tweet-parser 1.13.2

Goal:

Allow reliable parsing of Tweets delivered by the Gnip platform, in both activity-streams and original formats.

Status:

This package can be installed by cloning the repo and using pip install -e ., or by using pip install tweet_parser.

As of version 1.0.5, the package works with Python 2 and 3, and the API should be relatively stable. Recommended to use the more recent release. Current release is 1.13.2

Currently, this parser does not explicitly support Public API Twitter data.

Usage:

This package is intended to be used as a Python module inside your other Tweet-related code. An example Python program (after pip installing the package) would be:

from tweet_parser.tweet import Tweet
from tweet_parser.tweet_parser_errors import NotATweetError
import fileinput
import json

for line in fileinput.FileInput("gnip_tweet_data.json"):
    try:
        tweet_dict = json.loads(line)
        tweet = Tweet(tweet_dict)
    except (json.JSONDecodeError,NotATweetError):
        pass
    print(tweet.created_at_string, tweet.all_text)

I’ve also added simple command-line utility:

python tools/parse_tweets.py -f"gnip_tweet_data.json" -c"created_at_string,all_text"

Testing:

A Python test_tweet_parser.py package exists in test/.

The most important thing that it tests is the equivalence of outputs when comparing both activity-streams input and original-format input. Any new getter will be tested by running test$ python test_tweet_parser.py, as the test checks every method attached to the Tweet object, for every test tweet stored in test/tweet_payload_examples. For any cases where it is expected that the outputs are different (e.g., outputs that depend on poll options), conditional statements should be added to this test.

An option also exists for run-time checking of Tweet payload formats. This compares the set of all Tweet field keys to a superset of all possible keys, as well as a minimum set of all required keys, to make sure that each newly loaded Tweet fits those parameters. This shouldn’t be run every time you load Tweets (for one, it’s slow), but is implemented to use as a periodic check against Tweet format changes. This option is enabled with --do_format_validation on the command line, and by setting the keyword argument do_format_validation to True when initializing a Tweet object.

Contributing

Submit bug reports or feature requests through GitHub Issues, with self-contained minimum working examples where appropriate.

To contribute code, fork this repo, create your own local feature branch, make your changes, test them, and submit a pull request to the master branch. The contribution guidelines specified in the pandas documentation are a great reference.

When you submit a change, change the version number. For bug fixes and non-breaking changes that do not affect the top-level Tweet object API (fixing a bug or changing the internals of a getter while package naming/structure remains the same), increment the last number (X.Y.Z -> X.Y.Z+1) in setup.py. For changes that do affect the top-level Tweet object API (e.g., adding a new getter), increment the middle number (X.Y.Z -> X.Y+1.0).

(env) [tweet_parser]$ pip install -e .
(env) [tweet_parser]$ cd test/; python test_tweet_parser.py; cd -

(env) [tweet_parser]$ pip install -e .
(env) [tweet_parser]$ python tools/parse_tweets.py -f test/tweet_payload_examples/activity_streams_examples.json -c <your new field>