logging.debug('Found 0 unique terms, skipping %s' % (
parse_data['meta']['name']
))
return
# Add the language and file data to the unique terms list
# We dont need to check for these terms being unique because this
# is the only way they can contained in this context
unique_terms.extend([
[parse_data['meta']['name'], context_id_map['file']],
[parse_data['meta']['language'], context_id_map['language']]
])
# Construct our document datastructure with its document tree
# and embed it properly!
tree_id = TreeModel.insert(tree)
doc = dict(
tree_id=tree_id,
signals=parse_data['tree'].signals,
meta=parse_data['meta'],
)
doc_id = DocumentModel.insert(doc)
# Record our TermID -> DocID mapping so we can atomically update
# the database later with these pointers in the final phase.
for term, context in unique_terms:
if term not in self.terms:
self.terms[term] = []
self.terms[term].append([doc_id, context])
def search (self, query, page=1, results_per_page=10):
"""
Returns a list of documents with their corresponding scores and
summaries that are specific to the givne query.
The data is returned in the following format:
(num_results, [(score, summary, document), ....,])
The input parameters should be as follows
query The query string
page The page number, must be a positive
non-zero integer
results_per_page The number of reuslts per page, non-zero
positive integer
"""
# Take query string and translate it into the format we require for
# fast processing which is [(context_id, term_id), ...]
tokens = query_translate(query)
# Retrieve the relevant documents for our given query tokens and
# their associated classifications.
start_time = time.time()
docs = self._get_documents(tokens)
logging.debug('%.2f seconds to retrieve and sort %d relevant docs' % (
time.time() - start_time,
len(docs)
))
# Remove duplicate documents from our results before we even bother
# scoring them or doing anytihng else.
docs = self._remove_duplicates(docs)
# Now that we have the matching documents we can now calculate the
# total score matching our query and then sort them form highest to
# lowest.
calculator = Calculator(tokens)
results = sorted([(calculator(doc), doc) for doc in docs],
key=itemgetter(0),
reverse=True
)
# Now segment our results into the section we will use and then
# retrieve their document trees for the summarization process. We
# only want to summarize the range of documents for the specific
# page we are rendering.
results = results[(page-1)*results_per_page:page*results_per_page]
trees = TreeModel.get([doc['tree_id'] for score, doc in results])
for score, doc in results:
doc['tree'] = trees[doc['tree_id']]
summarize = Summarizer(tokens)
start_time = time.time()
results = [
(score, summarize(document['tree']), document) \
for score, document in results
]
logging.debug('%.4f seconds to summarize and return results' % (
time.time() - start_time
))
return len(docs), results
