utkarsh134 commited on
Commit
bfe3429
·
verified ·
1 Parent(s): 30fec28

Update services/resume_service.py

Browse files
Files changed (1) hide show
  1. services/resume_service.py +80 -80
services/resume_service.py CHANGED
@@ -1,81 +1,81 @@
1
- import json
2
- from typing import Dict
3
-
4
- from core.doc_processor import extract_text_from_bytes, extract_text_from_docx
5
- from core.text_processor import chunk_text
6
- from core.vector_store import create_or_load_vectorstore, similarity_search
7
- from services.llm_service import get_llm, get_experience_prompt
8
-
9
-
10
- def analyze_resume_document(
11
- file_bytes: bytes, job_role: str, file_extension: str
12
- ) -> Dict:
13
- if file_extension.lower() == ".pdf":
14
- resume_text = extract_text_from_bytes(file_bytes)
15
- elif file_extension.lower() == ".docx":
16
- resume_text = extract_text_from_docx(file_bytes)
17
- else:
18
- raise ValueError(f"Unsupported file format: {file_extension}")
19
-
20
- if resume_text.startswith("Error"):
21
- raise ValueError(resume_text)
22
-
23
- chunks = chunk_text(resume_text)
24
-
25
- vector_store = create_or_load_vectorstore(chunks)
26
-
27
- return analyze_experience_with_llm(vector_store, resume_text, job_role)
28
-
29
-
30
- def analyze_experience_with_llm(vectorstore, resume_text, job_role):
31
- try:
32
- retrieved_docs = similarity_search(vectorstore, job_role)
33
- relevant_content = "\n".join([doc.page_content for doc in retrieved_docs])
34
-
35
- experience_prompt = get_experience_prompt()
36
- llm = get_llm()
37
- chain = experience_prompt | llm
38
-
39
- response = chain.invoke(
40
- {
41
- "resume_text": resume_text,
42
- "relevant_content": relevant_content,
43
- "job_role": job_role,
44
- }
45
- )
46
-
47
- response = response.content.strip("`")
48
- if response.startswith("json"):
49
- response = response[4:]
50
-
51
- parsed_response = json.loads(response.strip())
52
- cleaned_response = {
53
- "relevant_experiences": [],
54
- "experience_summary": parsed_response.get(
55
- "experience_summary", "No relevant experience found"
56
- ),
57
- }
58
-
59
- for exp in parsed_response.get("relevant_experiences", []):
60
- cleaned_exp = {
61
- "position": exp.get("position", ""),
62
- "company": exp.get("company", ""),
63
- "duration": exp.get("duration", ""),
64
- "relevant_experience": exp.get("relevant_experience", ""),
65
- }
66
- cleaned_response["relevant_experiences"].append(cleaned_exp)
67
-
68
- return cleaned_response
69
-
70
- except json.JSONDecodeError as e:
71
- print(f"JSON Parse Error: {str(e)}")
72
- return {
73
- "relevant_experiences": [],
74
- "experience_summary": "Could not parse experience details from response",
75
- }
76
- except Exception as e:
77
- print(f"Unexpected Error: {str(e)}")
78
- return {
79
- "relevant_experiences": [],
80
- "experience_summary": f"Error analyzing experience: {str(e)}",
81
  }
 
1
+ import json
2
+ from typing import Dict
3
+
4
+ from core.doc_processor import extract_text_from_bytes, extract_text_from_docx
5
+ from core.text_processor import chunk_text
6
+ from core.vector_store import create_or_load_vectorstore, similarity_search
7
+ from services.llm_service import get_llm, get_experience_prompt
8
+
9
+
10
+ def analyze_resume_document(
11
+ file_bytes: bytes, job_role: str, file_extension: str
12
+ ) -> Dict:
13
+ if file_extension.lower() == ".pdf":
14
+ resume_text = extract_text_from_bytes(file_bytes)
15
+ elif file_extension.lower() == ".docx":
16
+ resume_text = extract_text_from_docx(file_bytes)
17
+ else:
18
+ raise ValueError(f"Unsupported file format: {file_extension}")
19
+
20
+ if resume_text.startswith("Error"):
21
+ raise ValueError(resume_text)
22
+
23
+ chunks = chunk_text(resume_text)
24
+
25
+ vector_store = create_or_load_vectorstore(chunks)
26
+
27
+ return analyze_experience_with_llm(vector_store, resume_text, job_role)
28
+
29
+
30
+ def analyze_experience_with_llm(vectorstore, resume_text, job_role):
31
+ try:
32
+ retrieved_docs = similarity_search(vectorstore, job_role)
33
+ relevant_content = "\n".join([doc.page_content for doc in retrieved_docs])
34
+
35
+ experience_prompt = get_experience_prompt()
36
+ llm = get_llm()
37
+ chain = experience_prompt | llm
38
+
39
+ response = chain.invoke(
40
+ {
41
+ "resume_text": resume_text,
42
+ "relevant_content": relevant_content,
43
+ "job_role": job_role,
44
+ }
45
+ )
46
+
47
+ response = response.content.strip("`")
48
+ if response.startswith("json"):
49
+ response = response[4:]
50
+
51
+ parsed_response = json.loads(response.strip())
52
+ cleaned_response = {
53
+ "relevant_experiences": [],
54
+ "experience_summary": parsed_response.get(
55
+ "experience_summary", "No relevant experience found"
56
+ ),
57
+ }
58
+
59
+ for exp in parsed_response.get("relevant_experiences", []):
60
+ cleaned_exp = {
61
+ "position": exp.get("position", ""),
62
+ "company": exp.get("company", ""),
63
+ "duration": exp.get("duration", ""),
64
+ "relevant_experience": exp.get("relevant_experience", ""),
65
+ }
66
+ cleaned_response["relevant_experiences"].append(cleaned_exp)
67
+
68
+ return cleaned_response
69
+
70
+ except json.JSONDecodeError as e:
71
+ print(f"JSON Parse Error: {str(e)}")
72
+ return {
73
+ "relevant_experiences": [],
74
+ "experience_summary": "Could not parse experience details from response",
75
+ }
76
+ except Exception as e:
77
+ print(f"Unexpected Error: {str(e)}")
78
+ return {
79
+ "relevant_experiences": [],
80
+ "experience_summary": f"Error analyzing experience: {str(e)}",
81
  }