def setCorpus(self, corpus_id):
r = self.session.get(lsrh.getCorpusURL(corpus_id))
lsrh.checkStatus(r, (200,), "Corpus id not found on remote server.")
response = r.json()
self.prompt_id = lsrh.getIdFromURL(response["prompt"])
self.corpus_id = corpus_id
print 'Set corpus to id {} for prompt {}'.format(self.corpus_id, self.prompt_id)
def newCorpus(self, prompt_id):
description = raw_input("Description of corpus? ")
data = {"prompt": lsrh.getPromptURL(prompt_id), "description": description}
r = self.session.post(lsrh.getCorpusURL(), data=json.dumps(data), headers=self.headers)
lsrh.checkStatus(r, (201,), "Prompt id not found on remote server, or description was illegal.")
self.prompt_id = prompt_id
self.corpus_id = lsrh.getIdFromResponse(r)
print 'New corpus with id {} for prompt {}'.format(self.corpus_id, self.prompt_id)
def trainModel(self):
''' Train model based on the training answers for the current corpus.
Returns url of trained_model. '''
# create and start training task
data = {"corpus": lsrh.getCorpusURL(self.corpus_id)}
r = self.session.post(lsrh.getTrainingTaskURL(), data=json.dumps(data))
lsrh.checkStatus(r, (201,), "Failed to create training task")
trainingTaskData = r.json()
r2 = self.session.post(trainingTaskData["process"])
r3 = self.waitForTask(trainingTaskData["url"], "training task")
return r3.json()["trained_model"]
def uploadCorpus(self, prompt_id, fname):
''' Create a new corpus of training answers, given the prompt_id and a file containing
the corpus. '''
self.newCorpus(prompt_id)
r = self.uploadToS3('corpus', fname)
soup = bsoup(r.content)
key = soup.find('key').get_text()
data2 = {'corpus': lsrh.getCorpusURL(self.corpus_id), 's3_key': key, 'content_type': 'text/csv'}
# create and start upload task
r2 = self.session.post(lsrh.getCorpusUploadTaskURL(), data=json.dumps(data2))
corpusUploadTaskData = r2.json()
r3 = self.session.post(corpusUploadTaskData["process"])
lsrh.checkStatus(r3, (200, 202), "Queueing of corpus upload task failed.")
r4 = self.waitForTask(corpusUploadTaskData["url"], "corpus upload task")
return r3.json()["corpus"]
def addTrainingAnswer(self, essay):
data = {"corpus": lsrh.getCorpusURL(self.corpus_id), "text": essay.getText()}
r = self.session.post(lsrh.getTrainingAnswerURL(), data=json.dumps(data))
self.trainingAnswerMap[essay.essay_id] = lsrh.getIdFromResponse(r)
