# ====== A. Pembersihan Teks Ringan (Python) ======
import re

def clean_text(s: str):
    s = s.lower()
    s = re.sub(r"http\S+|www\S+", " ", s)
    s = re.sub(r"[^a-zA-Z\s]", " ", s)  # contoh: huruf latin saja
    s = re.sub(r"\s+", " ", s).strip()
    return s

sample = "Produk ini BAGUS bgt!!! Cek di https://toko.example/123 😊"
print(clean_text(sample))

# ====== B. BoW & TF‑IDF (contoh cepat) ======
# Jika perlu: !pip install scikit-learn
from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer

corpus = [
  "makan nasi di warung padang enak",
  "warung padang ramai sekali",
  "saya makan mie ayam enak",
  "mie ayam dan bakso enak",
]

cv = CountVectorizer(ngram_range=(1,1))
x_bow = cv.fit_transform(corpus)
print('Vocab BoW:', cv.get_feature_names_out())
print('Shape BoW:', x_bow.shape)

tfidf = TfidfVectorizer(ngram_range=(1,2), min_df=1)
x_tfidf = tfidf.fit_transform(corpus)
print('Vocab TF‑IDF (dengan bigram):', tfidf.get_feature_names_out()[:12])
print('Shape TF‑IDF:', x_tfidf.shape)

# ====== C. Baseline Multinomial Naive Bayes (sentimen mini) ======
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.naive_bayes import MultinomialNB
from sklearn.metrics import classification_report, confusion_matrix, accuracy_score

# Dataset mini (positif/negatif) — ganti/augment sesuai kebutuhan
texts = [
  'bagus sekali pelayanannya', 'sangat memuaskan', 'luar biasa',
  'buruk dan lambat', 'mengecewakan', 'tidak rekomendasi',
  'produk bagus kualitas oke', 'harga tidak sesuai', 'kualitas jelek',
  'pengiriman cepat', 'layanan ramah', 'salah kirim barang',
  'mantap enak', 'parah rusak', 'top banget',
]
labels = [1,1,1,0,0,0,1,0,0,1,1,0,1,0,1]  # 1=positif, 0=negatif

Xtr, Xte, ytr, yte = train_test_split(texts, labels, test_size=0.3, random_state=42, stratify=labels)
vec = TfidfVectorizer(ngram_range=(1,2), min_df=1)
Xtrv = vec.fit_transform(Xtr)
Xtev = vec.transform(Xte)

clf = MultinomialNB()
clf.fit(Xtrv, ytr)
yp = clf.predict(Xtev)

print('Akurasi:', accuracy_score(yte, yp))
print('Confusion Matrix:\n', confusion_matrix(yte, yp))
print(classification_report(yte, yp, digits=3))

# ====== D. Logistic Regression + Pipeline + GridSearch ======
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.pipeline import Pipeline
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report

# Contoh kategorisasi 2 kelas (ganti dengan data Anda)
texts = [
  'jadwal rapat proyek senin pagi', 'rapat ditunda minggu depan', 'proposal pendanaan inovasi',
  'diskon besar belanja akhir pekan', 'voucher promo belanja hemat', 'flash sale tengah malam'
]
labels = [0,0,0,1,1,1]  # 0=office, 1=promo

Xtr, Xte, ytr, yte = train_test_split(texts, labels, test_size=0.34, random_state=0, stratify=labels)
pipe = Pipeline([
  ('tfidf', TfidfVectorizer(ngram_range=(1,2), min_df=1)),
  ('clf', LogisticRegression(max_iter=1000))
])
param = {
  'tfidf__ngram_range': [(1,1),(1,2)],
  'clf__C': [0.1, 1, 5],
  'clf__penalty': ['l2']
}
cv = GridSearchCV(pipe, param, cv=3, n_jobs=-1)
cv.fit(Xtr, ytr)
print('Best params:', cv.best_params_)
print(classification_report(yte, cv.best_estimator_.predict(Xte)))

# ====== E. Dataset Lebih Besar: 20 Newsgroups (opsional) ======
# Jalankan di Colab (perlu download data). Pastikan koneksi internet tersedia di runtime Colab.
from sklearn.datasets import fetch_20newsgroups
from sklearn.model_selection import train_test_split
from sklearn.pipeline import Pipeline
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report

cats = ['sci.space','rec.sport.baseball']  # contoh dua kategori
news = fetch_20newsgroups(subset='all', categories=cats, remove=('headers','footers','quotes'))
Xtr, Xte, ytr, yte = train_test_split(news.data, news.target, test_size=0.25, random_state=42, stratify=news.target)

pipe = Pipeline([
  ('tfidf', TfidfVectorizer(ngram_range=(1,2), min_df=2, max_df=0.9)),
  ('clf', LogisticRegression(max_iter=2000))
])
pipe.fit(Xtr, ytr)
print(classification_report(yte, pipe.predict(Xte)))

# ====== F. Heatmap Confusion Matrix ======
import numpy as np
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix

true = [1,1,1,0,0,0,1,0,1,0]
pred = [1,1,0,0,0,1,1,0,1,0]
cm = confusion_matrix(true, pred)

plt.figure(figsize=(4,3))
plt.imshow(cm)
for (i,j),v in np.ndenumerate(cm):
    plt.text(j, i, str(v), ha='center', va='center')
plt.title('Confusion Matrix')
plt.xlabel('Prediksi'); plt.ylabel('Benar')
plt.colorbar(); plt.show()

# ====== G. Kuis 15 (cek mandiri) ======
qs=[
  ("Apa keunggulan TF‑IDF dibanding BoW murni?",
   {"a":"Meningkatkan bobot kata umum","b":"Menurunkan bobot kata umum & menaikkan kata spesifik","c":"Menghapus semua stopwords","d":"Selalu lebih cepat"},"b"),
  ("Model baseline cepat untuk teks adalah...",
   {"a":"KNN","b":"Decision Tree","c":"Multinomial Naive Bayes","d":"K‑Means"},"c"),
  ("Metrik yang cocok untuk data tak seimbang:",
   {"a":"Akurasi saja","b":"Macro‑F1/Recall","c":"MSE","d":"Silhouette"},"b"),
  ("Dalam LR, parameter C mengatur...",
   {"a":"Jumlah n‑gram","b":"Kekuatan regularisasi (kebalikan) ","c":"Jumlah epoch","d":"Stopwords"},"b"),
]
print('Kunci jawaban:')
for i,(_,__,ans) in enumerate(qs,1):
    print(f'Q{i}: {ans}')