cppcheck/addons/cppcheckdata.py

1356 lines
45 KiB
Python
Raw Normal View History

"""
cppcheckdata
This is a Python module that helps you access Cppcheck dump data.
License: No restrictions, use this as you need.
"""
import argparse
2019-06-24 19:28:52 +02:00
import json
2019-04-09 20:36:24 +02:00
import sys
from xml.etree import ElementTree
from fnmatch import fnmatch
EXIT_CODE = 0
current_dumpfile_suppressions = []
2021-07-08 22:03:27 +02:00
def _load_location(location, element):
"""Load location from element/dict"""
location.file = element.get('file')
line = element.get('line')
if line is None:
line = element.get('linenr')
if line is None:
line = '0'
location.linenr = int(line)
location.column = int(element.get('column', '0'))
class Location:
"""Utility location class"""
file = None
linenr = None
column = None
def __init__(self, element):
_load_location(self, element)
class Directive:
"""
Directive class. Contains information about each preprocessor directive in the source code.
Attributes:
str The directive line, with all C or C++ comments removed
file Name of (possibly included) file where directive is defined
linenr Line number in (possibly included) file where directive is defined
To iterate through all directives use such code:
@code
data = cppcheckdata.parsedump(...)
for cfg in data.configurations:
for directive in cfg.directives:
print(directive.str)
@endcode
"""
str = None
file = None
linenr = None
2021-07-08 22:03:27 +02:00
column = None
def __init__(self, element):
self.str = element.get('str')
2021-07-08 22:03:27 +02:00
_load_location(self, element)
def __repr__(self):
attrs = ["str", "file", "linenr"]
return "{}({})".format(
"Directive",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
2021-07-08 22:03:27 +02:00
class MacroUsage:
"""
Tracks preprocessor macro usage
"""
name = None # Macro name
file = None
linenr = None
column = None
2021-07-22 19:51:31 +02:00
usefile = None
uselinenr = None
usecolumn = None
2021-07-08 22:03:27 +02:00
def __init__(self, element):
self.name = element.get('name')
_load_location(self, element)
2021-07-22 19:51:31 +02:00
self.usefile = element.get('usefile')
self.useline = element.get('useline')
self.usecolumn = element.get('usecolumn')
2021-07-08 22:03:27 +02:00
def __repr__(self):
2021-07-22 19:51:31 +02:00
attrs = ["name", "file", "linenr", "column", "usefile", "useline", "usecolumn"]
2021-07-08 22:03:27 +02:00
return "{}({})".format(
2021-07-22 19:51:31 +02:00
"MacroUsage",
2021-07-08 22:03:27 +02:00
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
2021-07-22 08:46:28 +02:00
class PreprocessorIfCondition:
"""
Information about #if/#elif conditions
"""
file = None
linenr = None
column = None
E = None
result = None
def __init__(self, element):
_load_location(self, element)
self.E = element.get('E')
self.result = int(element.get('result'))
def __repr__(self):
attrs = ["file", "linenr", "column", "E", "result"]
return "{}({})".format(
"PreprocessorIfCondition",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
2017-06-04 22:51:48 +02:00
2017-04-16 09:11:20 +02:00
class ValueType:
"""
ValueType class. Contains (promoted) type information for each node in the AST.
"""
type = None
sign = None
bits = 0
2017-04-16 09:11:20 +02:00
constness = 0
pointer = 0
typeScopeId = None
typeScope = None
originalTypeName = None
def __init__(self, element):
self.type = element.get('valueType-type')
self.sign = element.get('valueType-sign')
bits = element.get('valueType-bits')
if bits:
self.bits = int(bits)
2017-04-16 09:11:20 +02:00
self.typeScopeId = element.get('valueType-typeScope')
self.originalTypeName = element.get('valueType-originalTypeName')
constness = element.get('valueType-constness')
if constness:
self.constness = int(constness)
else:
self.constness = 0
pointer = element.get('valueType-pointer')
if pointer:
self.pointer = int(pointer)
else:
self.pointer = 0
def __repr__(self):
attrs = ["type", "sign", "bits", "typeScopeId", "originalTypeName",
2021-09-06 20:13:15 +02:00
"constness", "pointer"]
return "{}({})".format(
"ValueType",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
2017-04-16 09:11:20 +02:00
def setId(self, IdMap):
self.typeScope = IdMap[self.typeScopeId]
2017-04-16 12:13:30 +02:00
def isIntegral(self):
2017-06-04 22:51:48 +02:00
return self.type in {'bool', 'char', 'short', 'int', 'long', 'long long'}
2017-04-16 12:13:30 +02:00
def isFloat(self):
2017-06-04 22:51:48 +02:00
return self.type in {'float', 'double', 'long double'}
def isEnum(self):
return self.typeScope and self.typeScope.type == "Enum"
class Token:
"""
Token class. Contains information about each token in the source code.
The CppcheckData.tokenlist is a list of Token items
C++ class: https://cppcheck.sourceforge.io/devinfo/doxyoutput/classToken.html
Attributes:
str Token string
next Next token in tokenlist. For last token, next is None.
previous Previous token in tokenlist. For first token, previous is None.
link Linked token in tokenlist. Each '(', '[' and '{' are linked to the
corresponding '}', ']' and ')'. For templates, the '<' is linked to
the corresponding '>'.
scope Scope information for this token. See the Scope class.
isName Is this token a symbol name
isNumber Is this token a number, for example 123, 12.34
isInt Is this token a int value such as 1234
2021-09-06 20:13:15 +02:00
isFloat Is this token a float value such as 12.34
isString Is this token a string literal such as "hello"
strlen string length for string literal
isChar Is this token a char literal such as 'x'
isOp Is this token a operator
isArithmeticalOp Is this token a arithmetic operator
isAssignmentOp Is this token a assignment operator
isComparisonOp Is this token a comparison operator
2016-07-25 18:03:44 +02:00
isLogicalOp Is this token a logical operator: && ||
2016-12-09 22:31:47 +01:00
isUnsigned Is this token a unsigned type
isSigned Is this token a signed type
isExpandedMacro Is this token a expanded macro token
isSplittedVarDeclComma Is this a comma changed to semicolon in a splitted variable declaration ('int a,b;' => 'int a; int b;')
isSplittedVarDeclEq Is this a '=' changed to semicolon in a splitted variable declaration ('int a=5;' => 'int a; a=5;')
2021-07-07 13:34:55 +02:00
isImplicitInt Is this token an implicit "int"?
varId varId for token, each variable has a unique non-zero id
variable Variable information for this token. See the Variable class.
function If this token points at a function call, this attribute has the Function
information. See the Function class.
values Possible/Known values of token
impossible_values Impossible values of token
2017-04-15 21:55:07 +02:00
valueType type information
typeScope type scope (token->type()->classScope)
astParent ast parent
astOperand1 ast operand1
astOperand2 ast operand2
file file name
linenr line number
column column
To iterate through all tokens use such code:
@code
data = cppcheckdata.parsedump(...)
for cfg in data.configurations:
code = ''
for token in cfg.tokenlist:
code = code + token.str + ' '
print(code)
@endcode
"""
2014-07-21 16:10:04 +02:00
Id = None
str = None
next = None
previous = None
2015-06-21 13:26:32 +02:00
linkId = None
link = None
2014-07-21 16:10:04 +02:00
scopeId = None
scope = None
2015-06-21 13:26:32 +02:00
isName = False
isNumber = False
isInt = False
isFloat = False
isString = False
2014-07-21 16:10:04 +02:00
strlen = None
2015-06-21 13:26:32 +02:00
isChar = False
isOp = False
isArithmeticalOp = False
isAssignmentOp = False
isComparisonOp = False
isLogicalOp = False
2016-12-09 22:31:47 +01:00
isUnsigned = False
isSigned = False
isExpandedMacro = False
isSplittedVarDeclComma = False
isSplittedVarDeclEq = False
2021-07-07 13:34:55 +02:00
isImplicitInt = False
2014-07-21 16:10:04 +02:00
varId = None
variableId = None
variable = None
functionId = None
function = None
valuesId = None
values = None
impossible_values = None
2017-04-15 21:55:07 +02:00
valueType = None
2015-08-21 10:55:19 +02:00
typeScopeId = None
typeScope = None
2014-07-21 16:10:04 +02:00
astParentId = None
astParent = None
astOperand1Id = None
astOperand1 = None
astOperand2Id = None
astOperand2 = None
2015-06-21 13:26:32 +02:00
2014-07-21 16:10:04 +02:00
file = None
linenr = None
column = None
def __init__(self, element):
2014-07-21 16:10:04 +02:00
self.Id = element.get('id')
self.str = element.get('str')
self.next = None
self.previous = None
self.scopeId = element.get('scope')
self.scope = None
type = element.get('type')
2014-07-21 12:44:53 +02:00
if type == 'name':
self.isName = True
2016-12-09 22:31:47 +01:00
if element.get('isUnsigned'):
self.isUnsigned = True
if element.get('isSigned'):
self.isSigned = True
2014-07-21 12:44:53 +02:00
elif type == 'number':
self.isNumber = True
if element.get('isInt'):
self.isInt = True
elif element.get('isFloat'):
self.isFloat = True
elif type == 'string':
self.isString = True
self.strlen = int(element.get('strlen'))
elif type == 'char':
self.isChar = True
elif type == 'op':
self.isOp = True
if element.get('isArithmeticalOp'):
self.isArithmeticalOp = True
elif element.get('isAssignmentOp'):
self.isAssignmentOp = True
elif element.get('isComparisonOp'):
self.isComparisonOp = True
elif element.get('isLogicalOp'):
self.isLogicalOp = True
if element.get('isExpandedMacro'):
self.isExpandedMacro = True
if element.get('isSplittedVarDeclComma'):
self.isSplittedVarDeclComma = True
if element.get('isSplittedVarDeclEq'):
self.isSplittedVarDeclEq = True
2021-07-07 13:34:55 +02:00
if element.get('isImplicitInt'):
self.isImplicitInt = True
2014-07-21 16:10:04 +02:00
self.linkId = element.get('link')
self.link = None
if element.get('varId'):
self.varId = int(element.get('varId'))
2014-07-21 16:10:04 +02:00
self.variableId = element.get('variable')
self.variable = None
self.functionId = element.get('function')
self.function = None
self.valuesId = element.get('values')
self.values = None
2017-04-16 09:11:20 +02:00
if element.get('valueType-type'):
self.valueType = ValueType(element)
else:
self.valueType = None
self.typeScopeId = element.get('type-scope')
self.typeScope = None
2014-07-21 16:10:04 +02:00
self.astParentId = element.get('astParent')
self.astParent = None
self.astOperand1Id = element.get('astOperand1')
2014-07-21 16:10:04 +02:00
self.astOperand1 = None
self.astOperand2Id = element.get('astOperand2')
2014-07-21 16:10:04 +02:00
self.astOperand2 = None
2021-07-08 22:03:27 +02:00
_load_location(self, element)
def __repr__(self):
attrs = ["Id", "str", "scopeId", "isName", "isUnsigned", "isSigned",
"isNumber", "isInt", "isFloat", "isString", "strlen",
"isChar", "isOp", "isArithmeticalOp", "isComparisonOp",
"isLogicalOp", "isExpandedMacro", "isSplittedVarDeclComma",
2021-07-07 13:34:55 +02:00
"isSplittedVarDeclEq", "isImplicitInt", "linkId", "varId",
"variableId", "functionId", "valuesId", "valueType",
"typeScopeId", "astParentId", "astOperand1Id", "file",
"linenr", "column"]
return "{}({})".format(
"Token",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
def setId(self, IdMap):
2014-07-21 16:10:04 +02:00
self.scope = IdMap[self.scopeId]
self.link = IdMap[self.linkId]
self.variable = IdMap[self.variableId]
self.function = IdMap[self.functionId]
self.values = []
self.impossible_values = []
if IdMap[self.valuesId]:
for v in IdMap[self.valuesId]:
if v.isImpossible():
self.impossible_values.append(v)
else:
self.values.append(v)
2021-08-22 05:55:58 +02:00
v.setId(IdMap)
self.typeScope = IdMap[self.typeScopeId]
2014-07-21 16:10:04 +02:00
self.astParent = IdMap[self.astParentId]
self.astOperand1 = IdMap[self.astOperand1Id]
self.astOperand2 = IdMap[self.astOperand2Id]
2017-04-16 09:11:20 +02:00
if self.valueType:
self.valueType.setId(IdMap)
2014-07-21 16:10:04 +02:00
def getValue(self, v):
"""
Get value if it exists
Returns None if it doesn't exist
"""
if not self.values:
return None
for value in self.values:
if value.intvalue == v:
return value
return None
2020-11-13 07:21:34 +01:00
def getKnownIntValue(self):
"""
If token has a known int value then return that.
Otherwise returns None
"""
if not self.values:
return None
for value in self.values:
if value.valueKind == 'known':
return value.intvalue
return None
2021-07-07 23:00:12 +02:00
def isUnaryOp(self, op):
return self.astOperand1 and (self.astOperand2 is None) and self.str == op
def isBinaryOp(self):
return self.astOperand1 and self.astOperand2
class Scope:
"""
Scope. Information about global scope, function scopes, class scopes, inner scopes, etc.
C++ class: https://cppcheck.sourceforge.io/devinfo/doxyoutput/classScope.html
Attributes
bodyStart The { Token for this scope
bodyEnd The } Token for this scope
className Name of this scope.
For a function scope, this is the function name;
For a class scope, this is the class name.
2019-07-13 15:09:50 +02:00
function If this scope belongs at a function call, this attribute
has the Function information. See the Function class.
type Type of scope: Global, Function, Class, If, While
"""
2014-07-21 16:10:04 +02:00
Id = None
bodyStartId = None
bodyStart = None
bodyEndId = None
bodyEnd = None
2014-07-21 16:10:04 +02:00
className = None
2019-07-13 15:09:50 +02:00
functionId = None
function = None
2017-04-08 19:33:26 +02:00
nestedInId = None
nestedIn = None
2014-07-21 16:10:04 +02:00
type = None
isExecutable = None
2020-11-13 21:03:02 +01:00
varlistId = None
varlist = None
2014-07-21 16:10:04 +02:00
def __init__(self, element):
self.Id = element.get('id')
self.className = element.get('className')
2019-07-13 15:09:50 +02:00
self.functionId = element.get('function')
self.function = None
self.bodyStartId = element.get('bodyStart')
self.bodyStart = None
self.bodyEndId = element.get('bodyEnd')
self.bodyEnd = None
self.nestedInId = element.get('nestedIn')
2014-07-21 16:10:04 +02:00
self.nestedIn = None
self.type = element.get('type')
self.isExecutable = (self.type in ('Function', 'If', 'Else', 'For', 'While', 'Do',
'Switch', 'Try', 'Catch', 'Unconditional', 'Lambda'))
2020-11-13 21:03:02 +01:00
self.varlistId = list()
self.varlist = list()
def __repr__(self):
attrs = ["Id", "className", "functionId", "bodyStartId", "bodyEndId",
"nestedInId", "nestedIn", "type", "isExecutable"]
return "{}({})".format(
"Scope",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
def setId(self, IdMap):
self.bodyStart = IdMap[self.bodyStartId]
self.bodyEnd = IdMap[self.bodyEndId]
self.nestedIn = IdMap[self.nestedInId]
2019-07-13 15:09:50 +02:00
self.function = IdMap[self.functionId]
2020-11-13 21:03:02 +01:00
for v in self.varlistId:
value = IdMap.get(v)
if value:
self.varlist.append(value)
class Function:
"""
Information about a function
C++ class:
https://cppcheck.sourceforge.io/devinfo/doxyoutput/classFunction.html
2019-07-13 15:09:50 +02:00
Attributes
argument Argument list
token Token in function implementation
2019-07-13 15:09:50 +02:00
tokenDef Token in function definition
isVirtual Is this function is virtual
isImplicitlyVirtual Is this function is virtual this in the base classes
2021-07-18 21:18:07 +02:00
isInlineKeyword Is inline keyword used
isStatic Is this function static?
"""
2014-07-21 16:10:04 +02:00
Id = None
argument = None
argumentId = None
token = None
tokenId = None
tokenDef = None
tokenDefId = None
2015-07-28 14:18:58 +02:00
name = None
2018-04-30 16:52:51 +02:00
type = None
isVirtual = None
isImplicitlyVirtual = None
2021-07-18 21:18:07 +02:00
isInlineKeyword = None
isStatic = None
nestedIn = None
2014-07-21 16:10:04 +02:00
def __init__(self, element, nestedIn):
2014-07-21 16:10:04 +02:00
self.Id = element.get('id')
self.tokenId = element.get('token')
self.tokenDefId = element.get('tokenDef')
2015-07-28 14:18:58 +02:00
self.name = element.get('name')
2018-04-30 16:52:51 +02:00
self.type = element.get('type')
2021-07-18 21:18:07 +02:00
self.isImplicitlyVirtual = element.get('isImplicitlyVirtual', 'false') == 'true'
self.isVirtual = element.get('isVirtual', 'false') == 'true'
self.isInlineKeyword = element.get('isInlineKeyword', 'false') == 'true'
self.isStatic = element.get('isStatic', 'false') == 'true'
self.nestedIn = nestedIn
2014-07-21 16:10:04 +02:00
self.argument = {}
self.argumentId = {}
2014-07-21 16:10:04 +02:00
def __repr__(self):
attrs = ["Id", "tokenId", "tokenDefId", "name", "type", "isVirtual",
2021-07-18 21:18:07 +02:00
"isImplicitlyVirtual", "isInlineKeyword", "isStatic", "argumentId"]
return "{}({})".format(
"Function",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
def setId(self, IdMap):
2014-12-22 16:03:43 +01:00
for argnr, argid in self.argumentId.items():
self.argument[argnr] = IdMap[argid]
self.token = IdMap.get(self.tokenId, None)
self.tokenDef = IdMap[self.tokenDefId]
2014-07-21 16:10:04 +02:00
class Variable:
"""
Information about a variable
C++ class:
https://cppcheck.sourceforge.io/devinfo/doxyoutput/classVariable.html
Attributes:
2018-06-19 08:50:32 +02:00
nameToken Name token in variable declaration
typeStartToken Start token of variable declaration
typeEndToken End token of variable declaration
access Global/Local/Namespace/Public/Protected/Public/Throw/Argument
2018-06-19 08:50:32 +02:00
scope Variable scope
isArgument Is this variable a function argument?
isArray Is this variable an array?
isClass Is this variable a class or struct?
isConst Is this variable a const variable?
isGlobal Is this variable a global variable?
2017-04-17 07:45:27 +02:00
isExtern Is this variable an extern variable?
isLocal Is this variable a local variable?
isPointer Is this variable a pointer
isReference Is this variable a reference
isStatic Is this variable static?
2018-04-12 20:23:50 +02:00
constness Variable constness (same encoding as ValueType::constness)
"""
2014-07-21 16:10:04 +02:00
Id = None
nameTokenId = None
nameToken = None
typeStartTokenId = None
2014-07-21 16:10:04 +02:00
typeStartToken = None
typeEndTokenId = None
typeEndToken = None
access = None
2018-06-19 08:50:32 +02:00
scopeId = None
scope = None
isArgument = False
isArray = False
isClass = False
isConst = False
2017-04-17 07:45:27 +02:00
isExtern = False
isGlobal = False
isLocal = False
isPointer = False
isReference = False
isStatic = False
2018-04-12 20:23:50 +02:00
constness = 0
def __init__(self, element):
2014-07-21 16:10:04 +02:00
self.Id = element.get('id')
self.nameTokenId = element.get('nameToken')
self.nameToken = None
self.typeStartTokenId = element.get('typeStartToken')
2014-07-21 16:10:04 +02:00
self.typeStartToken = None
self.typeEndTokenId = element.get('typeEndToken')
self.typeEndToken = None
self.access = element.get('access')
2018-06-19 08:50:32 +02:00
self.scopeId = element.get('scope')
self.scope = None
self.isArgument = (self.access and self.access == 'Argument')
self.isArray = element.get('isArray') == 'true'
self.isClass = element.get('isClass') == 'true'
self.isConst = element.get('isConst') == 'true'
self.isGlobal = (self.access and self.access == 'Global')
2017-04-17 07:45:27 +02:00
self.isExtern = element.get('isExtern') == 'true'
self.isLocal = (self.access and self.access == 'Local')
self.isPointer = element.get('isPointer') == 'true'
self.isReference = element.get('isReference') == 'true'
self.isStatic = element.get('isStatic') == 'true'
2018-04-12 20:23:50 +02:00
self.constness = element.get('constness')
if self.constness:
self.constness = int(self.constness)
def __repr__(self):
attrs = ["Id", "nameTokenId", "typeStartTokenId", "typeEndTokenId",
"access", "scopeId", "isArgument", "isArray", "isClass",
"isConst", "isGlobal", "isExtern", "isLocal", "isPointer",
"isReference", "isStatic", "constness", ]
return "{}({})".format(
"Variable",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
def setId(self, IdMap):
self.nameToken = IdMap[self.nameTokenId]
self.typeStartToken = IdMap[self.typeStartTokenId]
self.typeEndToken = IdMap[self.typeEndTokenId]
2018-06-19 08:50:32 +02:00
self.scope = IdMap[self.scopeId]
2021-07-07 10:58:13 +02:00
class TypedefInfo:
"""
TypedefInfo class -- information about typedefs
"""
name = None
used = None
2021-07-08 22:03:27 +02:00
file = None
linenr = None
column = None
2021-07-07 10:58:13 +02:00
def __init__(self, element):
self.name = element.get('name')
2021-07-08 22:03:27 +02:00
_load_location(self, element)
2021-07-07 10:58:13 +02:00
self.used = (element.get('used') == '1')
class Value:
"""
Value class
Attributes:
intvalue integer value
tokvalue token value
floatvalue float value
containerSize container size
condition condition where this Value comes from
valueKind 'known' or 'possible'
inconclusive Is value inconclusive?
"""
intvalue = None
tokvalue = None
floatvalue = None
containerSize = None
condition = None
valueKind = None
inconclusive = False
def isKnown(self):
return self.valueKind and self.valueKind == 'known'
def isPossible(self):
return self.valueKind and self.valueKind == 'possible'
def isImpossible(self):
return self.valueKind and self.valueKind == 'impossible'
def __init__(self, element):
self.intvalue = element.get('intvalue')
if self.intvalue:
self.intvalue = int(self.intvalue)
2021-08-22 05:55:58 +02:00
self._tokvalueId = element.get('tokvalue')
self.floatvalue = element.get('floatvalue')
self.containerSize = element.get('container-size')
self.condition = element.get('condition-line')
if self.condition:
self.condition = int(self.condition)
if element.get('known'):
self.valueKind = 'known'
elif element.get('possible'):
self.valueKind = 'possible'
elif element.get('impossible'):
self.valueKind = 'impossible'
if element.get('inconclusive'):
self.inconclusive = True
2021-08-22 05:55:58 +02:00
def setId(self, IdMap):
self.tokvalue = IdMap.get(self._tokvalueId)
def __repr__(self):
attrs = ["intvalue", "tokvalue", "floatvalue", "containerSize",
"condition", "valueKind", "inconclusive"]
return "{}({})".format(
"Value",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
class ValueFlow:
"""
ValueFlow::Value class
Each possible value has a ValueFlow::Value item.
Each ValueFlow::Value either has a intvalue or tokvalue
C++ class:
https://cppcheck.sourceforge.io/devinfo/doxyoutput/classValueFlow_1_1Value.html
Attributes:
values Possible values
"""
Id = None
values = None
def __init__(self, element):
2014-07-21 16:10:04 +02:00
self.Id = element.get('id')
self.values = []
def __repr__(self):
attrs = ["Id", "values"]
return "{}({})".format(
"ValueFlow",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
class Suppression:
"""
Suppression class
This class contains a suppression entry to suppress a warning.
Attributes
errorId The id string of the error to suppress, can be a wildcard
fileName The name of the file to suppress warnings for, can include wildcards
lineNumber The number of the line to suppress warnings from, can be 0 to represent any line
symbolName The name of the symbol to match warnings for, can include wildcards
"""
errorId = None
fileName = None
lineNumber = None
symbolName = None
def __init__(self, element):
self.errorId = element.get('errorId')
self.fileName = element.get('fileName')
self.lineNumber = element.get('lineNumber')
self.symbolName = element.get('symbolName')
def __repr__(self):
attrs = ['errorId' , "fileName", "lineNumber", "symbolName"]
return "{}({})".format(
"Suppression",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
def isMatch(self, file, line, message, errorId):
if ((self.fileName is None or fnmatch(file, self.fileName))
and (self.lineNumber is None or int(line) == int(self.lineNumber))
and (self.symbolName is None or fnmatch(message, '*'+self.symbolName+'*'))
and fnmatch(errorId, self.errorId)):
return True
return False
class Configuration:
"""
Configuration class
This class contains the directives, tokens, scopes, functions,
variables, value flows, and suppressions for one configuration.
Attributes:
name Name of the configuration, "" for default
directives List of Directive items
2021-07-08 22:03:27 +02:00
macro_usage List of used macros
2021-07-22 08:46:28 +02:00
preprocessor_if_conditions List of preprocessor if conditions that was evaluated during preprocessing
tokenlist List of Token items
scopes List of Scope items
functions List of Function items
variables List of Variable items
2016-07-25 18:03:44 +02:00
valueflow List of ValueFlow values
standards List of Standards values
"""
name = ''
directives = []
2021-07-08 22:03:27 +02:00
macro_usage = []
2021-07-22 08:46:28 +02:00
preprocessor_if_conditions = []
tokenlist = []
2014-07-21 16:10:04 +02:00
scopes = []
functions = []
variables = []
2021-07-07 10:58:13 +02:00
typedefInfo = []
valueflow = []
standards = None
2021-12-11 12:42:15 +01:00
clang_warnings = []
def __init__(self, name):
self.name = name
self.directives = []
2021-07-08 22:03:27 +02:00
self.macro_usage = []
2021-07-22 08:46:28 +02:00
self.preprocessor_if_conditions = []
self.tokenlist = []
2014-07-21 16:10:04 +02:00
self.scopes = []
self.functions = []
self.variables = []
2021-07-07 10:58:13 +02:00
self.typedefInfo = []
self.valueflow = []
self.standards = Standards()
2021-12-11 12:42:15 +01:00
self.clang_warnings = []
def set_tokens_links(self):
"""Set next/previous links between tokens."""
prev = None
for token in self.tokenlist:
token.previous = prev
if prev:
prev.next = token
prev = token
def set_id_map(self, arguments):
2021-07-11 17:23:33 +02:00
IdMap = {None: None, '0': None, '00000000': None, '0000000000000000': None, '0x0': None}
for token in self.tokenlist:
IdMap[token.Id] = token
for scope in self.scopes:
IdMap[scope.Id] = scope
for function in self.functions:
IdMap[function.Id] = function
for variable in self.variables:
IdMap[variable.Id] = variable
for variable in arguments:
IdMap[variable.Id] = variable
for values in self.valueflow:
IdMap[values.Id] = values.values
for token in self.tokenlist:
token.setId(IdMap)
for scope in self.scopes:
scope.setId(IdMap)
for function in self.functions:
function.setId(IdMap)
for variable in self.variables:
variable.setId(IdMap)
for variable in arguments:
variable.setId(IdMap)
def setIdMap(self, functions_arguments):
"""Set relationships between objects stored in this configuration.
:param functions_arguments: List of Variable objects which are function arguments
"""
self.set_tokens_links()
self.set_id_map(functions_arguments)
class Platform:
"""
Platform class
This class contains type sizes
Attributes:
name Name of the platform
char_bit CHAR_BIT value
short_bit SHORT_BIT value
int_bit INT_BIT value
long_bit LONG_BIT value
long_long_bit LONG_LONG_BIT value
pointer_bit POINTER_BIT value
"""
name = ''
char_bit = 0
short_bit = 0
int_bit = 0
long_bit = 0
long_long_bit = 0
pointer_bit = 0
def __init__(self, platformnode):
self.name = platformnode.get('name')
self.char_bit = int(platformnode.get('char_bit'))
self.short_bit = int(platformnode.get('short_bit'))
self.int_bit = int(platformnode.get('int_bit'))
self.long_bit = int(platformnode.get('long_bit'))
self.long_long_bit = int(platformnode.get('long_long_bit'))
self.pointer_bit = int(platformnode.get('pointer_bit'))
def __repr__(self):
attrs = ["name", "char_bit", "short_bit", "int_bit",
"long_bit", "long_long_bit", "pointer_bit"]
return "{}({})".format(
"Platform",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
class Standards:
"""
Standards class
This class contains versions of standards that were used for the cppcheck
Attributes:
c C Standard used
cpp C++ Standard used
posix If Posix was used
"""
c = ""
cpp = ""
posix = False
def set_c(self, node):
self.c = node.get("version")
def set_cpp(self, node):
self.cpp = node.get("version")
def set_posix(self, node):
self.posix = node.get("posix") is not None
def __repr__(self):
attrs = ["c", "cpp", "posix"]
return "{}({})".format(
"Standards",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
class CppcheckData:
"""
Class that makes cppcheck dump data available
Contains a list of Configuration instances
Attributes:
filename Path to Cppcheck dump file
rawTokens List of rawToken elements
suppressions List of Suppressions
files Source files for elements occurred in this configuration
To iterate through all configurations use such code:
@code
data = cppcheckdata.parsedump(...)
for cfg in data.configurations:
print('cfg: ' + cfg.name)
@endcode
To iterate through all tokens in each configuration use such code:
@code
data = cppcheckdata.parsedump(...)
for cfg in data.configurations:
print('cfg: ' + cfg.name)
code = ''
for token in cfg.tokenlist:
code = code + token.str + ' '
print(' ' + code)
@endcode
To iterate through all scopes (functions, types, etc) use such code:
@code
data = cppcheckdata.parsedump(...)
for cfg in data.configurations:
print('cfg: ' + cfg.name)
for scope in cfg.scopes:
print(' type:' + scope.type + ' name:' + scope.className)
@endcode
"""
def __init__(self, filename):
"""
:param filename: Path to Cppcheck dump file
"""
self.filename = filename
self.rawTokens = []
self.platform = None
self.suppressions = []
self.files = []
platform_done = False
rawtokens_done = False
suppressions_done = False
# Parse general configuration options from <dumps> node
# We intentionally don't clean node resources here because we
# want to serialize in memory only small part of the XML tree.
for event, node in ElementTree.iterparse(self.filename, events=('start', 'end')):
if platform_done and rawtokens_done and suppressions_done:
break
if node.tag == 'platform' and event == 'start':
self.platform = Platform(node)
platform_done = True
elif node.tag == 'rawtokens' and event == 'end':
for rawtokens_node in node:
if rawtokens_node.tag == 'file':
self.files.append(rawtokens_node.get('name'))
elif rawtokens_node.tag == 'tok':
tok = Token(rawtokens_node)
tok.file = self.files[int(rawtokens_node.get('fileIndex'))]
self.rawTokens.append(tok)
rawtokens_done = True
elif node.tag == 'suppressions' and event == 'end':
for suppressions_node in node:
self.suppressions.append(Suppression(suppressions_node))
suppressions_done = True
global current_dumpfile_suppressions
current_dumpfile_suppressions = self.suppressions
# Set links between rawTokens.
for i in range(len(self.rawTokens)-1):
self.rawTokens[i+1].previous = self.rawTokens[i]
self.rawTokens[i].next = self.rawTokens[i+1]
@property
def configurations(self):
"""
Return the list of all available Configuration objects.
"""
return list(self.iterconfigurations())
def iterconfigurations(self):
"""
Create and return iterator for the available Configuration objects.
The iterator loops over all Configurations in the dump file tree, in document order.
"""
cfg = None
cfg_arguments = [] # function arguments for Configuration node initialization
cfg_function = None
cfg_valueflow = None
2020-11-13 21:03:02 +01:00
# Iterating <varlist> in a <scope>.
iter_scope_varlist = False
2021-07-07 10:58:13 +02:00
# Iterating <typedef-info>
iter_typedef_info = False
# Use iterable objects to traverse XML tree for dump files incrementally.
# Iterative approach is required to avoid large memory consumption.
# Calling .clear() is necessary to let the element be garbage collected.
for event, node in ElementTree.iterparse(self.filename, events=('start', 'end')):
# Serialize new configuration node
if node.tag == 'dump':
if event == 'start':
cfg = Configuration(node.get('cfg'))
continue
elif event == 'end':
cfg.setIdMap(cfg_arguments)
yield cfg
cfg = None
cfg_arguments = []
2021-12-11 12:42:15 +01:00
elif node.tag == 'clang-warning' and event == 'start':
cfg.clang_warnings.append({'file': node.get('file'),
'line': int(node.get('line')),
'column': int(node.get('column')),
'message': node.get('message')})
# Parse standards
elif node.tag == "standards" and event == 'start':
continue
elif node.tag == 'c' and event == 'start':
cfg.standards.set_c(node)
elif node.tag == 'cpp' and event == 'start':
cfg.standards.set_cpp(node)
elif node.tag == 'posix' and event == 'start':
cfg.standards.set_posix(node)
# Parse directives list
elif node.tag == 'directive' and event == 'start':
cfg.directives.append(Directive(node))
2017-04-09 10:11:54 +02:00
2021-07-08 22:03:27 +02:00
# Parse macro usage
elif node.tag == 'macro' and event == 'start':
cfg.macro_usage.append(MacroUsage(node))
2021-07-22 08:46:28 +02:00
# Preprocessor #if/#elif condition
elif node.tag == "if-cond" and event == 'start':
cfg.preprocessor_if_conditions.append(PreprocessorIfCondition(node))
# Parse tokens
elif node.tag == 'tokenlist' and event == 'start':
continue
elif node.tag == 'token' and event == 'start':
cfg.tokenlist.append(Token(node))
# Parse scopes
elif node.tag == 'scopes' and event == 'start':
continue
elif node.tag == 'scope' and event == 'start':
cfg.scopes.append(Scope(node))
elif node.tag == 'varlist':
if event == 'start':
2020-11-13 21:03:02 +01:00
iter_scope_varlist = True
elif event == 'end':
2020-11-13 21:03:02 +01:00
iter_scope_varlist = False
# Parse functions
elif node.tag == 'functionList' and event == 'start':
continue
elif node.tag == 'function':
if event == 'start':
cfg_function = Function(node, cfg.scopes[-1])
continue
elif event == 'end':
cfg.functions.append(cfg_function)
cfg_function = None
# Parse function arguments
elif node.tag == 'arg' and event == 'start':
arg_nr = int(node.get('nr'))
arg_variable_id = node.get('variable')
cfg_function.argumentId[arg_nr] = arg_variable_id
# Parse variables
elif node.tag == 'var' and event == 'start':
2020-11-13 21:03:02 +01:00
if iter_scope_varlist:
cfg.scopes[-1].varlistId.append(node.get('id'))
else:
var = Variable(node)
if var.nameTokenId:
cfg.variables.append(var)
else:
cfg_arguments.append(var)
2021-07-07 10:58:13 +02:00
# Parse typedef info
elif node.tag == 'typedef-info':
iter_typedef_info = (event == 'start')
elif iter_typedef_info and node.tag == 'info' and event == 'start':
cfg.typedefInfo.append(TypedefInfo(node))
# Parse valueflows (list of values)
elif node.tag == 'valueflow' and event == 'start':
2017-04-09 10:11:54 +02:00
continue
elif node.tag == 'values':
if event == 'start':
cfg_valueflow = ValueFlow(node)
continue
elif event == 'end':
cfg.valueflow.append(cfg_valueflow)
cfg_valueflow = None
# Parse values
elif node.tag == 'value' and event == 'start':
cfg_valueflow.values.append(Value(node))
# Remove links to the sibling nodes
node.clear()
def __repr__(self):
attrs = ["configurations", "platform"]
return "{}({})".format(
"CppcheckData",
", ".join(("{}={}".format(a, repr(getattr(self, a))) for a in attrs))
)
# Get function arguments
def getArgumentsRecursive(tok, arguments):
if tok is None:
return
if tok.str == ',':
getArgumentsRecursive(tok.astOperand1, arguments)
getArgumentsRecursive(tok.astOperand2, arguments)
else:
arguments.append(tok)
def getArguments(ftok):
if (not ftok.isName) or (ftok.next is None) or ftok.next.str != '(':
return None
args = []
getArgumentsRecursive(ftok.next.astOperand2, args)
return args
def parsedump(filename):
"""
parse a cppcheck dump file
"""
return CppcheckData(filename)
def astIsFloat(token):
"""
Check if type of ast node is float/double
"""
if not token:
return False
if token.str == '.':
return astIsFloat(token.astOperand2)
2017-06-05 13:23:00 +02:00
if token.str in '+-*/%':
2017-06-04 22:51:48 +02:00
return astIsFloat(token.astOperand1) or astIsFloat(token.astOperand2)
if not token.variable:
# float literal?
if token.str[0].isdigit():
for c in token.str:
2014-07-21 16:10:04 +02:00
if c == 'f' or c == '.' or c == 'E':
return True
return False
typeToken = token.variable.typeStartToken
endToken = token.variable.typeEndToken
while typeToken != endToken:
if typeToken.str == 'float' or typeToken.str == 'double':
return True
typeToken = typeToken.next
if typeToken.str == 'float' or typeToken.str == 'double':
return True
return False
class CppCheckFormatter(argparse.HelpFormatter):
"""
Properly formats multiline argument helps
"""
def _split_lines(self, text, width):
# this is the RawTextHelpFormatter._split_lines
if text.startswith('R|'):
2016-07-23 13:44:05 +02:00
return text[2:].splitlines()
return argparse.HelpFormatter._split_lines(self, text, width)
2017-06-04 22:51:48 +02:00
def ArgumentParser():
"""
Returns an argparse argument parser with an already-added
argument definition for -t/--template
"""
parser = argparse.ArgumentParser(formatter_class=CppCheckFormatter)
parser.add_argument('-t', '--template', metavar='<text>',
2016-07-23 13:44:05 +02:00
default='{callstack}: ({severity}) {message}',
help="R|Format the error messages. E.g.\n"
"'{file}:{line},{severity},{id},{message}' or\n"
"'{file}({line}):({severity}) {message}' or\n"
"'{callstack} {message}'\n"
"Pre-defined templates: gcc, vs, edit")
parser.add_argument("dumpfile", nargs='*',
help="Path of dump files from cppcheck.")
parser.add_argument("--cli",
help="Addon is executed from Cppcheck",
action="store_true")
parser.add_argument("--file-list", metavar='<text>',
default=None,
help="file list in a text file")
parser.add_argument("-q", "--quiet",
help='do not print "Checking ..." lines',
action="store_true")
return parser
2017-06-04 22:51:48 +02:00
def get_files(args):
"""Return dump_files, ctu_info_files"""
2021-07-17 19:59:21 +02:00
all_files = args.dumpfile
if args.file_list:
with open(args.file_list, 'rt') as f:
for line in f.readlines():
2021-07-17 19:59:21 +02:00
all_files.append(line.rstrip())
dump_files = []
ctu_info_files = []
for f in all_files:
if f.endswith('.ctu-info'):
ctu_info_files.append(f)
else:
dump_files.append(f)
return dump_files, ctu_info_files
def simpleMatch(token, pattern):
for p in pattern.split(' '):
if not token or token.str != p:
return False
token = token.next
return True
def get_function_call_name_args(token):
"""Get function name and arguments for function call
name, args = get_function_call_name_args(tok)
"""
if token is None:
return None, None
if not token.isName or not token.scope.isExecutable:
return None, None
if not simpleMatch(token.next, '('):
return None, None
if token.function:
nametok = token.function.token
2021-07-30 15:53:10 +02:00
if nametok is None:
nametok = token.function.tokenDef
if token in (token.function.token, token.function.tokenDef):
return None, None
name = nametok.str
while nametok.previous and nametok.previous.previous and nametok.previous.str == '::' and nametok.previous.previous.isName:
name = nametok.previous.previous.str + '::' + name
nametok = nametok.previous.previous
scope = token.function.nestedIn
while scope:
if scope.className:
name = scope.className + '::' + name
scope = scope.nestedIn
else:
nametok = token
name = nametok.str
while nametok.previous and nametok.previous.previous and nametok.previous.str == '::' and nametok.previous.previous.isName:
name = nametok.previous.previous.str + '::' + name
nametok = nametok.previous.previous
return name, getArguments(token)
def is_suppressed(location, message, errorId):
for suppression in current_dumpfile_suppressions:
if suppression.isMatch(location.file, location.linenr, message, errorId):
return True
return False
2019-06-24 19:28:52 +02:00
def reportError(location, severity, message, addon, errorId, extra=''):
2019-04-09 20:36:24 +02:00
if '--cli' in sys.argv:
2019-06-24 19:28:52 +02:00
msg = { 'file': location.file,
'linenr': location.linenr,
'column': location.column,
2019-06-24 19:28:52 +02:00
'severity': severity,
'message': message,
'addon': addon,
'errorId': errorId,
'extra': extra}
sys.stdout.write(json.dumps(msg) + '\n')
2019-04-09 20:36:24 +02:00
else:
if is_suppressed(location, message, '%s-%s' % (addon, errorId)):
return
2019-04-14 08:54:53 +02:00
loc = '[%s:%i]' % (location.file, location.linenr)
2019-06-24 19:28:52 +02:00
if len(extra) > 0:
message += ' (' + extra + ')'
sys.stderr.write('%s (%s) %s [%s-%s]\n' % (loc, severity, message, addon, errorId))
global EXIT_CODE
EXIT_CODE = 1
2021-07-07 10:58:13 +02:00
def reportSummary(dumpfile, summary_type, summary_data):
# dumpfile ends with ".dump"
ctu_info_file = dumpfile[:-4] + "ctu-info"
with open(ctu_info_file, 'at') as f:
msg = {'summary': summary_type, 'data': summary_data}
f.write(json.dumps(msg) + '\n')