14908152066

Committed 08 May 2025 01:49PM UTC coverage: 61.309% (-5.7%) from 67.057%

Build # 14908152066

Build Type

push

github

Committed by

alerque

Commit Message

chore(registries): Touchup command registry deprecation shims

Run Details

1 of 2 new or added lines in 2 files covered. (50.0%)

1379 existing lines in 46 files now uncovered.

13556 of 22111 relevant lines covered (61.31%)

11834.23 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

60.81

/languages/unicode.lua

local base = require("languages.base")

local language = pl.class(base)
language._name = "unicode"

function language:_init (typesetter)
   base._init(self, typesetter)
end

function language:setupNodeMaker ()
   self.nodemaker = require("languages.unicode-nodemaker")
end

local icu = require("justenoughicu")

function language:_numberingMethod (input)
   return ("numberTo%s"):format(icu.case(input, "und", "title"))
end

-- Decent subset from unum.h
local icuStyles = {
   default = 0, -- UNUM_PATTERN_DECIMAL
   decimal = 1, -- UNUM_DECIMAL
   string = 5, -- UNUM_SPELLOUT
   ordinal = 6, -- UNUM_ORDINAL
}

-- Numbering system for which we _know_ that ICU doesn't have a
-- default(0) format style rule (i.e. spits out latin)
-- This table is just an optimization to avoid calling ICU twice when this
-- occurs, e.g. "roman" may be quite frequent as a numbering system.
local icuStyleBypass = {
   roman = true,
}

local icuFormat = function (num, lang, options)
   -- Consistency: further below we'll concatenate those, and an empty
   -- string is likely a user mistake.
   if not lang and not options.system then
      SU.warn("Number formatting needs a language or a numbering system")
      return tonumber(num)
   end

   -- ICU format style (enum)
   options.style = options.style or "default"
   local icustyle = options.style and icuStyles[options.style]
   if not icustyle then
      SU.warn("Number formatting style is unrecognized (using default as fallback)")
      icustyle = 0
   end

   -- ICU locale: see  https://unicode-org.github.io/icu/userguide/locale/
   -- Ex. "en", "en-US", "sr-Latn"...
   local iculocale = lang or ""
   -- ICU keyword for a numbering system specifier: @numbers=xxxx
   -- The specifiers are defined here:
   -- https://github.com/unicode-org/cldr/blob/main/common/bcp47/number.xml
   if options.system then
      options.system = options.system:lower()
      iculocale = iculocale .. "@numbers=" .. options.system
      if icuStyleBypass[options.system] then
         icustyle = 1
      end
   end

   local ok, result = pcall(icu.format_number, num, iculocale, icustyle)
   if ok and options.system and icustyle == 0 and options.system ~= "latn" and result == tostring(num) then
      -- There are valid cases where "@numbers=xxxx" with default(0) and decimal(1) styles both work.
      -- Typically, with num=1234
      --   "@numbers=arab" in default(0) --> ١٢٣٤
      --   "@numbers=arab", in decimal(1) --> ١٬٢٣٤
      -- But in many cases, ICU may fallback to latin, e.g. take "roman" (or "grek")
      --   "@numbers=roman" in default(0) --> 1234
      --   "@numbers=roman" in default(1) --> MCCXXXIV
      -- Be user friendly and attempt honoring the script.
      ok, result = pcall(icu.format_number, num, "@numbers=" .. options.system, 1)
   end
   if not ok then
      SU.warn("Number formatting failed: " .. tostring(result))
   end
   return tostring(ok and result or num)
end

--- Formats a number according to options, and optional case
--- Options:
--- - system: a numbering system string, e.g. "latn" (= "arabic"), "roman", "arab", etc.
---   With the addition of "alpha".
---   Casing is guessed from the system (e.g. roman, Roman, ROMAN) unless specified
--- - style: a format style string, i.e. "default", "decimal", "ordinal", "string")
---   E.g. in English and latin script:   1234        1,234     1,124th    one thousand...
---   Possibly extended by additional language-specific formatting rules.
--- Obviously, some combinations of system, style, and case won't do anything of worth.
function language:formatNumber (num, options, case)
   if math.abs(num) > 9223372036854775807 then
      SU.warn("Integers larger than 64 bits do not reproduce properly in all formats")
   end
   options = options or {}
   -- "arabic" is the weirdly name, but quite friendly, used e.g. in counters and
   -- in several other places, let's keep it as a compatibility alias.
   if options.system == "arabic" then
      options.system = "latn"
   end
   local system = options.system
   -- Attempt to derive a case from the casing of system if none otherwise specified
   if not case then
      if system then
         if system:match("^%l") then
            case = "lower"
         elseif system:match("^.%l") then
            case = "title"
         else
            case = "upper"
         end
      else
         case = "lower"
      end
   end
   -- Now that we're done guessing the case, normalize system to lower for further use
   system = system and system:lower()
   local lang = system and system == "roman" and "la" or self._name
   local style = options.style
   local result
   if style and type(getmetatable(self)[self:_numberingMethod(style)]) == "function" then
      result = self[self:_numberingMethod(style)](self, num, options)
   elseif system and type(self[self:_numberingMethod(system)]) == "function" then
      result = self[self:_numberingMethod(system)](self, num, options)
   else
      result = icuFormat(num, lang, options)
   end
   return icu.case(result, lang, case)
end

-- Alpha is a special case (a numbering system, though this table is for formatting style hooks normally)
function language:numberToAlpha (num)
   local out = ""
   local a = string.byte("a")
   repeat
      num = num - 1
      out = string.char(num % 26 + a) .. out
      num = (num - num % 26) / 26
   until num < 1
   return out
end

-- Greek is another special case:
-- There are books where one wants to number items with Greek letters in sequence, e.g. annotations in biblical material
-- etc. as in "α β γ δ ε ζ η θ ι κ λ μ ν ξ ο π ρ σ τ υ φ χ ψ ω". We can't use ICU "grek" or "greklow" numbering systems
-- because they are non-arithmetical, e.g. 6 is a digamma (ϝ´), 11 is iota alpha (ια´), etc. and they are also all
-- followed by a numeric marker ("keraia").
function language:numberToGreek (num)
   local out = ""
   local a = SU.codepoint("α") -- alpha
   if num < 18 then
      -- alpha to rho
      out = luautf8.char(num + a - 1)
   elseif num < 25 then
      -- sigma to omega (unicode has two sigmas here, we skip one)
      out = luautf8.char(num + a)
   else
      -- Don't try to be too clever
      SU.error("Greek numbering is only supported up to 24")
   end
   return out
end

return language

1	local base = require("languages.base")	42✔
2
3	local language = pl.class(base)	42✔
4	language._name = "unicode"	42✔
5
6	function language:_init (typesetter)	42✔
7	base._init(self, typesetter)	133✔
8	end
9
10	function language:setupNodeMaker ()	42✔
11	self.nodemaker = require("languages.unicode-nodemaker")	131✔
12	end
13
14	local icu = require("justenoughicu")	42✔
15
16	function language:_numberingMethod (input)	42✔
17	return ("numberTo%s"):format(icu.case(input, "und", "title"))	96✔
18	end
19
20	-- Decent subset from unum.h
21	local icuStyles = {	42✔
22	default = 0, -- UNUM_PATTERN_DECIMAL
23	decimal = 1, -- UNUM_DECIMAL
24	string = 5, -- UNUM_SPELLOUT
25	ordinal = 6, -- UNUM_ORDINAL
26	}
27
28	-- Numbering system for which we _know_ that ICU doesn't have a
29	-- default(0) format style rule (i.e. spits out latin)
30	-- This table is just an optimization to avoid calling ICU twice when this
31	-- occurs, e.g. "roman" may be quite frequent as a numbering system.
32	local icuStyleBypass = {	42✔
33	roman = true,
34	}
35
36	local icuFormat = function (num, lang, options)
37	-- Consistency: further below we'll concatenate those, and an empty
38	-- string is likely a user mistake.
39	if not lang and not options.system then	96✔
40	SU.warn("Number formatting needs a language or a numbering system")	×
41	return tonumber(num)	×
42	end
43
44	-- ICU format style (enum)
45	options.style = options.style or "default"	96✔
46	local icustyle = options.style and icuStyles[options.style]	96✔
47	if not icustyle then	96✔
48	SU.warn("Number formatting style is unrecognized (using default as fallback)")	×
49	icustyle = 0	×
50	end
51
52	-- ICU locale: see https://unicode-org.github.io/icu/userguide/locale/
53	-- Ex. "en", "en-US", "sr-Latn"...
54	local iculocale = lang or ""	96✔
55	-- ICU keyword for a numbering system specifier: @numbers=xxxx
56	-- The specifiers are defined here:
57	-- https://github.com/unicode-org/cldr/blob/main/common/bcp47/number.xml
58	if options.system then	96✔
59	options.system = options.system:lower()	192✔
60	iculocale = iculocale .. "@numbers=" .. options.system	96✔
61	if icuStyleBypass[options.system] then	96✔
UNCOV 62	icustyle = 1	×
63	end
64	end
65
66	local ok, result = pcall(icu.format_number, num, iculocale, icustyle)	96✔
67	if ok and options.system and icustyle == 0 and options.system ~= "latn" and result == tostring(num) then	96✔
68	-- There are valid cases where "@numbers=xxxx" with default(0) and decimal(1) styles both work.
69	-- Typically, with num=1234
70	-- "@numbers=arab" in default(0) --> ١٢٣٤
71	-- "@numbers=arab", in decimal(1) --> ١٬٢٣٤
72	-- But in many cases, ICU may fallback to latin, e.g. take "roman" (or "grek")
73	-- "@numbers=roman" in default(0) --> 1234
74	-- "@numbers=roman" in default(1) --> MCCXXXIV
75	-- Be user friendly and attempt honoring the script.
UNCOV 76	ok, result = pcall(icu.format_number, num, "@numbers=" .. options.system, 1)	×
77	end
78	if not ok then	96✔
79	SU.warn("Number formatting failed: " .. tostring(result))	×
80	end
81	return tostring(ok and result or num)	96✔
82	end
83
84	--- Formats a number according to options, and optional case
85	--- Options:
86	--- - system: a numbering system string, e.g. "latn" (= "arabic"), "roman", "arab", etc.
87	--- With the addition of "alpha".
88	--- Casing is guessed from the system (e.g. roman, Roman, ROMAN) unless specified
89	--- - style: a format style string, i.e. "default", "decimal", "ordinal", "string")
90	--- E.g. in English and latin script: 1234 1,234 1,124th one thousand...
91	--- Possibly extended by additional language-specific formatting rules.
92	--- Obviously, some combinations of system, style, and case won't do anything of worth.
93	function language:formatNumber (num, options, case)	42✔
94	if math.abs(num) > 9223372036854775807 then	96✔
95	SU.warn("Integers larger than 64 bits do not reproduce properly in all formats")	×
96	end
97	options = options or {}	96✔
98	-- "arabic" is the weirdly name, but quite friendly, used e.g. in counters and
99	-- in several other places, let's keep it as a compatibility alias.
100	if options.system == "arabic" then	96✔
101	options.system = "latn"	96✔
102	end
103	local system = options.system	96✔
104	-- Attempt to derive a case from the casing of system if none otherwise specified
105	if not case then	96✔
106	if system then	96✔
107	if system:match("^%l") then	96✔
108	case = "lower"	96✔
UNCOV 109	elseif system:match("^.%l") then	×
110	case = "title"	×
111	else
UNCOV 112	case = "upper"	×
113	end
114	else
115	case = "lower"	×
116	end
117	end
118	-- Now that we're done guessing the case, normalize system to lower for further use
119	system = system and system:lower()	192✔
120	local lang = system and system == "roman" and "la" or self._name	96✔
121	local style = options.style	96✔
122	local result
123	if style and type(getmetatable(self)[self:_numberingMethod(style)]) == "function" then	96✔
124	result = self[self:_numberingMethod(style)](self, num, options)	×
125	elseif system and type(self[self:_numberingMethod(system)]) == "function" then	192✔
UNCOV 126	result = self[self:_numberingMethod(system)](self, num, options)	×
127	else
128	result = icuFormat(num, lang, options)	192✔
129	end
130	return icu.case(result, lang, case)	96✔
131	end
132
133	-- Alpha is a special case (a numbering system, though this table is for formatting style hooks normally)
134	function language:numberToAlpha (num)	42✔
UNCOV 135	local out = ""	×
UNCOV 136	local a = string.byte("a")	×
137	repeat
UNCOV 138	num = num - 1	×
UNCOV 139	out = string.char(num % 26 + a) .. out	×
UNCOV 140	num = (num - num % 26) / 26	×
UNCOV 141	until num < 1	×
UNCOV 142	return out	×
143	end
144
145	-- Greek is another special case:
146	-- There are books where one wants to number items with Greek letters in sequence, e.g. annotations in biblical material
147	-- etc. as in "α β γ δ ε ζ η θ ι κ λ μ ν ξ ο π ρ σ τ υ φ χ ψ ω". We can't use ICU "grek" or "greklow" numbering systems
148	-- because they are non-arithmetical, e.g. 6 is a digamma (ϝ´), 11 is iota alpha (ια´), etc. and they are also all
149	-- followed by a numeric marker ("keraia").
150	function language:numberToGreek (num)	42✔
151	local out = ""	×
152	local a = SU.codepoint("α") -- alpha	×
153	if num < 18 then	×
154	-- alpha to rho
155	out = luautf8.char(num + a - 1)	×
156	elseif num < 25 then	×
157	-- sigma to omega (unicode has two sigmas here, we skip one)
158	out = luautf8.char(num + a)	×
159	else
160	-- Don't try to be too clever
161	SU.error("Greek numbering is only supported up to 24")	×
162	end
163	return out	×
164	end
165
166	return language	42✔

sile-typesetter / sile / 14908152066

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous