Update app.py

app.py

@@ -59,57 +59,50 @@ def extract_text(uploaded_file):
     if not uploaded_file:
         return ""
     return extract_text_from_pdf(uploaded_file) if uploaded_file.type == "application/pdf" else extract_text_from_image(uploaded_file)
+
 def parse_resume(resume_text):
-    """Extract key information from resume text"""
-    parsed_info = {
-        'education': [],
-        'skills': [],
-        'experience': [],
-        'projects': [],
-        'publications': []
+    """Extract key information from resume text using improved parsing"""
+    sections = {
+        'education': ['Education:', 'EDUCATION', 'Academic Background'],
+        'experience': ['Experience:', 'EXPERIENCE', 'Work History', 'Employment'],
+        'skills': ['Skills:', 'SKILLS', 'Technical Skills', 'Technologies'],
+        'projects': ['Projects:', 'PROJECTS', 'Key Projects'],
+        'publications': ['Publications:', 'PUBLICATIONS', 'Research Papers']
     }
     
-    # Find education details
-    edu_markers = ['Education:', 'EDUCATION', 'Academic Background']
-    exp_markers = ['Experience:', 'EXPERIENCE', 'Work History', 'Employment']
-    skill_markers = ['Skills:', 'SKILLS', 'Technical Skills', 'Technologies']
-    proj_markers = ['Projects:', 'PROJECTS', 'Key Projects']
-    pub_markers = ['Publications:', 'PUBLICATIONS', 'Research Papers']
+    parsed_info = {key: '' for key in sections}
     
-    # Helper function to extract section content
-    def extract_section(text, start_markers, end_markers):
-        content = []
-        for start in start_markers:
-            start_idx = text.find(start)
-            if start_idx != -1:
-                section_start = start_idx + len(start)
-                section_end = len(text)
-                
-                # Find the next section marker
-                for end in end_markers:
-                    next_section = text.find(end, section_start)
-                    if next_section != -1:
-                        section_end = min(section_end, next_section)
-                
-                section_content = text[section_start:section_end].strip()
-                content.append(section_content)
-        
-        return '\n'.join(content)
+    # Convert text to lines for better parsing
+    lines = resume_text.split('\n')
+    current_section = None
+    section_content = []
     
-    # Extract sections
-    all_markers = edu_markers + exp_markers + skill_markers + proj_markers + pub_markers
+    for line in lines:
+        line = line.strip()
+        if not line:
+            continue
+            
+        # Check if this line is a section header
+        for section, headers in sections.items():
+            if any(header.lower() in line.lower() for header in headers):
+                if current_section:
+                    parsed_info[current_section] = '\n'.join(section_content)
+                current_section = section
+                section_content = []
+                break
+        else:
+            if current_section:
+                section_content.append(line)
     
-    parsed_info['education'] = extract_section(resume_text, edu_markers, all_markers)
-    parsed_info['experience'] = extract_section(resume_text, exp_markers, all_markers)
-    parsed_info['skills'] = extract_section(resume_text, skill_markers, all_markers)
-    parsed_info['projects'] = extract_section(resume_text, proj_markers, all_markers)
-    parsed_info['publications'] = extract_section(resume_text, pub_markers, all_markers)
+    # Add the last section
+    if current_section and section_content:
+        parsed_info[current_section] = '\n'.join(section_content)
     
     return parsed_info
 
 def extract_professor_details(text):
-    professor_pattern = r"(Dr\.|Professor|Prof\.?)\s+([A-Z][a-z]+\s[A-Z][a-z]+)"
-    university_pattern = r"(University|Institute|College|School of [A-Z][A-Za-z\s]+)"
+    professor_pattern = r"(Dr\.|Professor|Prof\.?)\s+([A-Z][a-z]+(?:\s[A-Z][a-z]+)*)"
+    university_pattern = r"(University|Institute|College|School) of [A-Z][A-Za-z\s]+"
     
     professor_match = re.search(professor_pattern, text)
     university_match = re.search(university_pattern, text)
@@ -117,85 +110,50 @@ def extract_professor_details(text):
     return (professor_match.group(0) if professor_match else "Not Found",
             university_match.group(0) if university_match else "Not Found")
 
-def clean_email_output(email_text):
-    """Clean and format email content"""
-    start_idx = email_text.find("Dear")
-    if start_idx == -1:
-        start_idx = 0
+def clean_output(text, type_="general"):
+    """Unified cleaning function for all document types"""
+    if not text:
+        return ""
         
-    end_markers = ["Best regards,", "Sincerely,", "Yours sincerely,", "Kind regards,"]
-    end_idx = len(email_text)
-    for marker in end_markers:
-        idx = email_text.find(marker)
-        if idx != -1:
-            end_idx = email_text.find("\n\n", idx) if email_text.find("\n\n", idx) != -1 else len(email_text)
-            break
-    
-    email_content = email_text[start_idx:end_idx].strip()
-    
-    if "Phone:" in email_text or "Email:" in email_text:
-        contact_info = "\n\n" + "\n".join([
-            line for line in email_text[end_idx:].split("\n")
-            if any(info in line for info in ["Phone:", "Email:"])
-        ]).strip()
-        email_content += contact_info
+    # Common start markers
+    start_markers = {
+        "email": ["Dear"],
+        "cover_letter": ["Dear", "To Whom", "Hiring"],
+        "research_statement": ["Research Statement", "Statement of Research"],
+        "sop": ["Statement of Purpose", "Personal Statement"]
+    }
     
-    return email_content
-
-def clean_cover_letter_output(letter_text):
-    """Clean and format cover letter content"""
-    start_markers = ["Dear", "To Whom", "Hiring"]
-    start_idx = len(letter_text)
-    for marker in start_markers:
-        idx = letter_text.find(marker)
-        if idx != -1:
-            start_idx = min(start_idx, idx)
+    # Common end markers
+    end_markers = ["Best regards,", "Sincerely,", "Yours sincerely,", "Kind regards,", "Thank you"]
     
-    end_markers = ["Sincerely,", "Best regards,", "Yours truly,", "Regards,"]
-    end_idx = len(letter_text)
-    for marker in end_markers:
-        idx = letter_text.find(marker)
+    # Find start of content
+    start_idx = 0
+    relevant_starts = start_markers.get(type_, start_markers["email"])
+    for marker in relevant_starts:
+        idx = text.find(marker)
         if idx != -1:
-            end_idx = letter_text.find("\n\n", idx) if letter_text.find("\n\n", idx) != -1 else len(letter_text)
+            start_idx = idx
             break
     
-    return letter_text[start_idx:end_idx].strip()
-
-def clean_research_statement_output(statement_text):
-    """Clean and format research statement content"""
-    # Remove common headers
-    headers = ["Research Statement", "Statement of Research", "Research Interests"]
-    cleaned_text = statement_text
-    for header in headers:
-        if cleaned_text.startswith(header):
-            cleaned_text = cleaned_text[len(header):].lstrip(":\n")
-    
-    # Remove any trailing references or bibliography sections
-    end_markers = ["References", "Bibliography", "Citations"]
+    # Find end of content
+    end_idx = len(text)
     for marker in end_markers:
-        idx = cleaned_text.find(marker)
+        idx = text.find(marker)
         if idx != -1:
-            cleaned_text = cleaned_text[:idx].strip()
+            end_idx = text.find("\n\n", idx) if text.find("\n\n", idx) != -1 else len(text)
+            break
     
-    return cleaned_text.strip()
-
-def clean_sop_output(sop_text):
-    """Clean and format Statement of Purpose content"""
-    # Remove common headers
-    headers = ["Statement of Purpose", "Personal Statement", "Academic Statement"]
-    cleaned_text = sop_text
-    for header in headers:
-        if cleaned_text.startswith(header):
-            cleaned_text = cleaned_text[len(header):].lstrip(":\n")
+    cleaned_text = text[start_idx:end_idx].strip()
     
-    # Remove any trailing sections
-    end_markers = ["Thank you", "References", "Additional Information"]
-    for marker in end_markers:
-        idx = cleaned_text.find(marker)
-        if idx != -1:
-            cleaned_text = cleaned_text[:idx].strip()
+    # Add contact information for emails
+    if type_ == "email" and ("Phone:" in text or "Email:" in text):
+        contact_info = "\n\n" + "\n".join([
+            line for line in text[end_idx:].split("\n")
+            if any(info in line for info in ["Phone:", "Email:"])
+        ]).strip()
+        cleaned_text += contact_info
     
-    return cleaned_text.strip()
+    return cleaned_text
 
 # Initialize session state
 if 'generated_content' not in st.session_state:
@@ -206,18 +164,18 @@ if 'generated_content' not in st.session_state:
         'sop': None
     }
 
-# Template Definitions
+# Template Definitions (simplified and standardized)
 templates = {
-    'email': PromptTemplate.from_template("""
+    'email': """
 Write ONLY a formal cold email for a research position.
 Start with 'Dear Professor' and end with a signature.
 
 Use these specific details from the CV:
-Education: {education}
-Relevant Experience: {experience}
-Key Skills: {skills}
-Notable Projects: {projects}
-Publications: {publications}
+{education}
+{experience}
+{skills}
+{projects}
+{publications}
 
 Additional Context:
 Professor: {professor_name}
@@ -227,83 +185,73 @@ Why This Lab: {reason}
 
 Guidelines:
 1. Keep the email concise (max 400 words)
-2. Focus on the most relevant experience and skills that match the lab's research
-3. Mention 1-2 specific projects or publications that align with the lab's work
-4. Include a clear statement of interest and why you're a good fit
+2. Focus on the most relevant experience and skills
+3. Mention 1-2 specific projects that align with the lab's work
+4. Include a clear statement of interest
 5. End with your contact information
-"""),
-    
-    'cover_letter': PromptTemplate.from_template("""
-Write ONLY a professional cover letter.
-Use these specific details from the CV:
-Education: {education}
-Relevant Experience: {experience}
-Technical Skills: {skills}
-Notable Projects: {projects}
-Publications: {publications}
-
-Position Details:
-Job Title: {job_title}
-Company: {company}
+""",
+    'cover_letter': """
+Write ONLY a professional cover letter for {job_title} at {company}.
+Use these specific details:
+{education}
+{experience}
+{skills}
+{projects}
+
 Required Skills: {key_skills}
 
 Guidelines:
 1. Start with a formal greeting
-2. Focus on experiences and skills that directly match the job requirements
-3. Provide specific examples from your projects and work history
-4. Demonstrate how your background makes you an ideal candidate
-5. End with a professional closing
-"""),
-    
-    'research_statement': PromptTemplate.from_template("""
-Write ONLY a research statement focused on your academic journey and research goals.
-Use these specific details from your background:
-Education: {education}
-Research Experience: {experience}
-Technical Skills: {skills}
-Research Projects: {projects}
-Publications: {publications}
-
-Additional Context:
-Research Background: {research_background}
-Key Projects: {key_projects}
+2. Focus on experiences matching job requirements
+3. Provide specific examples
+4. Show why you're an ideal candidate
+5. End professionally
+""",
+    'research_statement': """
+Write ONLY a research statement focused on your academic journey and future goals.
+Background:
+{education}
+{experience}
+{skills}
+{projects}
+{publications}
+
+Research Focus:
+{key_projects}
 Future Goals: {future_goals}
 
 Guidelines:
-1. Describe your research journey and motivation
-2. Highlight key research achievements and findings
-3. Connect past work to future research goals
-4. Demonstrate technical expertise and methodological knowledge
-5. End with your vision for future contributions to the field
-"""),
-    
-    'sop': PromptTemplate.from_template("""
-Write ONLY a Statement of Purpose (SOP).
-Use these specific details from your background:
-Education: {education}
-Research Experience: {experience}
-Technical Skills: {skills}
-Notable Projects: {projects}
-Publications: {publications}
-
-Additional Context:
+1. Describe your research journey
+2. Highlight key achievements
+3. Connect past work to future goals
+4. Show technical expertise
+5. Present your research vision
+""",
+    'sop': """
+Write ONLY a Statement of Purpose (SOP) for graduate studies.
+Background:
+{education}
+{experience}
+{skills}
+{projects}
+{publications}
+
+Context:
 Motivation: {motivation}
-Academic Goals: {academic_background}
-Research Interests: {research_experiences}
-Career Objectives: {career_goals}
+Career Goals: {career_goals}
 Program Interest: {why_this_program}
 
 Guidelines:
-1. Tell a coherent story about your academic journey
-2. Connect your background to your future goals
-3. Demonstrate why you're prepared for graduate study
-4. Show alignment between your interests and the program
-5. Make a compelling case for why you should be admitted
-""")
+1. Tell your academic journey
+2. Connect background to goals
+3. Show preparation for graduate study
+4. Demonstrate program alignment
+5. Make a compelling case
+"""
 }
 
-
-# Create LangChain instances
+# Convert templates to PromptTemplate objects
+templates = {k: PromptTemplate.from_template(v) for k, v in templates.items()}
 chains = {key: LLMChain(llm=llm, prompt=template) for key, template in templates.items()}
 
 # Sidebar for Input Collection
@@ -311,7 +259,12 @@ with st.sidebar:
     st.subheader("📝 Input Details")
     job_opening_text = st.text_area("Job/Research Opening Details", height=150)
     cv_resume_file = st.file_uploader("Upload CV/Resume", type=["pdf", "png", "jpg", "jpeg"])
-    cv_resume_text = extract_text(cv_resume_file)
+    cv_resume_text = extract_text(cv_resume_file) if cv_resume_file else ""
+
+# Parse resume once for all tabs
+resume_info = parse_resume(cv_resume_text) if cv_resume_text else {
+    'education': '', 'experience': '', 'skills': '', 'projects': '', 'publications': ''
+}
 
 # Tab Layout
 tab1, tab2, tab3, tab4 = st.tabs(["Cold Email", "Cover Letter", "Research Statement", "SOP"])
@@ -326,48 +279,43 @@ with tab1:
         if job_opening_text and cv_resume_text:
             with st.spinner("Generating..."):
                 try:
-                    # Parse resume information
-                    resume_info = parse_resume(cv_resume_text)
-                    
-                    # Generate email with parsed information
                     generated_email = chains['email'].run({
+                        **resume_info,
                         "professor_name": professor_name,
                         "university_name": university_name,
                         "research_interests": research_interests,
-                        "reason": reason,
-                        "education": resume_info['education'],
-                        "experience": resume_info['experience'],
-                        "skills": resume_info['skills'],
-                        "projects": resume_info['projects'],
-                        "publications": resume_info['publications']
+                        "reason": reason
                     })
-                    st.session_state.generated_content['email'] = clean_email_output(generated_email)
+                    st.session_state.generated_content['email'] = clean_output(generated_email, "email")
                 except Exception as e:
                     st.error(f"Generation error: {e}")
         else:
             st.error("Please provide all required inputs")
+    
+    if st.session_state.generated_content['email']:
+        st.markdown('<div class="output-container">', unsafe_allow_html=True)
+        st.markdown(st.session_state.generated_content['email'])
+        st.download_button("Download Email", st.session_state.generated_content['email'],
+                         file_name="email.txt", key="email_download")
+        st.markdown('</div>', unsafe_allow_html=True)
 
 # Cover Letter Tab
 with tab2:
     job_title = st.text_input("Job Title")
     company_name = university_name if university_name != "Not Found" else st.text_input("Company/University")
-    key_skills = st.text_input("Key Skills")
+    key_skills = st.text_input("Key Skills Required")
     
     if st.button("Generate Cover Letter", key="cover_letter_btn"):
         if job_opening_text and cv_resume_text:
             with st.spinner("Generating..."):
                 try:
-                    resume_info = parse_resume(cv_resume_text)
                     generated_letter = chains['cover_letter'].run({
+                        **resume_info,
                         "job_title": job_title,
                         "company": company_name,
-                        "key_skills": key_skills,
-                        "reason": reason,
-                        "skills": resume_info['skills'],
-                        "education": resume_info['education'],
-                        "experience": resume_info['experience']
+                        "key_skills": key_skills
                     })
-                    st.session_state.generated_content['cover_letter'] = clean_cover_letter_output(generated_letter)
+                    st.session_state.generated_content['cover_letter'] = clean_output(generated_letter, "cover_letter")
                 except Exception as e:
                     st.error(f"Generation error: {e}")
         else:
@@ -382,28 +330,23 @@ with tab2:
 
 # Research Statement Tab
 with tab3:
-    
     key_projects = st.text_input("Key Research Projects")
     future_goals = st.text_input("Future Research Goals")
     
     if st.button("Generate Research Statement", key="research_stmt_btn"):
-        with st.spinner("Generating..."):
-            try:
-                resume_info = parse_resume(cv_resume_text)
-                generated_statement = chains['research_statement'].run({
-                    "reason": reason,
-                    "education": resume_info['education'],
-                    "experience": resume_info['experience'],
-                    "skills": resume_info['skills'],
-                    "projects": resume_info['projects'],
-                    "publications": resume_info['publications'],
-                    "research_background": resume_info['publications'],
-                    "key_projects": key_projects,
-                    "future_goals": future_goals
-                })
-                st.session_state.generated_content['research_statement'] = clean_research_statement_output(generated_statement)
-            except Exception as e:
-                st.error(f"Generation error: {e}")
+        if cv_resume_text:
+            with st.spinner("Generating..."):
+                try:
+                    generated_statement = chains['research_statement'].run({
+                        **resume_info,
+                        "key_projects": key_projects,
+                        "future_goals": future_goals
+                    })
+                    st.session_state.generated_content['research_statement'] = clean_output(generated_statement, "research_statement")
+                except Exception as e:
+                    st.error(f"Generation error: {e}")
+        else:
+            st.error("Please upload your CV/Resume")
 
     if st.session_state.generated_content['research_statement']:
         st.markdown('<div class="output-container">', unsafe_allow_html=True)
@@ -415,29 +358,24 @@ with tab3:
 # SOP Tab
 with tab4:
     motivation = st.text_input("Motivation for Graduate Studies")
-    
-    
     career_goals = st.text_input("Career Goals")
     why_this_program = st.text_input("Why This Program")
     
     if st.button("Generate SOP", key="sop_btn"):
-        with st.spinner("Generating..."):
-            try:
-                resume_info = parse_resume(cv_resume_text)
-                generated_sop = chains['sop'].run({
-                    "motivation": motivation,
-                    "academic_background": resume_info['education'],
-                    "research_experiences": resume_info['publications'],
-                    "career_goals": career_goals,
-                    "why_this_program": why_this_program,
-                    "experience": resume_info['experience'],
-                    "skills": resume_info['skills'],
-                    "projects": resume_info['projects']
-
-                })
-                st.session_state.generated_content['sop'] = clean_sop_output(generated_sop)
-            except Exception as e:
-                st.error(f"Generation error: {e}")
+        if cv_resume_text:
+            with st.spinner("Generating..."):
+                try:
+                    generated_sop = chains['sop'].run({
+                        **resume_info,
+                        "motivation": motivation,
+                        "career_goals": career_goals,
+                        "why_this_program": why_this_program
+                    })
+                    st.session_state.generated_content['sop'] = clean_output(generated_sop, "sop")
+                except Exception as e:
+                    st.error(f"Generation error: {e}")
+        else:
+            st.error("Please upload your CV/Resume")
 
     if st.session_state.generated_content['sop']:
         st.markdown('<div class="output-container">', unsafe_allow_html=True)