Update app.py
Browse files
app.py
CHANGED
|
@@ -15,82 +15,11 @@ co_rerank = cohere.ClientV2(os.environ.get("COHERE_API_KEY"))
|
|
| 15 |
|
| 16 |
vectored = None
|
| 17 |
|
| 18 |
-
dataid = requests.get("https://www.ryhintl.com/dbjson/getjson?sqlcmd=select * from company_matters")
|
| 19 |
-
|
| 20 |
-
data_str = dataid.content.decode('utf-8')
|
| 21 |
-
data = json.loads(data_str)
|
| 22 |
-
|
| 23 |
-
'''def decode_text(data):
|
| 24 |
-
for item in data:
|
| 25 |
-
item['text'] = item['text'].encode('latin1').decode('utf-8') # latin1でエンコード、UTF-8でデコード
|
| 26 |
-
return data'''
|
| 27 |
-
|
| 28 |
-
# デコード関数
|
| 29 |
-
def decode_text(data):
|
| 30 |
-
for item in data:
|
| 31 |
-
try:
|
| 32 |
-
# latin1 でデコードし、utf-8に変換
|
| 33 |
-
item['text'] = item['text'].encode('latin1').decode('utf-8')
|
| 34 |
-
except UnicodeDecodeError as e:
|
| 35 |
-
print(f"エラー: {e}")
|
| 36 |
-
item['text'] = "[デコード失敗]"
|
| 37 |
-
return data
|
| 38 |
-
|
| 39 |
-
'''def upload_html_to_server(file_path, url):
|
| 40 |
-
with open(file_path, "rb") as file:
|
| 41 |
-
# ファイルをアップロードする
|
| 42 |
-
response = requests.post(url, files={"file": file})
|
| 43 |
-
return response.status_code, response.text
|
| 44 |
-
|
| 45 |
-
def upload_text_to_server(text, url):
|
| 46 |
-
# テキストを送信
|
| 47 |
-
response = requests.post(url, data={"content": text})
|
| 48 |
-
return response.status_code, response.text
|
| 49 |
-
|
| 50 |
-
# 自然言語にする関数
|
| 51 |
-
def json_to_text(data):
|
| 52 |
-
final_context = ""
|
| 53 |
-
for item in data:
|
| 54 |
-
text = (
|
| 55 |
-
f"タイトル: {item['caption']}\n"
|
| 56 |
-
f"表題: {item['title']}\n"
|
| 57 |
-
f"詳細: {item['content']}\n"
|
| 58 |
-
f"日付: {item['date']}\n"
|
| 59 |
-
)
|
| 60 |
-
final_context += text + "\n" # 結果を連結
|
| 61 |
-
|
| 62 |
-
#last_content = "<html>\n<head>\n<title>会議議事録</title>\n</head>\n<body>\n"+final_context+"</body>\n</html>"
|
| 63 |
-
last_content = final_context
|
| 64 |
-
|
| 65 |
-
file_name = "output_cm.html"
|
| 66 |
-
with open(file_name, "w", encoding="utf-8") as file:
|
| 67 |
-
file.write(last_content)
|
| 68 |
-
|
| 69 |
-
# サーバーURLを指定
|
| 70 |
-
server_url = "https://www.ryhintl.com/company_matters.php" # 実際のアップロード先のURLに置き換えてね
|
| 71 |
-
|
| 72 |
-
# アップロード実行
|
| 73 |
-
status, response_text = upload_text_to_server(last_content, server_url)
|
| 74 |
-
print(f"アップロードのステータス: {status}")
|
| 75 |
-
print(f"レスポンス: {response_text}")
|
| 76 |
-
|
| 77 |
-
return final_context
|
| 78 |
-
|
| 79 |
-
# 実行例
|
| 80 |
-
result = json_to_text(data)
|
| 81 |
-
|
| 82 |
-
'''
|
| 83 |
-
|
| 84 |
raw_documents = [
|
| 85 |
-
{"title": "バグダードの戦い", "url": "https://ja.wikipedia.org/wiki/バグダードの戦い"},
|
| 86 |
{"title": "2006年トリノオリンピック", "url": "https://ja.wikipedia.org/wiki/2006年トリノオリンピック"},
|
| 87 |
-
{"title": "
|
| 88 |
-
{"title": "えひめ丸事故", "url": "https://ja.wikipedia.org/wiki/えひめ丸事故"},
|
| 89 |
-
{"title": "会議議事録", "url": "https://www.ryhintl.com/reqfiles/company_matters/cm_output.html"}
|
| 90 |
]
|
| 91 |
|
| 92 |
-
#raw_documents = [{"title": "会議議事録", "url": "https://www.ryhintl.com/reqfiles/company_matters/cm_output.html"}]
|
| 93 |
-
|
| 94 |
# あなたのクラスとロジックをここに統合します
|
| 95 |
class Vectorstore:
|
| 96 |
|
|
@@ -207,8 +136,6 @@ vectorstore = Vectorstore(raw_documents)
|
|
| 207 |
# Gradioの関数
|
| 208 |
def search(query):
|
| 209 |
results = vectorstore.retrieve(query)
|
| 210 |
-
decoded_data = decode_text(results)
|
| 211 |
-
print("decoded_data:",decoded_data)
|
| 212 |
return "\n\n".join([f"**Title**: {r['title']}\n**Text**: {r['text']}\n**URL**: {r['url']}" for r in decoded_data])
|
| 213 |
|
| 214 |
# Gradioインターフェース
|
|
|
|
| 15 |
|
| 16 |
vectored = None
|
| 17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
raw_documents = [
|
|
|
|
| 19 |
{"title": "2006年トリノオリンピック", "url": "https://ja.wikipedia.org/wiki/2006年トリノオリンピック"},
|
| 20 |
+
{"title": "生成AI", "url": "https://ja.wikipedia.org/wiki/生成的人工知能"}
|
|
|
|
|
|
|
| 21 |
]
|
| 22 |
|
|
|
|
|
|
|
| 23 |
# あなたのクラスとロジックをここに統合します
|
| 24 |
class Vectorstore:
|
| 25 |
|
|
|
|
| 136 |
# Gradioの関数
|
| 137 |
def search(query):
|
| 138 |
results = vectorstore.retrieve(query)
|
|
|
|
|
|
|
| 139 |
return "\n\n".join([f"**Title**: {r['title']}\n**Text**: {r['text']}\n**URL**: {r['url']}" for r in decoded_data])
|
| 140 |
|
| 141 |
# Gradioインターフェース
|