diff options
Diffstat (limited to 'src/fontloader/misc')
| -rw-r--r-- | src/fontloader/misc/fontloader-l-lpeg.lua | 40 | ||||
| -rw-r--r-- | src/fontloader/misc/fontloader-util-str.lua | 28 |
2 files changed, 51 insertions, 17 deletions
diff --git a/src/fontloader/misc/fontloader-l-lpeg.lua b/src/fontloader/misc/fontloader-l-lpeg.lua index 4aadadb..55a0d89 100644 --- a/src/fontloader/misc/fontloader-l-lpeg.lua +++ b/src/fontloader/misc/fontloader-l-lpeg.lua @@ -10,6 +10,8 @@ if not modules then modules = { } end modules ['l-lpeg'] = { -- if i can use new features like capture / 2 and .B (at first sight the xml -- parser is some 5% slower) +-- lpeg.P("abc") is faster than lpeg.P("a") * lpeg.P("b") * lpeg.P("c") + -- a new lpeg fails on a #(1-P(":")) test and really needs a + P(-1) -- move utf -> l-unicode @@ -19,7 +21,7 @@ lpeg = require("lpeg") -- The latest lpeg doesn't have print any more, and even the new ones are not -- available by default (only when debug mode is enabled), which is a pitty as --- as it helps nailign down bottlenecks. Performance seems comparable: some 10% +-- as it helps nailing down bottlenecks. Performance seems comparable: some 10% -- slower pattern compilation, same parsing speed, although, -- -- local p = lpeg.C(lpeg.P(1)^0 * lpeg.P(-1)) @@ -841,7 +843,6 @@ local function make(t) local function making(t) local p = p_false local keys = sortedkeys(t) --- local okay = t[""] for i=1,#keys do local k = keys[i] if k ~= "" then @@ -850,8 +851,6 @@ local function make(t) p = p + P(k) * p_true elseif v == false then -- can't happen --- elseif okay then --- p = p + P(k) * (making(v) + p_true) else p = p + P(k) * making(v) end @@ -872,8 +871,6 @@ local function make(t) p = p + P(k) * p_true elseif v == false then -- can't happen --- elseif v[""] then --- p = p + P(k) * (making(v) + p_true) else p = p + P(k) * making(v) end @@ -882,6 +879,33 @@ local function make(t) return p end +local function collapse(t,x) + if type(t) ~= "table" then + return t, x + else + local n = next(t) + if n == nil then + return t, x + elseif next(t,n) == nil then + -- one entry + local k = n + local v = t[k] + if type(v) == "table" then + return collapse(v,x..k) + else + return v, x .. k + end + else + local tt = { } + for k, v in next, t do + local vv, kk = collapse(v,k) + tt[kk] = vv + end + return tt, x + end + end +end + function lpeg.utfchartabletopattern(list) -- goes to util-lpg local tree = { } local n = #list @@ -955,10 +979,14 @@ function lpeg.utfchartabletopattern(list) -- goes to util-lpg end end end +-- collapse(tree,"") -- needs testing, maybe optional, slightly faster because P("x")*P("X") seems slower than P"(xX") (why) -- inspect(tree) return make(tree) end +-- local t = { "start", "stoep", "staart", "paard" } +-- local p = lpeg.Cs((lpeg.utfchartabletopattern(t)/string.upper + 1)^1) + -- local t = { "a", "abc", "ac", "abe", "abxyz", "xy", "bef","aa" } -- local p = lpeg.Cs((lpeg.utfchartabletopattern(t)/string.upper + 1)^1) diff --git a/src/fontloader/misc/fontloader-util-str.lua b/src/fontloader/misc/fontloader-util-str.lua index a677a82..de4a87e 100644 --- a/src/fontloader/misc/fontloader-util-str.lua +++ b/src/fontloader/misc/fontloader-util-str.lua @@ -44,7 +44,12 @@ end if not number then number = { } end -- temp hack for luatex-fonts -local stripper = patterns.stripzeros +local stripper = patterns.stripzeros +local newline = patterns.newline +local endofstring = patterns.endofstring +local whitespace = patterns.whitespace +local spacer = patterns.spacer +local spaceortab = patterns.spaceortab local function points(n) n = tonumber(n) @@ -62,12 +67,12 @@ number.basepoints = basepoints -- str = " \n \ntest \n test\ntest " -- print("["..string.gsub(string.collapsecrlf(str),"\n","+").."]") -local rubish = patterns.spaceortab^0 * patterns.newline -local anyrubish = patterns.spaceortab + patterns.newline +local rubish = spaceortab^0 * newline +local anyrubish = spaceortab + newline local anything = patterns.anything -local stripped = (patterns.spaceortab^1 / "") * patterns.newline +local stripped = (spaceortab^1 / "") * newline local leading = rubish^0 / "" -local trailing = (anyrubish^1 * patterns.endofstring) / "" +local trailing = (anyrubish^1 * endofstring) / "" local redundant = rubish^3 / "\n" local pattern = Cs(leading * (trailing + redundant + stripped + anything)^0) @@ -129,7 +134,7 @@ local pattern = return "" end end - + patterns.newline * Cp() / function(position) + + newline * Cp() / function(position) extra, start = 0, position end + patterns.anything @@ -162,11 +167,6 @@ end -- return str -- end -local newline = patterns.newline -local endofstring = patterns.endofstring -local whitespace = patterns.whitespace -local spacer = patterns.spacer - local space = spacer^0 local nospace = space/"" local endofline = nospace * newline @@ -1117,3 +1117,9 @@ local pattern = function string.optionalquoted(str) return lpegmatch(pattern,str) or str end + +local pattern = Cs((newline / os.newline + 1)^0) + +function string.replacenewlines(str) + return lpegmatch(pattern,str) +end |