File: //usr/lib64/python2.6/site-packages/mx/Misc/ConfigFile.py
""" ConfigFile -- Interface to simple template based configuration files.
The format of these files is as follows:
globalvar = 2
[ABC]
a = 1
b = abc.html
c = text with spaces
[DEF]
a = 2
b = a + 3
c = a string
Entries in square brackets indicate new subsections. Global
variables may be set prior to starting any subsection. Emtpy lines
and lines starting with '#' (comments) are ignored. Indentation is
not necessary; lines can start at any column.
Entries may span multiple lines by using '\' continuations
at the line ends, e.g.
[Continuation]
a = first line\
second line
The lines are stripped of any white space before removing the
trailing '\' and concatenating them. Comment lines are removed as
well.
To parse these files, a template in form of a class including
subclasses (identifying the subsections) must be given to the
reader. This template defines which sections and attributes are
known. All others are rejected. A sample template for the above
looks like this:
from mx.Misc.ConfigFile import *
class Template(ConfigNamespace):
global = IntegerEntry()
class ABC(ConfigSection):
a = IntegerEntry('0')
b = StringEntry('default.html')
c = StringEntry()
class DEF(ConfigSection):
a = IntegerEntry('0')
b = EvalEntry('0')
c = StringEntry('default value')
The main class representing the global namespace of the ini file
must have ConfigNamespace as baseclass. It may contain any number
of ConfigSection subclasses each defining parsed attributes.
The ini file may only contain entries defined in this template.
In case an unknown section is found, the main class is looked up
for a ConfigSection subclass with name 'DefaultSection'. This
section object is then taken as template for a new section
of the given name.
The same feature is available for attributes in sections. In case
a given attribute is not found in the templates, the section's
'DefaultAttribute' attribute is deepcopied and the used as
attribute template.
Needs mxTools to be installed.
XXX Add support for long strings (including embedded control
characters and spanning multiple lines)
Copyright (c) 2000, Marc-Andre Lemburg; All Rights Reserved.
Copyright (c) 2000-2008, eGenix.com Software GmbH; All Rights Reserved.
See the documentation for further information on copyrights,
or contact the author. All Rights Reserved.
"""
import string,copy,re,types,exceptions,sys,os
import mx.Tools.NewBuiltins
from mx.Tools import freeze
import Namespace
# Print debug information ?
_debug = 0
### Errors
class Error(exceptions.StandardError):
pass
### Helpers
shell_var = re.compile('(?:\$([a-zA-Z0-9._]+)|\${([^}]*)})')
def expand_vars(text,locals,globals,
split=string.split,shell_var=shell_var):
""" Expands all variables of the form $var or ${var} using
the dictionaries locals,globals,os.environ in that order
as database.
Default value is the empty string, just like for shell
environment variables.
"""
while 1:
m = shell_var.search(text)
if not m:
break
g = m.groups()
varpath = split(g[0] or g[1],'.')
varname = varpath[0]
# Find object with name varname
value = locals.get(varname,None)
if value is None:
value = globals.get(varname,None)
if value is None:
value = os.environ.get(varname,'')
# Now get the requested attribute
for attr in varpath[1:]:
try:
value = getattr(value,attr)
except AttributeError:
raise Error,'attribute "%s" undefined' % attr
# Reformat text by inserting value as string
l,r = m.span()
if isinstance(value,Entry):
entry = value
if entry.value is NotGiven:
entry.parse(entry.default,locals,globals)
value = entry.value
text = text[:l] + str(value) + text[r:]
return text
### Template entry fields
class Entry:
""" Converts the value to another datatype depending on the
.converters attribute.
Each of those functions is applied to the value. The first one
to not raise an exception succeeds.
"""
# Value
value = NotGiven
# Default value
default = ''
# Converters to be tried; order is important the first successful
# one wins
converters = (int,float,str)
def __init__(self,default=NotGiven):
if default is not NotGiven:
self.default = default
def parse(self,text,locals,globals):
""" Parse text according to the namespace dictionaries
locals and globals.
Sets self.value and returns the parsed value as well.
"""
text = self.pre_process(str(text),locals,globals)
for converter in self.converters:
try:
value = converter(text)
break
except:
lastexc = sys.exc_info()[1]
else:
# Reraise the last exception
raise lastexc
lastexc = None
self.value = value = self.post_process(value,locals,globals)
return value
def pre_process(self,text,locals,globals,
strip=string.strip,expand_vars=expand_vars):
""" Preprocess the text value.
Default behaviour is to strip the text and then
apply variable expansion.
"""
return expand_vars(strip(text),locals,globals)
def post_process(self,value,locals,globals):
""" Postprocess the value after conversion has been
applied.
The default behaviour is to leave it as it is.
"""
return value
def __str__(self):
""" Return a stringified version of self.value.
The default value is parsed in case no value has yet been
set.
"""
if self.value is NotGiven:
self.parse(self.default,{},{})
return str(self.value)
###
def integer_value(value):
""" Convert value to an integer.
Takes base indicators into account, such as 0x for base-16,
0 for base-7.
"""
return int(value, 0)
class IntegerEntry(Entry):
""" Converts the value to an integer.
"""
converters = (integer_value,)
class FloatEntry(Entry):
""" Converts the value to a float.
"""
converters = (float,)
class NumericEntry(Entry):
""" Converts the value to a number (integer or float).
"""
converters = (integer_value, float)
class StringEntry(Entry):
""" Converts the value to a string.
"""
converters = (str,)
def comma_list(text,
strip=string.strip,split=string.split,map=map,tuple=tuple):
return map(strip,split(text,','))
def spaced_list(text,
strip=string.strip,split=string.split,map=map,tuple=tuple):
return map(strip,split(text))
class TupleEntry(Entry):
""" Converts the value to a tuple of strings.
value must be a string with entries separated by ','. Entries
are split at each occurance of ',' and then stripped of
surrounding spaces.
"""
converters = (comma_list,)
def post_process(self,value,locals,globals):
return tuple(value)
# Alias
StringTupleEntry = TupleEntry
CommaTupleEntry = TupleEntry
class ListEntry(Entry):
""" Converts the value to a list of strings.
value must be a string with entries separated by ','. Entries
are split at each occurance of ',' and then stripped of
surrounding spaces.
"""
converters = (comma_list,)
# Alias
StringListEntry = ListEntry
CommaListEntry = ListEntry
class SpacedTupleEntry(TupleEntry):
""" Converts the value to a tuple of strings.
value must be a string with entries separated by
whitespace. Entries are split at each occurance of whitespace;
surrounding spaces is removed.
"""
converters = (spaced_list,)
class SpacedListEntry(ListEntry):
""" Converts the value to a list of strings.
value must be a string with entries separated by
whitespace. Entries are split at each occurance of whitespace;
surrounding spaces is removed.
"""
converters = (spaced_list,)
class FileEntry(Entry):
""" Entry field for an OS file.
"""
converters = (str,)
class PathEntry(Entry):
""" Entry field for an OS path.
Paths will always end with os.sep if given.
"""
converters = (str,)
def post_process(self,value,locals,globals):
if value and value[-1] != os.sep:
value = value + os.sep
return value
class AbsoluteFileEntry(Entry):
""" Entry field for an absolute OS pathname.
The pathname will always start with os.sep if given.
"""
converters = (str,)
def post_process(self,value,locals,globals):
if value and value[0] != os.sep:
value = os.sep + value
return value
class AbsolutePathEntry(Entry):
""" Entry field for an absolute OS path.
The pathname will always start and end with os.sep if given.
"""
converters = (str,)
def post_process(self,value,locals,globals):
if value:
if value[0] != os.sep:
value = os.sep + value
if value[-1] != os.sep:
value = value + os.sep
return value
class RelativePathEntry(PathEntry):
""" Entry field for relative paths.
The path stored is the result of joining the given parameter
with a basepath. basepath is subject to variable expansion at
parsing time.
"""
def __init__(self,basepath,default=NotGiven):
self.basepath = basepath
PathEntry.__init__(self,default)
def post_process(self,value,locals,globals):
return os.path.join(expand_vars(self.basepath,locals,globals),value)
class RelativeFileEntry(FileEntry):
""" Entry field for relative pathnames to files.
The path stored is the result of joining the given parameter
with a basepath. basepath is subject to variable expansion at
parsing time.
"""
def __init__(self,basepath,default=NotGiven):
self.basepath = basepath
PathEntry.__init__(self,default)
def post_process(self,value,locals,globals):
return os.path.join(expand_vars(self.basepath,locals,globals),value)
class ExistingPathEntry(PathEntry):
""" Checks value to point to an existing OS path.
Raises an exception if the path does not exist.
"""
def post_process(self,value,locals,globals):
if value and value[-1] != os.sep:
value = value + os.sep
if not os.path.exists(value):
raise Error,'non existing path "%s"' % value
return value
class ExistingFileEntry(FileEntry):
""" Checks value to point to an existing file.
Raises an exception if the file does not exist.
"""
def post_process(self,value,locals,globals):
if not value or not os.path.exists(value):
raise Error,'non existing file "%s"' % value
return value
class ExistingRelativePathEntry(RelativePathEntry):
""" Entry field for relative paths.
The path stored is the result of joining the given parameter
with a basepath. Raises an exception if the path does not
exist.
"""
def post_process(self,value,locals,globals):
value = os.path.join(expand_vars(self.basepath,locals,globals),value)
if not os.path.exists(value):
raise Error,'non existing path "%s"' % value
return value
class URLEntry(StringEntry):
""" Entry field for URLs.
The field takes a default value as argument.
Needs mxHTMLTools to be installed.
"""
def post_process(self,value,locals,globals):
from mx import HTMLTools
return HTMLTools.URL(value)
class RelativeURLEntry(URLEntry):
""" Entry field for relative URLs.
A base URL can be set which is then urljoined with the
value given. The field also takes a default value.
Needs mxHTMLTools to be installed.
"""
def __init__(self,baseurl='',default=NotGiven):
self.baseurl = baseurl
StringEntry.__init__(self,default)
def post_process(self,value,locals,globals):
from mx import HTMLTools
return HTMLTools.urljoin(expand_vars(self.baseurl,locals,globals),
value)
class AbsoluteURLEntry(RelativeURLEntry):
""" Entry field for absolute URLs.
A base URL can be set which is then urljoined with the
value given. The field also takes a default value as second
argument.
"""
def post_process(self,value,locals,globals):
url = RelativeURLEntry.post_process(self,value,locals,globals)
if not url.absolute:
raise Error,'need an absolute URL: "%s"' % url
return url
class DateEntry(Entry):
""" Date entry field.
The value is stored using a DateTime instance. A time part is
ignored.
Needs mxDateTime to be installed.
"""
def post_process(self,value,locals,globals):
from mx.DateTime.Parser import DateFromString
return DateFromString(value)
class DateTimeEntry(Entry):
""" Date/time entry field.
The value is stored using a DateTime instance.
Needs mxDateTime to be installed.
"""
def post_process(self,value,locals,globals):
from mx.DateTime.Parser import DateTimeFromString
return DateTimeFromString(value)
class TimeEntry(Entry):
""" Time entry field.
The value is stored using a DateTimeDelta instance. A date
part is ignored.
Needs mxDateTime to be installed.
"""
def post_process(self,value,locals,globals):
from mx.DateTime.Parser import TimeFromString
return TimeFromString(value)
class DateTimeDeltaEntry(Entry):
""" Date/time delta entry field.
The value is stored using a DateTimeDelta instance.
Needs mxDateTime to be installed.
"""
def post_process(self,value,locals,globals):
from mx.DateTime.Parser import DateTimeDeltaFromString
return DateTimeDeltaFromString(value)
class RelativeDateTimeEntry(Entry):
""" Relative date/time entry field.
The value is stored using a RelativeDateTime instance.
Needs mxDateTime to be installed.
"""
def post_process(self,value,locals,globals):
from mx.DateTime.Parser import RelativeDateTimeFromString
return RelativeDateTimeFromString(value)
class EvalEntry(Entry):
""" Allows simple calculations to be done using the
current locals (the symbols defined in the class
where the entry is located) and globals.
"""
def parse(self,text,locals,globals,
eval=eval):
self.value = eval(text,locals,globals)
return self.value
class SafeEvalEntry(Entry):
""" Allows simple calculations to be done using the
current locals (the symbols defined in the class
where the entry is located) and globals.
Builtins are not available.
"""
def parse(self,text,locals,globals,
reval=reval):
self.value = reval(text,locals,globals)
return self.value
class DictEntry(Entry):
""" Allows the construction of a Python dictionary.
The entries value is taken as list of key: value pairs which
are evaluated in the current locals (the symbols defined in
the class where the entry is located) and globals.
Parsing is left to the Python interpreter. The needed curly
brackets {} are added by the parsing method.
Sample:
dict = 'a': (1,2,3), 'b': (3,4,5)
"""
def parse(self,text,locals,globals):
self.value = eval('{' + text + '}',locals,globals)
return self.value
### Namespace containers
class ConfigNamespace(Namespace.Namespace):
# This attribute may be set to allow new sections to be
# created by the ini-file editor. The parser will use this
# class to instantiate those sections.
DefaultSection = None
# This attribute may be set to allow new attributes to be
# created by the ini-file editor. The parser will use this
# object as template for the new attributes by deepcopying
# it.
DefaultAttribute = None
def __init__(self,
getattr=getattr,
skip_types=(types.ModuleType,
types.IntType,
types.StringType,
types.NoneType,
),
del_types=(types.FunctionType,
types.MethodType,
types.BuiltinFunctionType,
types.BuiltinMethodType,),
deepcopy=copy.deepcopy,ClassType=types.ClassType):
# Localize class attributes in the instance dict
classobj = self.__class__
freeze(classobj)
dict = self.__dict__
dict.update(classobj.__dict__)
# Fixup attributes
for name,obj in dict.items():
if name[:1] == '_ ':
# Delete all special attributes
del dict[name]
objtype = type(obj)
if objtype is ClassType:
# Handle classes
if name != 'DefaultSection' and \
issubclass(obj,ConfigSection):
# Instantiate ConfigSections
dict[name] = obj()
# leave all others as classes
elif objtype in del_types:
# Delete these types
del dict[name]
elif objtype in skip_types:
# Skip these types
pass
else:
# Deepcopy everything else
try:
dict[name] = deepcopy(obj)
except copy.error,why:
raise Error,\
'namespace entry "%s" could not be copied: %s' % \
(name,why)
#
# Dictionary like interface (we hide the internally used attributes)
#
def _items(self):
items = []
append = items.append
for k,v in self.__dict__.items():
if k[:1] == '_' or \
k in ('DefaultSection', 'DefaultAttribute', 'baseobj'):
continue
append((k,v))
return items
def _keys(self):
items = []
append = items.append
for k,v in self.__dict__.items():
if k[:1] == '_' or \
k in ('DefaultSection', 'DefaultAttribute', 'baseobj'):
continue
append(k)
return items
def _values(self):
items = []
append = items.append
for k,v in self.__dict__.items():
if k[:1] == '_' or \
k in ('DefaultSection', 'DefaultAttribute', 'baseobj'):
continue
append(v)
return items
def _dictionary(self):
d = {}
d.update(self.__dict__)
# Remove internal attributes
for k in d.keys():
if k[:1] == '_' or \
k in ('DefaultSection', 'DefaultAttribute', 'baseobj'):
del d[k]
return d
class ConfigSection(ConfigNamespace):
pass
###
parse_section = re.compile('\[([a-zA-Z_][a-zA-Z_0-9]*)\]')
parse_setattr = re.compile('([a-zA-Z_][a-zA-Z_0-9]*)[ \t]*=[ \t]*(.*)')
class ConfigFile:
""" Configuration file reader.
Takes a template (a ConfigNamespace subclass) as input which
defines sections using classes. Sections may include instances
of Entry as attributes. These are then used to process the
file input.
The template is (deep-)copied and placed into the instance
variable .data (a ConfigNamespace instance) prior to reading
the file. Section classes are replaced with ConfigSection
instances.
After successful reading the file, the configuration
information is available through this variable.
Errors are indicated by exceptions of type Error. These always
have values (line_number, explanation) where line_number is 0
for errors which do not refer to one specific line in the
file.
"""
data = None # configuration namespace
def __init__(self,template):
self.template = template
self.reset()
def reset(self,
deepcopy=copy.deepcopy,ClassType=types.ClassType,
ModuleType=types.ModuleType):
""" Reset the object to its initial state.
This initializes self.data to a template instance.
"""
self.data = self.template()
def read(self,file,
strip=string.strip,deepcopy=copy.deepcopy):
""" Read and parse the open file.
The configuration is stored in the instance variable
.data.
"""
context = {}
sectionname = ''
data = self.data # Configuration object
section = data # Current section object
globals = data.__dict__ # Global namespace
locals = globals # Local namespace
lineno = 0
if _debug:
print 'Reading config file:'
while 1:
line = file.readline()
lineno = lineno + 1
if not line:
break
line = strip(line)
if not line or line[0] == '#':
# Comment or emtpy line: ignore
continue
if _debug:
print ' read %s' % repr(line)
# Continuation
current_lineno = lineno
while line[-1] == '\\':
nextline = file.readline()
lineno = lineno + 1
if not nextline:
# EOF: end of continuation
line = line[:-1]
break
nextline = strip(nextline)
if not nextline:
# Empty line: end of continuation
line = line[:-1]
break
if nextline[0] == '#':
# Comment: ignore
continue
if _debug:
print ' read continuation %s' % repr(nextline)
# Add nextline to line
line = line[:-1] + nextline
if _debug:
print ' processing %s' % repr(line)
m = parse_section.match(line)
if m:
# Start a new section
sectionname = m.groups()[0]
section = getattr(data,sectionname,None)
if section is None:
if data.DefaultSection:
# Instantiate default section for use as
# section object and register in data
section = data.DefaultSection()
setattr(data,sectionname,section)
else:
raise Error,(current_lineno,
'unknown section "%s"' %
sectionname)
#print 'Switched to section:',sectionname
locals = section.__dict__
continue
m = parse_setattr.match(line)
if m:
# Add a new attribute
name,value = m.groups()
entry = locals.get(name,None)
if sectionname:
attrname = sectionname + '.' + name
else:
attrname = name
if entry is None:
if section.DefaultAttribute:
# Instantiate default attribute for use as
# attribute object and register in section
entry = deepcopy(section.DefaultAttribute)
setattr(section,name,entry)
else:
raise Error,(current_lineno,
'unknown attribute: "%s"' % attrname)
if not isinstance(entry, Entry):
# Since we replace the parsed entries with their parsed
# value any occurance of non-Entry instances point
# to a duplicate definition of an entry
raise Error,(current_lineno,
'duplicate attribute definition for "%s"' %
attrname)
try:
entry.parse(value,locals,globals)
except:
raise Error,(current_lineno,
'invalid attribute value for "%s" (%s: %s)' %
(attrname,
sys.exc_info()[0],sys.exc_info()[1]))
# Replace with parsed value
locals[name] = entry.value
continue
# The line doesn't have a valid syntax: ignore it
if _debug:
print 'Ignoring invalid config file line %i: %s' % \
(current_lineno,
repr(line))
# Check for missing entries and replace with parsed values
for name,obj in globals.items():
if isinstance(obj,Entry):
if obj.value is not NotGiven:
globals[name] = obj.value
else:
try:
obj.parse(obj.default,locals,globals)
globals[name] = obj.value
except:
raise Error,\
(0,
'invalid default attribute '
'value for global "%s" (%s)' %
(name,str(sys.exc_info()[1])))
elif isinstance(obj,ConfigSection):
locals = obj.__dict__
sectionname = name
for name,obj in locals.items():
if isinstance(obj,Entry):
if obj.value is not NotGiven:
locals[name] = obj.value
else:
try:
obj.parse(obj.default,locals,globals)
locals[name] = obj.value
except:
import traceback; traceback.print_exc()
raise Error,\
(0,
'invalid default attribute '
'value for local "%s" (%s)' %
(name,str(sys.exc_info()[1])))
### Test
if __name__ == '__main__':
class Template(ConfigNamespace):
import sys
globalvar = IntegerEntry(1)
# New attributes in the global section are strings
DefaultAttribute = StringEntry('')
class ABC(ConfigSection):
a = IntegerEntry('0')
b = StringEntry('default.html')
c = StringEntry()
d = StringTupleEntry('a,b')
path = ExistingPathEntry('.')
# Use this as template for new attributes
DefaultAttribute = RelativePathEntry('..','')
class DEF(ConfigSection):
a = IntegerEntry('0')
b = EvalEntry('0')
c = StringEntry('default value')
d = RelativePathEntry('$ABC.b','HOME')
g = IntegerEntry('0')
continuation = CommaTupleEntry('')
databases = DictEntry('')
timeout = TimeEntry('0:15:00.23')
# Default section
DefaultSection = ABC
text = """\
# Luckily we don't have to pay attention to proper indentation
# or whether strings have quotes or not... the template knows
# what to do.
globalvar = 2
# New global attribute
test = text
[ABC]
a = 1
b = abc.html
c = text with spaces
d = heinz, kunz, philipp
path = /tmp
# A new attribute
new = .cshrc
[DEF]
a = 2
b = a + 3
c = $PWD
d = home
continuation = first line, \\
# Comment should not hurt
second line, \\
# Invalid syntax:
third line \\
databases = 'iODBC': (sys,sys.exit,('c',1,2)), \\
'Adabas': (sys,sys.exit,('x','y','z'))
[NewSection]
a = 3
"""
import cStringIO
f = cStringIO.StringIO(text)
cf = ConfigFile(Template)
cf.read(f)