Version History¶

2.3.0 (2019-04-18)¶

Add pamqp.encode.support_deprecated_rabbitmq() function to limit data types available when encoding field-tables for older RabbitMQ versions.

Change pamqp.encode.timestamp() to allow for numeric/epoch timestamps (#14 - mosquito)
Change pamqp.frame.frame_parts() to a public method (#15 - mosquito)
Cleanup of code to pass configured flake8 tests
Add support for 8-bit unsigned integer values in pamqp.encode.table_integer()

Change raising a DeprecationWarning exception to using warnings.warn for deprecated AMQP methods (#13 - dzen)

Change Python versions supported to 2.7 and 3.4+
Always decode field table keys as strings (#6)
- This may be a breaking change means in Python3 keys will always be type str for short strings. This includes frame values and field table values.
- In Python 2.7 if a short-string (key, frame field value, etc) has UTF-8 characters in it, it will be a unicode object.
Combine test coverage across all Python versions
Fix range for signed short integer (#7)
Fix guards for usage of unsigned short usage in pamqp.encode (#7)
Fix encoding and decoding of unsigned short (#7)
Add support for unsigned short integer and long integer in field tables (#10)
Address edge case of small value in long type (#8)
Address long string encoding inconsistency (#9)
Cleanup unicode object & conditionals in py3 (#9)
Add pamqp.exceptions.PAMQPException as a base class for pamqp specific exceptions (#4)
Fix decoding of void values in a field table or array

Remove UTF-8 encoding from byte_array (#2)
Fix AMQP Field Tables / Basic.Properties headers behavior:
- Field names per spec should not exceed 128 bytes
- long-strings should not be utf-8 encoded (only short-strings boggle)
Ensure that field table long strings are not coerced to UTF-8 as specified in AMQP 0-9-1

If a string is passed in as a long string in a field table and it contains UTF-8 characters it will be UTF-8 encoded
Move AMQP Methods in specification.py to slotted classes
Change Basic.Properties to a slotted class
Instead of class level attributes with the same name as obj attributes, prefix class attributes for data types with an underscore
Add new class method type() for Basic.Properties for accessing data type
Add new class method type() for AMQP methods for accessing data type
Change Basic.Properties.attributes to Basic.Properties.attributes(), returning the list of slotted attributes
Fix a typo for booleans in the method mapping for table decoding
Frame.__getitem__ will now raise a KeyError instead of None for an invalid attribute
PropertiesBase no longer checks to see if an attribute is set for contains
Adds new specification tests
More efficiently handle the frame end character in Python 3

Cleanup how UTF-8 is handled in decoding strings
Ensure that field tables (headers property, etc) can use keys with utf-8 data
Address missing and mis-aligned AMQP-0-9-1 field table decoding with the field type indicators from the RabbitMQ protocol errata page
Fix a encoding by type bug introduced with 1.4 having to do with bytearrays
Be explicit about needing a class id in the ContentHeader
Update the tests to reflect the unicode changes
Clean up the tests

Fix a long standing bug for non-specified responses for RabbitMQ AMQP extensions
Refactor adding bytearrays and recoding complexity
Add bytearray support (#1 and gmr/rabbitpy#48)
Change encode/decode type errors from ValueError to TypeError exceptions
Remove separate codecs for Python 2 & 3
Move codecs from pamqp.codec.encode and pamqp.codec.decode to pamqp.encode and pamqp.decode
Deprecate pamqp.codec
Remove weird imports from top level __init__.py, not sure what I was thinking there
Clean up codegen a bit to make it more PYTHON3 compatible
Update codegen/include for new codec and PYTHON2/PYTHON3 behavior
Update documentation
Distribution updates:
- Let travis upload to pypi
- Add wheel distribution
- Update supported python versions
- Update classifiers