9322841962

Committed 31 May 2024 06:41PM UTC coverage: 47.763% (-8.9%) from 56.674%

Build # 9322841962

Build Type

push

github

Committed by

web-flow

Commit Message

Merge 71aa6f2a1 into bb89a7e79

Run Details

8081 of 16919 relevant lines covered (47.76%)

246.83 hits per line

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

74.31

/languages/unicode.lua

local icu = require("justenoughicu")

local chardata = require("char-def")

SILE.settings:declare({
   parameter = "languages.fixedNbsp",
   type = "boolean",
   default = false,
   help = "Whether to treat U+00A0 (NO-BREAK SPACE) as a fixed-width space",
})

SILE.nodeMakers.base = pl.class({

   _init = function (self, options)
      self.contents = {}
      self.options = options
      self.token = ""
      self.lastnode = false
      self.lasttype = false
   end,

   makeToken = function (self)
      if #self.contents > 0 then
         coroutine.yield(SILE.shaper:formNnode(self.contents, self.token, self.options))
         SU.debug("tokenizer", "Token:", self.token)
         self.contents = {}
         self.token = ""
         self.lastnode = "nnode"
      end
   end,

   addToken = function (self, char, item)
      self.token = self.token .. char
      table.insert(self.contents, item)
   end,

   makeGlue = function (self, item)
      if SILE.settings:get("typesetter.obeyspaces") or self.lastnode ~= "glue" then
         SU.debug("tokenizer", "Space node")
         coroutine.yield(SILE.shaper:makeSpaceNode(self.options, item))
      end
      self.lastnode = "glue"
      self.lasttype = "sp"
   end,

   makePenalty = function (self, p)
      if self.lastnode ~= "penalty" and self.lastnode ~= "glue" then
         coroutine.yield(SILE.types.node.penalty({ penalty = p or 0 }))
      end
      self.lastnode = "penalty"
   end,

   makeNonBreakingSpace = function (self)
      -- Unicode Line Breaking Algorithm (UAX 14) specifies that U+00A0
      -- (NO-BREAK SPACE) is expanded or compressed like a normal space.
      coroutine.yield(SILE.types.node.kern(SILE.shaper:measureSpace(self.options)))
      self.lastnode = "glue"
      self.lasttype = "sp"
   end,

   iterator = function (_, _)
      SU.error("Abstract function nodemaker:iterator called", true)
   end,

   charData = function (_, char)
      local cp = SU.codepoint(char)
      if not chardata[cp] then
         return {}
      end
      return chardata[cp]
   end,

   isActiveNonBreakingSpace = function (self, char)
      return self:isNonBreakingSpace(char) and not SILE.settings:get("languages.fixedNbsp")
   end,

   isBreaking = function (self, char)
      return self.breakingTypes[self:charData(char).linebreak]
   end,

   isNonBreakingSpace = function (self, char)
      local c = self:charData(char)
      return c.contextname and c.contextname == "nobreakspace"
   end,

   isPunctuation = function (self, char)
      return self.puctuationTypes[self:charData(char).category]
   end,

   isSpace = function (self, char)
      return self.spaceTypes[self:charData(char).linebreak]
   end,

   isQuote = function (self, char)
      return self.quoteTypes[self:charData(char).linebreak]
   end,

   isWord = function (self, char)
      return self.wordTypes[self:charData(char).linebreak]
   end,
})

SILE.nodeMakers.unicode = pl.class(SILE.nodeMakers.base)

SILE.nodeMakers.unicode.breakingTypes = { ba = true, zw = true }
SILE.nodeMakers.unicode.puctuationTypes = { po = true }
SILE.nodeMakers.unicode.quoteTypes = {} -- quote linebreak category is ambiguous depending on the language
SILE.nodeMakers.unicode.spaceTypes = { sp = true }
SILE.nodeMakers.unicode.wordTypes = { cm = true }

function SILE.nodeMakers.unicode:dealWith (item)
   local char = item.text
   local cp = SU.codepoint(char)
   local thistype = chardata[cp] and chardata[cp].linebreak
   if self:isSpace(item.text) then
      self:makeToken()
      self:makeGlue(item)
   elseif self:isActiveNonBreakingSpace(item.text) then
      self:makeToken()
      self:makeNonBreakingSpace()
   elseif self:isBreaking(item.text) then
      self:addToken(char, item)
      self:makeToken()
      self:makePenalty(0)
   elseif self:isQuote(item.text) then
      self:addToken(char, item)
      self:makeToken()
   elseif self.lasttype and (thistype and thistype ~= self.lasttype and not self:isWord(thistype)) then
      self:addToken(char, item)
   else
      self:letterspace()
      self:addToken(char, item)
   end
   self.lasttype = thistype
end

function SILE.nodeMakers.unicode:handleInitialGlue (items)
   local i = 1
   while i <= #items do
      local item = items[i]
      if self:isSpace(item.text) then
         self:makeGlue(item)
      else
         break
      end
      i = i + 1
   end
   return i, items
end

function SILE.nodeMakers.unicode:letterspace ()
   if not SILE.settings:get("document.letterspaceglue") then
      return
   end
   if self.token then
      self:makeToken()
   end
   if self.lastnode and self.lastnode ~= "glue" then
      local w = SILE.settings:get("document.letterspaceglue").width
      SU.debug("tokenizer", "Letter space glue:", w)
      coroutine.yield(SILE.types.node.kern({ width = w }))
      self.lastnode = "glue"
      self.lasttype = "sp"
   end
end

function SILE.nodeMakers.unicode.isICUBreakHere (_, chunks, item)
   return chunks[1] and (item.index >= chunks[1].index)
end

function SILE.nodeMakers.unicode:handleICUBreak (chunks, item)
   -- The ICU library has told us there is a breakpoint at
   -- this index. We need to...
   local bp = chunks[1]
   -- ... remove this breakpoint (and any out of order ones)
   -- from the ICU breakpoints array so that chunks[1] is
   -- the next index point for comparison against the string...
   while chunks[1] and item.index >= chunks[1].index do
      table.remove(chunks, 1)
   end
   -- ...decide which kind of breakpoint we have here and
   -- handle it appropriately.
   if bp.type == "word" then
      self:handleWordBreak(item)
   elseif bp.type == "line" then
      self:handleLineBreak(item, bp.subtype)
   end
   return chunks
end

function SILE.nodeMakers.unicode:handleWordBreak (item)
   self:makeToken()
   if self:isSpace(item.text) then
      -- Spacing word break
      self:makeGlue(item)
   elseif self:isActiveNonBreakingSpace(item.text) then
      -- Non-breaking space word break
      self:makeNonBreakingSpace()
   else
      -- a word break which isn't a space
      self:addToken(item.text, item)
   end
end

function SILE.nodeMakers.unicode:_handleWordBreakRepeatHyphen (item)
   -- According to some language rules, when a break occurs at an explicit hyphen,
   -- the hyphen gets repeated at the beginning of the new line
   if item.text == "-" then
      self:addToken(item.text, item)
      self:makeToken()
      if self.lastnode ~= "discretionary" then
         coroutine.yield(SILE.nodefactory.discretionary({
            postbreak = SILE.shaper:createNnodes("-", self.options),
         }))
         self.lastnode = "discretionary"
      end
   else
      SILE.nodeMakers.unicode.handleWordBreak(self, item)
   end
end

function SILE.nodeMakers.unicode:handleLineBreak (item, subtype)
   -- Because we are in charge of paragraphing, we
   -- will override space-type line breaks, and treat
   -- them just as ordinary word spaces.
   if self:isSpace(item.text) or self:isActiveNonBreakingSpace(item.text) then
      self:handleWordBreak(item)
      return
   end
   -- But explicit line breaks we will turn into
   -- soft and hard breaks.
   self:makeToken()
   self:makePenalty(subtype == "soft" and 0 or -1000)
   local char = item.text
   self:addToken(char, item)
   local cp = SU.codepoint(char)
   self.lasttype = chardata[cp] and chardata[cp].linebreak
end

function SILE.nodeMakers.unicode:_handleLineBreakRepeatHyphen (item, subtype)
   if self.lastnode == "discretionary" then
      -- Initial word boundary after a discretionary:
      -- Bypass it and just deal with the token.
      self:dealWith(item)
   else
      SILE.nodeMakers.unicode.handleLineBreak(self, item, subtype)
   end
end

function SILE.nodeMakers.unicode:iterator (items)
   local fulltext = ""
   for i = 1, #items do
      fulltext = fulltext .. items[i].text
   end
   local chunks = { icu.breakpoints(fulltext, self.options.language) }
   table.remove(chunks, 1)
   return coroutine.wrap(function ()
      local i
      i, self.items = self:handleInitialGlue(items)
      for j = i, #items do
         self.i = j
         self.item = self.items[self.i]
         if self:isICUBreakHere(chunks, self.item) then
            chunks = self:handleICUBreak(chunks, self.item)
         else
            self:dealWith(self.item)
         end
      end
      self:makeToken()
   end)
end

1	local icu = require("justenoughicu")	7✔
2
3	local chardata = require("char-def")	7✔
4
5	SILE.settings:declare({	7✔
6	parameter = "languages.fixedNbsp",
7	type = "boolean",
8	default = false,
9	help = "Whether to treat U+00A0 (NO-BREAK SPACE) as a fixed-width space",
10	})
11
12	SILE.nodeMakers.base = pl.class({	14✔
13
14	_init = function (self, options)
15	self.contents = {}	131✔
16	self.options = options	131✔
17	self.token = ""	131✔
18	self.lastnode = false	131✔
19	self.lasttype = false	131✔
20	end,
21
22	makeToken = function (self)
23	if #self.contents > 0 then	569✔
24	coroutine.yield(SILE.shaper:formNnode(self.contents, self.token, self.options))	708✔
25	SU.debug("tokenizer", "Token:", self.token)	354✔
26	self.contents = {}	354✔
27	self.token = ""	354✔
28	self.lastnode = "nnode"	354✔
29	end
30	end,
31
32	addToken = function (self, char, item)
33	self.token = self.token .. char	1,321✔
34	table.insert(self.contents, item)	1,321✔
35	end,
36
37	makeGlue = function (self, item)
38	if SILE.settings:get("typesetter.obeyspaces") or self.lastnode ~= "glue" then	438✔
39	SU.debug("tokenizer", "Space node")	215✔
40	coroutine.yield(SILE.shaper:makeSpaceNode(self.options, item))	430✔
41	end
42	self.lastnode = "glue"	219✔
43	self.lasttype = "sp"	219✔
44	end,
45
46	makePenalty = function (self, p)
47	if self.lastnode ~= "penalty" and self.lastnode ~= "glue" then	196✔
48	coroutine.yield(SILE.types.node.penalty({ penalty = p or 0 }))	×
49	end
50	self.lastnode = "penalty"	196✔
51	end,
52
53	makeNonBreakingSpace = function (self)
54	-- Unicode Line Breaking Algorithm (UAX 14) specifies that U+00A0
55	-- (NO-BREAK SPACE) is expanded or compressed like a normal space.
56	coroutine.yield(SILE.types.node.kern(SILE.shaper:measureSpace(self.options)))	×
57	self.lastnode = "glue"	×
58	self.lasttype = "sp"	×
59	end,
60
61	iterator = function (_, _)
62	SU.error("Abstract function nodemaker:iterator called", true)	×
63	end,
64
65	charData = function (_, char)
66	local cp = SU.codepoint(char)	5,169✔
67	if not chardata[cp] then	5,169✔
68	return {}	×
69	end
70	return chardata[cp]	5,169✔
71	end,
72
73	isActiveNonBreakingSpace = function (self, char)
74	return self:isNonBreakingSpace(char) and not SILE.settings:get("languages.fixedNbsp")	2,642✔
75	end,
76
77	isBreaking = function (self, char)
78	return self.breakingTypes[self:charData(char).linebreak]	2,182✔
79	end,
80
81	isNonBreakingSpace = function (self, char)
82	local c = self:charData(char)	1,321✔
83	return c.contextname and c.contextname == "nobreakspace"	1,321✔
84	end,
85
86	isPunctuation = function (self, char)
87	return self.puctuationTypes[self:charData(char).category]	×
88	end,
89
90	isSpace = function (self, char)
91	return self.spaceTypes[self:charData(char).linebreak]	3,328✔
92	end,
93
94	isQuote = function (self, char)
95	return self.quoteTypes[self:charData(char).linebreak]	2,182✔
96	end,
97
98	isWord = function (self, char)
99	return self.wordTypes[self:charData(char).linebreak]	4✔
100	end,
101	})	7✔
102
103	SILE.nodeMakers.unicode = pl.class(SILE.nodeMakers.base)	14✔
104
105	SILE.nodeMakers.unicode.breakingTypes = { ba = true, zw = true }	7✔
106	SILE.nodeMakers.unicode.puctuationTypes = { po = true }	7✔
107	SILE.nodeMakers.unicode.quoteTypes = {} -- quote linebreak category is ambiguous depending on the language	7✔
108	SILE.nodeMakers.unicode.spaceTypes = { sp = true }	7✔
109	SILE.nodeMakers.unicode.wordTypes = { cm = true }	7✔
110
111	function SILE.nodeMakers.unicode:dealWith (item)	14✔
112	local char = item.text	1,091✔
113	local cp = SU.codepoint(char)	1,091✔
114	local thistype = chardata[cp] and chardata[cp].linebreak	1,091✔
115	if self:isSpace(item.text) then	2,182✔
116	self:makeToken()	×
117	self:makeGlue(item)	×
118	elseif self:isActiveNonBreakingSpace(item.text) then	2,182✔
119	self:makeToken()	×
120	self:makeNonBreakingSpace()	×
121	elseif self:isBreaking(item.text) then	2,182✔
122	self:addToken(char, item)	×
123	self:makeToken()	×
124	self:makePenalty(0)	×
125	elseif self:isQuote(item.text) then	2,182✔
126	self:addToken(char, item)	×
127	self:makeToken()	×
128	elseif self.lasttype and (thistype and thistype ~= self.lasttype and not self:isWord(thistype)) then	1,093✔
129	self:addToken(char, item)	4✔
130	else
131	self:letterspace()	1,089✔
132	self:addToken(char, item)	1,089✔
133	end
134	self.lasttype = thistype	1,091✔
135	end
136
137	function SILE.nodeMakers.unicode:handleInitialGlue (items)	14✔
138	local i = 1	131✔
139	while i <= #items do	142✔
140	local item = items[i]	135✔
141	if self:isSpace(item.text) then	270✔
142	self:makeGlue(item)	22✔
143	else
144	break
145	end
146	i = i + 1	11✔
147	end
148	return i, items	131✔
149	end
150
151	function SILE.nodeMakers.unicode:letterspace ()	14✔
152	if not SILE.settings:get("document.letterspaceglue") then	2,178✔
153	return	1,089✔
154	end
155	if self.token then	×
156	self:makeToken()	×
157	end
158	if self.lastnode and self.lastnode ~= "glue" then	×
159	local w = SILE.settings:get("document.letterspaceglue").width	×
160	SU.debug("tokenizer", "Letter space glue:", w)	×
161	coroutine.yield(SILE.types.node.kern({ width = w }))	×
162	self.lastnode = "glue"	×
163	self.lasttype = "sp"	×
164	end
165	end
166
167	function SILE.nodeMakers.unicode.isICUBreakHere (_, chunks, item)	14✔
168	return chunks[1] and (item.index >= chunks[1].index)	1,529✔
169	end
170
171	function SILE.nodeMakers.unicode:handleICUBreak (chunks, item)	14✔
172	-- The ICU library has told us there is a breakpoint at
173	-- this index. We need to...
174	local bp = chunks[1]	438✔
175	-- ... remove this breakpoint (and any out of order ones)
176	-- from the ICU breakpoints array so that chunks[1] is
177	-- the next index point for comparison against the string...
178	while chunks[1] and item.index >= chunks[1].index do	876✔
179	table.remove(chunks, 1)	876✔
180	end
181	-- ...decide which kind of breakpoint we have here and
182	-- handle it appropriately.
183	if bp.type == "word" then	438✔
184	self:handleWordBreak(item)	484✔
185	elseif bp.type == "line" then	196✔
186	self:handleLineBreak(item, bp.subtype)	196✔
187	end
188	return chunks	438✔
189	end
190
191	function SILE.nodeMakers.unicode:handleWordBreak (item)	14✔
192	self:makeToken()	242✔
193	if self:isSpace(item.text) then	484✔
194	-- Spacing word break
195	self:makeGlue(item)	416✔
196	elseif self:isActiveNonBreakingSpace(item.text) then	68✔
197	-- Non-breaking space word break
198	self:makeNonBreakingSpace()	×
199	else
200	-- a word break which isn't a space
201	self:addToken(item.text, item)	34✔
202	end
203	end
204
205	function SILE.nodeMakers.unicode:_handleWordBreakRepeatHyphen (item)	14✔
206	-- According to some language rules, when a break occurs at an explicit hyphen,
207	-- the hyphen gets repeated at the beginning of the new line
208	if item.text == "-" then	×
209	self:addToken(item.text, item)	×
210	self:makeToken()	×
211	if self.lastnode ~= "discretionary" then	×
212	coroutine.yield(SILE.nodefactory.discretionary({	×
213	postbreak = SILE.shaper:createNnodes("-", self.options),
214	}))
215	self.lastnode = "discretionary"	×
216	end
217	else
218	SILE.nodeMakers.unicode.handleWordBreak(self, item)	×
219	end
220	end
221
222	function SILE.nodeMakers.unicode:handleLineBreak (item, subtype)	14✔
223	-- Because we are in charge of paragraphing, we
224	-- will override space-type line breaks, and treat
225	-- them just as ordinary word spaces.
226	if self:isSpace(item.text) or self:isActiveNonBreakingSpace(item.text) then	588✔
227	self:handleWordBreak(item)	×
228	return	×
229	end
230	-- But explicit line breaks we will turn into
231	-- soft and hard breaks.
232	self:makeToken()	196✔
233	self:makePenalty(subtype == "soft" and 0 or -1000)	196✔
234	local char = item.text	196✔
235	self:addToken(char, item)	196✔
236	local cp = SU.codepoint(char)	196✔
237	self.lasttype = chardata[cp] and chardata[cp].linebreak	196✔
238	end
239
240	function SILE.nodeMakers.unicode:_handleLineBreakRepeatHyphen (item, subtype)	14✔
241	if self.lastnode == "discretionary" then	×
242	-- Initial word boundary after a discretionary:
243	-- Bypass it and just deal with the token.
244	self:dealWith(item)	×
245	else
246	SILE.nodeMakers.unicode.handleLineBreak(self, item, subtype)	×
247	end
248	end
249
250	function SILE.nodeMakers.unicode:iterator (items)	14✔
251	local fulltext = ""	131✔
252	for i = 1, #items do	1,671✔
253	fulltext = fulltext .. items[i].text	1,540✔
254	end
255	local chunks = { icu.breakpoints(fulltext, self.options.language) }	131✔
256	table.remove(chunks, 1)	131✔
257	return coroutine.wrap(function ()	131✔
258	local i
259	i, self.items = self:handleInitialGlue(items)	262✔
260	for j = i, #items do	1,660✔
261	self.i = j	1,529✔
262	self.item = self.items[self.i]	1,529✔
263	if self:isICUBreakHere(chunks, self.item) then	3,058✔
264	chunks = self:handleICUBreak(chunks, self.item)	876✔
265	else
266	self:dealWith(self.item)	1,091✔
267	end
268	end
269	self:makeToken()	131✔
270	end)
271	end

sile-typesetter / sile / 9322841962

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