app.py

import json
import numpy as np
import tensorflow as tf
from tensorflow.keras import layers
import sentencepiece as spm
import requests
from flask import Flask, request, Response, session, jsonify
from bs4 import BeautifulSoup
from huggingface_hub import hf_hub_download
import uuid
import os
import time
from collections import Counter

app = Flask(__name__, static_folder="static")
app.secret_key = os.urandom(32)

# =====================
# 모델/토크나이저 다운로드
# =====================
os.environ["HF_HOME"] = "/tmp/hf_cache"
hf_token = os.getenv("HF_TOKEN")

CHAT_MODEL_PATH = hf_hub_download(
    repo_id="AlphaCNN/LamCo",
    filename="LamCo.weights.h5",
    repo_type="model",
    token=hf_token
)
CHAT_TOKENIZER_PATH = hf_hub_download(
    repo_id="AlphaCNN/LamCo",
    filename="ko_unigram.model",
    repo_type="model",
    token=hf_token
)

print(CHAT_MODEL_PATH)
sp = spm.SentencePieceProcessor()
sp.load(CHAT_TOKENIZER_PATH)
pad_id = sp.piece_to_id("<pad>") or 0
start_id = sp.piece_to_id("<start>") or 1
end_id = sp.piece_to_id("<end>") or 2
unk_id = sp.piece_to_id("<unk>") or 3
sep_id = sp.piece_to_id("<sep>")
vocab_size = sp.get_piece_size()
max_len = 125

def text_to_ids(text):
    return sp.encode(text, out_type=int)

def ids_to_text(ids):
    return sp.decode(ids)

class SwiGLU(layers.Layer):
    def __init__(self, d_model, f_d=8/3):
        super().__init__()
        hidden_dim = int(d_model * f_d)
        self.proj = layers.Dense(hidden_dim * 2, use_bias=False, dtype='float32')
        self.out = layers.Dense(d_model, use_bias=False, dtype='float32')

    def call(self, x):
        x_val, x_gate = tf.split(self.proj(x), 2, axis=-1)
        return self.out(x_val * tf.nn.silu(x_gate))

class DilatedConvLayer(layers.Layer):
    def __init__(self, d_model, dilation_rate, dropout_rate=0.1):
        super().__init__()
        self.conv = layers.Conv1D(
            filters=d_model,
            kernel_size=3,
            dilation_rate=dilation_rate,
            padding='causal',
            use_bias=True,
            kernel_initializer='he_normal',
            dtype='float32'
        )
        self.ln = layers.LayerNormalization(epsilon=1e-5, dtype='float32')
        self.dropout = layers.Dropout(dropout_rate)

    def call(self, x, training=False):
        residual = x
        x = self.conv(x)
        x = self.ln(x + residual)
        x = self.dropout(x, training=training)
        return x

class Lamko(tf.keras.Model):
    def __init__(self, vocab_size, max_seq_len, d_model, n_layers, dropout_rate=0.1):
        super().__init__()
        self.token_embedding = layers.Embedding(vocab_size, d_model, dtype='float32')
        self.pos_embedding = layers.Embedding(max_seq_len, d_model, dtype='float32')

        self.blocks = []
        for i in range(n_layers):
            self.blocks.append(DilatedConvLayer(d_model, 2 ** i, dropout_rate))
            if (i + 1) % 3 == 0:
                self.blocks.append(SwiGLU(d_model))
                self.blocks.append(layers.LayerNormalization(epsilon=1e-5, dtype='float32'))

        self.ln_f = layers.LayerNormalization(epsilon=1e-5, dtype='float32')

    def call(self, x, training=False):
        batch_size, seq_len = tf.shape(x)[0], tf.shape(x)[1]
        positions = tf.range(seq_len)[tf.newaxis, :]
        positions = tf.clip_by_value(positions, 0, self.pos_embedding.input_dim - 1)

        x = self.token_embedding(x) + self.pos_embedding(positions)

        for block in self.blocks:
            if isinstance(block, SwiGLU):
                x = x + block(x)
            else:
                x = block(x, training=training) if hasattr(block, 'training') else block(x)

        x = self.ln_f(x)
        logits = tf.matmul(x, self.token_embedding.weights[0], transpose_b=True)
        return logits

model = Lamko(vocab_size=vocab_size, max_seq_len=max_len, d_model=384, n_layers=9)
dummy_input = tf.zeros((1, max_len), dtype=tf.int32)
_ = model(dummy_input)
model.load_weights(CHAT_MODEL_PATH)
print("모델 가중치 로드 완료!")

@tf.function(input_signature=[
    tf.TensorSpec(shape=(1, None), dtype=tf.int32),  # input_ids
    tf.TensorSpec(shape=(vocab_size,), dtype=tf.int32),  # token_counts
    tf.TensorSpec(shape=(), dtype=tf.int32),  # current_length
    tf.TensorSpec(shape=(), dtype=tf.float32),  # temperature
    tf.TensorSpec(shape=(), dtype=tf.float32),  # repetition_penalty
    tf.TensorSpec(shape=(), dtype=tf.float32),  # top_p
    tf.TensorSpec(shape=(), dtype=tf.int32),  # top_k
    tf.TensorSpec(shape=(), dtype=tf.int32),  # min_len
    tf.TensorSpec(shape=(), dtype=tf.int32),  # step
])
def generate_step(input_ids, token_counts, current_length, temperature, repetition_penalty, top_p, top_k, min_len, step):
    pad_len = max_len - tf.shape(input_ids)[1]
    input_padded = tf.pad(input_ids, [[0,0],[0,pad_len]], constant_values=pad_id)
    logits = model(input_padded, training=False)
    next_logits = logits[0, current_length - 1]

    penalty = tf.pow(repetition_penalty, tf.cast(token_counts, tf.float32))
    next_logits = next_logits / penalty

    # 최소 길이와 pad 마스킹
    if current_length < min_len:
        next_logits = tf.tensor_scatter_nd_update(next_logits, [[end_id]], [-1e9])
    next_logits = tf.tensor_scatter_nd_update(next_logits, [[pad_id]], [-1e9])

    # top-k 필터링
    if top_k > 0:
        kth_val = tf.math.top_k(next_logits, k=top_k).values[-1]
        mask = next_logits < kth_val
        next_logits = tf.where(mask, -1e9, next_logits)

    # top-p (nucleus) 필터링 + temperature
    next_logits = next_logits / temperature
    probs = tf.nn.softmax(next_logits)
    sorted_probs, sorted_idx = tf.math.top_k(probs, k=vocab_size)
    cum_probs = tf.cumsum(sorted_probs)
    cutoff_mask = cum_probs <= top_p
    cutoff_idx = tf.reduce_sum(tf.cast(cutoff_mask, tf.int32)) + 1
    cutoff_idx = tf.minimum(cutoff_idx, vocab_size)
    filtered_idx = sorted_idx[:cutoff_idx]
    filtered_probs = sorted_probs[:cutoff_idx]
    filtered_probs = filtered_probs / tf.reduce_sum(filtered_probs)

    # 🔹 50%는 argmax, 50%는 샘플링
    rand_val = tf.random.uniform([], 0, 1)
    def sample():
        sampled_id = tf.random.categorical(tf.math.log([filtered_probs]), 1)[0,0]
        return filtered_idx[sampled_id]
    def argmax():
        return filtered_idx[tf.argmax(filtered_probs)]
    sampled_id = tf.cond(rand_val < 0.5536, argmax, sample)
    sampled_id = tf.cast(sampled_id, tf.int32)

    # token_counts 업데이트
    token_counts = tf.tensor_scatter_nd_add(token_counts, [[sampled_id]], [1])
    return sampled_id, token_counts


# =====================
# 스트리밍 생성기 (CPU 최적화 버전)
# =====================
def generate_text_streaming(model, prompt, max_len=115, max_gen=100,
                            temperature=0.75, min_len=20,
                            repetition_penalty=1.2, top_p=0.9, top_k=50):
    model_input = text_to_ids(f"<start> {prompt} <sep>")
    model_input = model_input[:max_len]
    generated = list(model_input)
    start_output_idx = len(model_input)

    # TF 변수로 토큰 카운트 관리
    token_counts_np = np.zeros(vocab_size, dtype=np.int32)
    for t in generated:
        token_counts_np[t] += 1
    token_counts = tf.Variable(token_counts_np, dtype=tf.int32)

    prev_decoded = ""

    for step in range(max_gen):
        input_tensor = tf.expand_dims(generated, axis=0)  # [1, seq_len]

        sampled_id, token_counts = generate_step(
            input_tensor,
            token_counts,
            tf.constant(len(generated), dtype=tf.int32),
            tf.constant(temperature, dtype=tf.float32),
            tf.constant(repetition_penalty, dtype=tf.float32),
            tf.constant(top_p, dtype=tf.float32),
            tf.constant(top_k, dtype=tf.int32),
            tf.constant(min_len, dtype=tf.int32),
            tf.constant(step, dtype=tf.int32)
        )

        sampled_id = int(sampled_id.numpy())
        generated.append(sampled_id)

        # 디코딩은 출력 시점에만
        if len(generated) > start_output_idx:
            decoded_full = sp.decode(generated[start_output_idx:])
            decoded_full = decoded_full.replace("▁", " ").strip()
            for t in ["<start>", "<sep>", "<end>"]:
                decoded_full = decoded_full.replace(t, "")
            decoded_full = decoded_full.lstrip(",!?.는은 ")

            new_output = decoded_full[len(prev_decoded):]
            if new_output:
                yield new_output
                prev_decoded = decoded_full

            # 종료 조건
            if len(generated) >= min_len and (sampled_id == end_id or decoded_full.endswith(('.', '!', '?'))):
                break

token_map = {
    "하이": "안녕하세요!",
    "ㅎㅇ": "안녕하세요!",
    "하이~": "안녕하세요!",
    "안녕": "안녕하세요!",
    "안녕!": "안녕하세요!",
    "잘가": "잘가. 나중에 보자",
    "잘 가": "잘 가. 나중에 보자"
}

def preprocess_text(text):
    for key, val in token_map.items():
        text = text.replace(key, val)
    return text

# =====================
@app.route('/')
def index():
    return app.send_static_file('index.html')

@app.route('/api/search')
def search_api():
    query = request.args.get("query", "").strip()
    if not query:
        return jsonify({"results": []})

    search_url = f"https://ko.wikipedia.org/w/index.php?search={query}"
    headers = {"User-Agent": "Mozilla/5.0"}
    resp = requests.get(search_url, headers=headers)
    soup = BeautifulSoup(resp.text, "html.parser")

    results = []

    # 1. 검색 결과 리스트가 있는 경우
    search_items = soup.select(".mw-search-result-heading a")
    if search_items:
        for item in search_items[:5]:
            title = item.text
            link = "https://ko.wikipedia.org" + item.get("href")
            snippet_tag = item.find_parent().find("div", class_="searchresult")
            snippet = snippet_tag.text.strip() if snippet_tag else ""
            results.append({"title": title, "link": link, "snippet": snippet})

    # 2. 검색어와 정확히 일치하는 페이지로 바로 이동한 경우
    elif soup.select("#firstHeading"):
        title = soup.select_one("#firstHeading").text.strip()
        link = resp.url
        # 문서 첫 번째 단락 추출
        content_paragraph = soup.select_one(".mw-parser-output > p")
        snippet = content_paragraph.text.strip() if content_paragraph else ""
        results.append({"title": title, "link": link, "snippet": snippet})

    return jsonify({"results": results})

@app.before_request
def ensure_user_id():
    if 'user_id' not in session:
        session['user_id'] = str(uuid.uuid4())

@app.route('/api/chat', methods=['GET','POST'])
def chat_api():
    user_msg = (request.json.get("message") if request.method=="POST" else request.args.get("message") or "").strip()
    if not user_msg:
        return Response((f'data: {{"error":"메시지를 입력해주세요."}}\n\n' for _ in range(1)),
                        mimetype='text/event-stream')
    user_id = session['user_id']

    user_msg = preprocess_text(user_msg)

    def gen():
        try:
            # 외부 검색 제거, search_result는 항상 빈 문자열
            search_result = ""
            for token in generate_text_streaming(
            model, user_msg,
            max_len=max_len,
            max_gen=115,
            temperature=0.8,
            min_len=10,
            repetition_penalty=1.1,
            top_p=0.9,
            top_k=5
        ):
                safe_token = json.dumps(token)
                yield f'data: {{"char":{safe_token}}}\n\n'

            yield 'data: {"done":true}\n\n'
        except Exception as e:
            yield f'data: {{"error":{json.dumps(str(e))}}}\n\n'

    return Response(gen(), mimetype='text/event-stream')

if __name__=="__main__":
    app.run(host="0.0.0.0", port=7860)