app.py

import os
import re
import json
import requests
import gradio as gr
from pypdf import PdfReader
from docx import Document

OPENWEBNINJA_API_KEY = os.getenv("OPENWEBNINJA_API_KEY")

GEMINI_CHAT_URL = "https://api.openwebninja.com/gemini/chat"
JSEARCH_URL = "https://api.openwebninja.com/jsearch/search-v2"


def extract_text(file):
    if file is None:
        return ""

    path = file if isinstance(file, str) else file.name

    print("FILE PATH:", path)

    try:
        # Try PDF first
        try:
            reader = PdfReader(path)

            text = ""
            for page in reader.pages:
                page_text = page.extract_text()
                if page_text:
                    text += page_text + "\n"

            if text.strip():
                print("Detected PDF")
                return text

        except:
            pass

        # Try DOCX
        try:
            doc = Document(path)

            text = "\n".join(
                p.text for p in doc.paragraphs
            )

            if text.strip():
                print("Detected DOCX")
                return text

        except:
            pass

        # Try TXT
        try:
            with open(path, "r", encoding="utf-8", errors="ignore") as f:
                text = f.read()

            if text.strip():
                print("Detected TXT")
                return text

        except:
            pass

    except Exception as e:
        print("TEXT EXTRACTION ERROR:", e)

    return ""


def clean_text(text):
    text = text.replace("\n", " ")
    text = re.sub(r"([a-z0-9])([A-Z])", r"\1 \2", text)
    text = re.sub(r"\s+", " ", text)
    return text.strip()


def call_openwebninja_gemini(prompt):
    headers = {
        "X-API-Key": OPENWEBNINJA_API_KEY,
        "Content-Type": "application/json"
    }

    payloads = [
        {"message": prompt},
        {"prompt": prompt},
        {"messages": [{"role": "user", "content": prompt}]}
    ]

    for payload in payloads:
        try:
            response = requests.post(
                GEMINI_CHAT_URL,
                headers=headers,
                json=payload,
                timeout=60
            )

            print("GEMINI STATUS:", response.status_code)
            print("GEMINI RESPONSE:", response.text[:1000])

            if response.status_code == 200:
                return response.text

        except Exception as e:
            print("Gemini API error:", e)

    return ""


def extract_json(text):
    if not text:
        return None

    try:
        data = json.loads(text)

        # OpenWebNinja Gemini format
        if isinstance(data, dict):
            if "data" in data and isinstance(data["data"], dict):
                reply_text = data["data"].get("reply_text")
                if reply_text:
                    return extract_json(reply_text)

            for key in ["reply_text", "response", "text", "message", "content", "answer"]:
                if key in data and isinstance(data[key], str):
                    return extract_json(data[key])

            return data

    except:
        pass

    match = re.search(r"\{[\s\S]*\}", text)

    if match:
        try:
            return json.loads(match.group(0))
        except:
            return None

    return None

def analyze_resume_with_ai(resume_text):
    prompt = f"""

You are an AI resume analysis assistant.

Analyze this resume/CV from any career field.

Return ONLY valid JSON in this exact structure:

{{

  "candidate_field": "",

  "seniority_level": "",

  "education": [],

  "work_experience": [],

  "technical_skills": [],

  "soft_skills": [],

  "certifications": [],

  "suitable_job_titles": [],

  "job_search_queries": []

}}

Rules:

Rules:

- The resume can be from any field.

- Generate 15 to 20 suitable_job_titles.

- Include junior, graduate, entry-level and related roles.

- Include alternative titles and synonyms.

- Generate 10 to 15 diverse job_search_queries.

- job_search_queries must not be empty.

- Return at least 15 suitable_job_titles.

- Do not include explanations outside JSON.

Resume:

{resume_text[:7000]}

"""

    raw = call_openwebninja_gemini(prompt)
    profile = extract_json(raw)

    if not profile:
        return None, raw

    # Make sure all expected keys exist
    profile.setdefault("candidate_field", "")
    profile.setdefault("seniority_level", "")
    profile.setdefault("education", [])
    profile.setdefault("work_experience", [])
    profile.setdefault("technical_skills", [])
    profile.setdefault("soft_skills", [])
    profile.setdefault("certifications", [])
    profile.setdefault("suitable_job_titles", [])
    expanded = []

    for role in profile["suitable_job_titles"]:

        expanded.extend([
            role,
            f"Junior {role}",
            f"Graduate {role}",
            f"Entry Level {role}"
        ])

    profile["suitable_job_titles"] = list(
        dict.fromkeys(expanded)
    )
    profile.setdefault("job_search_queries", [])

    # Fallback query generation
    if not profile["job_search_queries"]:
        fallback_queries = []

        for title in profile["suitable_job_titles"]:
            fallback_queries.append(title)

        if profile["candidate_field"]:
            fallback_queries.append(profile["candidate_field"])

        for skill in profile["technical_skills"][:4]:
            fallback_queries.append(skill)

        # Remove duplicates
        seen = set()
        clean_queries = []

        for q in fallback_queries:
            q = str(q).strip()
            if q and q.lower() not in seen:
                seen.add(q.lower())
                clean_queries.append(q)

        profile["job_search_queries"] = clean_queries[:8]

    return profile, raw

def clean_query(query, location):
    query = query.replace("jobs in", "")
    query = query.replace("Jobs in", "")
    query = query.replace(location, "")
    return " ".join(query.split()).strip()


def search_jobs(query, location):
    headers = {
        "X-API-Key": OPENWEBNINJA_API_KEY
    }

    cleaned = clean_query(query, location)
    full_query = f"{cleaned} jobs in {location}"

    response = requests.get(
        JSEARCH_URL,
        params={"query": full_query},
        headers=headers,
        timeout=30
    )

    print("JSEARCH QUERY:", full_query)
    print("JSEARCH STATUS:", response.status_code)
    print("JSEARCH RESPONSE:", response.text[:1000])

    if response.status_code != 200:
        return []

    data = response.json()

    jobs = (
        data.get("jobs")
        or data.get("data")
        or data.get("results")
        or data.get("items")
        or []
    )

    if isinstance(jobs, dict):
        jobs = (
            jobs.get("jobs")
            or jobs.get("results")
            or jobs.get("items")
            or []
        )

    return jobs if isinstance(jobs, list) else []


def get_job_field(job, *keys):
    for key in keys:
        if job.get(key):
            return job.get(key)

    for nested_key in ["job", "company", "employer", "details"]:
        nested = job.get(nested_key)
        if isinstance(nested, dict):
            for key in keys:
                if nested.get(key):
                    return nested.get(key)

    return ""


def score_job(profile, job):
    skills = profile.get("technical_skills", []) + profile.get("soft_skills", [])
    titles = profile.get("suitable_job_titles", [])

    job_text = json.dumps(job).lower()
    matched = []

    score = 0

    for skill in skills:
        if skill.lower() in job_text:
            matched.append(skill)
            score += 8

    for title in titles:
        if title.lower() in job_text:
            score += 15

    score = min(score, 100)

    return score, matched


def analyze_resume(file, location):
    try:

        print("FILE OBJECT:", file)
        print("FILE TYPE:", type(file))


        if not OPENWEBNINJA_API_KEY:
            return "❌ OPENWEBNINJA_API_KEY is missing in Hugging Face Secrets."
            
        resume_text = clean_text(extract_text(file))

        print("EXTRACTED LENGTH:", len(resume_text))
        print("EXTRACTED TEXT:", resume_text[:500])

        if not resume_text:
            return f"""

        ❌ Could not extract text from the resume.

        File:

        {file}

        Type:

        {type(file)}

        """
        profile, raw_ai = analyze_resume_with_ai(resume_text)

        if not profile:
            return f"❌ AI could not analyze the resume.\n\nRaw response:\n\n```text\n{raw_ai}\n```"

        queries = profile.get("job_search_queries", [])

        if not queries:
            return f"""❌ AI did not generate job search queries.

        

        Raw AI response:

        

        ```text

        {raw_ai}

        {json.dumps(profile, indent=2)}

        ```"""
        output = "# CareerMatch AI Results\n\n"
        output += f"**Field:** {profile.get('candidate_field', 'Not detected')}\n\n"
        output += f"**Seniority:** {profile.get('seniority_level', 'Not detected')}\n\n"
        output += f"**Skills:** {', '.join(profile.get('technical_skills', [])) or 'Not detected'}\n\n"
        output += f"**Suitable Jobs:** {', '.join(profile.get('suitable_job_titles', [])) or 'Not detected'}\n\n"
        output += f"**Search Queries:** {', '.join(queries)}\n\n---\n\n"

        all_jobs = []

        for query in queries[:15]:
            jobs = search_jobs(query, location)
            output += f"Search: `{clean_query(query, location)} jobs in {location}` → {len(jobs)} jobs found\n\n"

            for job in jobs[:10]:
                if isinstance(job, dict):
                    all_jobs.append(job)

        if not all_jobs:
            return output + "❌ No jobs found from JSearch."

        unique_jobs = []
        seen = set()

        for job in all_jobs:
            title = get_job_field(job, "title", "job_title", "name")
            company = get_job_field(job, "company_name", "company", "employer_name")
            key = f"{title}-{company}"

            if key not in seen:
                seen.add(key)
                unique_jobs.append(job)

        ranked = []

        for job in unique_jobs[:10]:
            score, matched = score_job(profile, job)
            ranked.append((score, matched, job))

        ranked.sort(key=lambda x: x[0], reverse=True)

        output += "\n---\n\n## Real Job Matches\n\n"

        for score, matched, job in ranked:
            title = get_job_field(job, "title", "job_title", "name") or "Unknown Job"
            company = get_job_field(job, "company_name", "company", "employer_name") or "Unknown Company"
            job_location = get_job_field(job, "location", "job_location", "formatted_location") or "Not specified"
            url = get_job_field(job, "url", "apply_link", "job_apply_link", "link") or "#"

            output += f"### {title} — {score}% Match\n\n"
            output += f"**Company:** {company}\n\n"
            output += f"**Location:** {job_location}\n\n"
            output += f"**Matched Skills:** {', '.join(matched) if matched else 'Not detected'}\n\n"
            output += f"[Apply Here]({url})\n\n---\n\n"

        return output

    except Exception as e:
        return f"❌ Error occurred:\n\n```text\n{type(e).__name__}: {str(e)}\n```"


demo = gr.Interface(
    fn=analyze_resume,
    inputs=[
        gr.File(
            label="Upload Resume/CV",
            type="filepath"
        ),
        gr.Textbox(label="Job Location", value="Mauritius")
    ],
    outputs=gr.Markdown(label="Results"),
    title="CareerMatch AI",
    description="Upload any resume and get real job matches using OpenWebNinja Gemini + JSearch."
)


if __name__ == "__main__":
    demo.launch()