Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -6,6 +6,11 @@ import google.generativeai as genai
|
|
| 6 |
import tempfile
|
| 7 |
import base64
|
| 8 |
from concurrent.futures import ThreadPoolExecutor
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9 |
|
| 10 |
# Gemini APIの設定 (環境変数から取得)
|
| 11 |
GOOGLE_API_KEY = os.environ.get("GOOGLE_API_KEY")
|
|
@@ -13,12 +18,6 @@ if not GOOGLE_API_KEY:
|
|
| 13 |
raise ValueError("環境変数 'GOOGLE_API_KEY' が設定されていません。")
|
| 14 |
genai.configure(api_key=GOOGLE_API_KEY)
|
| 15 |
|
| 16 |
-
import logging
|
| 17 |
-
|
| 18 |
-
# ロギング設定
|
| 19 |
-
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
| 20 |
-
|
| 21 |
-
|
| 22 |
|
| 23 |
def split_pdf(pdf_path, output_dir, pages_per_chunk=5):
|
| 24 |
"""PDFを指定ページ数ごとに分割する関数"""
|
|
@@ -60,7 +59,7 @@ def ocr_pdf_with_gemini(pdf_path):
|
|
| 60 |
pdf_base64 = encode_pdf_to_base64(pdf_path)
|
| 61 |
|
| 62 |
# Geminiモデルの設定
|
| 63 |
-
model = genai.GenerativeModel('gemini-1.5-pro') #
|
| 64 |
|
| 65 |
# プロンプトの設定
|
| 66 |
prompt = """
|
|
@@ -83,37 +82,48 @@ def ocr_pdf_with_gemini(pdf_path):
|
|
| 83 |
# 結果を返す
|
| 84 |
return response.text
|
| 85 |
except Exception as e:
|
| 86 |
-
|
| 87 |
-
return f"エラーが発生しました: {e}"
|
| 88 |
|
| 89 |
|
| 90 |
-
def process_pdf(pdf_file):
|
| 91 |
"""PDFファイルを処理するメイン関数"""
|
| 92 |
-
logging.info(f"Received file: {pdf_file.name if hasattr(pdf_file, 'name') else pdf_file}")
|
| 93 |
|
| 94 |
# 一時ディレクトリを作成
|
| 95 |
with tempfile.TemporaryDirectory() as temp_dir:
|
| 96 |
-
# Gradioから渡されるのはNamedStringオブジェクトなので、.nameでファイルパスを取得
|
| 97 |
temp_pdf_path = pdf_file.name
|
| 98 |
-
|
| 99 |
-
|
| 100 |
logging.info(f"Temporary PDF path: {temp_pdf_path}")
|
| 101 |
|
| 102 |
-
# PDFを分割
|
| 103 |
split_pdf_paths = split_pdf(temp_pdf_path, temp_dir)
|
| 104 |
logging.info(f"Split PDF paths: {split_pdf_paths}")
|
|
|
|
| 105 |
|
| 106 |
# 並列処理でOCR変換
|
| 107 |
markdown_results = []
|
| 108 |
with ThreadPoolExecutor() as executor:
|
| 109 |
-
|
| 110 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
|
|
|
|
|
|
|
| 112 |
# 結果を結合
|
| 113 |
combined_markdown = "\n\n".join(markdown_results)
|
|
|
|
|
|
|
| 114 |
|
| 115 |
return combined_markdown
|
| 116 |
|
|
|
|
| 117 |
# Gradioインターフェースの作成
|
| 118 |
def create_interface():
|
| 119 |
with gr.Blocks() as demo:
|
|
@@ -124,10 +134,22 @@ def create_interface():
|
|
| 124 |
pdf_input = gr.File(label="PDFファイルをアップロード", file_types=[".pdf"])
|
| 125 |
|
| 126 |
with gr.Row():
|
| 127 |
-
convert_btn = gr.Button("変換開始")
|
|
|
|
|
|
|
|
|
|
| 128 |
|
| 129 |
with gr.Row():
|
| 130 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
|
| 132 |
convert_btn.click(
|
| 133 |
fn=process_pdf,
|
|
@@ -135,6 +157,31 @@ def create_interface():
|
|
| 135 |
outputs=markdown_output
|
| 136 |
)
|
| 137 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 138 |
return demo
|
| 139 |
|
| 140 |
|
|
|
|
| 6 |
import tempfile
|
| 7 |
import base64
|
| 8 |
from concurrent.futures import ThreadPoolExecutor
|
| 9 |
+
import logging
|
| 10 |
+
import time # インポート追加
|
| 11 |
+
|
| 12 |
+
# ロギング設定
|
| 13 |
+
logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
|
| 14 |
|
| 15 |
# Gemini APIの設定 (環境変数から取得)
|
| 16 |
GOOGLE_API_KEY = os.environ.get("GOOGLE_API_KEY")
|
|
|
|
| 18 |
raise ValueError("環境変数 'GOOGLE_API_KEY' が設定されていません。")
|
| 19 |
genai.configure(api_key=GOOGLE_API_KEY)
|
| 20 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
|
| 22 |
def split_pdf(pdf_path, output_dir, pages_per_chunk=5):
|
| 23 |
"""PDFを指定ページ数ごとに分割する関数"""
|
|
|
|
| 59 |
pdf_base64 = encode_pdf_to_base64(pdf_path)
|
| 60 |
|
| 61 |
# Geminiモデルの設定
|
| 62 |
+
model = genai.GenerativeModel('gemini-1.5-pro') # または利用可能な他のモデル
|
| 63 |
|
| 64 |
# プロンプトの設定
|
| 65 |
prompt = """
|
|
|
|
| 82 |
# 結果を返す
|
| 83 |
return response.text
|
| 84 |
except Exception as e:
|
| 85 |
+
logging.error(f"Error during Gemini API call: {e}")
|
| 86 |
+
return f"エラーが発生しました: {e}"
|
| 87 |
|
| 88 |
|
| 89 |
+
def process_pdf(pdf_file, progress=gr.Progress()):
|
| 90 |
"""PDFファイルを処理するメイン関数"""
|
| 91 |
+
logging.info(f"Received file: {pdf_file.name if hasattr(pdf_file, 'name') else pdf_file}")
|
| 92 |
|
| 93 |
# 一時ディレクトリを作成
|
| 94 |
with tempfile.TemporaryDirectory() as temp_dir:
|
|
|
|
| 95 |
temp_pdf_path = pdf_file.name
|
|
|
|
|
|
|
| 96 |
logging.info(f"Temporary PDF path: {temp_pdf_path}")
|
| 97 |
|
| 98 |
+
# PDFを分割
|
| 99 |
split_pdf_paths = split_pdf(temp_pdf_path, temp_dir)
|
| 100 |
logging.info(f"Split PDF paths: {split_pdf_paths}")
|
| 101 |
+
progress(0.2, desc="PDFを分割中...") # 進捗更新
|
| 102 |
|
| 103 |
# 並列処理でOCR変換
|
| 104 |
markdown_results = []
|
| 105 |
with ThreadPoolExecutor() as executor:
|
| 106 |
+
# futureオブジェクトのリストを作成し、進捗を追跡
|
| 107 |
+
futures = [executor.submit(ocr_pdf_with_gemini, path) for path in split_pdf_paths]
|
| 108 |
+
for i, future in enumerate(futures):
|
| 109 |
+
try:
|
| 110 |
+
result = future.result()
|
| 111 |
+
markdown_results.append(result)
|
| 112 |
+
progress(0.2 + 0.6 * (i + 1) / len(futures), desc="OCR処理中...") # 進捗更新
|
| 113 |
+
except Exception as e:
|
| 114 |
+
logging.error(f"Error processing split PDF: {e}")
|
| 115 |
+
markdown_results.append(f"分割PDFの処理中にエラーが発生しました: {e}")
|
| 116 |
|
| 117 |
+
logging.info(f"Markdown results length: {len(markdown_results)}")
|
| 118 |
+
progress(0.8, desc="結果を結合中...")# 進捗更新
|
| 119 |
# 結果を結合
|
| 120 |
combined_markdown = "\n\n".join(markdown_results)
|
| 121 |
+
progress(1.0, desc="完了") # 進捗更新
|
| 122 |
+
time.sleep(0.5) #完了表示のため少し待つ
|
| 123 |
|
| 124 |
return combined_markdown
|
| 125 |
|
| 126 |
+
|
| 127 |
# Gradioインターフェースの作成
|
| 128 |
def create_interface():
|
| 129 |
with gr.Blocks() as demo:
|
|
|
|
| 134 |
pdf_input = gr.File(label="PDFファイルをアップロード", file_types=[".pdf"])
|
| 135 |
|
| 136 |
with gr.Row():
|
| 137 |
+
convert_btn = gr.Button("変換開始", variant="primary", elem_id="convert-button") # variantとelem_idを追加
|
| 138 |
+
|
| 139 |
+
with gr.Row():
|
| 140 |
+
markdown_output = gr.Textbox(label="変換結果", lines=10, max_lines=20) # MarkdownからTextboxに変更、行数を指定
|
| 141 |
|
| 142 |
with gr.Row():
|
| 143 |
+
copy_btn = gr.Button("クリップボードにコピー")
|
| 144 |
+
download_btn = gr.Button("ダウンロード")
|
| 145 |
+
|
| 146 |
+
# スタイル設定 (CSS)
|
| 147 |
+
demo.load(None, None, None, _js="""
|
| 148 |
+
() => {
|
| 149 |
+
document.getElementById('convert-button').style.backgroundColor = 'orange';
|
| 150 |
+
}
|
| 151 |
+
""")
|
| 152 |
+
|
| 153 |
|
| 154 |
convert_btn.click(
|
| 155 |
fn=process_pdf,
|
|
|
|
| 157 |
outputs=markdown_output
|
| 158 |
)
|
| 159 |
|
| 160 |
+
# クリップボードにコピー
|
| 161 |
+
copy_btn.click(
|
| 162 |
+
None,
|
| 163 |
+
markdown_output,
|
| 164 |
+
[],
|
| 165 |
+
js=f"(x) => {{ navigator.clipboard.writeText(x); }}",
|
| 166 |
+
)
|
| 167 |
+
|
| 168 |
+
# ダウンロード
|
| 169 |
+
download_btn.click(
|
| 170 |
+
None,
|
| 171 |
+
markdown_output,
|
| 172 |
+
[],
|
| 173 |
+
js=f"""(x) =>{{
|
| 174 |
+
const blob = new Blob([x], {{type: 'text/markdown;charset=utf-8'}});
|
| 175 |
+
const url = URL.createObjectURL(blob);
|
| 176 |
+
const a = document.createElement('a');
|
| 177 |
+
a.href = url;
|
| 178 |
+
a.download = 'converted.md';
|
| 179 |
+
document.body.appendChild(a);
|
| 180 |
+
a.click();
|
| 181 |
+
document.body.removeChild(a);
|
| 182 |
+
URL.revokeObjectURL(url);
|
| 183 |
+
}}"""
|
| 184 |
+
)
|
| 185 |
return demo
|
| 186 |
|
| 187 |
|