diff options
Diffstat (limited to 'lualibs-util-str.lua')
| -rw-r--r-- | lualibs-util-str.lua | 287 |
1 files changed, 207 insertions, 80 deletions
diff --git a/lualibs-util-str.lua b/lualibs-util-str.lua index f04f0e5..52c48ba 100644 --- a/lualibs-util-str.lua +++ b/lualibs-util-str.lua @@ -20,16 +20,24 @@ local utfchar, utfbyte = utf.char, utf.byte ----- loadstripped = utilities.lua.loadstripped ----- setmetatableindex = table.setmetatableindex --- local loadstripped = _LUAVERSION < 5.2 and load or function(str) --- return load(dump(load(str),true)) -- it only makes sense in luajit and luatex where we have a stipped load --- end +local loadstripped = nil -local loadstripped = function(str,shortcuts) - if shortcuts then - return load(dump(load(str),true),nil,nil,shortcuts) - else - return load(dump(load(str),true)) +if _LUAVERSION < 5.2 then + + loadstripped = function(str,shortcuts) + return load(str) + end + +else + + loadstripped = function(str,shortcuts) + if shortcuts then + return load(dump(load(str),true),nil,nil,shortcuts) + else + return load(dump(load(str),true)) + end end + end -- todo: make a special namespace for the formatter @@ -39,10 +47,12 @@ if not number then number = { } end -- temp hack for luatex-fonts local stripper = patterns.stripzeros local function points(n) + n = tonumber(n) return (not n or n == 0) and "0pt" or lpegmatch(stripper,format("%.5fpt",n/65536)) end local function basepoints(n) + n = tonumber(n) return (not n or n == 0) and "0bp" or lpegmatch(stripper,format("%.5fbp", n*(7200/7227)/65536)) end @@ -144,17 +154,105 @@ end -- print(strings.tabtospace(t[k])) -- end -function strings.striplong(str) -- strips all leading spaces - str = gsub(str,"^%s*","") - str = gsub(str,"[\n\r]+ *","\n") - return str +-- todo: lpeg + +-- function strings.striplong(str) -- strips all leading spaces +-- str = gsub(str,"^%s*","") +-- str = gsub(str,"[\n\r]+ *","\n") +-- 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 + +local stripend = (whitespace^1 * endofstring)/"" + +local normalline = (nospace * ((1-space*(newline+endofstring))^1) * nospace) + +local stripempty = endofline^1/"" +local normalempty = endofline^1 +local singleempty = endofline * (endofline^0/"") +local doubleempty = endofline * endofline^-1 * (endofline^0/"") + +local stripstart = stripempty^0 + +local p_prune_normal = Cs ( stripstart * ( stripend + normalline + normalempty )^0 ) +local p_prune_collapse = Cs ( stripstart * ( stripend + normalline + doubleempty )^0 ) +local p_prune_noempty = Cs ( stripstart * ( stripend + normalline + singleempty )^0 ) +local p_retain_normal = Cs ( ( normalline + normalempty )^0 ) +local p_retain_collapse = Cs ( ( normalline + doubleempty )^0 ) +local p_retain_noempty = Cs ( ( normalline + singleempty )^0 ) + +-- function striplines(str,prune,collapse,noempty) +-- if prune then +-- if noempty then +-- return lpegmatch(p_prune_noempty,str) or str +-- elseif collapse then +-- return lpegmatch(p_prune_collapse,str) or str +-- else +-- return lpegmatch(p_prune_normal,str) or str +-- end +-- else +-- if noempty then +-- return lpegmatch(p_retain_noempty,str) or str +-- elseif collapse then +-- return lpegmatch(p_retain_collapse,str) or str +-- else +-- return lpegmatch(p_retain_normal,str) or str +-- end +-- end +-- end + +local striplinepatterns = { + ["prune"] = p_prune_normal, + ["prune and collapse"] = p_prune_collapse, -- default + ["prune and no empty"] = p_prune_noempty, + ["retain"] = p_retain_normal, + ["retain and collapse"] = p_retain_collapse, + ["retain and no empty"] = p_retain_noempty, +} + +strings.striplinepatterns = striplinepatterns + +function strings.striplines(str,how) + return str and lpegmatch(how and striplinepatterns[how] or p_prune_collapse,str) or str end --- local template = string.striplong([[ +strings.striplong = strings.striplines -- for old times sake + +-- local str = table.concat( { +-- " ", +-- " aap", +-- " noot mies", +-- " ", +-- " ", +-- " zus wim jet", +-- "zus wim jet", +-- " zus wim jet", +-- " ", +-- }, "\n") + +-- local str = table.concat( { +-- " aaaa", +-- " bb", +-- " cccccc", +-- }, "\n") + +-- for k, v in table.sortedhash(utilities.strings.striplinepatterns) do +-- logs.report("stripper","method: %s, result: [[%s]]",k,utilities.strings.striplines(str,k)) +-- end + +-- inspect(strings.striplong([[ -- aaaa -- bb -- cccccc --- ]]) +-- ]])) function strings.nice(str) str = gsub(str,"[:%-+_]+"," ") -- maybe more @@ -305,60 +403,58 @@ local template = [[ return function(%s) return %s end ]] --- local environment = { --- lpeg = lpeg, --- type = type, --- string = string, --- number = number, --- table = table, --- utf = utf, --- } --- --- local preamble = [[ --- local type = type --- local tostring = tostring --- local tonumber = tonumber --- local format = string.format --- local concat = table.concat --- local signed = number.signed --- local points = number.points --- local basepoints = number.basepoints --- local utfchar = utf.char --- local utfbyte = utf.byte --- local lpegmatch = lpeg.match --- local nspaces = string.nspaces --- local tracedchar = string.tracedchar --- local autosingle = string.autosingle --- local autodouble = string.autodouble --- local sequenced = table.sequenced --- local formattednumber = number.formatted --- local sparseexponent = number.sparseexponent --- ]] - -local environment = { - global = global or _G, - lpeg = lpeg, - type = type, - tostring = tostring, - tonumber = tonumber, - format = string.format, - concat = table.concat, - signed = number.signed, - points = number.points, - basepoints = number.basepoints, - utfchar = utf.char, - utfbyte = utf.byte, - lpegmatch = lpeg.match, - nspaces = string.nspaces, - tracedchar = string.tracedchar, - autosingle = string.autosingle, - autodouble = string.autodouble, - sequenced = table.sequenced, - formattednumber = number.formatted, - sparseexponent = number.sparseexponent, -} +local preamble, environment = "", { } + +if _LUAVERSION < 5.2 then + + preamble = [[ +local lpeg=lpeg +local type=type +local tostring=tostring +local tonumber=tonumber +local format=string.format +local concat=table.concat +local signed=number.signed +local points=number.points +local basepoints= number.basepoints +local utfchar=utf.char +local utfbyte=utf.byte +local lpegmatch=lpeg.match +local nspaces=string.nspaces +local tracedchar=string.tracedchar +local autosingle=string.autosingle +local autodouble=string.autodouble +local sequenced=table.sequenced +local formattednumber=number.formatted +local sparseexponent=number.sparseexponent + ]] + +else + + environment = { + global = global or _G, + lpeg = lpeg, + type = type, + tostring = tostring, + tonumber = tonumber, + format = string.format, + concat = table.concat, + signed = number.signed, + points = number.points, + basepoints = number.basepoints, + utfchar = utf.char, + utfbyte = utf.byte, + lpegmatch = lpeg.match, + nspaces = string.nspaces, + tracedchar = string.tracedchar, + autosingle = string.autosingle, + autodouble = string.autodouble, + sequenced = table.sequenced, + formattednumber = number.formatted, + sparseexponent = number.sparseexponent, + } -local preamble = "" +end -- -- -- @@ -412,7 +508,7 @@ local format_i = function(f) if f and f ~= "" then return format("format('%%%si',a%s)",f,n) else - return format("format('%%i',a%s)",n) + return format("format('%%i',a%s)",n) -- why not just tostring() end end @@ -428,6 +524,11 @@ local format_f = function(f) return format("format('%%%sf',a%s)",f,n) end +local format_F = function(f) + n = n + 1 + return format("((a%s == 0 and '0') or (a%s == 1 and '1') or format('%%%sf',a%s))",n,n,f,n) +end + local format_g = function(f) n = n + 1 return format("format('%%%sg',a%s)",f,n) @@ -701,7 +802,7 @@ local builder = Cs { "start", V("!") -- new + V("s") + V("q") + V("i") + V("d") - + V("f") + V("g") + V("G") + V("e") + V("E") + + V("f") + V("F") + V("g") + V("G") + V("e") + V("E") + V("x") + V("X") + V("o") -- + V("c") @@ -736,6 +837,7 @@ local builder = Cs { "start", ["i"] = (prefix_any * P("i")) / format_i, -- %i => regular %i (integer) ["d"] = (prefix_any * P("d")) / format_d, -- %d => regular %d (integer) ["f"] = (prefix_any * P("f")) / format_f, -- %f => regular %f (float) + ["F"] = (prefix_any * P("F")) / format_F, -- %F => regular %f (float) but 0/1 check ["g"] = (prefix_any * P("g")) / format_g, -- %g => regular %g (float) ["G"] = (prefix_any * P("G")) / format_G, -- %G => regular %G (float) ["e"] = (prefix_any * P("e")) / format_e, -- %e => regular %e (float) @@ -810,7 +912,8 @@ local function make(t,str) f = loadstripped(p)() else n = 0 - p = lpegmatch(builder,str,1,"..",t._extensions_) -- after this we know n + -- p = lpegmatch(builder,str,1,"..",t._extensions_) -- after this we know n + p = lpegmatch(builder,str,1,t._connector_,t._extensions_) -- after this we know n if n > 0 then p = format(template,preamble,t._preamble_,arguments[n],p) -- print("builder 2 >",p) @@ -869,14 +972,28 @@ strings.formatters = { } -- table (metatable) in which case we could better keep a count and -- clear that table when a threshold is reached -function strings.formatters.new() - local e = { } -- better make a copy as we can overload - for k, v in next, environment do - e[k] = v +-- _connector_ is an experiment + +if _LUAVERSION < 5.2 then + + function strings.formatters.new(noconcat) + local t = { _type_ = "formatter", _connector_ = noconcat and "," or "..", _extensions_ = { }, _preamble_ = preamble, _environment_ = { } } + setmetatable(t, { __index = make, __call = use }) + return t end - local t = { _extensions_ = { }, _preamble_ = "", _environment_ = e, _type_ = "formatter" } - setmetatable(t, { __index = make, __call = use }) - return t + +else + + function strings.formatters.new(noconcat) + local e = { } -- better make a copy as we can overload + for k, v in next, environment do + e[k] = v + end + local t = { _type_ = "formatter", _connector_ = noconcat and "," or "..", _extensions_ = { }, _preamble_ = "", _environment_ = e } + setmetatable(t, { __index = make, __call = use }) + return t + end + end -- function strings.formatters.new() @@ -921,9 +1038,19 @@ patterns.luaquoted = Cs(Cc('"') * ((1-S('"\n'))^1 + P('"')/'\\"' + P('\n')/'\\n" -- add(formatters,"tex", [[lpegmatch(texescape,%s)]],[[local texescape = lpeg.patterns.texescape]]) -- add(formatters,"lua", [[lpegmatch(luaescape,%s)]],[[local luaescape = lpeg.patterns.luaescape]]) -add(formatters,"xml", [[lpegmatch(xmlescape,%s)]],{ xmlescape = lpeg.patterns.xmlescape }) -add(formatters,"tex", [[lpegmatch(texescape,%s)]],{ texescape = lpeg.patterns.texescape }) -add(formatters,"lua", [[lpegmatch(luaescape,%s)]],{ luaescape = lpeg.patterns.luaescape }) +if _LUAVERSION < 5.2 then + + add(formatters,"xml",[[lpegmatch(xmlescape,%s)]],"local xmlescape = lpeg.patterns.xmlescape") + add(formatters,"tex",[[lpegmatch(texescape,%s)]],"local texescape = lpeg.patterns.texescape") + add(formatters,"lua",[[lpegmatch(luaescape,%s)]],"local luaescape = lpeg.patterns.luaescape") + +else + + add(formatters,"xml",[[lpegmatch(xmlescape,%s)]],{ xmlescape = lpeg.patterns.xmlescape }) + add(formatters,"tex",[[lpegmatch(texescape,%s)]],{ texescape = lpeg.patterns.texescape }) + add(formatters,"lua",[[lpegmatch(luaescape,%s)]],{ luaescape = lpeg.patterns.luaescape }) + +end -- -- yes or no: -- |