Module:Unicode data/testcases/documentation

(refresh)

test get_block_range:

	Text	Expected	Actual
	Basic Latin	U+0000–U+007F	U+0000–U+007F
	blah	nil	nil

test get_entry_title:

	Text	Expected	Actual
	U+0023: #	Unsupported titles/Number sign	Unsupported titles/Number sign
	U+0020:	Unsupported titles/Space	Unsupported titles/Space

test is_assigned:

	Text	Expected	Actual
	U+0061: a	true	true
	U+0378	false	false
	U+40000	false	false

test is_combining:

	Text	Expected	Actual
	U+0060: `	false	false
	U+0300: ◌̀	true	true
	U+0378	false	false
	U+DC00	false	false

test is_printable:

	Text	Expected	Actual
	U+0000	false	false
	U+0020:	true	true
	U+0061: a	true	true

test is_valid_pagename:

	Text	Expected	Actual
	#	false	false
		false	false
	word	true	true

test is_whitespace:

	Text	Expected	Actual
	U+0020:	true	true
	U+0061: a	false	false

test lookup_block:

	Text	Expected	Actual
	U+0064: d	Basic Latin	Basic Latin
	U+030B: ◌̋	Combining Diacritical Marks	Combining Diacritical Marks
	U+03A3: Σ	Greek and Coptic	Greek and Coptic
	U+3175: ㅵ	Hangul Compatibility Jamo	Hangul Compatibility Jamo
	U+AC01: 각	Hangul Syllables	Hangul Syllables
	U+10FFFF	Supplementary Private Use Area-B	Supplementary Private Use Area-B

test lookup_category:

	Text	Expected	Actual
	U+0009	Cc	Cc
	U+0020:	Zs	Zs
	U+005B: [	Ps	Ps
	U+005D: ]	Pe	Pe
	U+005E: ^	Sk	Sk
	U+0041: A	Lu	Lu
	U+00AD	Cf	Cf
	U+00BE: ¾	No	No
	U+00AB: «	Pi	Pi
	U+00BB: »	Pf	Pf
	U+0300: ◌̀	Mn	Mn
	U+0488: ҈	Me	Me
	U+0663: ٣	Nd	Nd
	U+5B50: 子	Lo	Lo
	U+1FAE: ᾮ	Lt	Lt
	U+1B44: ◌᭄	Mc	Mc
	U+2208: ∈	Sm	Sm
	U+203F: ‿	Pc	Pc
	U+21B9: ↹	So	So
	U+2E17: ⸗	Pd	Pd
	U+2167: Ⅷ	Nl	Nl
	U+2028	Zl	Zl
	U+2029	Zp	Zp
	U+309E: ゞ	Lm	Lm
	U+D800	Cs	Cs
	U+FFE1: ￡	Sc	Sc
	U+FFFF	Cn	Cn
	U+100000	Co	Co

test lookup_image:

	Text	Expected	Actual
	U+203D: ‽
	U+30A2: ア
	U+0B85: அ

test lookup_name:

	Text	Expected	Actual
	U+0000	<control-0000>	<control-0000>
	U+007F	<control-007F>	<control-007F>
	U+00C1: Á	LATIN CAPITAL LETTER A WITH ACUTE	LATIN CAPITAL LETTER A WITH ACUTE
	U+0300: ◌̀	COMBINING GRAVE ACCENT	COMBINING GRAVE ACCENT
	U+0378	<reserved-0378>	<reserved-0378>
	U+1B44: ◌᭄	BALINESE ADEG ADEG	BALINESE ADEG ADEG
	U+1F71: ά	GREEK SMALL LETTER ALPHA WITH OXIA	GREEK SMALL LETTER ALPHA WITH OXIA
	U+3555: 㕕	CJK UNIFIED IDEOGRAPH-3555	CJK UNIFIED IDEOGRAPH-3555
	U+AC01: 각	HANGUL SYLLABLE GAG	HANGUL SYLLABLE GAG
	U+D5FF: 헿	HANGUL SYLLABLE HEH	HANGUL SYLLABLE HEH
	U+DC00	<surrogate-DC00>	<surrogate-DC00>
	U+EEEE	<private-use-EEEE>	<private-use-EEEE>
	U+FDD1	<noncharacter-FDD1>	<noncharacter-FDD1>
	U+FFFD: �	REPLACEMENT CHARACTER	REPLACEMENT CHARACTER
	U+FFFF	<noncharacter-FFFF>	<noncharacter-FFFF>
	U+1F4A9: 💩	PILE OF POO	PILE OF POO
	U+E0000	<reserved-E0000>	<reserved-E0000>
	U+F0F0F	<private-use-F0F0F>	<private-use-F0F0F>
	U+10FFFF	<noncharacter-10FFFF>	<noncharacter-10FFFF>

test lookup_script:

	Text	Expected	Actual
	U+005B: [	Zyyy	Zyyy
	U+0041: A	Latn	Latn
	U+4E00: 一	Hani	Hani
	U+0300: ◌̀	Zinh	Zinh
	U+03CE: ώ	Grek	Grek
	U+1F66: ὦ	Grek	Grek
	U+0416: Ж	Cyrl	Cyrl
	U+0478: Ѹ	Cyrl	Cyrl
	U+A651: ꙑ	Cyrl	Cyrl
	U+10D0: ა	Geor	Geor
	U+10A0: Ⴀ	Geor	Geor
	U+2D00: ⴀ	Geor	Geor
	U+0021: !	Zyyy	Zyyy
	U+2F82B: 北	Hani	Hani