Spaces:
Sleeping
Sleeping
Update utils/mistral.py
Browse files- utils/mistral.py +28 -19
utils/mistral.py
CHANGED
@@ -370,6 +370,15 @@ def extract_link_details(text):
|
|
370 |
links_RE = [link for link in links_RE if not any(email in link for email in emails)]
|
371 |
|
372 |
return links_RE
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
373 |
|
374 |
def process_resume_data(file_path):
|
375 |
resume_text, hyperlinks = extract_text_based_on_format(file_path)
|
@@ -410,33 +419,33 @@ def process_resume_data(file_path):
|
|
410 |
# Combine both personal and professional details into a structured output
|
411 |
result = {
|
412 |
"personal": {
|
413 |
-
"name": per_data.get('personal', {}).get('name', None),
|
414 |
-
"contact": per_data.get('personal', {}).get('contact_number', None),
|
415 |
-
"email": per_data.get('personal', {}).get('email', None),
|
416 |
-
"location": per_data.get('personal', {}).get('address', None),
|
417 |
-
"linkedin": linkedin_links,
|
418 |
-
"github": github_links,
|
419 |
-
"other_links": hyperlinks
|
420 |
},
|
421 |
"professional": {
|
422 |
-
"technical_skills": pro_data.get('professional', {}).get('technical_skills', None),
|
423 |
-
"non_technical_skills": pro_data.get('professional', {}).get('non_technical_skills', None),
|
424 |
-
"tools": pro_data.get('professional', {}).get('tools', None),
|
425 |
"experience": [
|
426 |
{
|
427 |
-
"company": pro_data.get('professional', {}).get('companies_worked_at', None),
|
428 |
-
"projects": pro_data.get('professional', {}).get('projects', None),
|
429 |
-
"role": pro_data.get('professional', {}).get('roles', None),
|
430 |
-
"years": pro_data.get('professional', {}).get('experience', None),
|
431 |
-
"project_experience": pro_data.get('professional', {}).get('projects_experience', None)
|
432 |
}
|
433 |
],
|
434 |
"education": [
|
435 |
{
|
436 |
-
"qualification": Edu_data.get('educational', {}).get('qualifications', None),
|
437 |
-
"university": Edu_data.get('educational', {}).get('university', None),
|
438 |
-
"course": Edu_data.get('educational', {}).get('courses', None),
|
439 |
-
"certificate": Edu_data.get('educational', {}).get('certifications', None)
|
440 |
}
|
441 |
]
|
442 |
}
|
|
|
370 |
links_RE = [link for link in links_RE if not any(email in link for email in emails)]
|
371 |
|
372 |
return links_RE
|
373 |
+
|
374 |
+
# For handling multiple data
|
375 |
+
def normalize_data(value):
|
376 |
+
"""Replace empty lists with None and convert strings to lists."""
|
377 |
+
if value == []:
|
378 |
+
return None
|
379 |
+
elif isinstance(value, str):
|
380 |
+
return [value]
|
381 |
+
return value
|
382 |
|
383 |
def process_resume_data(file_path):
|
384 |
resume_text, hyperlinks = extract_text_based_on_format(file_path)
|
|
|
419 |
# Combine both personal and professional details into a structured output
|
420 |
result = {
|
421 |
"personal": {
|
422 |
+
"name": normalize_data(per_data.get('personal', {}).get('name', None)),
|
423 |
+
"contact": normalize_data(per_data.get('personal', {}).get('contact_number', None)),
|
424 |
+
"email": normalize_data(per_data.get('personal', {}).get('email', None)),
|
425 |
+
"location": normalize_data(per_data.get('personal', {}).get('address', None)),
|
426 |
+
"linkedin": normalize_data(linkedin_links),
|
427 |
+
"github": normalize_data(github_links),
|
428 |
+
"other_links": normalize_data(hyperlinks)
|
429 |
},
|
430 |
"professional": {
|
431 |
+
"technical_skills": normalize_data(pro_data.get('professional', {}).get('technical_skills', None)),
|
432 |
+
"non_technical_skills": normalize_data(pro_data.get('professional', {}).get('non_technical_skills', None)),
|
433 |
+
"tools": normalize_data(pro_data.get('professional', {}).get('tools', None)),
|
434 |
"experience": [
|
435 |
{
|
436 |
+
"company": normalize_data(pro_data.get('professional', {}).get('companies_worked_at', None)),
|
437 |
+
"projects": normalize_data(pro_data.get('professional', {}).get('projects', None)),
|
438 |
+
"role": normalize_data(pro_data.get('professional', {}).get('roles', None)),
|
439 |
+
"years": normalize_data(pro_data.get('professional', {}).get('experience', None)),
|
440 |
+
"project_experience": normalize_data(pro_data.get('professional', {}).get('projects_experience', None))
|
441 |
}
|
442 |
],
|
443 |
"education": [
|
444 |
{
|
445 |
+
"qualification": normalize_data(Edu_data.get('educational', {}).get('qualifications', None)),
|
446 |
+
"university": normalize_data(Edu_data.get('educational', {}).get('university', None)),
|
447 |
+
"course": normalize_data(Edu_data.get('educational', {}).get('courses', None)),
|
448 |
+
"certificate": normalize_data(Edu_data.get('educational', {}).get('certifications', None))
|
449 |
}
|
450 |
]
|
451 |
}
|