initial 0.1.9 release commit

LexPredict · Jul 3, 2018 · 1997efd · 1997efd
1 parent b6df9a0
commit 1997efd
Show file tree

Hide file tree

Showing 103 changed files with 703 additions and 228 deletions.
diff --git a/.gitignore b/.gitignore
@@ -111,3 +111,6 @@ ENV/
 .mypy_cache/
 
 benchmarks
+lexnlp/extract/en/contracts/data/*.model
+/.pytest_cache/
+
diff --git a/MANIFEST.in b/MANIFEST.in
@@ -2,6 +2,7 @@ include README.md
 include python-requirements.txt
 recursive-include lexnlp *.pickle
 recursive-include lexnlp/extract/en/addresses *.json *.txt *.xml
+recursive-include lexnlp/extract/en/contracts *.model.part.*
 recursive-include lexnlp *.csv
 recursive-include libs *
 recursive-include scripts *
diff --git a/lexnlp/__init__.py b/lexnlp/__init__.py
@@ -2,9 +2,9 @@
 import os
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
 

diff --git a/lexnlp/config/__init__.py b/lexnlp/config/__init__.py
@@ -1,6 +1,6 @@
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
diff --git a/lexnlp/config/en/__init__.py b/lexnlp/config/en/__init__.py
@@ -1,6 +1,6 @@
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
diff --git a/lexnlp/config/en/company_types.py b/lexnlp/config/en/company_types.py
@@ -13,9 +13,9 @@
 
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
 

diff --git a/lexnlp/config/en/geoentities_config.py b/lexnlp/config/en/geoentities_config.py
@@ -3,9 +3,9 @@
 """
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
 

diff --git a/lexnlp/config/stanford.py b/lexnlp/config/stanford.py
@@ -6,9 +6,9 @@
 from lexnlp import get_lib_path
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
 

diff --git a/lexnlp/extract/__init__.py b/lexnlp/extract/__init__.py
@@ -1,6 +1,6 @@
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
diff --git a/lexnlp/extract/en/__init__.py b/lexnlp/extract/en/__init__.py
@@ -1,6 +1,6 @@
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
diff --git a/lexnlp/extract/en/addresses/address_features.py b/lexnlp/extract/en/addresses/address_features.py
@@ -15,7 +15,7 @@
 from dateutil import parser as dateparser
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
 __version__ = "0.1.6"
 __maintainer__ = "LexPredict, LLC"

diff --git a/lexnlp/extract/en/addresses/addresses.py b/lexnlp/extract/en/addresses/addresses.py
@@ -12,7 +12,7 @@
 from lexnlp.extract.en.addresses import address_features
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
 __version__ = "0.1.6"
 __maintainer__ = "LexPredict, LLC"

diff --git a/lexnlp/extract/en/addresses/tests/test_addresses.py b/lexnlp/extract/en/addresses/tests/test_addresses.py
@@ -2,7 +2,7 @@
 from lexnlp.tests import lexnlp_tests
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
 __version__ = "0.1.6"
 __maintainer__ = "LexPredict, LLC"

diff --git a/lexnlp/extract/en/amounts.py b/lexnlp/extract/en/amounts.py
@@ -35,9 +35,9 @@
 from num2words import num2words
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
 

diff --git a/lexnlp/extract/en/citations.py b/lexnlp/extract/en/citations.py
@@ -12,9 +12,9 @@
 from reporters_db import EDITIONS, REPORTERS
 
 __author__ = "ContraxSuite, LLC; LexPredict, LLC"
-__copyright__ = "Copyright 2015-2017, ContraxSuite, LLC"
+__copyright__ = "Copyright 2015-2018, ContraxSuite, LLC"
 __license__ = "https://github.com/LexPredict/lexpredict-lexnlp/blob/master/LICENSE"
-__version__ = "0.1.8"
+__version__ = "0.1.9"
 __maintainer__ = "LexPredict, LLC"
 __email__ = "support@contraxsuite.com"
 
@@ -44,28 +44,31 @@ def get_citations(text, return_source=False, as_dict=False) -> Generator:
 
     for source_text, volume, reporter, page, page2, court, year\
             in CITATION_PTN_RE.findall(text):
-        reporter_data = REPORTERS[EDITIONS[reporter]]
-        reporter_full_name = ''
-        if len(reporter_data) == 1:
-            reporter_full_name = reporter_data[0]['name']
-        elif year:
-            for period_data in reporter_data:
-                if reporter in period_data['editions']:
-                    start = period_data['editions'][reporter]['start'].year
-                    end = period_data['editions'][reporter]['end']
-                    if (end and start <= int(year) <= end.year) or start <= int(year):
-                        reporter_full_name = period_data['name']
-        item = (int(volume),
-                reporter,
-                reporter_full_name,
-                int(page),
-                page2 or None,
-                court.strip(', ') or None,
-                int(year) if year.isdigit() else None)
-        if return_source:
-            item += (source_text.strip(),)
-        if as_dict:
-            keys = ['volume', 'reporter', 'reporter_full_name',
-                    'page', 'page2', 'court', 'year', 'citation_str']
-            item = {keys[n]: val for n, val in enumerate(item)}
-        yield item
+        try:
+            reporter_data = REPORTERS[EDITIONS[reporter]]
+            reporter_full_name = ''
+            if len(reporter_data) == 1:
+                reporter_full_name = reporter_data[0]['name']
+            elif year:
+                for period_data in reporter_data:
+                    if reporter in period_data['editions']:
+                        start = period_data['editions'][reporter]['start'].year
+                        end = period_data['editions'][reporter]['end']
+                        if (end and start <= int(year) <= end.year) or start <= int(year):
+                            reporter_full_name = period_data['name']
+            item = (int(volume),
+                    reporter,
+                    reporter_full_name,
+                    int(page),
+                    page2 or None,
+                    court.strip(', ') or None,
+                    int(year) if year.isdigit() else None)
+            if return_source:
+                item += (source_text.strip(),)
+            if as_dict:
+                keys = ['volume', 'reporter', 'reporter_full_name',
+                        'page', 'page2', 'court', 'year', 'citation_str']
+                item = {keys[n]: val for n, val in enumerate(item)}
+            yield item
+        except KeyError:
+            pass