{"repo": "centyuan/pydemo", "files": {"PyExample/OcrGather/clients/xfyun_ocr.py": {"b8aa48d1868c924389f65310169c2cae4fbce2ff": {"t": 1785674330.869733, "n": 4327, "h": "16bf86525886f24f", "c": "from datetime import datetime\nfrom wsgiref.handlers import format_date_time\nfrom time import mktime\nimport hashlib\nimport base64\nimport hmac\nfrom urllib.parse import urlencode\nimport json\nimport requests\n\n\nclass AssembleHeaderException(Exception):\n    def __init__(self, msg):\n        self.message = msg\n\n\ndef sha256base64(data):\n    sha256 = hashlib.sha256()\n    sha256.update(data)\n    digest = base64.b64encode(sha256.digest()).decode(encoding='utf-8')\n    return digest\n\n\ndef parse_url(requset_url):\n    \"\"\"\n    解析url\n    :param requset_url:\n    :return:\n    \"\"\"\n    stidx = requset_url.index(\"://\")\n    host = requset_url[stidx + 3:]\n    schema = requset_url[:stidx + 3]\n    edidx = host.index(\"/\")\n    if edidx <= 0:\n        raise AssembleHeaderException(\"invalid request url:\" + requset_url)\n    path = host[edidx:]\n    host = host[:edidx]\n    return host, path, schema\n\n\ndef assemble_ws_auth_url(requset_url, method=\"POST\", api_key=\"\", api_secret=\"\"):\n    \"\"\"\n    鉴权 并返回request url\n    :param requset_url:\n    :param method:\n    :param api_key:\n    :param api_secret:\n    :return:\n    \"\"\"\n    host, path, schema = parse_url(requset_url)\n    date = format_date_time(mktime(datetime.now().timetuple()))\n    # date = \"Thu, 12 Dec 2019 01:57:27 GMT\"\n    signature_origin = \"host: {}\\ndate: {}\\n{} {} HTTP/1.1\".format(host, date, method, path)\n    \"\"\"\n    signature原始字段(signature_origin)则为：\n        host: api.xf-yun.com\n        date: Wed, 11 Aug 2021 06:55:18 GMT\n        POST /v1/private/sf8e6aca1 HTTP/1.1\n    \"\"\"\n    # 生成摘要\n    signature_sha = hmac.new(api_secret.encode('utf-8'), signature_origin.encode('utf-8'),\n                             digestmod=hashlib.sha256).digest()\n    # 对摘要编码\n    signature_sha = base64.b64encode(signature_sha).decode(encoding='utf-8')\n    authorization_origin = \"api_key=\\\"%s\\\", algorithm=\\\"%s\\\", headers=\\\"%s\\\", signature=\\\"%s\\\"\" % (\n        api_key, \"hmac-sha256\", \"host date request-line\", signature_sha)\n    authorization = base64.b64encode(authorization_origin.encode('utf-8')).decode(encoding='utf-8')\n    values = {\n        \"host\": host,\n        \"date\": date,\n        \"authorization\": authorization\n    }\n    return requset_url + \"?\" + urlencode(values)\n\n\ndef XFyunApi(app_id, api_key, secret_key, file_path=None, file_data=None):\n    \"\"\"\n    :param app_id:\n    :param api_key:\n    :param secret_key:\n    :param file_path:\n    :param file_data:\n    :return:\n    \"\"\"\n    APPId = app_id  # 控制台获取\n    APISecret = secret_key  # 控制台获取\n    APIKey = api_key  # 控制台获取\n    imageBytes = None\n    if file_path:\n        with open(file_path, \"rb\") as f:\n            imageBytes = f.read()\n    if imageBytes:\n        url = 'https://api.xf-yun.com/v1/private/sf8e6aca1'\n        body = {\n            \"header\": {\n                \"app_id\": APPId,\n                \"status\": 3\n            },\n            \"parameter\": {\n                \"sf8e6aca1\": {\n                    \"category\": \"ch_en_public_cloud\",\n                    \"result\": {\n                        \"encoding\": \"utf8\",\n                        \"compress\": \"raw\",\n                        \"format\": \"json\"\n                    }\n                }\n            },\n            \"payload\": {\n                \"sf8e6aca1_data_1\": {\n                    \"encoding\": \"jpg\",\n                    \"image\": str(base64.b64encode(imageBytes), 'UTF-8'),\n                    \"status\": 3\n                }\n            }\n        }\n        request_url = assemble_ws_auth_url(url, \"POST\", APIKey, APISecret)\n\n        headers = {'content-type': \"application/json\", 'host': 'api.xf-yun.com', 'app_id': APPId}\n        response = requests.post(request_url, data=json.dumps(body), headers=headers)\n        tempResult = json.loads(response.content.decode())\n\n        finalResult = base64.b64decode(tempResult['payload']['result']['text']).decode()\n        finalResult = finalResult.replace(\" \", \"\").replace(\"\\n\", \"\").replace(\"\\t\", \"\").strip()\n        text = ''\n        for item in eval(finalResult)['pages'][0]['lines'][0]['words']:\n            text += item['content']\n        return text\n\n\nif __name__ == '__main__':\n    app_id = \"eb78ba1a\"  # 控制台获取\n    secret_key = \"ZmM3Mjc4NjcyNjcyYmJiNGZmNmU5NmNh\"  # 控制台获取\n    api_key = \"618642319dd26346780f676e9f9dfad9\"  # 控制台获取\n    text = XFyunApi(app_id, api_key, secret_key, file_path='../resource/ocr_test.png')\n    print(text)\n"}}}}