Release 0.4

.gitignore

@@ -2,6 +2,7 @@
 __pycache__/
 *.py[cod]
 *$py.class
+.pytest_cache
 
 # C extensions
 *.so

.travis.yml

@@ -4,6 +4,9 @@ python:
 - '3.6'
 install:
 - pip install ".[testing]"
+- pip install ".[nlp]"
+- pip install ".[s3driver]"
+- python -m nltk.downloader punkt stopwords wordnet
 script: pytest
 deploy:
   provider: pypi

quantgov/__init__.py

@@ -3,11 +3,14 @@
 
 __all__ = [
     'corpora',
+    'corpus',
     'estimator',
     'project',
     'utils',
 ]
 
-from .corpora.utils import load_driver
+from . import corpora  # Backwards compatibility
 
-__version__ = '0.3.3'
+from .utils import load_driver
+
+__version__ = '0.4.0'

quantgov/__main__.py

@@ -15,7 +15,7 @@
 import requests
 
 import quantgov
-import quantgov.corpora.builtins
+import quantgov.corpus.builtins
 
 from pathlib import Path
 
@@ -40,7 +40,7 @@ def parse_args():
     # Corpus command
     corpus = subparsers.add_parser('corpus')
     corpus_subcommands = corpus.add_subparsers(dest='subcommand')
-    for command, builtin in quantgov.corpora.builtins.commands.items():
+    for command, builtin in quantgov.corpus.builtins.commands.items():
         subcommand = corpus_subcommands.add_parser(
             command, help=builtin.cli.help)
         subcommand.add_argument(
@@ -118,6 +118,9 @@ def parse_args():
     estimate.add_argument(
         '--probability', action='store_true',
         help='output probabilities instead of predictions')
+    estimate.add_argument(
+        '--precision', default=4, type=int,
+        help='number of decimal places to round the probabilities')
     estimate.add_argument(
         '-o', '--outfile',
         type=lambda x: open(x, 'w', newline='', encoding='utf-8'),
@@ -161,7 +164,7 @@ def start_component(args):
 def run_corpus_builtin(args):
     driver = quantgov.load_driver(args.corpus)
     writer = csv.writer(args.outfile)
-    builtin = quantgov.corpora.builtins.commands[args.subcommand]
+    builtin = quantgov.corpus.builtins.commands[args.subcommand]
     func_args = {i: j for i, j in vars(args).items()
                  if i not in {'command', 'subcommand', 'outfile', 'corpus'}}
     writer.writerow(driver.index_labels + builtin.get_columns(func_args))
@@ -187,7 +190,7 @@ def run_estimator(args):
     elif args.subcommand == "estimate":
         quantgov.estimator.estimate(
             args.vectorizer, args.model, args.corpus, args.probability,
-            args.outfile
+            args.precision, args.outfile
         )
 
 

quantgov/corpora/__init__.py

@@ -1,9 +1,18 @@
-from .structures import (
+import warnings
+
+from ..corpus import (
     Document,
     CorpusStreamer,
     CorpusDriver,
     FlatFileCorpusDriver,
     RecursiveDirectoryCorpusDriver,
     NamePatternCorpusDriver,
-    IndexDriver
+    IndexDriver,
+    S3Driver,
+    S3DatabaseDriver
 )
+
+warnings.warn(
+    ("quantgov.corpora has been moved to quantgov.corpus and will be removed"
+     " in a future version."),
+    DeprecationWarning)

quantgov/corpus/__init__.py

@@ -0,0 +1,11 @@
+from .structures import (
+    Document,
+    CorpusStreamer,
+    CorpusDriver,
+    FlatFileCorpusDriver,
+    RecursiveDirectoryCorpusDriver,
+    NamePatternCorpusDriver,
+    IndexDriver,
+    S3Driver,
+    S3DatabaseDriver
+)