def test_resume_basic_user_info_extraction(self): first_name, last_name, phone_number, email = get_basics(settings.TESTDATA_DIRS + 'resume_extraction/ideal_resume_data.txt') first_name_2, last_name_2, phone_number_2, email_2 = get_basics(settings.TESTDATA_DIRS + 'resume_extraction/resume_without_name.txt') first_name_3, last_name_3, phone_number_3, email_3 = get_basics(settings.TESTDATA_DIRS + 'resume_extraction/resume_without_email.txt') first_name_4, last_name_4, phone_number_4, email_4 = get_basics(settings.TESTDATA_DIRS + 'resume_extraction/resume_with_11_digit_phone_number.txt') first_name_5, last_name_5, phone_number_5, email_5 = get_basics(settings.TESTDATA_DIRS + 'resume_extraction/resume_with_wrong_email.txt') self.assertNotEqual(email_5, '*****@*****.**') self.assertNotEqual(email_5, '*****@*****.**') self.assertEqual(email_5, '') self.assertIn(first_name_4, 'anubhav') self.assertIn(last_name_4, 'singh') self.assertNotEqual(phone_number_4, '+916976318701') self.assertIn(email_4, '*****@*****.**') self.assertEqual(first_name_3, 'anubhav') self.assertEqual(last_name_3, 'singh') self.assertEqual(phone_number_3, '+917697631870') self.assertEqual(email_3, '') self.assertEqual(first_name, 'anubhav') self.assertEqual(last_name, 'singh') self.assertEqual(phone_number, '+917697631870') self.assertEqual(email, '*****@*****.**') self.assertEqual(first_name_2, '') self.assertEqual(last_name_2, '') self.assertEqual(phone_number_2, '+917697631870') self.assertEqual(email_2, '*****@*****.**')
def post(self, request, *args, **kwargs):
form = UploadResumeForm(request.POST, request.FILES)
if not form.is_valid():
return Response(form.errors)
f = request.FILES.get('file', None)
if not f:
return Response(
{'message': 'Cannot proceed further. File is missing.'},
status=status.HTTP_400_BAD_REQUEST)
# Basic file storage on a local filesystem.
fs = FileSystemStorage()
# Save file.
filename = fs.save("temp+" + f.name.lower(), f)
# Absolute path to the directory that will hold the files.
path = fs.location + '/' + filename
# Get basic resume details.
first_name, last_name, phone_number, email = get_basics(path)
# Delete file as it was temporary.
fs.delete(filename)
return Response(
{
'email': email,
'last_name': last_name,
'first_name': first_name,
'phone_number': phone_number,
'status': 'Successfully parsed.'
},
status=status.HTTP_200_OK)
def test_get_basics(self): first_name, last_name, phone_number, email = get_basics(settings.TESTDATA_DIRS + 'view_tests/nda.pdf') self.assertEqual(first_name, 'anubhav') self.assertEqual(phone_number, '+917697631870') self.assertEqual(email, '*****@*****.**') self.assertEqual(last_name, 'singh')
def test_website_score(self):
urls = get_urls(self.test_website_score)
category_urls = url_categories(urls)
# self.assertEqual('http://imgur.com/', category_urls['others'][0])
url_response = get_url_response(category_urls)
resume = Resume.objects.create(parse_status=0)
for item in url_response:
resume_urls = Url.objects.filter(url=item['name'])
if resume_urls.exists():
resume_url = resume_urls[0]
resume_url.category = item['type']
resume_url.save()
else:
resume_url = Url.objects.create(url=item['name'],
category=item['type'])
resume.urls.add(resume_url)
first_name, last_name, phone_number, email = get_basics(
settings.TESTDATA_DIRS +
'score_calculation/test_website_score.txt')
self.assertEqual(email, '*****@*****.**')
apply_website_score = mock.Mock(return_value=(None, None, None))
(website_activity_score, website_reputation_score,
website_contribution_score) = apply_website_score(
category_urls, resume, email)
self.assertEqual(website_activity_score, None)
self.assertEqual(website_contribution_score, None)
self.assertEqual(website_reputation_score, None)
def post(self, request, *args, **kwargs):
form = UploadResumeForm(request.POST, request.FILES)
if form.is_valid():
get_file = request.FILES['file']
file_name = get_file.name.lower()
name, ext = os.path.splitext(file_name)
uploaded_file_name = default_storage.save(
"%s" % uuid4() + ext, ContentFile(get_file.read()))
path = default_storage.open(uploaded_file_name).name
first_name, last_name, phone_number, email = get_basics(path)
return Response(
{
'status': 'Successfully Parsed',
'first_name': first_name,
'last_name': last_name,
'phone_number': phone_number,
'email': email
},
status=status.HTTP_200_OK)
return JsonResponse(form.errors)
def test_itunes_store_score(self):
urls = get_urls(self.test_itunes_score)
category_urls = url_categories(urls)
url_response = get_url_response(category_urls)
self.assertIn(
'https://itunes.apple.com/in/app/whatsapp-messenger/id310633997?mt=8',
category_urls['apps'])
first_name, last_name, email, phone_number = get_basics(
settings.TESTDATA_DIRS + 'score_calculation/test_itunes_score.txt')
self.assertEqual(first_name, 'anubhav')
resume = Resume.objects.create(parse_status=0)
for item in url_response:
resume_urls = Url.objects.filter(url=item['name'])
if resume_urls.exists():
resume_url = resume_urls[0]
resume_url.category = item['type']
resume_url.save()
else:
resume_url = Url.objects.create(url=item['name'],
category=item['type'])
resume.urls.add(resume_url)
calculate_average_mobile_contrib_score = mock.Mock(return_value=(.3,
.7,
.05))
apply_itunes_score(category_urls, first_name, resume)
avg_mobile_apps_activity_score, avg_mobile_apps_reputation_score, avg_mobile_apps_contribution_score = calculate_average_mobile_contrib_score(
resume)
self.assertEqual(avg_mobile_apps_contribution_score, .05)
self.assertEqual(avg_mobile_apps_reputation_score, .7)
self.assertEqual(avg_mobile_apps_activity_score, .3)
def parse_resume(path,
text,
resume_id,
skills,
file_name,
hash_value,
callback_client=False):
content_list = text.lower().split()
# Get resume instance.
resume_details = Resume.objects.get(id=resume_id)
# Categorising urls
categories_url = url_categories(get_urls(text))
# Getting custom url response using get_url_response
url_response = get_url_response(categories_url)
# Get basic details.
first_name, last_name, phone_number, email = get_basics(path)
# Consider basic details only if resume_details for the same is non-nil.
if resume_details.email:
email = resume_details.email
if resume_details.first_name:
first_name = resume_details.first_name
if resume_details.last_name:
last_name = resume_details.last_name
if resume_details.phone_number:
phone_number = resume_details.phone_number
# Url Instance
for item in url_response:
resume_urls = Url.objects.filter(url=item['name'])
if resume_urls.exists():
resume_url = resume_urls[0]
resume_url.category = item['type']
resume_url.save()
else:
resume_url = Url.objects.create(url=item['name'],
category=item['type'])
resume_details.urls.add(resume_url)
# Skills Matching Score
(skill_match_score, skills_matched,
skills_not_matched) = get_skill_matching_score(skills, text)
# Definition of scores
github_contribution_score = 0
github_activity_score = 0
github_reputation_score = 0
bit_bucket_contribution_score = 0
bit_bucket_activity_score = 0
bit_bucket_reputation_score = 0
stackoverflow_contribution_score = 0
stackoverflow_reputation_score = 0
stackoverflow_activity_score = 0
# StackOverflow Score
stackoverflow_user_id = get_stackoverflow_userid(text)
if stackoverflow_user_id:
(stackoverflow_activity_score, stackoverflow_reputation_score,
stackoverflow_contribution_score) = apply_stackoverflow_score(
stackoverflow_user_id, resume_details)
# GitHub Score
github_username = get_github_username(text)
if github_username:
(github_activity_score, github_reputation_score,
github_contribution_score) = apply_github_score(
github_username, resume_details)
# Blog score
apply_blog_score(categories_url, resume_details)
# BitBucket Score
contribution_urls = categories_url['contributions']
bit_bucket_url = get_bit_bucket_url(contribution_urls)
if bit_bucket_url == 'No Url Found' or bit_bucket_url is None:
pass
else:
bit_bucket_user_name = get_bitbucket_username(bit_bucket_url)
(bit_bucket_activity_score, bit_bucket_reputation_score,
bit_bucket_contribution_score) = apply_bitbucket_score(
bit_bucket_user_name, resume_details)
# MobileApp Database saving and score calculations
# Play Store - Total Score
apply_play_store_app_score(categories_url, resume_details)
# ITunes - Total Score
apply_itunes_score(categories_url, first_name, resume_details)
# Website Score
(website_activity_score, website_reputation_score,
website_contribution_score) = apply_website_score(categories_url,
resume_details, email)
save_resume_skills(resume_details, skills_matched, skills_not_matched)
# Work Experience
features = FeatureExtraction()
work_experience = features.get_work_experience(text)
# Blog
(average_blog_activity_score, average_blog_reputation_score,
average_blog_contribution_score) = calculate_blog_scores(resume_details)
# Website
(average_website_activity_score, average_website_reputation_score,
average_website_contribution_score
) = calculate_website_scores(resume_details)
# average mobile contribution score
(average_mobile_app_activity_score, average_mobile_app_reputation_score,
average_mobile_app_contribution_score
) = calculate_average_mobile_contrib_score(resume_details)
# Activity Scores
blog_activity_score = average_blog_activity_score
website_activity_score = average_website_activity_score
mobile_app_activity_score = average_mobile_app_activity_score
# Contributions Scores
blog_contribution_score = average_blog_contribution_score
website_contribution_score = average_website_contribution_score
mobile_app_contribution_score = average_mobile_app_contribution_score
# Reputation Scores
blog_reputation_score = average_blog_reputation_score
website_reputation_score = average_website_reputation_score
mobile_app_reputation_score = average_mobile_app_reputation_score
# Total Contribution Score
coding_total_contribution_score = (github_contribution_score +
bit_bucket_contribution_score +
stackoverflow_contribution_score +
mobile_app_contribution_score)
social_total_contribution_score = (blog_contribution_score +
website_contribution_score)
# Total Activity score
coding_total_activity_score = (github_activity_score +
stackoverflow_activity_score +
bit_bucket_activity_score +
mobile_app_activity_score)
social_total_activity_score = blog_activity_score + website_activity_score
# Total Reputation Score
coding_total_reputation_score = (github_reputation_score +
stackoverflow_reputation_score +
bit_bucket_reputation_score +
mobile_app_reputation_score)
social_total_reputation_score = (blog_reputation_score +
website_reputation_score)
# Total Coding score 2.5 out of 5
total_coding_score = (coding_total_contribution_score +
coding_total_reputation_score +
coding_total_activity_score)
# Total Social Score 1.5 out 5
total_social_score = (social_total_contribution_score +
social_total_activity_score +
social_total_reputation_score)
# Total Skill Matching Score 1 out 5
total_skill_score = skill_match_score
# Saving to Scores Model
# saving total contribution score
coding_score_instance = Score.objects.create(type=Score.TYPES.coding,
score=total_coding_score)
resume_details.scores.add(coding_score_instance)
social_score_instance = Score.objects.create(type=Score.TYPES.social,
score=total_social_score)
resume_details.scores.add(social_score_instance)
skill_score_instance = Score.objects.create(
type=Score.TYPES.skill_matching, score=total_skill_score)
resume_details.scores.add(skill_score_instance)
total_ranking = total_coding_score + total_social_score + total_skill_score
# Extracting Location, Company and Institution Names
extract_features = ExtractFeatures()
locations = extract_features.get_location(text)
companies = extract_features.get_company_names(text)
institutions = extract_features.get_institution_names(text)
for location in locations:
location_instance, created = Location.objects.get_or_create(
name=location)
resume_details.locations.add(location_instance)
for company in companies:
company_instance, created = Company.objects.get_or_create(name=company)
resume_details.companies.add(company_instance)
for institution in institutions:
institution_instance, created = Institution.objects.get_or_create(
name=institution)
resume_details.institutions.add(institution_instance)
# Extract skills from provided text.
# NOTE: As per now we're getting top 1000 tags from SO to extract skills.
# With those skills we're getting intersection with list of content text.
# Need to find better solution to do so [Bloom filter, et cetera]
import re
content_list = map(lambda x: re.sub('[^0-9a-zA-Z\.]+', '', x),
content_list)
# print(content_list)
# content_list = map(lambda x: x.replace(',', ''), content_list)
matched_skills = list(skillset.intersection(content_list))
for skill in matched_skills:
skill_instance, created = Skill.objects.get_or_create(name=skill)
rskills = ResumeSkills(resume=resume_details, skill=skill_instance)
rskills.save()
# Resume
file_name = check_file_name_length(file_name)
resume_details.first_name = first_name
resume_details.last_name = last_name
resume_details.phone_number = phone_number
resume_details.parse_status = Resume.STATUS.processed
resume_details.file_name = file_name
resume_details.content_hash = hash_value
resume_details.content = text
resume_details.email = email
resume_details.resume_location = path
resume_details.experience = work_experience
resume_details.total_score = total_ranking
resume_details.save()
if callback_client:
resp = callback_internal_client(resume_details)
if resp.status_code != requests.codes.ok:
print(
"ERROR: Unable to callback to internal client for resume: %s".
format(str(resume_id)))
return "Resume Processed"