7232859119

Committed 16 Dec 2023 03:49PM UTC coverage: 66.878% (-7.7%) from 74.62%

Build # 7232859119

Build Type

push

github

Committed by

web-flow

Commit Message

Merge 05d75c2a3 into 8686730e4

Run Details

0 of 4 new or added lines in 1 file covered. (0.0%)

1201 existing lines in 56 files now uncovered.

10550 of 15775 relevant lines covered (66.88%)

3347.52 hits per line

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

92.97

/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.nodefactory.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.nodefactory.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,

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

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

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

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

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

  })

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

SILE.nodeMakers.unicode.isWordType = { cm = true }
SILE.nodeMakers.unicode.isSpaceType = { sp = true }
SILE.nodeMakers.unicode.isBreakingType = { ba = true, zw = true }
SILE.nodeMakers.unicode.isPunctuationType = { po = true }
SILE.nodeMakers.unicode.isQuoteType = {} -- quote linebreak category is ambiguous depending on the language

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.isWordType[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.nodefactory.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: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: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")	71✔
2
3	local chardata = require("char-def")	71✔
4
5	SILE.settings:declare({	71✔
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({	142✔
13
14	_init = function (self, options)
15	self.contents = {}	3,983✔
16	self.options = options	3,983✔
17	self.token = ""	3,983✔
18	self.lastnode = false	3,983✔
19	self.lasttype = false	3,983✔
20	end,
21
22	makeToken = function (self)
23	if #self.contents > 0 then	10,663✔
24	coroutine.yield(SILE.shaper:formNnode(self.contents, self.token, self.options))	14,854✔
25	SU.debug("tokenizer", "Token:", self.token)	7,427✔
26	self.contents = {}	7,427✔
27	self.token = ""	7,427✔
28	self.lastnode = "nnode"	7,427✔
29	end
30	end,
31
32	addToken = function (self, char, item)
33	self.token = self.token .. char	24,832✔
34	table.insert(self.contents, item)	24,832✔
35	end,
36
37	makeGlue = function (self, item)
38	if SILE.settings:get("typesetter.obeyspaces") or self.lastnode ~= "glue" then	6,414✔
39	SU.debug("tokenizer", "Space node")	3,162✔
40	coroutine.yield(SILE.shaper:makeSpaceNode(self.options, item))	6,324✔
41	end
42	self.lastnode = "glue"	3,207✔
43	self.lasttype = "sp"	3,207✔
44	end,
45
46	makePenalty = function (self, p)
47	if self.lastnode ~= "penalty" and self.lastnode ~= "glue" then	3,053✔
48	coroutine.yield( SILE.nodefactory.penalty({ penalty = p or 0 }) )	78✔
49	end
50	self.lastnode = "penalty"	3,053✔
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.nodefactory.kern(SILE.shaper:measureSpace(self.options)))	87✔
57	self.lastnode = "glue"	29✔
58	self.lasttype = "sp"	29✔
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)	99,726✔
67	if not chardata[cp] then return {} end	99,726✔
68	return chardata[cp]	98,874✔
69	end,
70
71	isPunctuation = function (self, char)
72	return self.isPunctuationType[self:charData(char).category]	×
73	end,
74
75	isSpace = function (self, char)
76	return self.isSpaceType[self:charData(char).linebreak]	64,144✔
77	end,
78
79	isNonBreakingSpace = function (self, char)
80	local c = self:charData(char)	24,890✔
81	return c.contextname and c.contextname == "nobreakspace"	24,890✔
82	end,
83
84	isActiveNonBreakingSpace = function (self, char)
85	return self:isNonBreakingSpace(char) and not SILE.settings:get("languages.fixedNbsp")	49,741✔
86	end,
87
88	isBreaking = function (self, char)
89	return self.isBreakingType[self:charData(char).linebreak]	42,776✔
90	end,
91	isQuote = function (self, char)
92	return self.isQuoteType[self:charData(char).linebreak]	42,752✔
93	end
94
95	})	71✔
96
97	SILE.nodeMakers.unicode = pl.class(SILE.nodeMakers.base)	142✔
98
99	SILE.nodeMakers.unicode.isWordType = { cm = true }	71✔
100	SILE.nodeMakers.unicode.isSpaceType = { sp = true }	71✔
101	SILE.nodeMakers.unicode.isBreakingType = { ba = true, zw = true }	71✔
102	SILE.nodeMakers.unicode.isPunctuationType = { po = true }	71✔
103	SILE.nodeMakers.unicode.isQuoteType = {} -- quote linebreak category is ambiguous depending on the language	71✔
104
105	function SILE.nodeMakers.unicode:dealWith (item)	142✔
106	local char = item.text	21,416✔
107	local cp = SU.codepoint(char)	21,416✔
108	local thistype = chardata[cp] and chardata[cp].linebreak	21,416✔
109	if self:isSpace(item.text) then	42,832✔
110	self:makeToken()	28✔
111	self:makeGlue(item)	56✔
112	elseif self:isActiveNonBreakingSpace(item.text) then	42,776✔
113	self:makeToken()	×
114	self:makeNonBreakingSpace()	×
115	elseif self:isBreaking(item.text) then	42,776✔
116	self:addToken(char, item)	12✔
117	self:makeToken()	12✔
118	self:makePenalty(0)	24✔
119	elseif self:isQuote(item.text) then	42,752✔
UNCOV 120	self:addToken(char, item)	×
UNCOV 121	self:makeToken()	×
122	elseif self.lasttype and (thistype and thistype ~= self.lasttype and not self.isWordType[thistype]) then	21,376✔
123	self:addToken(char, item)	200✔
124	else
125	self:letterspace()	21,276✔
126	self:addToken(char, item)	21,276✔
127	end
128	self.lasttype = thistype	21,416✔
129	end
130
131	function SILE.nodeMakers.unicode:handleInitialGlue (items)	142✔
132	local i = 1	3,981✔
133	while i <= #items do	4,051✔
134	local item = items[i]	4,034✔
135	if self:isSpace(item.text) then self:makeGlue(item) else break end	8,138✔
136	i = i + 1	70✔
137	end
138	return i, items	3,981✔
139	end
140
141	function SILE.nodeMakers.unicode:letterspace ()	142✔
142	if not SILE.settings:get("document.letterspaceglue") then return end	42,552✔
143	if self.token then self:makeToken() end	55✔
144	if self.lastnode and self.lastnode ~= "glue" then	55✔
145	local w = SILE.settings:get("document.letterspaceglue").width	102✔
146	SU.debug("tokenizer", "Letter space glue:", w)	51✔
147	coroutine.yield(SILE.nodefactory.kern({ width = w }))	102✔
148	self.lastnode = "glue"	51✔
149	self.lasttype = "sp"	51✔
150	end
151	end
152
153	function SILE.nodeMakers.unicode.isICUBreakHere (_, chunks, item)	142✔
154	return chunks[1] and (item.index >= chunks[1].index)	27,998✔
155	end
156
157	function SILE.nodeMakers.unicode:handleICUBreak (chunks, item)	142✔
158	-- The ICU library has told us there is a breakpoint at
159	-- this index. We need to...
160	local bp = chunks[1]	6,581✔
161	-- ... remove this breakpoint (and any out of order ones)
162	-- from the ICU breakpoints array so that chunks[1] is
163	-- the next index point for comparison against the string...
164	while chunks[1] and item.index >= chunks[1].index do	13,164✔
165	table.remove(chunks, 1)	13,166✔
166	end
167	-- ...decide which kind of breakpoint we have here and
168	-- handle it appropriately.
169	if bp.type == "word" then	6,581✔
170	self:handleWordBreak(item)	7,076✔
171	elseif bp.type == "line" then	3,043✔
172	self:handleLineBreak(item, bp.subtype)	3,043✔
173	end
174	return chunks	6,581✔
175	end
176
177	function SILE.nodeMakers.unicode:handleWordBreak (item)	142✔
178	self:makeToken()	3,536✔
179	if self:isSpace(item.text) then	7,072✔
180	-- Spacing word break
181	self:makeGlue(item)	6,218✔
182	elseif self:isActiveNonBreakingSpace(item.text) then	854✔
183	-- Non-breaking space word break
184	self:makeNonBreakingSpace()	58✔
185	else
186	-- a word break which isn't a space
187	self:addToken(item.text, item)	398✔
188	end
189	end
190
191	function SILE.nodeMakers.unicode:handleLineBreak (item, subtype)	142✔
192	-- Because we are in charge of paragraphing, we
193	-- will override space-type line breaks, and treat
194	-- them just as ordinary word spaces.
195	if self:isSpace(item.text) or self:isActiveNonBreakingSpace(item.text) then	9,123✔
196	self:handleWordBreak(item)	×
197	return	×
198	end
199	-- But explicit line breaks we will turn into
200	-- soft and hard breaks.
201	self:makeToken()	3,041✔
202	self:makePenalty(subtype == "soft" and 0 or -1000)	3,041✔
203	local char = item.text	3,041✔
204	self:addToken(char, item)	3,041✔
205	local cp = SU.codepoint(char)	3,041✔
206	self.lasttype = chardata[cp] and chardata[cp].linebreak	3,041✔
207	end
208
209	function SILE.nodeMakers.unicode:iterator (items)	142✔
210	local fulltext = ""	3,981✔
211	for i = 1, #items do	32,049✔
212	fulltext = fulltext .. items[i].text	28,068✔
213	end
214	local chunks = { icu.breakpoints(fulltext, self.options.language) }	3,981✔
215	table.remove(chunks, 1)	3,981✔
216	return coroutine.wrap(function ()	3,981✔
217	local i
218	i, self.items = self:handleInitialGlue(items)	7,962✔
219	for j = i, #items do	31,979✔
220	self.i = j	27,998✔
221	self.item = self.items[self.i]	27,998✔
222	if self:isICUBreakHere(chunks, self.item) then	55,996✔
223	chunks = self:handleICUBreak(chunks, self.item)	13,162✔
224	else
225	self:dealWith(self.item)	21,417✔
226	end
227	end
228	self:makeToken()	3,981✔
229	end)
230	end

sile-typesetter / sile / 7232859119

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