66 increase tests

hamlet/theses/fixtures/theses.json

@@ -1 +1 @@
-[{"model": "theses.thesis", "pk": 32600, "fields": {"title": "Induced polarization of metallic minerals : a study of its chemical basis.", "degree": "Ph.D. / Sc.D.", "url": "https://dspace.mit.edu/bitstream/1721.1/17134/2/02361749-MIT.pdf", "year": 1976, "identifier": 17134, "unextractable": true, "_vector": null, "department": [287]}}, {"model": "theses.thesis", "pk": 43703, "fields": {"title": "Clock division as a power saving strategy in a system constrained by high transmission frequency and low data rate", "degree": "Master's degree", "url": "https://dspace.mit.edu/bitstream/1721.1/33360/2/62413893-MIT.pdf", "year": 2005, "identifier": 33360, "unextractable": false, "_vector": null, "department": [275]}}, {"model": "theses.thesis", "pk": 60330, "fields": {"title": "Architecture for ultra-low power multi-channel transmitters for Body Area Networks using RF resonators", "degree": "Master's degree", "url": "https://dspace.mit.edu/bitstream/1721.1/66473/2/756456998-MIT.pdf", "year": 2011, "identifier": 66473, "unextractable": false, "_vector": null, "department": [275]}}, {"model": "theses.thesis", "pk": 76265, "fields": {"title": "Ultra low power, high sensitivity secure wake-up receiver for the Internet of Things", "degree": "Master's degree", "url": "https://dspace.mit.edu/bitstream/1721.1/111908/1/1005140973-MIT.pdf", "year": 2017, "identifier": 111908, "unextractable": false, "_vector": null, "department": [275]}}]
+[{"model": "theses.thesis", "pk": 32600, "fields": {"title": "Induced polarization of metallic minerals : a study of its chemical basis.", "degree": "Ph.D. / Sc.D.", "url": "https://dspace.mit.edu/bitstream/1721.1/17134/2/02361749-MIT.pdf", "year": 1976, "identifier": 17134, "unextractable": true, "department": [287]}}, {"model": "theses.thesis", "pk": 43703, "fields": {"title": "Clock division as a power saving strategy in a system constrained by high transmission frequency and low data rate", "degree": "Master's degree", "url": "https://dspace.mit.edu/bitstream/1721.1/33360/2/62413893-MIT.pdf", "year": 2005, "identifier": 33360, "unextractable": false, "department": [275]}}, {"model": "theses.thesis", "pk": 60330, "fields": {"title": "Architecture for ultra-low power multi-channel transmitters for Body Area Networks using RF resonators", "degree": "Master's degree", "url": "https://dspace.mit.edu/bitstream/1721.1/66473/2/756456998-MIT.pdf", "year": 2011, "identifier": 66473, "unextractable": false, "department": [275]}}, {"model": "theses.thesis", "pk": 76265, "fields": {"title": "Ultra low power, high sensitivity secure wake-up receiver for the Internet of Things", "degree": "Master's degree", "url": "https://dspace.mit.edu/bitstream/1721.1/111908/1/1005140973-MIT.pdf", "year": 2017, "identifier": 111908, "unextractable": false, "department": [275]}}]

hamlet/theses/migrations/0007_remove_thesis__vector.py

@@ -0,0 +1,19 @@
+# -*- coding: utf-8 -*-
+# Generated by Django 1.11.10 on 2018-02-06 21:14
+from __future__ import unicode_literals
+
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('theses', '0006_auto_20171120_1935'),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name='thesis',
+            name='_vector',
+        ),
+    ]

hamlet/theses/models.py

@@ -56,7 +56,7 @@ def clean_metadata(namestring):
                    ', Nav.E. Massachusetts Institute of Technology',
                    ', Nav. E. Massachusetts Institute of Technology',
                    ', M.B.A. Massachusetts Institute of Technology',
-                   ' Massachusetts Institute of Technology']
+                   ', Massachusetts Institute of Technology']
 
         for deg in degrees:
             names = [name.replace(deg, '') for name in names]
@@ -94,9 +94,7 @@ def clean_metadata(deptstring):
 
     @staticmethod
     def get_or_create_from_metadata(metadata):
-        print(metadata)
         clean = Department.clean_metadata(metadata)
-        print(clean)
         dept, _ = Department.objects.get_or_create(name=clean)
         return dept
 
@@ -136,15 +134,6 @@ class Thesis(models.Model):
         help_text='Will be set to True if attempts to extract text from '
             'the pdf failed; such theses are not part of the neural net, '
             'and cannot be used in data visualization.')
-    # Contains the inferred vector for a document. Always needed for getting
-    # neural net data on documents outside of the training set. Documents
-    # inside the training set use their labels where possible, but cannot
-    # always do so (e.g. if being compared to documents outside the training
-    # set).
-    # The inferred vector is a numpy array; we use pickle to serialize it into
-    # bytes before storing it here. Access via the vector property, which
-    # unpickles.
-    _vector = models.BinaryField(blank=True, null=True)
 
     def __str__(self):
         return self.title
@@ -153,10 +142,6 @@ def __str__(self):
     def label(self):
         return '1721.1-{}.txt'.format(self.identifier)
 
-    @cached_property
-    def vector(self):
-        return pickle.loads(self._vector)
-
     @cached_property
     def authors(self):
         contribs = Contribution.objects.filter(

hamlet/theses/tests/test_models.py

@@ -0,0 +1,176 @@
+from django.test import TestCase
+
+from ..models import Person, Department, Thesis
+
+
+class PersonTestCase(TestCase):
+    fixtures = ['theses.json', 'departments.json', 'authors.json',
+                'contributions.json']
+
+    def test_clean_metadata_remove_degrees_1(self):
+        base = 'Kofi Annan, S.M. Massachusetts Institute of Technology'
+        self.assertEqual(['Kofi Annan'], Person.clean_metadata(base))
+
+    def test_clean_metadata_remove_degrees_2(self):
+        base = 'Somebody, M. Eng. Massachusetts Institute of Technology'
+        self.assertEqual(['Somebody'], Person.clean_metadata(base))
+
+    def test_clean_metadata_remove_degrees_3(self):
+        base = 'Ronald McNair, Ph. D. Massachusetts Institute of Technology'
+        self.assertEqual(['Ronald McNair'], Person.clean_metadata(base))
+
+    def test_clean_metadata_remove_degrees_4(self):
+        base = 'Boaty McBoatface, Nav. E. Massachusetts Institute of Technology'  # noqa
+        self.assertEqual(['Boaty McBoatface'], Person.clean_metadata(base))
+
+    def test_clean_metadata_remove_degrees_5(self):
+        base = 'Boaty von Boatface, Nav.E. Massachusetts Institute of Technology'  # noqa
+        self.assertEqual(['Boaty von Boatface'], Person.clean_metadata(base))
+
+    def test_clean_metadata_remove_degrees_6(self):
+        base = 'The Man, M.B.A. Massachusetts Institute of Technology'
+        self.assertEqual(['The Man'], Person.clean_metadata(base))
+
+    def test_clean_metadata_remove_degrees_7(self):
+        base = 'Buzz Aldrin, Massachusetts Institute of Technology'
+        self.assertEqual(['Buzz Aldrin'], Person.clean_metadata(base))
+
+    def test_clean_metadata_multiple_people(self):
+        base = 'Dr. Jekyll and Mr. Hyde'
+        self.assertEqual(['Dr. Jekyll', 'Mr. Hyde'],
+                         Person.clean_metadata(base))
+
+
+class DepartmentTestCase(TestCase):
+    fixtures = ['theses.json', 'departments.json', 'authors.json',
+                'contributions.json']
+
+    def test_clean_metadata_1(self):
+        base = 'Massachusetts Institute of Technology. Department of Basketweaving'  # noqa
+        self.assertEqual('Department of Basketweaving',
+                         Department.clean_metadata(base))
+
+    def test_clean_metadata_2(self):
+        base = 'Dept. of Basketweaving'
+        self.assertEqual('Department of Basketweaving',
+                         Department.clean_metadata(base))
+
+    def test_get_or_create_from_metadata(self):
+        # A department is created.
+        base = 'Department of Basketweaving'
+        dept = Department.get_or_create_from_metadata(base)
+        self.assertEqual(dept.name, 'Department of Basketweaving')
+
+        # New departments are not created when we are fed metadata we already
+        # know about - instead we fetch an existing one.
+        base = 'Department of Basketweaving'
+        dept2 = Department.get_or_create_from_metadata(base)
+        self.assertEqual(dept.pk, dept2.pk)
+
+        base = 'Dept. of Basketweaving'
+        dept3 = Department.get_or_create_from_metadata(base)
+        self.assertEqual(dept.pk, dept3.pk)
+
+
+class ThesisTestCase(TestCase):
+    fixtures = ['theses.json', 'departments.json', 'authors.json',
+                'contributions.json']
+
+    def test_add_single_new_person(self):
+        name = 'Whitfield Diffie'
+        thesis = Thesis.objects.first()
+
+        assert not Person.objects.filter(name=name)
+
+        thesis.add_people([name])
+        # Don't use thesis.authors - since it's a cached property, it may not
+        # have updated.
+        author_names = [person.name for person in thesis.authors.all()]
+
+        assert name in author_names
+
+    def test_add_multiple_new_people(self):
+        name1 = 'Limor Fried'
+        name2 = 'Shirley Jackson'
+        thesis = Thesis.objects.first()
+
+        assert not Person.objects.filter(name=name1)
+        assert not Person.objects.filter(name=name2)
+
+        thesis.add_people([name1, name2])
+        author_names = [person.name for person in thesis.authors.all()]
+
+        assert name1 in author_names
+        assert name2 in author_names
+
+    def test_add_single_known_person(self):
+        name = 'Irene Pepperberg'
+        person = Person.objects.get_or_create(name=name)
+        thesis = Thesis.objects.first()
+
+        assert person not in thesis.authors
+
+        thesis.add_people([name])
+        author_names = [person.name for person in thesis.authors.all()]
+
+        assert name in author_names
+
+    def test_add_multiple_known_people(self):
+        name1 = 'Oliver R. Smoot'
+        name2 = 'Tom Magliozzi'
+        name3 = 'Ray Magliozzi'
+        thesis = Thesis.objects.first()
+
+        oliver, _ = Person.objects.get_or_create(name=name1)
+        tom, _ = Person.objects.get_or_create(name=name2)
+        ray, _ = Person.objects.get_or_create(name=name3)
+
+        assert oliver not in thesis.authors
+        assert tom not in thesis.authors
+        assert ray not in thesis.authors
+
+        thesis.add_people([name1, name2, name3])
+
+        assert oliver in thesis.authors.all()
+        assert tom in thesis.authors.all()
+        assert ray in thesis.authors.all()
+
+    def test_add_an_advisor(self):
+        name = 'Ellen Spertus'
+        thesis = Thesis.objects.first()
+
+        assert not Person.objects.filter(name=name)
+
+        thesis.add_people([name], author=False)
+        advisor_names = [person.name for person in thesis.advisors.all()]
+
+        assert name in advisor_names
+
+    def test_add_department(self):
+        dept = 'Department of Baconology'
+        thesis = Thesis.objects.first()
+        thesis.add_departments([dept])
+
+        dept_names = [dept.name for dept in thesis.department.all()]
+        assert dept in dept_names
+
+    def test_add_departments(self):
+        dept1 = 'Department of Transwarp Technologies'
+        dept2 = 'Department of Marshmallow Engineering'
+        thesis = Thesis.objects.first()
+        thesis.add_departments([dept1, dept2])
+
+        dept_names = [dept.name for dept in thesis.department.all()]
+        assert dept1 in dept_names
+        assert dept2 in dept_names
+
+    def test_add_known_department(self):
+        name = 'Course Eleventy-One'
+        dept, _ = Department.objects.get_or_create(name=name)
+        thesis = Thesis.objects.first()
+
+        assert dept not in thesis.department.all()
+
+        thesis.add_departments([name])
+
+        assert dept in thesis.department.all()

hamlet/theses/tests.py → hamlet/theses/tests/test_views.py

@@ -3,9 +3,9 @@
 from django.core.urlresolvers import reverse
 from django.test import Client, RequestFactory, TestCase, override_settings
 
-from .forms import AuthorAutocompleteForm, TitleAutocompleteForm
-from .models import Thesis, Person, Contribution
-from . import views
+from ..forms import AuthorAutocompleteForm, TitleAutocompleteForm
+from ..models import Thesis, Person, Contribution
+from .. import views
 
 
 # See http://tech.novapost.fr/django-unit-test-your-views-en.html .
@@ -54,12 +54,10 @@ def test_suggestion_context(self):
         assert 'suggestions' in response.context
 
         pks = [t.pk for t in response.context['suggestions']]
-        assert 43703 in pks
         assert 60330 in pks
 
         content = str(response.content)
 
-        assert 'Clock division as a power saving strategy in a system constrained by high transmission frequency and low data rate' in content  # noqa
         assert 'Architecture for ultra-low power multi-channel transmitters for Body Area Networks using RF resonators' in content  # noqa
 
     def test_get_correct_object(self):