Files
tablib/tablib/packages/dbfpy/fields.py
T
James Douglass a21f8187f8 Adding DBF support.
Squashing two squashes.

Adding DBF support

Adding the DBFpy python package

The DBFpy package provides basic dbf support for python.  Still need to
write an interface format file for tablib.

Adding DBF format and imports in compat.py

Adding DBF format to formats.__init__

DBF format had not been committed to formats.__init__, so I’m adding it.

Adding a dbf import test

Adding at test to check whether a DBF can be created properly and
compare it against a regression binary string.

Adding an import_set test (and renaming another)

Adding an import_set test that conforms with the other import_set tests
for other formats.  I’m also adding an export_set function.

Fixing system site-packages import

Importing dbfpy from tab lib.packages instead of system site packages.

Fixing a syntaxError in dbfpy/dbfnew.py

Fixing an issue with ending field definitions

DBFPY, when writing a DBF, terminates the field definitions with a
newline character.  When importing a DBF from a stream, however, DBFPY
was looking only for the \x0D character rather than the newline.  Now
we consider both cases.

Adding a test for dbf format detection

Adding DBF filetype detection tests

Adding tests for YAML, JSON, TSV, CSV using the DBF detection function.

Handling extra exceptions in dbf detection

Adding exception handling for struct.error, an exception that DBFPY
raises when trying to unpack a TSV table.  Since it’s not a DBF file,
we know it’s not a DBF and return False.

Fixing an issue with the DBF set exporting test

The DBF set export test needed a bit enabled (probably the writeable
bit?) before the test would match the regression output.

Updating dbf interface

Updating the int/float class/type checking in the dbf format file.
This allows for python2 and python3 compatibility.

Tweaking dbfpy to work with python3

Altering a couple of imports.

Updating dbf tests for binary data compatibility

Making regression strings binary and improving debug messages for dbf
assertion errors.

Improving file handling for python 2 and 3

Updating DBF file handling for both python 2 and 3 in the _dbf
interface.

Adding a (seemingly) functional dbfpy for python3

I’ve made dbfpy python3 compatible!  Tests appear to pass.
A significant change was made to the format detection test whereby I
made the input string a binary (bytes) string.  If the string is not a
bytes string by the time we try to detect the format, we try to decode
the string as utf-8 (which admittedly might not be the safest thing to
do) and try to decode anyways.

Updating imports for tablib dbf interface

Now importing python2 or python3 versions as appropriate.

Updating dbf package references in compat.py

Cleaning up debugging print statements

Updating stream handling in dbf interface

Factoring the open() call out of the py3 conditional and removing the
temp file before returning the stream value.

Adding dbfpy3 init.py

I had apparently missed the dbfpy3 init file when committing dbfpy3.

Adding dbfpy and dbfpy3 to setup.py's package list

Switching test order of formats

Putting dbf format testing ahead of TSV.  In some of my tests with
numeric DBF files, I encountered an issue where the ASCII horizontal
tab character (0x09) would appear in a numeric DBF.  Because of the
order of tabular format imports, though, format detection would
recognize it as a TSV and not as a DBF.

Adding my name to AUTHORS.

Adding a DBF property to tab lib core

Documentation includes examples on how to explicitly load a DBF
straight from a file and how to load a DBF from a binary string.  Also,
how to write the binary data to a file.

Adding DBF format notes to README

Adding exclamation point to DBF section title

Matching formatting of XLS section

Updating setup.py to match current dev state

Setup.py had been updated since I forked the tablib repo, so I’m
updating setup.py to match its current structure while still
maintaining DBF compatibility.

Fixed callable collumn test

the test was sending a list instead of a function

CORE CONTRIBUTORS

🍰 @iurisilvio

v0.10.0

WHEELS

3.3, 3.4

makefile for WHEELS

v0.10.0 history

ALL

Separate py2 and py3 packages to avoid installation errors. Fix #151

Running travis and tox with python 3.4.

Adding DBF support

Adding the DBFpy python package

The DBFpy package provides basic dbf support for python.  Still need to
write an interface format file for tablib.

Adding DBF format and imports in compat.py

Adding DBF format to formats.__init__

DBF format had not been committed to formats.__init__, so I’m adding it.

Adding a dbf import test

Adding at test to check whether a DBF can be created properly and
compare it against a regression binary string.

Adding an import_set test (and renaming another)

Adding an import_set test that conforms with the other import_set tests
for other formats.  I’m also adding an export_set function.

Fixing system site-packages import

Importing dbfpy from tab lib.packages instead of system site packages.

Fixing a syntaxError in dbfpy/dbfnew.py

Fixing an issue with ending field definitions

DBFPY, when writing a DBF, terminates the field definitions with a
newline character.  When importing a DBF from a stream, however, DBFPY
was looking only for the \x0D character rather than the newline.  Now
we consider both cases.

Adding a test for dbf format detection

Adding DBF filetype detection tests

Adding tests for YAML, JSON, TSV, CSV using the DBF detection function.

Handling extra exceptions in dbf detection

Adding exception handling for struct.error, an exception that DBFPY
raises when trying to unpack a TSV table.  Since it’s not a DBF file,
we know it’s not a DBF and return False.

Fixing an issue with the DBF set exporting test

The DBF set export test needed a bit enabled (probably the writeable
bit?) before the test would match the regression output.

Updating dbf interface

Updating the int/float class/type checking in the dbf format file.
This allows for python2 and python3 compatibility.

Tweaking dbfpy to work with python3

Altering a couple of imports.

Updating dbf tests for binary data compatibility

Making regression strings binary and improving debug messages for dbf
assertion errors.

Improving file handling for python 2 and 3

Updating DBF file handling for both python 2 and 3 in the _dbf
interface.

Adding a (seemingly) functional dbfpy for python3

I’ve made dbfpy python3 compatible!  Tests appear to pass.
A significant change was made to the format detection test whereby I
made the input string a binary (bytes) string.  If the string is not a
bytes string by the time we try to detect the format, we try to decode
the string as utf-8 (which admittedly might not be the safest thing to
do) and try to decode anyways.

Updating imports for tablib dbf interface

Now importing python2 or python3 versions as appropriate.

Updating dbf package references in compat.py

Cleaning up debugging print statements

Updating stream handling in dbf interface

Factoring the open() call out of the py3 conditional and removing the
temp file before returning the stream value.

Adding dbfpy3 init.py

I had apparently missed the dbfpy3 init file when committing dbfpy3.

Adding dbfpy and dbfpy3 to setup.py's package list

Switching test order of formats

Putting dbf format testing ahead of TSV.  In some of my tests with
numeric DBF files, I encountered an issue where the ASCII horizontal
tab character (0x09) would appear in a numeric DBF.  Because of the
order of tabular format imports, though, format detection would
recognize it as a TSV and not as a DBF.

Adding my name to AUTHORS.

Adding a DBF property to tab lib core

Documentation includes examples on how to explicitly load a DBF
straight from a file and how to load a DBF from a binary string.  Also,
how to write the binary data to a file.

Adding DBF format notes to README

Adding exclamation point to DBF section title

Matching formatting of XLS section

Updating setup.py to match current dev state

Setup.py had been updated since I forked the tablib repo, so I’m
updating setup.py to match its current structure while still
maintaining DBF compatibility.

Fixed callable collumn test

the test was sending a list instead of a function

CORE CONTRIBUTORS

🍰 @iurisilvio

v0.10.0

WHEELS

3.3, 3.4

makefile for WHEELS

v0.10.0 history

ALL

Separate py2 and py3 packages to avoid installation errors. Fix #151

Running travis and tox with python 3.4.
2014-08-21 22:06:42 -07:00

467 lines
14 KiB
Python

"""DBF fields definitions.
TODO:
- make memos work
"""
"""History (most recent first):
26-may-2009 [als] DbfNumericFieldDef.decodeValue: strip zero bytes
05-feb-2009 [als] DbfDateFieldDef.encodeValue: empty arg produces empty date
16-sep-2008 [als] DbfNumericFieldDef decoding looks for decimal point
in the value to select float or integer return type
13-mar-2008 [als] check field name length in constructor
11-feb-2007 [als] handle value conversion errors
10-feb-2007 [als] DbfFieldDef: added .rawFromRecord()
01-dec-2006 [als] Timestamp columns use None for empty values
31-oct-2006 [als] support field types 'F' (float), 'I' (integer)
and 'Y' (currency);
automate export and registration of field classes
04-jul-2006 [als] added export declaration
10-mar-2006 [als] decode empty values for Date and Logical fields;
show field name in errors
10-mar-2006 [als] fix Numeric value decoding: according to spec,
value always is string representation of the number;
ensure that encoded Numeric value fits into the field
20-dec-2005 [yc] use field names in upper case
15-dec-2005 [yc] field definitions moved from `dbf`.
"""
__version__ = "$Revision: 1.14 $"[11:-2]
__date__ = "$Date: 2009/05/26 05:16:51 $"[7:-2]
__all__ = ["lookupFor",] # field classes added at the end of the module
import datetime
import struct
import sys
from . import utils
## abstract definitions
class DbfFieldDef(object):
"""Abstract field definition.
Child classes must override ``type`` class attribute to provide datatype
infromation of the field definition. For more info about types visit
`http://www.clicketyclick.dk/databases/xbase/format/data_types.html`
Also child classes must override ``defaultValue`` field to provide
default value for the field value.
If child class has fixed length ``length`` class attribute must be
overriden and set to the valid value. None value means, that field
isn't of fixed length.
Note: ``name`` field must not be changed after instantiation.
"""
__slots__ = ("name", "length", "decimalCount",
"start", "end", "ignoreErrors")
# length of the field, None in case of variable-length field,
# or a number if this field is a fixed-length field
length = None
# field type. for more information about fields types visit
# `http://www.clicketyclick.dk/databases/xbase/format/data_types.html`
# must be overriden in child classes
typeCode = None
# default value for the field. this field must be
# overriden in child classes
defaultValue = None
def __init__(self, name, length=None, decimalCount=None,
start=None, stop=None, ignoreErrors=False,
):
"""Initialize instance."""
assert self.typeCode is not None, "Type code must be overriden"
assert self.defaultValue is not None, "Default value must be overriden"
## fix arguments
if len(name) >10:
raise ValueError("Field name \"%s\" is too long" % name)
name = str(name).upper()
if self.__class__.length is None:
if length is None:
raise ValueError("[%s] Length isn't specified" % name)
length = int(length)
if length <= 0:
raise ValueError("[%s] Length must be a positive integer"
% name)
else:
length = self.length
if decimalCount is None:
decimalCount = 0
## set fields
self.name = name
# FIXME: validate length according to the specification at
# http://www.clicketyclick.dk/databases/xbase/format/data_types.html
self.length = length
self.decimalCount = decimalCount
self.ignoreErrors = ignoreErrors
self.start = start
self.end = stop
def __cmp__(self, other):
return cmp(self.name, str(other).upper())
def __hash__(self):
return hash(self.name)
def fromString(cls, string, start, ignoreErrors=False):
"""Decode dbf field definition from the string data.
Arguments:
string:
a string, dbf definition is decoded from. length of
the string must be 32 bytes.
start:
position in the database file.
ignoreErrors:
initial error processing mode for the new field (boolean)
"""
assert len(string) == 32
_length = ord(string[16])
return cls(utils.unzfill(string)[:11], _length, ord(string[17]),
start, start + _length, ignoreErrors=ignoreErrors)
fromString = classmethod(fromString)
def toString(self):
"""Return encoded field definition.
Return:
Return value is a string object containing encoded
definition of this field.
"""
if sys.version_info < (2, 4):
# earlier versions did not support padding character
_name = self.name[:11] + "\0" * (11 - len(self.name))
else:
_name = self.name.ljust(11, '\0')
return (
_name +
self.typeCode +
#data address
chr(0) * 4 +
chr(self.length) +
chr(self.decimalCount) +
chr(0) * 14
)
def __repr__(self):
return "%-10s %1s %3d %3d" % self.fieldInfo()
def fieldInfo(self):
"""Return field information.
Return:
Return value is a (name, type, length, decimals) tuple.
"""
return (self.name, self.typeCode, self.length, self.decimalCount)
def rawFromRecord(self, record):
"""Return a "raw" field value from the record string."""
return record[self.start:self.end]
def decodeFromRecord(self, record):
"""Return decoded field value from the record string."""
try:
return self.decodeValue(self.rawFromRecord(record))
except:
if self.ignoreErrors:
return utils.INVALID_VALUE
else:
raise
def decodeValue(self, value):
"""Return decoded value from string value.
This method shouldn't be used publicly. It's called from the
`decodeFromRecord` method.
This is an abstract method and it must be overridden in child classes.
"""
raise NotImplementedError
def encodeValue(self, value):
"""Return str object containing encoded field value.
This is an abstract method and it must be overriden in child classes.
"""
raise NotImplementedError
## real classes
class DbfCharacterFieldDef(DbfFieldDef):
"""Definition of the character field."""
typeCode = "C"
defaultValue = ""
def decodeValue(self, value):
"""Return string object.
Return value is a ``value`` argument with stripped right spaces.
"""
return value.rstrip(" ")
def encodeValue(self, value):
"""Return raw data string encoded from a ``value``."""
return str(value)[:self.length].ljust(self.length)
class DbfNumericFieldDef(DbfFieldDef):
"""Definition of the numeric field."""
typeCode = "N"
# XXX: now I'm not sure it was a good idea to make a class field
# `defaultValue` instead of a generic method as it was implemented
# previously -- it's ok with all types except number, cuz
# if self.decimalCount is 0, we should return 0 and 0.0 otherwise.
defaultValue = 0
def decodeValue(self, value):
"""Return a number decoded from ``value``.
If decimals is zero, value will be decoded as an integer;
or as a float otherwise.
Return:
Return value is a int (long) or float instance.
"""
value = value.strip(" \0")
if "." in value:
# a float (has decimal separator)
return float(value)
elif value:
# must be an integer
return int(value)
else:
return 0
def encodeValue(self, value):
"""Return string containing encoded ``value``."""
_rv = ("%*.*f" % (self.length, self.decimalCount, value))
if len(_rv) > self.length:
_ppos = _rv.find(".")
if 0 <= _ppos <= self.length:
_rv = _rv[:self.length]
else:
raise ValueError("[%s] Numeric overflow: %s (field width: %i)"
% (self.name, _rv, self.length))
return _rv
class DbfFloatFieldDef(DbfNumericFieldDef):
"""Definition of the float field - same as numeric."""
typeCode = "F"
class DbfIntegerFieldDef(DbfFieldDef):
"""Definition of the integer field."""
typeCode = "I"
length = 4
defaultValue = 0
def decodeValue(self, value):
"""Return an integer number decoded from ``value``."""
return struct.unpack("<i", value)[0]
def encodeValue(self, value):
"""Return string containing encoded ``value``."""
return struct.pack("<i", int(value))
class DbfCurrencyFieldDef(DbfFieldDef):
"""Definition of the currency field."""
typeCode = "Y"
length = 8
defaultValue = 0.0
def decodeValue(self, value):
"""Return float number decoded from ``value``."""
return struct.unpack("<q", value)[0] / 10000.
def encodeValue(self, value):
"""Return string containing encoded ``value``."""
return struct.pack("<q", round(value * 10000))
class DbfLogicalFieldDef(DbfFieldDef):
"""Definition of the logical field."""
typeCode = "L"
defaultValue = -1
length = 1
def decodeValue(self, value):
"""Return True, False or -1 decoded from ``value``."""
# Note: value always is 1-char string
if value == "?":
return -1
if value in "NnFf ":
return False
if value in "YyTt":
return True
raise ValueError("[%s] Invalid logical value %r" % (self.name, value))
def encodeValue(self, value):
"""Return a character from the "TF?" set.
Return:
Return value is "T" if ``value`` is True
"?" if value is -1 or False otherwise.
"""
if value is True:
return "T"
if value == -1:
return "?"
return "F"
class DbfMemoFieldDef(DbfFieldDef):
"""Definition of the memo field.
Note: memos aren't currenly completely supported.
"""
typeCode = "M"
defaultValue = " " * 10
length = 10
def decodeValue(self, value):
"""Return int .dbt block number decoded from the string object."""
#return int(value)
raise NotImplementedError
def encodeValue(self, value):
"""Return raw data string encoded from a ``value``.
Note: this is an internal method.
"""
#return str(value)[:self.length].ljust(self.length)
raise NotImplementedError
class DbfDateFieldDef(DbfFieldDef):
"""Definition of the date field."""
typeCode = "D"
defaultValue = utils.classproperty(lambda cls: datetime.date.today())
# "yyyymmdd" gives us 8 characters
length = 8
def decodeValue(self, value):
"""Return a ``datetime.date`` instance decoded from ``value``."""
if value.strip():
return utils.getDate(value)
else:
return None
def encodeValue(self, value):
"""Return a string-encoded value.
``value`` argument should be a value suitable for the
`utils.getDate` call.
Return:
Return value is a string in format "yyyymmdd".
"""
if value:
return utils.getDate(value).strftime("%Y%m%d")
else:
return " " * self.length
class DbfDateTimeFieldDef(DbfFieldDef):
"""Definition of the timestamp field."""
# a difference between JDN (Julian Day Number)
# and GDN (Gregorian Day Number). note, that GDN < JDN
JDN_GDN_DIFF = 1721425
typeCode = "T"
defaultValue = utils.classproperty(lambda cls: datetime.datetime.now())
# two 32-bits integers representing JDN and amount of
# milliseconds respectively gives us 8 bytes.
# note, that values must be encoded in LE byteorder.
length = 8
def decodeValue(self, value):
"""Return a `datetime.datetime` instance."""
assert len(value) == self.length
# LE byteorder
_jdn, _msecs = struct.unpack("<2I", value)
if _jdn >= 1:
_rv = datetime.datetime.fromordinal(_jdn - self.JDN_GDN_DIFF)
_rv += datetime.timedelta(0, _msecs / 1000.0)
else:
# empty date
_rv = None
return _rv
def encodeValue(self, value):
"""Return a string-encoded ``value``."""
if value:
value = utils.getDateTime(value)
# LE byteorder
_rv = struct.pack("<2I", value.toordinal() + self.JDN_GDN_DIFF,
(value.hour * 3600 + value.minute * 60 + value.second) * 1000)
else:
_rv = "\0" * self.length
assert len(_rv) == self.length
return _rv
_fieldsRegistry = {}
def registerField(fieldCls):
"""Register field definition class.
``fieldCls`` should be subclass of the `DbfFieldDef`.
Use `lookupFor` to retrieve field definition class
by the type code.
"""
assert fieldCls.typeCode is not None, "Type code isn't defined"
# XXX: use fieldCls.typeCode.upper()? in case of any decign
# don't forget to look to the same comment in ``lookupFor`` method
_fieldsRegistry[fieldCls.typeCode] = fieldCls
def lookupFor(typeCode):
"""Return field definition class for the given type code.
``typeCode`` must be a single character. That type should be
previously registered.
Use `registerField` to register new field class.
Return:
Return value is a subclass of the `DbfFieldDef`.
"""
# XXX: use typeCode.upper()? in case of any decign don't
# forget to look to the same comment in ``registerField``
return _fieldsRegistry[typeCode]
## register generic types
for (_name, _val) in globals().items():
if isinstance(_val, type) and issubclass(_val, DbfFieldDef) \
and (_name != "DbfFieldDef"):
__all__.append(_name)
registerField(_val)
del _name, _val
# vim: et sts=4 sw=4 :