Ars/ai_functions.py

from pdfminer.high_level import extract_text_to_fp
from io import BytesIO
from Ars.objects import ai,ResumeData,AutomationRiskResult,AutomationRiskInput,RealWorldQuestion,SkillDepthResult,SkillDepthInput,BreakDownByDomainUpdate,FlaggedRiskAreasUpdate,BoostSuggestionsUpdate,AICollabReadinessInput
from Ars.embedDoc import search_pinecone_text
from fastapi import UploadFile

def extract_text_from_bytes(pdf_bytes: bytes) -> str:
    output_string = BytesIO()
    with BytesIO(pdf_bytes) as input_stream:
        extract_text_to_fp(input_stream, output_string)
    return output_string.getvalue().decode()

            
async def resume_analysis(upload_file:UploadFile) -> ResumeData:
    contents = await upload_file.read()
    resume = extract_text_from_bytes(pdf_bytes=contents)
    if resume:
        prompt = f"""
        You are an AI resume parser. Read the following resume and extract the following structured information from the resume below if any was found:

        1. Work experience details: job titles and descriptions.(max3)
        2. List of technical skills (technologies, tools, platforms, etc.).(max4)
        3. Soft and transferable skills: 
            - Leadership and collaboration
            - Critical thinking and problem-solving
            - Communication skills
            - Cross-functional/interdisciplinary experience
            - Initiative and adaptability
        4. Career path details:
            - Current or intended role/field
            - Industry and sector context
            - Career trajectory trends
        5. Project experience (if any): project titles, descriptions, role.(max3)
        6. Evidence of upskilling (optional): certifications, courses, learning projects, hackathons.

        Return the extracted data as a JSON object that matches this schema:

        Here is the resume:

        {resume}
        """
        result =ai.chat(prompt=prompt,output_schema=ResumeData)
        return result





def calculate_automation_risk(resume:ResumeData):
    def generate_real_world_data(resume:ResumeData):
        prompt=f"Generata a search query using the resume details below I want to check some text that has data on future jobs reports using semantic searches  Here is the resume data: {resume} "
        result = ai.chat(prompt=prompt,output_schema=RealWorldQuestion)
        search_result =search_pinecone_text(query_text=result.question)

        return search_result

    real_world_data=generate_real_world_data(resume=resume)
    prompt = f"""
    You are an Automation Risk Calculator AI. Your task is to analyze the user's resume and any provided real-world data to estimate how automatable their skills and job titles are.

    For each of the following factors, please evaluate on a scale from 0 to 5, where:

    0 = Not present or irrelevant  
    1 = Very low presence or impact  
    2 = Low presence or impact  
    3 = Moderate presence or impact  
    4 = High presence or impact  
    5 = Very high presence or impact

    Factors to score:

    - repetitiveness of tasks  
    - creativity required  
    - emotional intelligence needed  
    - reliance on data-driven tasks  
    - physical/manual work involved  
    - level of autonomous decision-making  
    - need for strategic thinking  
    - importance of collaboration  
    - current use of AI tools in the role  
    - recent upskilling or adaptability  

    The real-world data may contain labor market trends, automation probabilities, or other relevant insights — use it if helpful.

    After scoring all factors, calculate an overall automation risk score between 0 and 100, where:  
    - 100 means very low automation risk (skills are highly resilient to automation),  
    - 0 means very high automation risk (skills are highly automatable).

    Return only the final integer risk score (0 to 100) — no explanations or extra text.

    Here is the resume:

    (
    {resume}
    )

    Here is the real-world data:

    (
    {real_world_data}
    )
    """
    result = ai.chat(prompt=prompt, output_schema=AutomationRiskInput)
    return result



def calculate_skill_depth(resume:ResumeData):


    prompt = f"""
You are a Skill Depth Calculator AI. Your task is to analyze the user's resume data and score various skill-related factors.

For each of the following factors, please score on a scale from 0 to 5, where:

0 = Not present or no depth  
1 = Very shallow experience or presence  
2 = Basic or limited experience  
3 = Moderate experience or involvement  
4 = Strong experience or senior level  
5 = Expert or highly advanced experience  

Factors to score include (but are not limited to):

- Years of experience per skill  
- Seniority level in roles held  
- Number and relevance of certifications  
- Breadth of skills (variety and diversity)  
- Depth in core technical skills  
- Depth in leadership or management skills  
- Involvement in complex projects  
- Contributions to strategic initiatives  
- Frequency of skill usage in recent roles  
- Evidence of continuous learning or upskilling  
- Cross-functional collaboration skills  
- Recognition or awards related to skills  
- Public speaking or training experience  
- Publications or patents (if any)  
- Industry-specific expertise depth  
- Mentoring or coaching experience  
- Ability to innovate using skills  
- Adaptability to new technologies  
- Problem-solving skills depth  
- Communication skills related to technical content  



Here is the resume:

(
{resume}
)
"""

    result=ai.chat(prompt=prompt,output_schema=SkillDepthInput)
    return result



def calculate_Ai_collab_readiness(resume:ResumeData):


    prompt = f"""
You are an AI Collaboration Readiness Calculator. Your task is to read the following resume and assess how ready the individual is to **collaborate with AI tools**, not just compete against them.

Score the user on the following 10 readiness traits, each on a scale from 0 to 5:

0 = Not mentioned or no evidence  
1 = Very minimal evidence or weak relevance  
2 = Some mention, limited depth  
3 = Moderate use or awareness  
4 = Strong use or understanding  
5 = Expert use or deep integration

Traits to evaluate:

- Mention of AI tools (e.g., ChatGPT, GitHub Copilot, Midjourney, etc.)
- Adaptability to AI-enhanced workflows
- Willingness to learn and adopt AI tools
- Understanding of AI ethics and responsible use
- Demonstrated collaboration with AI (not just use)
- Use of AI for problem-solving
- Creativity in applying AI to new tasks or contexts
- Speed in learning and applying new AI technologies
- Communication skills related to explaining or using AI
- Integration of AI into work/study processes


Here is the resume:

(
{resume}
)
"""

    result=ai.chat(prompt=prompt,output_schema=AICollabReadinessInput)
    return result






def generate_domain_breakdown(resume:ResumeData,**helpful_key_data):


    prompt = f"""
You have one clear objective: analyze the following resume and produce a domain skills breakdown based on the taxonomy below.

**Task:**  
Classify and score the candidate’s skills/domains on a scale from 0 to 100, where 0 means no proficiency and 100 means expert-level proficiency.

**Example output format:**  
- Technical: 65  
- Creative: 80  
- Strategy: 75  
- Collaboration: 60

**Domains to consider (but not limited to):**  
- Technical (e.g., software development, programming)  
- Creative (e.g., design, content creation)  
- Strategy (e.g., planning, business analysis)  
- Collaboration (e.g., teamwork, communication)  

**Requirements:**  
- Return only the domain names and their scores as shown in the example.  
- Scores must be integers between 0 and 100.  
- Provide 4–6 domain categories relevant to the resume content.  
- Use your best judgment based on the resume to assign scores reflecting demonstrated skills and experience.

Here is the resume to analyze:

({resume})


**Analysis data to aid  evaluation**:
**{helpful_key_data}**
"""

    result=ai.chat(prompt=prompt,output_schema=BreakDownByDomainUpdate)
    return result





def generate_flagged_risk_areas(resume:ResumeData,**helpful_key_data):


    prompt = f"""

**You are an AI Resume Risk Evaluator.**
Your job is to review the following resume and flag potential *career resilience risks*. Focus strictly on:

1. **Missing recent/up-to-date certifications**
2. **Lack of AI or automation tool usage**
3. **Dependence on easily automatable tasks** (e.g., repetitive data entry, basic spreadsheets)

For each issue identified, return **a concise 1–2 sentence explanation** with clear language.

**Response Format (Examples)**:

* "Heavy reliance on spreadsheet tasks, which are easily automated."
* "No evidence of recent certifications to stay current in the field."
* "Resume lacks any mention of AI or automation tools."

**Resume to evaluate**:

```
({resume})  
```

**Analysis data to aid  evaluation**:
**{helpful_key_data}**

"""
    result=ai.chat(prompt=prompt,output_schema=FlaggedRiskAreasUpdate)
    return result




def generate_boost_suggestion(resume:ResumeData,**helpful_key_data):


    prompt = f"""
You are a Resume Enhancement AI. Analyze the resume below and generate 2 to 4 clear, actionable suggestions to help the user strengthen their profile by addressing skill or experience gaps.

Focus on practical, AI-driven improvements such as:
- Mastering relevant AI tools
- Enrolling in advanced or specialized courses
- Expanding the complexity or impact of projects
- Obtaining up-to-date certifications

**Examples**:
- "Learn ChatGPT to boost your coding efficiency."
- "Enhance your Python skills through an advanced course."
- "Add recent certifications to demonstrate continuous growth."
- "Integrate AI tools into your project portfolio."

**Guidelines**:
- Suggestions must be specific, relevant, and directly tied to the resume content.
- Keep each suggestion concise (1–2 sentences).
- Avoid generic advice; prioritize actionable, targeted improvements.

Resume to analyze:

({resume})
**Analysis data to aid  evaluation**:
**{helpful_key_data}**
"""

    result=ai.chat(prompt=prompt,output_schema=BoostSuggestionsUpdate)
    return result