9307175333

Committed 30 May 2024 06:08PM UTC coverage: 70.644% (-3.5%) from 74.124%

Build # 9307175333

Build Type

push

github

Committed by

web-flow

Commit Message

Merge b18390e74 into 70ff5c335

Run Details

1910 of 2592 new or added lines in 108 files covered. (73.69%)

901 existing lines in 52 files now uncovered.

12203 of 17274 relevant lines covered (70.64%)

6112.16 hits per line

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

0.0

/languages/ug.lua

-- Thanks to Tom Milo of Decotype for challenging me to implement
-- Uyghur support, providing the methodology and basic algorithms,
-- and preventing me from leaving the Granshan 2015 conference until
-- I had most of this working.

-- Uyghur is Turkish, right?
SILE.languageSupport.loadLanguage("tr")

local chardata = require("char-def")

SILE.settings:declare({
   parameter = "languages.ug.hyphenoffset",
   help = "Space added between text and hyphen",
   type = "glue or nil",
   default = SILE.types.node.glue("1pt"),
})

local transliteration = {
   -- I'm going to pretend that normalisation isn't a problem
   { al = "ئا", la = "a", lapa = "^a" },
   { al = "ا", la = "a" },
   { al = "ب", la = "b" },
   { al = "پ", la = "p" },
   { al = "ت", la = "t" },
   { al = "ج", la = "c" },
   { al = "چ", la = "ç" },
   { al = "د", la = "d" },
   { al = "ر", la = "r" },
   { al = "ژ", la = "j" },
   { al = "ز", la = "z" },
   { al = "ش", la = "ş" },
   { al = "س", la = "s" },
   { al = "غ", la = "ğ" },
   { al = "ف", la = "f" },
   { al = "ق", la = "q" },
   { al = "ك", la = "k" },
   { al = "گ", la = "g" },
   { al = "ڭ", la = "ġ" },
   { al = "ل", la = "l" },
   { al = "م", la = "m" },
   { al = "ن", la = "n" },
   { al = "ھ", la = "h" },
   { al = "ۋ", la = "w" },
   { al = "ئې", la = "e", lapa = "^e" },
   { al = "ې", la = "e" },
   { al = "ئە", la = "ä", lapa = "^ä" },
   { al = "ە", la = "ä" },
   { al = "ئى", la = "i", lapa = "^i" },
   { al = "ى", la = "i" },
   { al = "ي", la = "y" },
   { al = "ئو", la = "o", lapa = "^o" },
   { al = "و", la = "o" },
   { al = "ئۇ", la = "u", lapa = "^u" },
   { al = "ۇ", la = "u" },
   { al = "ئۆ", la = "ö", lapa = "^ö" },
   { al = "ۆ", la = "ö" },
   { al = "ئۈ", la = "ü", lapa = "^ü" },
   { al = "ۈ", la = "ü" },
   { al = "خ", la = "x" },
}

local arabicToLatin = function (s)
   for i = 1, #transliteration do
      s = s:gsub(transliteration[i].al, transliteration[i].la)
   end
   return s
end

local latinToArabic = function (s, useLapa)
   for i = 1, #transliteration do
      if useLapa then
         s = s:lower():gsub(transliteration[i].lapa or transliteration[i].la, transliteration[i].al)
      elseif not transliteration[i].lapa then
         s = s:lower():gsub(transliteration[i].la, transliteration[i].al)
      end
   end
   return s
end

local zwj = SU.utf8charfromcodepoint("U+200D")
-- local zwnj = SU.utf8charfromcodepoint("U+200C")

-- local dropLast = function(t)
--   local bt = SU.splitUtf8(t)
--   local n = ""
--   for i = 1,#bt-1 do n = n..bt[i] end
--   return n
-- end

-- local dropFirst = function(t)
--   local bt = SU.splitUtf8(t)
--   local n = ""
--   for i = 2,#bt do n = n..bt[i] end
--   return n
-- end

local lastjoinable = function (t)
   t = SU.splitUtf8(t)
   local last = t[#t]
   local jointype = chardata[SU.codepoint(last)] and chardata[SU.codepoint(last)].arabic
   local joinable = jointype == "d" or jointype == "l"
   return joinable
end

-- local firstjoinable = function (t)
--   local t = SU.splitUtf8(t)
--   local first = t[1]
--   local jointype = chardata[SU.codepoint(first)] and chardata[SU.codepoint(first)].arabic
--   local joinable = jointype == "d" or jointype=="r"
--   return joinable
-- end

-- function debugUyghur(word)
--   SILE.languageSupport.loadLanguage("ug")
--   print(SILE.showHyphenationPoints(word,"ug"))
--   local items = SILE._hyphenate(SILE.hyphenators["ug"],word)
--   print(reorderHyphenations(items,true))
-- end

SILE.hyphenator.languages.ug = function (n)
   local latin = arabicToLatin(n.text)
   SU.debug("uyghur", "Original:", n.text, "->", latin, "->")
   local state = n.options
   -- Make "Turkish" nodes
   local newoptions = pl.tablex.deepcopy(n.options)
   newoptions.language = "lt"
   if not SILE.hyphenators.lt then
      SILE.hyphenate(SILE.shaper:createNnodes(latin, newoptions))
   end
   local items = SILE._hyphenate(SILE.hyphenators["lt"], latin)
   if #items == 1 then
      SU.debug("uyghur", latin, "No hyphenation points")
      return { n }
   end
   -- Choose 1. Aim to split in middle.
   SU.debug("uyghur", function ()
      return SU.concat(items, "/") .. " -> "
   end)
   local splitpoint = math.ceil(#items / 2)
   local nitems = { "", "" }
   for i = 1, #items do
      if i <= splitpoint then
         nitems[1] = nitems[1] .. items[i]
      else
         nitems[2] = nitems[2] .. items[i]
      end
   end
   items = nitems
   SU.debug("uyghur", items[1], "/", items[2])
   state.language = "ug"
   items[1] = latinToArabic(items[1])
   items[2] = latinToArabic(items[2])
   local hyphen = SILE.settings:get("font.hyphenchar")
   local prebreak = SILE.shaper:createNnodes(items[1] .. (lastjoinable(items[1]) and zwj or ""), state)
   if SILE.settings:get("languages.ug.hyphenoffset") then
      local w = SILE.settings:get("languages.ug.hyphenoffset").width
      prebreak[#prebreak + 1] = SILE.types.node.kern({ width = w })
   end
   local hnodes = SILE.shaper:createNnodes(hyphen, state)
   prebreak[#prebreak + 1] = hnodes[1]
   local postbreak = SILE.shaper:createNnodes((lastjoinable(items[1]) and zwj or "") .. items[2], state)
   local d = SILE.types.node.discretionary({
      replacement = { n },
      prebreak = prebreak,
      postbreak = postbreak,
   })
   return { SILE.types.node.zerohbox(), d }
end

1	-- Thanks to Tom Milo of Decotype for challenging me to implement
2	-- Uyghur support, providing the methodology and basic algorithms,
3	-- and preventing me from leaving the Granshan 2015 conference until
4	-- I had most of this working.
5
6	-- Uyghur is Turkish, right?
7	SILE.languageSupport.loadLanguage("tr")	×
8
9	local chardata = require("char-def")	×
10
11	SILE.settings:declare({	×
12	parameter = "languages.ug.hyphenoffset",
13	help = "Space added between text and hyphen",
14	type = "glue or nil",
15	default = SILE.types.node.glue("1pt"),
16	})
17
18	local transliteration = {	×
19	-- I'm going to pretend that normalisation isn't a problem
20	{ al = "ئا", la = "a", lapa = "^a" },	×
21	{ al = "ا", la = "a" },	×
22	{ al = "ب", la = "b" },	×
23	{ al = "پ", la = "p" },	×
24	{ al = "ت", la = "t" },	×
25	{ al = "ج", la = "c" },	×
26	{ al = "چ", la = "ç" },	×
27	{ al = "د", la = "d" },	×
28	{ al = "ر", la = "r" },	×
29	{ al = "ژ", la = "j" },	×
30	{ al = "ز", la = "z" },	×
31	{ al = "ش", la = "ş" },	×
32	{ al = "س", la = "s" },	×
33	{ al = "غ", la = "ğ" },	×
34	{ al = "ف", la = "f" },	×
35	{ al = "ق", la = "q" },	×
36	{ al = "ك", la = "k" },	×
37	{ al = "گ", la = "g" },	×
38	{ al = "ڭ", la = "ġ" },	×
39	{ al = "ل", la = "l" },	×
40	{ al = "م", la = "m" },	×
41	{ al = "ن", la = "n" },	×
42	{ al = "ھ", la = "h" },	×
43	{ al = "ۋ", la = "w" },	×
44	{ al = "ئې", la = "e", lapa = "^e" },	×
45	{ al = "ې", la = "e" },	×
46	{ al = "ئە", la = "ä", lapa = "^ä" },	×
47	{ al = "ە", la = "ä" },	×
48	{ al = "ئى", la = "i", lapa = "^i" },	×
49	{ al = "ى", la = "i" },	×
50	{ al = "ي", la = "y" },	×
51	{ al = "ئو", la = "o", lapa = "^o" },	×
52	{ al = "و", la = "o" },	×
53	{ al = "ئۇ", la = "u", lapa = "^u" },	×
54	{ al = "ۇ", la = "u" },	×
55	{ al = "ئۆ", la = "ö", lapa = "^ö" },	×
56	{ al = "ۆ", la = "ö" },	×
57	{ al = "ئۈ", la = "ü", lapa = "^ü" },	×
58	{ al = "ۈ", la = "ü" },	×
59	{ al = "خ", la = "x" },	×
60	}
61
62	local arabicToLatin = function (s)
63	for i = 1, #transliteration do	×
64	s = s:gsub(transliteration[i].al, transliteration[i].la)	×
65	end
66	return s	×
67	end
68
69	local latinToArabic = function (s, useLapa)
70	for i = 1, #transliteration do	×
71	if useLapa then	×
72	s = s:lower():gsub(transliteration[i].lapa or transliteration[i].la, transliteration[i].al)	×
73	elseif not transliteration[i].lapa then	×
74	s = s:lower():gsub(transliteration[i].la, transliteration[i].al)	×
75	end
76	end
77	return s	×
78	end
79
80	local zwj = SU.utf8charfromcodepoint("U+200D")	×
81	-- local zwnj = SU.utf8charfromcodepoint("U+200C")
82
83	-- local dropLast = function(t)
84	-- local bt = SU.splitUtf8(t)
85	-- local n = ""
86	-- for i = 1,#bt-1 do n = n..bt[i] end
87	-- return n
88	-- end
89
90	-- local dropFirst = function(t)
91	-- local bt = SU.splitUtf8(t)
92	-- local n = ""
93	-- for i = 2,#bt do n = n..bt[i] end
94	-- return n
95	-- end
96
97	local lastjoinable = function (t)
98	t = SU.splitUtf8(t)	×
99	local last = t[#t]	×
100	local jointype = chardata[SU.codepoint(last)] and chardata[SU.codepoint(last)].arabic	×
101	local joinable = jointype == "d" or jointype == "l"	×
102	return joinable	×
103	end
104
105	-- local firstjoinable = function (t)
106	-- local t = SU.splitUtf8(t)
107	-- local first = t[1]
108	-- local jointype = chardata[SU.codepoint(first)] and chardata[SU.codepoint(first)].arabic
109	-- local joinable = jointype == "d" or jointype=="r"
110	-- return joinable
111	-- end
112
113	-- function debugUyghur(word)
114	-- SILE.languageSupport.loadLanguage("ug")
115	-- print(SILE.showHyphenationPoints(word,"ug"))
116	-- local items = SILE._hyphenate(SILE.hyphenators["ug"],word)
117	-- print(reorderHyphenations(items,true))
118	-- end
119
120	SILE.hyphenator.languages.ug = function (n)	×
121	local latin = arabicToLatin(n.text)	×
122	SU.debug("uyghur", "Original:", n.text, "->", latin, "->")	×
123	local state = n.options	×
124	-- Make "Turkish" nodes
125	local newoptions = pl.tablex.deepcopy(n.options)	×
126	newoptions.language = "lt"	×
127	if not SILE.hyphenators.lt then	×
128	SILE.hyphenate(SILE.shaper:createNnodes(latin, newoptions))	×
129	end
130	local items = SILE._hyphenate(SILE.hyphenators["lt"], latin)	×
131	if #items == 1 then	×
132	SU.debug("uyghur", latin, "No hyphenation points")	×
133	return { n }	×
134	end
135	-- Choose 1. Aim to split in middle.
136	SU.debug("uyghur", function ()	×
137	return SU.concat(items, "/") .. " -> "	×
138	end)
139	local splitpoint = math.ceil(#items / 2)	×
140	local nitems = { "", "" }	×
141	for i = 1, #items do	×
142	if i <= splitpoint then	×
143	nitems[1] = nitems[1] .. items[i]	×
144	else
145	nitems[2] = nitems[2] .. items[i]	×
146	end
147	end
148	items = nitems	×
149	SU.debug("uyghur", items[1], "/", items[2])	×
150	state.language = "ug"	×
151	items[1] = latinToArabic(items[1])	×
152	items[2] = latinToArabic(items[2])	×
153	local hyphen = SILE.settings:get("font.hyphenchar")	×
154	local prebreak = SILE.shaper:createNnodes(items[1] .. (lastjoinable(items[1]) and zwj or ""), state)	×
155	if SILE.settings:get("languages.ug.hyphenoffset") then	×
156	local w = SILE.settings:get("languages.ug.hyphenoffset").width	×
NEW 157	prebreak[#prebreak + 1] = SILE.types.node.kern({ width = w })	×
158	end
159	local hnodes = SILE.shaper:createNnodes(hyphen, state)	×
160	prebreak[#prebreak + 1] = hnodes[1]	×
161	local postbreak = SILE.shaper:createNnodes((lastjoinable(items[1]) and zwj or "") .. items[2], state)	×
NEW 162	local d = SILE.types.node.discretionary({	×
163	replacement = { n },
164	prebreak = prebreak,
165	postbreak = postbreak,
166	})
NEW 167	return { SILE.types.node.zerohbox(), d }	×
168	end

sile-typesetter / sile / 9307175333

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