12313034533

Committed 13 Dec 2024 09:28AM UTC coverage: 60.234% (-0.7%) from 60.941%

Build # 12313034533

Build Type

push

github

Committed by

web-flow

Commit Message

Merge 5a7694dff into d737b2656

Run Details

9 of 25 new or added lines in 5 files covered. (36.0%)

145 existing lines in 16 files now uncovered.

12801 of 21252 relevant lines covered (60.23%)

2545.46 hits per line

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

44.83

/core/utilities/numbers.lua

--- Number formatting utilities.
--- @module SU.numbers

local icu = require("justenoughicu")

--- @type formatNumber
-- Language-specific number formatters add functions to this table,
-- see e.g. languages/eo.lua
local formatNumber = {
   und = {

      -- Alpha is a special case (a numbering system, though this table is for
      -- formatting style hooks normally)
      alpha = function (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").
      greek = function (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,
   },
}

-- 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 = not options.style and "default" or options.style
   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

setmetatable(formatNumber, {
   __call = function (self, num, options, case)
      -- 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 worth.
      if math.abs(num) > 9223372036854775807 then
         SU.warn("Integers larger than 64 bits do not reproduce properly in all formats")
      end
      options = options or {}

      -- BEGIN COMPATIBILITY SHIM
      if type(options) ~= "table" then
         -- It used to be a string aggregating both concepts.
         SU.deprecated(
            "Previous syntax of SU.formatNumber",
            "new syntax for SU.formatNumber",
            "0.14.6",
            "0.16.0",
            [[
               Previous syntax was SU.formatNumber(num, format[, case]) with a format string
               New syntax is SU.formatNumber(num, options[, case]) with an options table,
               possibly containing:

                 - system: a numbering system string

                   e.g. "latn" (= "arabic"), "roman", "arab", etc. with the addition of
                   "alpha" and "greek". Casing is taken into account (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.

               Note that the new syntax doesn't handle casing on the format style, for
               separation of concerns.
            ]]
         )
         if not case then
            if options:match("^%l") then
               case = "lower"
            elseif options:match("^.%l") then
               case = "title"
            else
               case = "upper"
            end
         end
         if options:lower() == "nth" then
            SU.deprecated("Format 'nth' in SU.formatNumber", "'ordinal' in SU.formatNumber", "0.14.6", "0.16.0")
            options = { style = "ordinal" }
         elseif options:lower() == "string" then
            options = { style = "string" }
         elseif options:lower() == "ordinal" and SILE.settings:get("document.language") == "tr" then
            SU.deprecated(
               "Format 'ordinal' in Turkish in SU.formatNumber",
               "'ordinal-string' in SU.formatNumber",
               "0.14.6",
               "0.16.0"
            )
            options = { style = "ordinal-string" }
         else
            options = { system = options }
         end
      end
      -- END COMPATIBILITY SHIM

      if options.system == "arabic" then
         -- "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.
         options.system = "latn"
      end

      local system = options.system
      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
      system = system and system:lower()

      local lang = system and system == "roman" and "la" or SILE.settings:get("document.language")
      local style = options.style
      local result
      if self[lang] and style and type(self[lang][style]) == "function" then
         -- Language specific hooks exists, use them...
         result = self[lang][style](num, options)
      elseif style and type(self["und"][style]) == "function" then
         -- Global specific hooks exists: use them...
         result = self.und[system](num, options)
      elseif system and type(self["und"][system]) == "function" then
         -- TRICK: Notably, special case for "alpha" and "greek"
         result = system and self.und[system](num, options)
      else
         --- Otherwise, rely on ICU...
         result = icuFormat(num, lang, options)
      end
      return icu.case(result, lang, case)
   end,
})

return formatNumber

1	--- Number formatting utilities.
2	--- @module SU.numbers
3
4	local icu = require("justenoughicu")	97✔
5
6	--- @type formatNumber
7	-- Language-specific number formatters add functions to this table,
8	-- see e.g. languages/eo.lua
9	local formatNumber = {	97✔
10	und = {	97✔
11
12	-- Alpha is a special case (a numbering system, though this table is for
13	-- formatting style hooks normally)
14	alpha = function (num)
UNCOV 15	local out = ""	×
UNCOV 16	local a = string.byte("a")	×
17	repeat
UNCOV 18	num = num - 1	×
UNCOV 19	out = string.char(num % 26 + a) .. out	×
UNCOV 20	num = (num - num % 26) / 26	×
UNCOV 21	until num < 1	×
UNCOV 22	return out	×
23	end,
24	-- Greek is another special case:
25	-- There are books where one wants to number items with Greek letters in
26	-- sequence, e.g. annotations in biblical material etc.
27	-- 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".
28	-- We can't use ICU "grek" or "greklow" numbering systems because they are
29	-- non-arithmetical, e.g. 6 is a digamma (ϝ´), 11 is iota alpha (ια´), etc.
30	-- and they are also all followed by a numeric marker ("keraia").
31	greek = function (num)
32	local out = ""	×
33	local a = SU.codepoint("α") -- alpha	×
34	if num < 18 then	×
35	-- alpha to rho
36	out = luautf8.char(num + a - 1)	×
37	elseif num < 25 then	×
38	-- sigma to omega (unicode has two sigmas here, we skip one)
39	out = luautf8.char(num + a)	×
40	else
41	-- Don't try to be too clever
42	SU.error("Greek numbering is only supported up to 24")	×
43	end
44	return out	×
45	end,
46	},	97✔
47	}
48
49	-- Decent subset from unum.h
50	local icuStyles = {	97✔
51	default = 0, -- UNUM_PATTERN_DECIMAL
52	decimal = 1, -- UNUM_DECIMAL
53	string = 5, -- UNUM_SPELLOUT
54	ordinal = 6, -- UNUM_ORDINAL
55	}
56
57	-- Numbering system for which we _know_ that ICU doesn't have a
58	-- default(0) format style rule (i.e. spits out latin)
59	-- This table is just an optimization to avoid calling ICU twice when this
60	-- occurs, e.g. "roman" may be quite frequent as a numbering system.
61	local icuStyleBypass = {	97✔
62	roman = true,
63	}
64
65	local icuFormat = function (num, lang, options)
66	-- Consistency: further below we'll concatenate those, and an empty
67	-- string is likely a user mistake.
68	if not lang and not options.system then	108✔
69	SU.warn("Number formatting needs a language or a numbering system")	×
70	return tonumber(num)	×
71	end
72
73	-- ICU format style (enum)
74	options.style = not options.style and "default" or options.style	108✔
75	local icustyle = options.style and icuStyles[options.style]	108✔
76	if not icustyle then	108✔
77	SU.warn("Number formatting style is unrecognized (using default as fallback)")	×
78	icustyle = 0	×
79	end
80
81	-- ICU locale: see https://unicode-org.github.io/icu/userguide/locale/
82	-- Ex. "en", "en-US", "sr-Latn"...
83	local iculocale = lang or ""	108✔
84	-- ICU keyword for a numbering system specifier: @numbers=xxxx
85	-- The specifiers are defined here:
86	-- https://github.com/unicode-org/cldr/blob/main/common/bcp47/number.xml
87	if options.system then	108✔
88	options.system = options.system:lower()	216✔
89	iculocale = iculocale .. "@numbers=" .. options.system	108✔
90	if icuStyleBypass[options.system] then	108✔
UNCOV 91	icustyle = 1	×
92	end
93	end
94
95	local ok, result = pcall(icu.format_number, num, iculocale, icustyle)	108✔
96	if ok and options.system and icustyle == 0 and options.system ~= "latn" and result == tostring(num) then	108✔
97	-- There are valid cases where "@numbers=xxxx" with default(0) and decimal(1) styles both work.
98	-- Typically, with num=1234
99	-- "@numbers=arab" in default(0) --> ١٢٣٤
100	-- "@numbers=arab", in decimal(1) --> ١٬٢٣٤
101	-- But in many cases, ICU may fallback to latin, e.g. take "roman" (or "grek")
102	-- "@numbers=roman" in default(0) --> 1234
103	-- "@numbers=roman" in default(1) --> MCCXXXIV
104	-- Be user friendly and attempt honoring the script.
UNCOV 105	ok, result = pcall(icu.format_number, num, "@numbers=" .. options.system, 1)	×
106	end
107	if not ok then	108✔
108	SU.warn("Number formatting failed: " .. tostring(result))	×
109	end
110	return tostring(ok and result or num)	108✔
111	end
112
113	setmetatable(formatNumber, {	194✔
114	__call = function (self, num, options, case)
115	-- Formats a number according to options, and optional case
116	-- Options:
117	-- - system: a numbering system string, e.g. "latn" (= "arabic"), "roman", "arab", etc.
118	-- With the addition of "alpha".
119	-- Casing is guessed from the system (e.g. roman, Roman, ROMAN) unless specified
120	-- - style: a format style string, i.e. "default", "decimal", "ordinal", "string")
121	-- E.g. in English and latin script: 1234 1,234 1,124th one thousand...
122	-- Possibly extended by additional language-specific formatting rules.
123	-- Obviously, some combinations of system, style and case won't do anything worth.
124	if math.abs(num) > 9223372036854775807 then	108✔
125	SU.warn("Integers larger than 64 bits do not reproduce properly in all formats")	×
126	end
127	options = options or {}	108✔
128
129	-- BEGIN COMPATIBILITY SHIM
130	if type(options) ~= "table" then	108✔
131	-- It used to be a string aggregating both concepts.
132	SU.deprecated(	×
133	"Previous syntax of SU.formatNumber",
134	"new syntax for SU.formatNumber",
135	"0.14.6",
136	"0.16.0",
137	[[	×
138	Previous syntax was SU.formatNumber(num, format[, case]) with a format string
139	New syntax is SU.formatNumber(num, options[, case]) with an options table,
140	possibly containing:
141
142	- system: a numbering system string
143
144	e.g. "latn" (= "arabic"), "roman", "arab", etc. with the addition of
145	"alpha" and "greek". Casing is taken into account (e.g. roman, Roman,
146	ROMAN) unless specified.
147
148	- style: a format style string
149
150	i.e. "default", "decimal", "ordinal", "string"). E.g. in English and latin
151	script: 1234 1,234 1,124th one thousand ...
152	Possibly extended by additional language-specific formatting rules.
153
154	Note that the new syntax doesn't handle casing on the format style, for
155	separation of concerns.
156	]]	×
157	)
158	if not case then	×
159	if options:match("^%l") then	×
160	case = "lower"	×
161	elseif options:match("^.%l") then	×
162	case = "title"	×
163	else
164	case = "upper"	×
165	end
166	end
167	if options:lower() == "nth" then	×
168	SU.deprecated("Format 'nth' in SU.formatNumber", "'ordinal' in SU.formatNumber", "0.14.6", "0.16.0")	×
169	options = { style = "ordinal" }	×
170	elseif options:lower() == "string" then	×
171	options = { style = "string" }	×
172	elseif options:lower() == "ordinal" and SILE.settings:get("document.language") == "tr" then	×
173	SU.deprecated(	×
174	"Format 'ordinal' in Turkish in SU.formatNumber",
175	"'ordinal-string' in SU.formatNumber",
176	"0.14.6",
177	"0.16.0"
178	)
179	options = { style = "ordinal-string" }	×
180	else
181	options = { system = options }	×
182	end
183	end
184	-- END COMPATIBILITY SHIM
185
186	if options.system == "arabic" then	108✔
187	-- "arabic" is the weirdly name, but quite friendly, used e.g. in counters and
188	-- in several other places, let's keep it as a compatibility alias.
189	options.system = "latn"	108✔
190	end
191
192	local system = options.system	108✔
193	if not case then	108✔
194	if system then	108✔
195	if system:match("^%l") then	108✔
196	case = "lower"	108✔
UNCOV 197	elseif system:match("^.%l") then	×
198	case = "title"	×
199	else
UNCOV 200	case = "upper"	×
201	end
202	else
203	case = "lower"	×
204	end
205	end
206	system = system and system:lower()	216✔
207
208	local lang = system and system == "roman" and "la" or SILE.settings:get("document.language")	108✔
209	local style = options.style	108✔
210	local result
211	if self[lang] and style and type(self[lang][style]) == "function" then	108✔
212	-- Language specific hooks exists, use them...
213	result = self[lang][style](num, options)	×
214	elseif style and type(self["und"][style]) == "function" then	108✔
215	-- Global specific hooks exists: use them...
216	result = self.und[system](num, options)	×
217	elseif system and type(self["und"][system]) == "function" then	108✔
218	-- TRICK: Notably, special case for "alpha" and "greek"
UNCOV 219	result = system and self.und[system](num, options)	×
220	else
221	--- Otherwise, rely on ICU...
222	result = icuFormat(num, lang, options)	216✔
223	end
224	return icu.case(result, lang, case)	108✔
225	end,
226	})
227
228	return formatNumber	97✔

sile-typesetter / sile / 12313034533

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