File size: 4,683 Bytes
4f67c97
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8f35593
4f67c97
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
import io
import re
import os
import torch
import PyPDF2
from fastapi import FastAPI, UploadFile, File, HTTPException
from fastapi.middleware.cors import CORSMiddleware
from sentence_transformers import SentenceTransformer, util

app = FastAPI()

# ==============================
# CORS (Allow semua untuk testing)
# ==============================
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_methods=["*"],
    allow_headers=["*"],
)

# ==============================
# LOAD MODEL
# ==============================

REPO_ID = "lilcoderi/cv-matcher-model"

try:
    model = SentenceTransformer(REPO_ID)
    model.eval()
except Exception as e:
    raise RuntimeError(f"Gagal load model: {str(e)}")

THRESHOLD = 0.60

# ==============================
# REGEX OPTIMIZED
# ==============================
RE_CLEAN = re.compile(r'[•\-*●▪◦☑]')
RE_SPACES = re.compile(r'\s+')
RE_NON_ALPHA = re.compile(r'[^\w\s]')

# ==============================
# TEXT PREPROCESSING
# ==============================

def clean_text(text: str) -> str:
    text = text.lower()
    text = RE_CLEAN.sub(' ', text)
    text = text.encode("ascii", "ignore").decode()
    text = RE_NON_ALPHA.sub(' ', text)
    return RE_SPACES.sub(' ', text).strip()


def standardize_education(text: str) -> str:
    edu_map = {
        r'\b(sarjana|s1|strata 1|universitas|politeknik|institut)\b': 's1',
        r'\b(diploma 3|d3|ahli madya)\b': 'd3',
        r'\b(sma|smk|stm|smu|ma|sekolah menengah)\b': 'sma_smk',
    }
    for pattern, replacement in edu_map.items():
        text = re.sub(pattern, replacement, text)
    return text


def clean_job_description(text: str) -> str:
    noise_patterns = [
        r'we are hiring',
        r'send us your cv',
        r'kirim cv anda',
        r'subjek:.*',
        r'lowongan ini dibuka sampai.*',
        r'format pdf'
    ]
    for pattern in noise_patterns:
        text = re.sub(pattern, '', text, flags=re.IGNORECASE)
    return text


# ==============================
# PDF READER
# ==============================

def extract_text_from_pdf(file_bytes, max_pages=3):
    try:
        pdf_reader = PyPDF2.PdfReader(io.BytesIO(file_bytes))
        text = ""
        pages_to_read = min(len(pdf_reader.pages), max_pages)

        for i in range(pages_to_read):
            content = pdf_reader.pages[i].extract_text()
            if content:
                text += content + " "

        return text.strip()

    except Exception:
        raise HTTPException(status_code=400, detail="Gagal membaca file PDF")


# ==============================
# HEALTH CHECK (penting buat HF)
# ==============================

@app.get("/")
def root():
    return {"status": "CV Matcher API Running"}


# ==============================
# MAIN ENDPOINT
# ==============================

@app.post("/match")
async def match_cvs(
    job_file: UploadFile = File(...),
    cv_files: list[UploadFile] = File(...)
):

    # ---------- JOB DESCRIPTION ----------
    job_raw = extract_text_from_pdf(await job_file.read(), max_pages=5)
    job_cleaned = clean_job_description(job_raw)
    job_final = standardize_education(clean_text(job_cleaned))

    if not job_final:
        raise HTTPException(status_code=400, detail="Job description kosong")

    # ---------- CV PROCESS ----------
    cv_texts_processed = []
    filenames = []

    for cv in cv_files:
        content = await cv.read()
        raw_text = extract_text_from_pdf(content, max_pages=3)
        processed_text = standardize_education(clean_text(raw_text))

        if processed_text:
            cv_texts_processed.append(processed_text)
            filenames.append(cv.filename)

    if not cv_texts_processed:
        raise HTTPException(status_code=400, detail="Tidak ada CV yang valid")

    # ---------- EMBEDDING ----------
    with torch.no_grad():
        job_embedding = model.encode(
            job_final,
            convert_to_tensor=True,
            normalize_embeddings=True
        )

        cv_embeddings = model.encode(
            cv_texts_processed,
            convert_to_tensor=True,
            normalize_embeddings=True
        )

        scores = util.cos_sim(job_embedding, cv_embeddings)[0]

    # ---------- RESULT ----------
    results = []

    for i in range(len(filenames)):
        score_val = float(scores[i])

        results.append({
            "filename": filenames[i],
            "score": round(score_val, 4),
            "percentage": round(score_val * 100, 2),
            "status": "Cocok" if score_val >= THRESHOLD else "Tidak Cocok"
        })

    results.sort(key=lambda x: x['score'], reverse=True)

    return {"results": results}