# above qName definition allows double :: and excludes non-ascii letters
# qName = Regex("[_A-Za-z\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD]"
# r"[_\-\."
# "\xB7A-Za-z0-9\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u0300-\u036F\u203F-\u2040]*"
# "[:]?"
# r"[_\-\."
# "\xB7A-Za-z0-9\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u0300-\u036F\u203F-\u2040]*")
ncName = Word(alphas + '_',alphanums + '_-.')
prefixOp = Literal(":")
decimalPoint = Literal('.')
exponentLiteral = CaselessLiteral('e')
plusorminusLiteral = Literal('+') | Literal('-')
digits = Word(nums)
integerLiteral = Combine( Optional(plusorminusLiteral) + digits )
decimalFractionLiteral = Combine( Optional(plusorminusLiteral) + decimalPoint + digits )
infLiteral = Combine( Optional(plusorminusLiteral) + Literal("INF") )
nanLiteral = Literal("NaN")
floatLiteral = ( Combine( integerLiteral +
( ( decimalPoint + Optional(digits) + exponentLiteral + integerLiteral ) |
( exponentLiteral + integerLiteral ) )
) |
Combine( decimalFractionLiteral + exponentLiteral + integerLiteral ) |
infLiteral | nanLiteral )
decimalLiteral = ( Combine( integerLiteral + decimalPoint + Optional(digits) ) |
decimalFractionLiteral )
#emptySequence = Literal( "(" ) + Literal( ")" )
lParen = Literal( "(" )
def compileSphinxGrammar(cntlr):
global isGrammarCompiled, sphinxProg, lineno
if isGrammarCompiled:
return sphinxProg
debugParsing = True
cntlr.showStatus(_("Compiling Sphinx Grammar"))
if sys.version[0] >= '3':
# python 3 requires modified parser to allow release of global objects when closing DTS
from arelle.pyparsing.pyparsing_py3 import (
Word, Keyword, alphas, Literal, CaselessLiteral, Combine, Optional,
nums, Or, Forward, Group, ZeroOrMore, StringEnd, alphanums,
ParserElement, quotedString, delimitedList, Suppress, Regex,
FollowedBy, lineno)
else:
from pyparsing import (Word, Keyword, alphas, Literal, CaselessLiteral,
Combine, Optional, nums, Or, Forward, Group,
ZeroOrMore, StringEnd, alphanums, ParserElement,
quotedString, delimitedList, Suppress, Regex,
FollowedBy, lineno)
ParserElement.enablePackrat()
"""
the pyparsing parser constructs are defined in this method to prevent the need to compile
the grammar when the plug in is loaded (which is likely to be when setting up GUI
menus or command line parser).
instead the grammar is compiled the first time that any sphinx needs to be parsed
only the sphinxExpression (result below) needs to be global for the parser
"""
# define grammar
sphinxComment = Regex(
r"/(?:\*(?:[^*]*\*+)+?/|/[^\n]*(?:\n[^\n]*)*?(?:(?<!\\)|\Z))"
).setParseAction(compileComment)
variableRef = Regex(
"[$]" # variable prefix
# localname part
"([A-Za-z\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD_]"
"[A-Za-z0-9\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u0300-\u036F\u203F-\u2040\xB7_.-]*)"
)
qName = Regex(
"([A-Za-z\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD_]"
"[A-Za-z0-9\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u0300-\u036F\u203F-\u2040\xB7_.-]*:)?"
# localname or wildcard-localname part
"([A-Za-z\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD_]"
"[A-Za-z0-9\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u0300-\u036F\u203F-\u2040\xB7_.-]*|[*])"
)
ncName = Regex(
"([A-Za-z\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD_]"
"[A-Za-z0-9\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD\u0300-\u036F\u203F-\u2040\xB7_.-]*)"
).setName("ncName").setDebug(debugParsing)
#annotationName = Word("@",alphanums + '_-.').setName("annotationName").setDebug(debugParsing)
annotationName = Regex(
"@[A-Za-z\xC0-\xD6\xD8-\xF6\xF8-\xFF\u0100-\u02FF\u0370-\u037D\u037F-\u1FFF\u200C-\u200D\u2070-\u218F\u2C00-\u2FEF\u3001-\uD7FF\uF900-\uFDCF\uFDF0-\uFFFD_]\w*"
).setName("annotationName").setDebug(debugParsing)
decimalPoint = Literal('.')
exponentLiteral = CaselessLiteral('e')
plusorminusLiteral = Literal('+') | Literal('-')
digits = Word(nums)
integerLiteral = Combine(Optional(plusorminusLiteral) + digits)
decimalFractionLiteral = Combine(
Optional(plusorminusLiteral) + decimalPoint + digits)
infLiteral = Combine(Optional(plusorminusLiteral) + Literal("INF"))
nanLiteral = Literal("NaN")
floatLiteral = (Combine(integerLiteral + (
(decimalPoint + Optional(digits) + exponentLiteral + integerLiteral) |
(exponentLiteral + integerLiteral) |
(decimalPoint + Optional(digits))))
| Combine(decimalFractionLiteral + exponentLiteral +
integerLiteral) | decimalFractionLiteral
| infLiteral | nanLiteral)
#emptySequence = Literal( "(" ) + Literal( ")" )
lParen = Literal("(")
rParen = Literal(")")
lPred = Literal("[[") | Literal("[")
rPred = Literal("]]") | Literal("]")
commaOp = Literal(",")
ifOp = Keyword("if")
elseOp = Keyword("else")
forOp = Keyword("for")
inOp = Keyword("in")
withOp = Keyword("with")
notOp = Keyword("not")
valuesOp = Keyword("values")
andOp = Keyword("and")
orOp = Keyword("or")
neOp = Literal("!=")
leOp = Literal("<=")
ltOp = Literal("<")
geOp = Literal(">=")
gtOp = Literal(">")
eqOp = Literal("==")
compOp = leOp | ltOp | geOp | gtOp
plusOp = Literal("|+|") | Literal("|+") | Literal("+|") | Literal("+")
minusOp = Literal("|-|") | Literal("|-") | Literal("-|") | Literal("-")
plusMinusOp = (plusOp | minusOp).setParseAction(compileOp)
multOp = Literal("*")
divOp = Literal("/")
varAssign = Literal("=")
tagOp = Literal("#")
asOp = Keyword("as")
whereOp = Keyword("where")
wildOp = Literal("**") | Literal("*")
methodOp = Literal("::")
formulaOp = Literal(":=")
namespaceDeclaration = (
Literal("xmlns") + Optional(Suppress(Literal(":")) + ncName) +
Suppress(Literal("=")) + quotedString
).setParseAction(compileNamespaceDeclaration).ignore(sphinxComment)
annotationDeclaration = (
Suppress(Keyword("annotation")) + ncName +
Optional(Suppress(Keyword("as")) + ncName)
).setParseAction(compileAnnotationDeclaration).ignore(sphinxComment)
packageDeclaration = (
Suppress(Keyword("package")) +
ncName).setParseAction(compilePackageDeclaration).ignore(sphinxComment)
severity = (Suppress(Keyword("severity")) +
(ncName)).setParseAction(compileSeverity).ignore(sphinxComment)
expr = Forward()
atom = ((forOp - Suppress(lParen) - ncName - Suppress(inOp) - expr -
Suppress(rParen) - expr).setParseAction(compileFor) |
(ifOp - Suppress(lParen) - expr - Suppress(rParen) - expr -
Suppress(elseOp) - expr).setParseAction(compileIf) |
(ncName + Suppress(lParen) + Optional(
delimitedList(
ZeroOrMore((ncName + Optional(tagOp + Optional(ncName)) +
varAssign + expr + Suppress(Literal(";"))
).setParseAction(compileVariableAssignment)) +
Optional(ncName + varAssign) + expr)) +
Suppress(rParen)).setParseAction(compileFunctionReference) |
(floatLiteral).setParseAction(compileFloatLiteral) |
(integerLiteral).setParseAction(compileIntegerLiteral) |
(quotedString).setParseAction(compileStringLiteral) |
(Optional(qName) + lPred + Optional(
delimitedList(
((whereOp + expr) |
((qName | variableRef) +
Optional(tagOp + Optional(ncName)) + Optional(
(varAssign + (wildOp | expr) | (inOp + expr) |
(asOp + ncName + varAssign + wildOp +
Optional(whereOp + expr)))))
).setParseAction(compileHyperspaceAxis),
delim=';')) +
rPred).setParseAction(compileHyperspaceExpression) |
(variableRef).setParseAction(compileVariableReference) |
(qName).setParseAction(compileQname) |
(Suppress(lParen) - expr -
Optional(commaOp - Optional(expr - ZeroOrMore(commaOp - expr))) -
Suppress(rParen)).setParseAction(compileBrackets)
).ignore(sphinxComment)
atom.setName("atom").setDebug(debugParsing)
valueExpr = atom
taggedExpr = (valueExpr - Optional(tagOp - ncName)
).setParseAction(compileTagAssignment).ignore(sphinxComment)
methodExpr = (
(methodOp + ncName + ZeroOrMore(methodOp + taggedExpr)
).setParseAction(compileMethodReference) |
(ZeroOrMore(taggedExpr + methodOp) + taggedExpr)
).setParseAction(compileMethodReference).ignore(sphinxComment)
unaryExpr = (
Optional(plusMinusOp) +
methodExpr).setParseAction(compileUnaryOperation).ignore(sphinxComment)
negateExpr = (
Optional(notOp) +
unaryExpr).setParseAction(compileUnaryOperation).ignore(sphinxComment)
valuesExpr = (Optional(valuesOp) + negateExpr).setParseAction(
compileValuesIteration).ignore(sphinxComment)
method2Expr = (valuesExpr +
Optional(methodOp + methodExpr)).setParseAction(
compileMethodReference).ignore(sphinxComment)
multiplyExpr = (method2Expr +
Optional(multOp + method2Expr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
divideExpr = (multiplyExpr +
Optional(divOp + multiplyExpr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
addExpr = (divideExpr + Optional(plusOp + divideExpr)
).setParseAction(compileBinaryOperation).ignore(sphinxComment)
subtractExpr = (addExpr + Optional(minusOp + addExpr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
equalityExpr = (subtractExpr +
Optional(eqOp + subtractExpr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
inequalityExpr = (equalityExpr +
Optional(neOp + equalityExpr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
comparisonExpr = (inequalityExpr +
Optional(compOp + inequalityExpr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
andExpr = (comparisonExpr + Optional(andOp + comparisonExpr)
).setParseAction(compileBinaryOperation).ignore(sphinxComment)
orExpr = (andExpr + Optional(orOp + andExpr)
).setParseAction(compileBinaryOperation).ignore(sphinxComment)
formulaExpr = (orExpr + Optional(formulaOp + orExpr)).setParseAction(
compileBinaryOperation).ignore(sphinxComment)
withExpr = (Optional(withOp + Suppress(lParen) + expr + Suppress(rParen)) +
ZeroOrMore(
(ncName + Optional(tagOp + Optional(ncName)) + varAssign +
expr + Suppress(Literal(";"))).setParseAction(
compileVariableAssignment).ignore(sphinxComment)) +
formulaExpr).setParseAction(compileWith)
#parsedExpr = withExpr
#parsedExpr.setName("parsedExpr").setDebug(debugParsing)
#expr << parsedExpr
expr << withExpr
expr.setName("expr").setDebug(debugParsing)
annotation = (annotationName + Optional(
Suppress(lParen) + Optional(delimitedList(expr)) +
Suppress(rParen))).setParseAction(compileAnnotation).ignore(
sphinxComment).setName("annotation").setDebug(debugParsing)
constant = (Suppress(Keyword("constant")) + ncName +
Optional(tagOp + Optional(ncName)) + varAssign +
expr).setParseAction(compileConstant).ignore(sphinxComment)
functionDeclaration = (
(Keyword("function") | Keyword("macro")) + ncName + lParen +
Optional(delimitedList(ncName)) + rParen +
expr).setParseAction(compileFunctionDeclaration).ignore(sphinxComment)
message = (Suppress(Keyword("message")) +
expr).setParseAction(compileMessage)
preconditionDeclaration = (
Suppress(Keyword("precondition")) + ncName + expr + Optional(
Keyword("otherwise") + Keyword("raise") + ncName +
Optional(severity) + Optional(message))
).setParseAction(compilePreconditionDeclaration).ignore(sphinxComment)
assignedExpr = (ncName + Optional(tagOp + Optional(ncName)) + varAssign +
expr + Suppress(Literal(";"))).setParseAction(
compileVariableAssignment).ignore(sphinxComment)
precondition = (
Suppress(Keyword("require")) +
delimitedList(ncName)).setParseAction(compilePrecondition).ignore(
sphinxComment).setName("precondition").setDebug(debugParsing)
formulaRule = (Optional(precondition) + Keyword("formula") + ncName +
Optional(severity) + Optional((Keyword("bind") + expr)) +
ZeroOrMore(assignedExpr) + expr + Optional(message)
).setParseAction(compileFormulaRule).ignore(sphinxComment)
reportRule = (Optional(precondition) + Keyword("report") + ncName +
Optional(severity) + ZeroOrMore(assignedExpr) + expr +
Optional(message)
).setParseAction(compileReportRule).ignore(sphinxComment)
validationRule = (Optional(precondition) + Keyword("raise") + ncName +
Optional(severity) + ZeroOrMore(assignedExpr) + expr +
Optional(message)).setParseAction(
compileValidationRule).ignore(sphinxComment)
ruleBase = (
Optional(precondition) + Suppress(Keyword("rule-base")) + ZeroOrMore(
(Suppress(Keyword("transform")) +
(Keyword("namespace") + expr + Suppress(Keyword("to")) + expr) |
(Keyword("qname") + expr + Suppress(Keyword("to")) +
expr)).setParseAction(compileTransform))
).setParseAction(compileRuleBase).ignore(sphinxComment).setName(
"ruleBase").setDebug(debugParsing)
sphinxProg = (
ZeroOrMore(namespaceDeclaration | sphinxComment) +
ZeroOrMore(annotationDeclaration | annotation | constant
| preconditionDeclaration | packageDeclaration
| functionDeclaration | ruleBase | formulaRule | reportRule
| validationRule | sphinxComment)) + StringEnd()
sphinxProg.ignore(sphinxComment)
startedAt = time.time()
cntlr.modelManager.showStatus(_("initializing sphinx grammar"))
sphinxProg.parseString("// force initialization\n", parseAll=True)
from arelle.Locale import format_string
logMessage(
"INFO", "info",
format_string(cntlr.modelManager.locale,
_("Sphinx grammar initialized in %.2f secs"),
time.time() - startedAt))
isGrammarCompiled = True
return sphinxProg