cppcheck/addons/cppcheckdata.py

738 lines
24 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 xml.etree.ElementTree as ET
import argparse
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
def __init__(self, element):
self.str = element.get('str')
self.file = element.get('file')
self.linenr = element.get('linenr')
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
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')
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 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'}
class Token:
"""
Token class. Contains information about each token in the source code.
The CppcheckData.tokenlist is a list of Token items
C++ class: http://cppcheck.net/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
isFloat Is this token a int 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
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 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
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
2014-07-21 16:10:04 +02:00
varId = None
variableId = None
variable = None
functionId = None
function = None
valuesId = None
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
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
2014-07-21 16:10:04 +02:00
self.linkId = element.get('link')
self.link = None
self.varId = element.get('varId')
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
self.file = element.get('file')
self.linenr = element.get('linenr')
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 = IdMap[self.valuesId]
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
class Scope:
"""
Scope. Information about global scope, function scopes, class scopes, inner scopes, etc.
C++ class: http://cppcheck.net/devinfo/doxyoutput/classScope.html
Attributes
classStart The { Token for this scope
classEnd 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.
type Type of scope: Global, Function, Class, If, While
"""
2014-07-21 16:10:04 +02:00
Id = None
classStartId = None
2014-07-21 16:10:04 +02:00
classStart = None
classEndId = None
classEnd = None
className = None
2017-04-08 19:33:26 +02:00
nestedInId = None
nestedIn = None
2014-07-21 16:10:04 +02:00
type = None
def __init__(self, element):
self.Id = element.get('id')
self.className = element.get('className')
self.classStartId = element.get('classStart')
2014-07-21 16:10:04 +02:00
self.classStart = None
self.classEndId = element.get('classEnd')
self.classEnd = None
self.nestedInId = element.get('nestedIn')
2014-07-21 16:10:04 +02:00
self.nestedIn = None
self.type = element.get('type')
def setId(self, IdMap):
self.classStart = IdMap[self.classStartId]
self.classEnd = IdMap[self.classEndId]
self.nestedIn = IdMap[self.nestedInId]
class Function:
"""
Information about a function
C++ class:
http://cppcheck.net/devinfo/doxyoutput/classFunction.html
"""
2014-07-21 16:10:04 +02:00
Id = None
argument = None
argumentId = None
tokenDef = None
tokenDefId = None
2015-07-28 14:18:58 +02:00
name = None
isVirtual = None
isImplicitlyVirtual = None
2014-07-21 16:10:04 +02:00
def __init__(self, element):
self.Id = element.get('id')
self.tokenDefId = element.get('tokenDef')
2015-07-28 14:18:58 +02:00
self.name = element.get('name')
isVirtual = element.get('isVirtual')
self.isVirtual = (isVirtual and isVirtual == 'true')
isImplicitlyVirtual = element.get('isImplicitlyVirtual')
self.isImplicitlyVirtual = (isImplicitlyVirtual and isImplicitlyVirtual == 'true')
2014-07-21 16:10:04 +02:00
self.argument = {}
self.argumentId = {}
for arg in element:
self.argumentId[int(arg.get('nr'))] = arg.get('variable')
2014-07-21 16:10:04 +02:00
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.tokenDef = IdMap[self.tokenDefId]
2014-07-21 16:10:04 +02:00
class Variable:
"""
Information about a variable
C++ class:
http://cppcheck.net/devinfo/doxyoutput/classVariable.html
Attributes:
nameToken name token in variable declaration
typeStartToken start token of variable declaration
typeEndToken end token of variable declaration
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?
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
isArgument = False
isArray = False
isClass = False
isConst = False
2017-04-17 07:45:27 +02:00
isExtern = 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.isArgument = element.get('isArgument') == 'true'
self.isArray = element.get('isArray') == 'true'
self.isClass = element.get('isClass') == 'true'
self.isConst = element.get('isConst') == 'true'
2017-04-17 07:45:27 +02:00
self.isExtern = element.get('isExtern') == 'true'
self.isLocal = element.get('isLocal') == 'true'
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 setId(self, IdMap):
self.nameToken = IdMap[self.nameTokenId]
self.typeStartToken = IdMap[self.typeStartTokenId]
self.typeEndToken = IdMap[self.typeEndTokenId]
class ValueFlow:
"""
ValueFlow::Value class
Each possible value has a ValueFlow::Value item.
Each ValueFlow::Value either has a intvalue or tokvalue
C++ class:
http://cppcheck.net/devinfo/doxyoutput/classValueFlow_1_1Value.html
Attributes:
values Possible values
"""
Id = None
values = None
class Value:
"""
Value class
2016-07-25 18:03:44 +02:00
Attributes:
intvalue integer value
tokvalue token value
condition condition where this Value comes from
"""
intvalue = None
tokvalue = None
condition = None
2014-07-21 16:10:04 +02:00
def __init__(self, element):
self.intvalue = element.get('intvalue')
if self.intvalue:
self.intvalue = int(self.intvalue)
self.tokvalue = element.get('tokvalue')
self.condition = element.get('condition-line')
if self.condition:
self.condition = int(self.condition)
def __init__(self, element):
2014-07-21 16:10:04 +02:00
self.Id = element.get('id')
self.values = []
for value in element:
self.values.append(ValueFlow.Value(value))
class Configuration:
"""
Configuration class
This class contains the directives, tokens, scopes, functions,
variables and value flows for one configuration.
Attributes:
name Name of the configuration, "" for default
directives List of Directive items
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
"""
name = ''
directives = []
tokenlist = []
2014-07-21 16:10:04 +02:00
scopes = []
functions = []
variables = []
valueflow = []
def __init__(self, confignode):
self.name = confignode.get('cfg')
self.directives = []
self.tokenlist = []
2014-07-21 16:10:04 +02:00
self.scopes = []
self.functions = []
self.variables = []
self.valueflow = []
arguments = []
for element in confignode:
if element.tag == 'directivelist':
for directive in element:
self.directives.append(Directive(directive))
if element.tag == 'tokenlist':
for token in element:
self.tokenlist.append(Token(token))
# set next/previous..
prev = None
for token in self.tokenlist:
token.previous = prev
if prev:
prev.next = token
prev = token
if element.tag == 'scopes':
for scope in element:
self.scopes.append(Scope(scope))
for functionList in scope:
if functionList.tag == 'functionList':
for function in functionList:
self.functions.append(Function(function))
if element.tag == 'variables':
for variable in element:
var = Variable(variable)
if var.nameToken:
self.variables.append(var)
else:
arguments.append(var)
if element.tag == 'valueflow':
for values in element:
self.valueflow.append(ValueFlow(values))
IdMap = {None: None, '0': None, '00000000': None, '0000000000000000': 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)
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'))
class CppcheckData:
"""
Class that makes cppcheck dump data available
Contains a list of Configuration instances
Attributes:
configurations List of Configurations
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
"""
2017-04-09 10:11:54 +02:00
rawTokens = []
platform = None
configurations = []
def __init__(self, filename):
self.configurations = []
data = ET.parse(filename)
2017-04-09 10:11:54 +02:00
for platformNode in data.getroot():
if platformNode.tag == 'platform':
self.platform = Platform(platformNode)
2017-04-09 10:11:54 +02:00
for rawTokensNode in data.getroot():
if rawTokensNode.tag != 'rawtokens':
continue
files = []
for node in rawTokensNode:
if node.tag == 'file':
files.append(node.get('name'))
elif node.tag == 'tok':
tok = Token(node)
tok.file = files[int(node.get('fileIndex'))]
self.rawTokens.append(tok)
2017-06-04 22:51:48 +02:00
for i in range(len(self.rawTokens) - 1):
self.rawTokens[i + 1].previous = self.rawTokens[i]
self.rawTokens[i].next = self.rawTokens[i + 1]
2017-04-09 10:11:54 +02:00
# root is 'dumps' node, each config has its own 'dump' subnode.
for cfgnode in data.getroot():
2017-06-04 22:51:48 +02:00
if cfgnode.tag == 'dump':
2017-04-09 10:11:54 +02:00
self.configurations.append(Configuration(cfgnode))
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")
return parser
2017-06-04 22:51:48 +02:00
2017-07-22 11:05:50 +02:00
def reportError(template, callstack=(), severity='', message='', id=''):
"""
Format an error message according to the template.
2016-07-23 13:44:05 +02:00
:param template: format string, or 'gcc', 'vs' or 'edit'.
:param callstack: e.g. [['file1.cpp',10],['file2.h','20'], ... ]
:param severity: e.g. 'error', 'warning' ...
:param id: message ID.
:param message: message text.
"""
# expand predefined templates
if template == 'gcc':
template = '{file}:{line}: {severity}: {message}'
elif template == 'vs':
template = '{file}({line}): {severity}: {message}'
elif template == 'edit':
template = '{file} +{line}: {severity}: {message}'
# compute 'callstack}, {file} and {line} replacements
2017-06-04 22:51:48 +02:00
stack = ' -> '.join('[' + f + ':' + str(l) + ']' for (f, l) in callstack)
file = callstack[-1][0]
line = str(callstack[-1][1])
# format message
return template.format(callstack=stack, file=file, line=line,
2016-07-23 13:44:05 +02:00
severity=severity, message=message, id=id)