def handle(self, *args, **options):
std_arrays = defaultdict(list)
doclist = []
canned_loader = CannedLoader()
for doc in canned_loader.iterate():
# Process this document
tm = TextManager(doc.text, doc.year)
# Trace progress message
self.stdout.write(doc.title + " " + doc.author)
for key, value in sorted(tm.profile_stats().items()):
std_arrays[key].append(value)
doclist.append((doc.author + ", " + doc.title, tm.profile_stats()))
mean_averages = {key: numpy.mean(values) for key, values in std_arrays.items()}
standard_deviations = {key: numpy.std(values) for key, values in std_arrays.items()}
doclist = [(title, _convert_to_zscores(stats, mean_averages, standard_deviations)) for title, stats in doclist]
for i, doc in enumerate(doclist):
title, stats = doc
rankings = []
for j, doc2 in enumerate(doclist):
if j == i:
continue
distance = _measure_distance(doc, doc2)
rankings.append((doc2[0], doc2[1], distance))
rankings.sort(key=lambda d: d[2])
self.stdout.write("\n\n" + title)
self.stdout.write(repr(stats))
for d in rankings[0:10]:
self.stdout.write("\t%s\t%f" % (d[0], d[2]))
self.stdout.write("\t\t" + repr(d[1]))
def prepare_canned_texts():
"""
Load the list of documents from the text file, process them,
and save the results as canned JSON objects in the Document database.
"""
# Truncate the existing Document database table
Document.objects.all().delete()
canned_loader = CannedLoader()
for doc in canned_loader.iterate():
# Process this document
tm = TextManager(doc.text, doc.year)
# Trace progress message
stdout.write('%s, %s\n' % (doc.author, doc.title))
# Trim fields where necessary
mlength = Document._meta.get_field('author').max_length
author = doc.author[:mlength]
mlength = Document._meta.get_field('authorsort').max_length
author_sort = doc.author_sort[:mlength]
mlength = Document._meta.get_field('title').max_length
title = doc.title[:mlength]
mlength = Document._meta.get_field('titlesort').max_length
title_sort = doc.title_sort[:mlength]
# Prepare database record for this document
dbdoc = Document(author=author,
authorsort=author_sort,
title=title,
titlesort=title_sort,
year=doc.year,
teaser=tm.teaser(),
text=tm.text,
lemmas=tm.lemmas_datastruct(formalism='json'),
tokens=tm.tokens_datastruct(formalism='json'),
randomsort=random.randint(0, 1000),)
# Save the database record
dbdoc.save()
