diff options
Diffstat (limited to 'Master/texmf-dist/tex/luatex/lualibs/lualibs-lpeg.lua')
-rw-r--r-- | Master/texmf-dist/tex/luatex/lualibs/lualibs-lpeg.lua | 27 |
1 files changed, 21 insertions, 6 deletions
diff --git a/Master/texmf-dist/tex/luatex/lualibs/lualibs-lpeg.lua b/Master/texmf-dist/tex/luatex/lualibs/lualibs-lpeg.lua index b33df96b7d9..399b3ad6501 100644 --- a/Master/texmf-dist/tex/luatex/lualibs/lualibs-lpeg.lua +++ b/Master/texmf-dist/tex/luatex/lualibs/lualibs-lpeg.lua @@ -82,7 +82,6 @@ setinspector(function(v) if lpegtype(v) then lpegprint(v) return true end end) lpeg.patterns = lpeg.patterns or { } -- so that we can share local patterns = lpeg.patterns - local anything = P(1) local endofstring = P(-1) local alwaysmatched = P(true) @@ -109,29 +108,42 @@ local space = P(" ") local period = P(".") local comma = P(",") -local utfbom_32_be = P('\000\000\254\255') -local utfbom_32_le = P('\255\254\000\000') -local utfbom_16_be = P('\255\254') -local utfbom_16_le = P('\254\255') -local utfbom_8 = P('\239\187\191') +local utfbom_32_be = P('\000\000\254\255') -- 00 00 FE FF +local utfbom_32_le = P('\255\254\000\000') -- FF FE 00 00 +local utfbom_16_be = P('\254\255') -- FE FF +local utfbom_16_le = P('\255\254') -- FF FE +local utfbom_8 = P('\239\187\191') -- EF BB BF local utfbom = utfbom_32_be + utfbom_32_le + utfbom_16_be + utfbom_16_le + utfbom_8 local utftype = utfbom_32_be * Cc("utf-32-be") + utfbom_32_le * Cc("utf-32-le") + utfbom_16_be * Cc("utf-16-be") + utfbom_16_le * Cc("utf-16-le") + utfbom_8 * Cc("utf-8") + alwaysmatched * Cc("utf-8") -- assume utf8 +local utfstricttype = utfbom_32_be * Cc("utf-32-be") + utfbom_32_le * Cc("utf-32-le") + + utfbom_16_be * Cc("utf-16-be") + utfbom_16_le * Cc("utf-16-le") + + utfbom_8 * Cc("utf-8") local utfoffset = utfbom_32_be * Cc(4) + utfbom_32_le * Cc(4) + utfbom_16_be * Cc(2) + utfbom_16_le * Cc(2) + utfbom_8 * Cc(3) + Cc(0) local utf8next = R("\128\191") +patterns.utfbom_32_be = utfbom_32_be +patterns.utfbom_32_le = utfbom_32_le +patterns.utfbom_16_be = utfbom_16_be +patterns.utfbom_16_le = utfbom_16_le +patterns.utfbom_8 = utfbom_8 + +patterns.utf_16_be_nl = P("\000\r\000\n") + P("\000\r") + P("\000\n") +patterns.utf_16_le_nl = P("\r\000\n\000") + P("\r\000") + P("\n\000") + patterns.utf8one = R("\000\127") patterns.utf8two = R("\194\223") * utf8next patterns.utf8three = R("\224\239") * utf8next * utf8next patterns.utf8four = R("\240\244") * utf8next * utf8next * utf8next patterns.utfbom = utfbom patterns.utftype = utftype +patterns.utfstricttype = utfstricttype patterns.utfoffset = utfoffset local utf8char = patterns.utf8one + patterns.utf8two + patterns.utf8three + patterns.utf8four @@ -454,6 +466,9 @@ function lpeg.replacer(one,two,makefunction,isutf) -- in principle we should sor end end +-- local pattern1 = P(1-P(pattern))^0 * P(pattern) : test for not nil +-- local pattern2 = (P(pattern) * Cc(true) + P(1))^0 : test for true (could be faster, but not much) + function lpeg.finder(lst,makefunction) -- beware: slower than find with 'patternless finds' local pattern if type(lst) == "table" then |