parser/ner_extractor.py

import spacy
import re
from difflib import get_close_matches

# Safe model load
try:
    nlp = spacy.load("en_core_web_sm")
except:
    from spacy.cli import download
    download("en_core_web_sm")
    nlp = spacy.load("en_core_web_sm")

# Common skills list (used for fuzzy matching)
common_skills = [
    "python", "sql", "java", "c++", "machine learning", "deep learning", "nlp", "tensorflow",
    "keras", "pandas", "numpy", "matplotlib", "power bi", "excel", "html", "css", "django", "flask",
    "scikit-learn", "seaborn", "fastapi", "pytorch"
]

edu_keywords = ["bachelor", "bs", "msc", "ms", "phd", "b.sc", "b.s.", "university", "institute", "college"]


def extract_skills(text):
    doc = nlp(text)
    # Extract noun phrases
    noun_phrases = set(chunk.text.lower().strip() for chunk in doc.noun_chunks)

    # Add common skills if fuzzy matched
    matched_common = set()
    words = set(text.lower().split())
    for skill in common_skills:
        if get_close_matches(skill.lower(), words, n=1, cutoff=0.85):
            matched_common.add(skill.lower())

    # Combine and clean
    all_skills = noun_phrases.union(matched_common)
    filtered = [s for s in all_skills if 2 <= len(s.split()) <= 5 and not s.startswith(("i ", "my ", "our "))]
    return list(set(filtered))


def extract_entities(text):
    doc = nlp(text)

    entities = {
        "Name": None,
        "Email": None,
        "Phone": None,
        "Skills": [],
        "Education": [],
        "Experience": [],
        "Certifications": [],
        "Projects": []
    }

    # ✅ Extract Name
    for ent in doc.ents:
        if ent.label_ == "PERSON":
            entities["Name"] = ent.text
            break

    # ✅ Extract Email & Phone
    entities["Email"] = next(iter(re.findall(r"[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+", text)), None)
    entities["Phone"] = next(iter(re.findall(r"\+?\d[\d\s\-]{8,15}", text)), None)

    # ✅ Extract Skills
    entities["Skills"] = extract_skills(text)

    # ✅ Section-based parsing
    current_section = None
    lines = text.split("\n")
    for line in lines:
        clean_line = line.strip()
        if not clean_line:
            continue
        line_lower = clean_line.lower()

        if "experience" in line_lower:
            current_section = "Experience"
        elif "project" in line_lower:
            current_section = "Projects"
        elif "certification" in line_lower:
            current_section = "Certifications"
        elif any(kw in line_lower for kw in edu_keywords):
            entities["Education"].append(clean_line)
        elif current_section:
            entities[current_section].append(clean_line)

    # ✅ Deduplicate list entries
    for key in entities:
        if isinstance(entities[key], list):
            entities[key] = list(set(entities[key]))

    return entities