-- -*- coding: utf-8 ; -*- -- -- This is file 'piton.lua' which is a part of the Latex package 'piton' -- Copyright (C) 2022 by F. Pantigny -- This file may be distributed and/or modified under the -- conditions of the Latex Project Public License, either -- version 1.3 of this license or (at your option) any later -- version. The latest version of this license is in: -- -- http://www.Latex-project.org/lppl.txt -- -- and version 1.3 or later is part of all distributions of -- Latex version 2005/12/01 or later. -- This file is part of the version 0.55 of the package 'piton' local P, S, V , C , Ct, Cc, Cf = lpeg.P, lpeg.S, lpeg.V, lpeg.C, lpeg.Ct, lpeg.Cc, lpeg.Cf --[[ By convention, a capture which provides as value a table (and not a string), provides, in fact, a string (the first element of the table) which is a formatting LaTeX instruction (it will be thrown back to TeX with normal catcodes (ant not ``other'' catcode for everybody).]] local function L(string) return lpeg.Cc ( { string } ) end local function K(pattern, style) return L ( "{\\PitonStyle{" .. style .. "}{" ) * lpeg.C(pattern) * L ( "}}" ) end --[[ The text in "escape" (between begin_escape and end_escape) is captured and put in a table (with only one component). Indeed, we have decided that a capture which is encapsulated in a table must be transmitted to TeX with the normal TeX catcodes.]] local Escape = P(begin_escape) * Ct ( C ( ( 1 - P(end_escape) ) ^ 1 ) ) * P(end_escape) local EOL = ( P "\r" ) * L ( "\\pitonEOL" ) lpeg.locale(lpeg) -- mandatory local alpha , digit , space , punct = lpeg.alpha , lpeg.digit , lpeg.space , lpeg.punct local letter = alpha + S"âàçéèêëïîôûüÂÀÇÉÈÊËÏÎÔÛÜ_" local alphanum = letter + digit local identifier = letter * alphanum ^ 0 local Identifier = C ( identifier ) local Space = C ( ( space - P "\r" ) ^ 1 ) local SkipSpace = C ( ( space - P "\r" ) ^ 0 ) local Punct = C ( punct ) local Number = K ( ( digit^1 * P "." * digit^0 + digit^0 * P "." * digit^1 + digit^1 ) * ( S "eE" * S "+-" ^ -1 * digit^1 ) ^ -1 + digit^1 , 'Number' ) local Word = C ( ( ( 1 - space ) - S "'\"\r[()]" - digit ) ^ 1 ) local Delim = C ( S "[()]" ) local Keyword = K ( P "assert" + P "break" + P "case" + P "continue" + P "del" + P "elif" + P "else" + P "except" + P "exec" + P "finally" + P "for" + P "from" + P "global" + P "if" + P "import" + P "lambda" + P "non local" + P "pass" + P "return" + P "try" + P "while" + P "with" + P "yield" + P "yield from" , 'Keyword' ) + K ( P "True" + P "False" + P "None" , 'Keyword.Constant' ) local Builtin = K ( P "__import__" + P "abs" + P "all" + P "any" + P "bin" + P "bool" + P "bytearray" + P "bytes" + P "chr" + P "classmethod" + P "compile" + P "complex" + P "delattr" + P "dict" + P "dir" + P "divmod" + P "enumerate" + P "eval" + P "filter" + P "float" + P "format" + P "frozenset" + P "getattr" + P "globals" + P "hasattr" + P "hash" + P "hex" + P "id" + P "input" + P "int" + P "isinstance" + P "issubclass" + P "iter" + P "len" + P "list" + P "locals" + P "map" + P "max" + P "memoryview" + P "min" + P "next" + P "object" + P "oct" + P "open" + P "ord" + P "pow" + P "print" + P "property" + P "range" + P "repr" + P "reversed" + P "round" + P "set" + P "setattr" + P "slice" + P "sorted" + P "staticmethod" + P "str" + P "sum" + P "super" + P "tuple" + P "type" + P "vars" + P "zip" , 'Name.Builtin' ) local Exception = K ( "ArithmeticError" + P "AssertionError" + P "AttributeError" + P "BaseException" + P "BufferError" + P "BytesWarning" + P "DeprecationWarning" + P "EOFError" + P "EnvironmentError" + P "Exception" + P "FloatingPointError" + P "FutureWarning" + P "GeneratorExit" + P "IOError" + P "ImportError" + P "ImportWarning" + P "IndentationError" + P "IndexError" + P "KeyError" + P "KeyboardInterrupt" + P "LookupError" + P "MemoryError" + P "NameError" + P "NotImplementedError" + P "OSError" + P "OverflowError" + P "PendingDeprecationWarning" + P "ReferenceError" + P "ResourceWarning" + P "RuntimeError" + P "RuntimeWarning" + P "StopIteration" + P "SyntaxError" + P "SyntaxWarning" + P "SystemError" + P "SystemExit" + P "TabError" + P "TypeError" + P "UnboundLocalError" + P "UnicodeDecodeError" + P "UnicodeEncodeError" + P "UnicodeError" + P "UnicodeTranslateError" + P "UnicodeWarning" + P "UserWarning" + P "ValueError" + P "VMSError" + P "Warning" + P "WindowsError" + P "ZeroDivisionError" + P "BlockingIOError" + P "ChildProcessError" + P "ConnectionError" + P "BrokenPipeError" + P "ConnectionAbortedError" + P "ConnectionRefusedError" + P "ConnectionResetError" + P "FileExistsError" + P "FileNotFoundError" + P "InterruptedError" + P "IsADirectoryError" + P "NotADirectoryError" + P "PermissionError" + P "ProcessLookupError" + P "TimeoutError" + P "StopAsyncIteration" + P "ModuleNotFoundError" + P "RecursionError" , 'Exception' ) local RaiseException = K ( P "raise" , 'Keyword' ) * SkipSpace * Exception * C ( P "(" ) local ExceptionInConsole = Exception * C ( ( 1 - P "\r" ) ^ 0 ) * EOL local Namespace = K ( P "from" , 'Keyword' ) * Space * K ( alphanum^1 , 'Name.Namespace' ) * ( Space * K ( P "import" , 'Keyword' ) ) ^ -1 local ImportAs = K ( P "import" , 'Keyword' ) * Space * K ( identifier , 'Name.Namespace' ) * ( SkipSpace * C ( P "," ) * SkipSpace * K ( identifier , 'Name.Namespace' ) ) ^ 0 * ( Space * K ( P "as" , 'Keyword' ) * Space * K ( identifier , 'Name.Namespace' ) ) ^ 0 local Class = K ( P "class" , 'Keyword' ) * Space * K ( identifier , 'Name.Class' ) local Decorator = K ( P "@" * letter^1 , 'Name.Decorator' ) local SingleShortInterpol = K ( P "{" , 'String.Interpol') * K ( ( 1 - S "}':" ) ^ 0 , 'Interpol.Inside' ) * C ( P ":" * (1 - S "}:'") ^ 0 ) ^ -1 * K ( P "}" , 'String.Interpol' ) local DoubleShortInterpol = K ( P "{" , 'String.Interpol' ) * K ( ( 1 - S "}\":" ) ^ 0 , 'Interpol.Inside' ) * ( K ( P ":" , 'String.Interpol' ) * C ( (1 - S "}:\"") ^ 0 ) ) ^ -1 * K ( P "}" , 'String.Interpol' ) local SingleLongInterpol = K ( P "{" , 'String.Interpol' ) * K ( ( 1 - S "}:\r" - P "'''" ) ^ 0 , 'Interpol.Inside' ) * C ( P ":" * (1 - S "}:\r" - P "'''" ) ^ 0 ) ^ -1 * K ( P "}" , 'String.Interpol' ) local DoubleLongInterpol = K ( P "{" , 'String.Interpol' ) * K ( ( 1 - S "}:\r" - P "\"\"\"" ) ^ 0 , 'Interpol.Inside' ) * C ( P ":" * (1 - S "}:\r" - P "\"\"\"" ) ^ 0 ) ^ -1 * K ( P "}" , 'String.Interpol' ) local SingleShortPureString = C ( ( P "\\'" + P "{{" + P "}}" + 1 - S "{}'" ) ^ 1 ) local DoubleShortPureString = C ( ( P "\\\"" + P "{{" + P "}}" + 1 - S "{}\"" ) ^ 1 ) local SingleLongPureString = C ( ( 1 - P "'''" - S "{}'\r" ) ^ 1 ) local DoubleLongPureString = C ( ( 1 - P "\"\"\"" - S "{}\"\r" ) ^ 1 ) local SingleShortString = L ( "{\\PitonStyle{String.Short}{" ) * ( C ( P "f'" + P "F'" ) * ( SingleShortInterpol + SingleShortPureString ) ^ 0 * C ( P "'" ) + C ( ( P "'" + P "r'" + P "R'" ) * ( P "\\'" + 1 - S "'\r" ) ^ 0 * P "'" ) ) * L ( "}}" ) local DoubleShortString = L ( "{\\PitonStyle{String.Short}{" ) * ( C ( P "f\"" + P "F\"" ) * ( DoubleShortInterpol + DoubleShortPureString ) ^ 0 * C ( P "\"" ) + C ( ( P "\"" + P "r\"" + P "R\"" ) * ( P "\\\"" + 1 - S "\"\r" ) ^ 0 * P "\"" ) ) * L ( "}}" ) local ShortString = SingleShortString + DoubleShortString local SingleLongString = L "{\\PitonStyle{String.Long}{" * ( C ( S "fF" * P "'''" ) * ( SingleLongInterpol + SingleLongPureString ) ^ 0 * L "}}" * ( EOL + L "{\\PitonStyle{String.Long}{" * ( SingleLongInterpol + SingleLongPureString ) ^ 0 * L "}}" * EOL ) ^ 0 * L "{\\PitonStyle{String.Long}{" * ( SingleLongInterpol + SingleLongPureString ) ^ 0 + C ( ( S "rR" ) ^ -1 * P "'''" * ( 1 - P "'''" - P "\r" ) ^ 0 ) * L "}}" * ( L "{\\PitonStyle{String.Long}{" * C ( ( 1 - P "'''" - P "\r" ) ^ 0 ) * L "}}" * EOL ) ^ 0 * L "{\\PitonStyle{String.Long}{" * C ( ( 1 - P "'''" - P "\r" ) ^ 0 ) ) * C ( P "'''" ) * L "}}" local DoubleLongString = L "{\\PitonStyle{String.Long}{" * ( C ( S "fF" * P "\"\"\"" ) * ( DoubleLongInterpol + DoubleLongPureString ) ^ 0 * L "}}" * ( EOL + L "{\\PitonStyle{String.Long}{" * ( DoubleLongInterpol + DoubleLongPureString ) ^ 0 * L "}}" * EOL ) ^ 0 * L "{\\PitonStyle{String.Long}{" * ( DoubleLongInterpol + DoubleLongPureString ) ^ 0 + C ( ( S "rR" ) ^ -1 * P "\"\"\"" * ( 1 - P "\"\"\"" - P "\r" ) ^ 0 ) * L "}}" * ( L "{\\PitonStyle{String.Long}{" * C ( ( 1 - P "\"\"\"" - P "\r" ) ^ 0 ) * L "}}" * EOL ) ^ 0 * L "{\\PitonStyle{String.Long}{" * C ( ( 1 - P "\"\"\"" - P "\r" ) ^ 0 ) ) * C ( P "\"\"\"" ) * L "}}" local LongString = SingleLongString + DoubleLongString local Expression = P { "E" , E = ( 1 - S "{}()[]\r," ) ^ 0 * ( ( P "{" * V "F" * P "}" + P "(" * V "F" * P ")" + P "[" * V "F" * P "]" ) * ( 1 - S "{}()[]\r," ) ^ 0 ) ^ 0 , F = ( 1 - S "{}()[]\r\"'" ) ^ 0 * ( ( P "'" * (P "\\'" + 1 - S"'\r" )^0 * P "'" + P "\"" * (P "\\\"" + 1 - S"\"\r" )^0 * P "\"" + P "{" * V "F" * P "}" + P "(" * V "F" * P ")" + P "[" * V "F" * P "]" ) * ( 1 - S "{}()[]\r\"'" ) ^ 0 ) ^ 0 , } local Param = SkipSpace * K ( identifier , '' ) * SkipSpace * ( K ( P "=" * Expression , 'InitialValues' ) + K ( P ":" , '' ) * SkipSpace * K ( letter^1 , 'Name.Type' )) + SkipSpace * K ( alphanum ^ 1 , '' ) * SkipSpace local Params = Param * ( K ( P "," , '' ) * Param ) ^ 0 local StringDoc = K ( P "\"\"\"" , 'String.Doc' ) * ( K ( (1 - P "\"\"\"" - P "\r" ) ^ 0 , 'String.Doc' ) * EOL ) ^ 0 * K ( ( 1 - P "\"\"\"" - P "\r" ) ^ 0 * P "\"\"\"" , 'String.Doc' ) + K ( P "'''" , 'String.Doc' ) * ( K ( (1 - P "'''" - P "\r")^0 , 'String.Doc' ) * EOL ) ^ 0 * K ( ( 1 - P "'''" - P "\r")^0 * P "'''" , 'String.Doc' ) local CommentMath = P "$" * K ( ( 1 - S "$\r" ) ^ 1 , 'Comment.Math' ) * P "$" local Comment = L ( "{\\pitonStyleComment {" ) * C ( P "#" ) * ( CommentMath + C ( ( 1 - S "$\r" ) ^ 1 ) ) ^ 0 * L ( "}}" ) * ( EOL + -1 ) local CommentLaTeX = P "##" * L "{\\PitonStyle{Comment.LaTeX}{" * Ct ( C ( ( 1 - P "\r" ) ^ 0 ) ) * L "}}" * ( EOL + -1 ) local DefFunction = K ( P "def" , 'Keyword' ) * Space * K ( identifier , 'Name.Function' ) * ( SkipSpace * K ( P "(" , '' ) * Params * K ( P ")" , '' ) ) ^ -1 * ( SkipSpace * K ( ( 1 - S ":\r" )^0 , 'Post.Function' ) * K ( P ":" , 'Keyword' ) * SkipSpace * ( EOL + CommentLaTeX + Comment ) * SkipSpace * StringDoc ) ^ -1 local ItemDict = ShortString * SkipSpace * C ( P ":" ) * K ( Expression , 'Dict.Value' ) local ItemOfSet = SkipSpace * ( ItemDict + ShortString ) * SkipSpace local Set = C ( P "{" ) * ItemOfSet * ( C ( P "," ) * ItemOfSet ) ^ 0 * C ( P "}" ) local Operator = K ( P "!=" + P "==" + P "<<" + P ">>" + S "-~+/*%=<>&.@|" , 'Operator') local OperatorWord = K ( P "in" + P "is" + P "and" + P "or" + P "not" , 'Operator.Word') local SyntaxPython = ( ( space - P "\r" ) ^0 * P "\r" ) ^ -1 * space ^ -1 * ( ( space^1 * -1 ) + EOL + Space + Escape + CommentLaTeX + LongString + Comment + ExceptionInConsole + Set + Delim + Class * ( Space + Punct + EOL ) + Namespace * ( Space + Punct + EOL ) + ImportAs + RaiseException + Keyword * ( Space + Punct + EOL ) + DefFunction + ShortString + Decorator * ( Space + Punct + EOL ) + Operator + OperatorWord * ( Space + Punct + EOL ) + Builtin * ( Space + Punct + EOL ) + Identifier + Number + Word ) ^0 * -1 local MinimalSyntax = P { "S" ; S = K ( (1 - P "\r" ) ^ 0 , '') + EOL * S } --- with cctab, all characters including spaces have catcode 12 local cctab = luatexbase.catcodetables.CatcodeTableOther function Parse(code) local t = Ct(SyntaxPython) : match(code) if t then else t = Ct(MinimalSyntax) : match(code) end for i = 1 , #t do if type(t[i]) == 'string' then tex.sprint(cctab, t[i]) else tex.sprint( t[i][1] ) end end end function ParseFile(name) s = '' for line in io.lines(name) do s = s .. '\r' .. line end Parse(s) end function define_gobble_syntax(n) GobbleSyntax = ( 1 - P "\r" ) ^ (-n) * C ( ( 1 - P "\r" ) ^0 ) * ( C ( P "\r" ) * ( 1 - P "\r" ) ^ (-n) * C ( ( 1 - P "\r" ) ^0 ) ) ^ 0 end function GobbleParse(code) local t = Ct(GobbleSyntax):match(code) local new_code = "" for i = 1 , #t do new_code = new_code .. t[i] end Parse(new_code) end function min(x,y) if x <= y then return x else return y end end function add(acc,new_value) return acc + new_value end --[[ The following LPEG returns as capture the minimal number of spaces at the beginning of the lines of code]] AutoGobbleSyntax = ( space ^ 0 * P "\r" ) ^ -1 * Cf ( ( ( P " " ) ^ 0 * P "\r" + Cf ( Cc(0) * ( P " " * Cc(1) ) ^ 0 , add ) * ( 1 - P " " ) * ( 1 - P "\r" ) ^ 0 * P "\r" ) ^ 0 * ( Cf ( Cc(0) * ( P " " * Cc(1) ) ^ 0 , add ) * ( 1 - P " " ) * ( 1 - P "\r" ) ^ 0 ) ^ -1 , min ) function AutoGobbleParse(code) local n = AutoGobbleSyntax:match(code) if n==0 then Parse(code) else define_gobble_syntax(n) GobbleParse(code) end end