This package provides useful class definitions for configuring scripts in the Gnip-Analysis-Pipeline package. The intention is that you work from a working directory (we'll call this "TEST"), and that both gnip_analysis_pipeline and gnip_analysis_config are installed as packages. Remember that these packages can be installed from the cloned repo location with:

[REPO] $ pip install -e .

According to the Gnip-Analysis-Pipeline docs, we configure enrichments by defining the enrichment_class_list variable in a configuration file.

The enrichments directory in this package contains files that define a base enrichment class along with some other helpful enrichment classes, including a simple example. To use the test enrichment from your working directory, you would create an enrichments configuration file (called my_enrichments.py):

from gnip_analysis_config.enrichments import test_enrichment

enrichment_class_list = [test_enrichment.TestEnrichment]

We can the enrich the Tweets in my_tweets.json as follows:

[TEST] $ cat my_tweets.json | tweet_enricher.py -c my_enrichments.py > my_enriched_tweets.json

To configure an NLP enrichment with NLTK, we provide nltk_enrichment.py, which can be configured like:

from gnip_analysis_config.enrichments import nltk_enrichment

enrichment_class_list = nltk_enrichment.nltk_enrichments_list

Notice that this module has conveniently defined the list of enrichment classes.

A custom enrichment class can be defined locally:

from gnip_analysis_config.enrichments import base_enrichment

class MyEnrichment(enrichment_base.BaseEnrichment):
    def enrichment_value(self,tweet):
        return "my_test_enrichment_value"

According to the Gnip-Analysis-Pipeline docs, we configure measurementss by defining the measurements_class_list variable in a configuration file.

The measurements directory in this package contains files that contain a variety of base/helper classes for construction measurement classes. To use the test measurement from your working directory, you would create an enrichments configuration file (called my_measurements.py):

from gnip_analysis_config.measurements.test_measurements import TweetCounter,ReTweetCounter

enrichment_class_list = [TweetCounter,ReTweetCounter]

We can the build time series from the Tweets in my_enriched_tweets.json as follows:

[TEST] $ cat my_enriched_tweets.json | tweet_time_series_builder.py -c my_measurements.py > time_series.csv

(Note that none of the enrichments we added in the previous section are required to build the specified time series.)

To construct a time series for each observed hashtag, we can define a class locally that inherits key functionality from classes in measurement_base.py:

from gnip_analysis_config.measurements.measurement_base import Counters

class HashtagCounters(Counters):
    def update(self,tweet):
        for item in tweet['twitter_entities']['hashtags']:
        # put a # in from of the term,
        # since they've been removed in the payload
        self.counters['#'+item['text']] += 1

measurement_class_list = [ HashtagCounters ]

See measurement_base.py for a full description of how to create custom measurement classes.

This package is a repository for common analysis tools. For example nlp/utils.py defines some common choices for NLP utility functions.