def save_tfidf_results(self,job_id):
db_collection = _get_db_collection()
job_info = db_collection.find_one({'_id': ObjectId(job_id)})
logger.info("Job %s: TF-IDF" % (job_id))
# tfidf them
filepaths = job_info['filepaths']
tfidf = samediff.analysis.tf_idf(filepaths)
cosine_similarity = samediff.analysis.cosine_similarity(filepaths)
# save the results back to the db (based on the job_number)
job_info['tfidf'] = tfidf
job_info['cosineSimilarity'] = cosine_similarity
# delete the raw input files
for path in job_info['filepaths']:
os.remove(path)
del job_info['filepaths']
job_info['status'] = 'complete'
db_collection.save(job_info)
# TODO: catch any exceptions and queue them up for retry attempts
# notify them with email
# TODO: Internationalize and put the text stuff into some kind of templating structure
name = job_info['email'].split('@')[0]
body = u'Dear %s, \n\nYour SameDiff job is ready at this URL: %s! \n\nSincerely, \n %s ' % ( name , job_info["results_url"], settings.get('mail','from_name'))
envelope = Envelope(
from_addr=(settings.get('mail','from_email'), settings.get('mail','from_name')),
to_addr=(job_info['email'], name),
subject=u'Your SameDiff job is ready!',
text_body=body)
envelope.send('mail.gandi.net', login=settings.get('mail','login'),
password=settings.get('mail','password'), tls=True)
app.config["UPLOADED_DOCS_DEST"] = TEMP_DIR
docs = UploadSet(name="docs", extensions=("txt"))
configure_uploads(app, (docs,))
patch_request_class(app, 4 * 1024 * 1024) # 4MB
# setup logging
logging.basicConfig(filename=os.path.join(base_dir, "../", "samediff.log"), level=logging.DEBUG)
logger = logging.getLogger(__name__)
logger.info("Temp Dir is %s" % TEMP_DIR)
db_client = MongoClient(get_mongo_uri())
app.db = db_client[settings.get("db", "name")]
app.db_collection = app.db[settings.get("db", "collection")]
@app.before_request
def before():
if request.view_args and "lang_code" in request.view_args:
if request.view_args["lang_code"] not in ("es", "en"):
return abort(404) # bail on invalid language
g.current_lang = request.view_args["lang_code"]
request.view_args.pop("lang_code")
@babel.localeselector
def get_locale():
return g.get("current_lang", "en")
from __future__ import absolute_import
import os
from celery import Celery
from samediff import settings
celery_app = Celery('samediff',
broker=settings.get('queue','broker_url'),
backend=settings.get('queue','backend_url'),
include=['samediff.tasks'])
# expire backend results in one hour
celery_app.conf.update(
CELERY_TASK_RESULT_EXPIRES=3600,
)
if __name__ == '__main__':
celery_app.start()
def _get_db_collection():
db_client = MongoClient(samediff.get_mongo_uri())
db = db_client[settings.get('db','name')]
db_collection = db[settings.get('db','collection')]
return db_collection
