Unicode version

Note: PR's upgrading the unicode version won't get merged, open an issue instead!

• Run nimble gen to check there are no changes to ./src/*_data.nim. If there are try an older Nim version and fix the generators accordingly
• Run nimble gen_tests to update all test data to current unicode version. The tests for a new unicode version run against the previous unicode version.
• Run tests and fix all failing tests. This should require just temporarily commenting out all checks for missing unicode points.
• Overwrite ./gen/UCD data with latest unicode UCD.
• Run nimble gen to generate the new data.
• Run tests. Add checks for missing unicode points back. A handful of unicode points may have change its data, check the unicode changelog page, make sure they are correct and skip them.
• Note: starting Unicode 15 they added multiple @missing lines which breaks the assumption of a default prop for missing CPs and these lines need to be parsed (see DerivedBidiClass for example). So if they add this to more files, the data gen need fixing. Look for lines containing # @missing with a range other than 0000..10FFFF. See Missing_Conventions

Tests

Initial tests were ran against [a dump of] Python's unicodedata module to ensure correctness. Also, the related libraries have their own custom tests (some of the test data is provided by the unicode consortium).

nimble test

Contributing

I plan to work on most missing related libraries (case folding, etc). If you would like to work in one of those, please let me know and I'll add it to the list. If you find the required database data is missing, either open an issue or a PR.

LICENSE

MIT