{"id":1588,"date":"2021-06-21T16:39:31","date_gmt":"2021-06-21T08:39:31","guid":{"rendered":"http:\/\/www.yatenglg.cn\/?p=1588"},"modified":"2022-10-18T16:34:39","modified_gmt":"2022-10-18T08:34:39","slug":"voc%e6%a0%bc%e5%bc%8f%e6%95%b0%e6%8d%ae%e9%9b%86xml%e6%96%87%e4%bb%b6%e6%93%8d%e4%bd%9c","status":"publish","type":"post","link":"http:\/\/www.yatenglg.cn\/blog\/?p=1588","title":{"rendered":"voc\u683c\u5f0f\u6570\u636e\u96c6xml\u6587\u4ef6\u64cd\u4f5c"},"content":{"rendered":"\n

1. \u4ecb\u7ecd<\/h2>\n\n\n\n

\u5305\u62ec\uff1a\u79fb\u52a8\uff0c\u590d\u5236\uff0c\u68c0\u67e5\u7a7a\u6587\u4ef6\u3001\u65e0\u76ee\u6807\u6587\u4ef6\uff0c\u66ff\u6362\u7c7b\u522b\u540d\u7b49\u3002\u53ef\u901a\u8fc7\u7ee7\u627fBaseOp\u7c7b\uff0c\u8f7b\u677e\u6269\u5c55\u5176\u4ed6\u529f\u80fd\u3002<\/p>\n\n\n\n

\"\"<\/figure><\/div>\n\n\n\n
\"\"<\/figure><\/div>\n\n\n\n
\"\"<\/figure><\/div>\n\n\n\n
\"\"<\/figure><\/div>\n\n\n\n
\"\"<\/figure><\/div>\n\n\n\n

2. \u4f8b\u5b50<\/h2>\n\n\n\n
# \u4f8b\u5b50-\u590d\u5236, \u5c06\/my_data\u4e0b\u6240\u6709\u7684xml\u6587\u4ef6\uff08\u5305\u62ec\u5b50\u6587\u4ef6\u5939\u4e0b\u7684\uff09\u590d\u5236\u5230 my_data_new\u76ee\u5f55\u4e0b\npython VocXMLOps.py --root \/home\/my_data --recursion --num_processes 5 copy --to_root \/home\/my_data_new\/\n\n# \u4f8b\u5b50-\u66ff\u6362\u7c7b\u522b\u540d, \u5c06\/my_data\u5f53\u524d\u76ee\u5f55\u4e0b\u7684xml\u6587\u4ef6\u4e2d\u7684 'Person'\u7c7b\u522b\u540d\u66ff\u6362\u4e3a'\u4eba'\npython VocXMLOps.py --root \/home\/my_data replace --old_name Person --new_name \u4eba<\/code><\/pre>\n\n\n\n
3. \u4ee3\u7801<\/h2>\n\n\n\n
import xml.etree.ElementTree as ET\nimport os\nimport shutil\nfrom multiprocessing import Pool\n\n\nclass BaseOp(object):\n    def __init__(self):\n        """\n        \u64cd\u4f5c\u7c7b\u57fa\u7c7b\n        """\n        pass\n\n    def run(self, file_path):\n        raise NotImplementedError\n\n\nclass IsInvaild(BaseOp):\n    def __init__(self, remove=False):\n        """\n        \u67e5\u627e\u65e0\u6548\u6216\u65e0\u76ee\u6807\u7684xml\u6587\u4ef6\n        """\n        super(IsInvaild, self).__init__()\n        self.remove = remove\n\n    def run(self, file_path):\n        try:\n            tree = ET.parse(file_path)\n            objs = tree.findall('object')\n            if len(objs) < 1:\n                if self.remove:\n                    os.remove(file_path)\n                    print('Found and remove no object xml file : {}'.format(file_path))\n                else:\n                    print('Found no object xml file : {}'.format(file_path))\n\n        except:\n            if self.remove:\n                try:\n                    os.remove(file_path)\n                    print('Found and remove invaild xml file : {}'.format(file_path))\n                except:\n                    print('Found invaild xml file : {}, but remove failed.'.format(file_path))\n            else:\n                print('Found invaild xml file : {}'.format(file_path))\n\n\nclass DelObject(BaseOp):\n    def __init__(self, del_object_names:list):\n        super(DelObject, self).__init__()\n        self.del_object_names = del_object_names\n\n    def run(self, file_path):\n        try:\n            tree = ET.parse(file_path)\n            root = tree.getroot()\n            objs = tree.findall('object')\n            for obj in objs:\n                name = obj.find('name').text\n                if name in self.del_object_names:\n                    root.remove(obj)\n\n            tree.write(file_path)\n            print("Del objs: {} from {}".format(self.del_object_names, file_path))\n        except:\n            pass\n\n\nclass MoveOp(BaseOp):\n    def __init__(self, to_root):\n        """\n        \u79fb\u52a8\u64cd\u4f5c\n        :param to_root:\n        """\n        super(MoveOp, self).__init__()\n        self.to_root = to_root\n\n    def run(self, file_path):\n        try:\n            to_path = os.path.join(self.to_root, os.path.split(file_path)[-1])\n            shutil.move(file_path, os.path.join(self.to_root, os.path.split(file_path)[-1]))\n            print("Move {} to {}".format(file_path, to_path))\n        except:\n            pass\n\n\nclass CopyOp(BaseOp):\n    def __init__(self, to_root):\n        """\n        \u590d\u5236\u64cd\u4f5c\n        :param to_root:\n        """\n        super(CopyOp, self).__init__()\n        self.to_root = to_root\n\n    def run(self, file_path):\n        try:\n            to_path = os.path.join(self.to_root, os.path.split(file_path)[-1])\n            shutil.copy(file_path, to_path)\n            print("Copy {} to {}".format(file_path, to_path))\n        except:\n            pass\n\n\nclass ReplaceNameOp(BaseOp):\n    def __init__(self, old_name, new_name):\n        """\n        xml\u7c7b\u522b\u540d\u79f0\u91cd\u547d\u540d\u64cd\u4f5c\n        :param old_name:\n        :param new_name:\n        """\n        super(ReplaceNameOp, self).__init__()\n        self.old_name = old_name\n        self.new_name = new_name\n\n    def run(self, file_path):\n        try:\n            tree = ET.parse(file_path)\n            objs = tree.findall('object')\n            for obj in objs:\n                name = obj.find('name').text\n                if name == self.old_name:\n                    obj.find('name').text = self.new_name\n            tree.write(file_path)\n            print("Replace {}".format(file_path))\n        except:\n            pass\n\n\nclass VOCParser(object):\n    def __init__(self, root:str, recursion=False, num_processes=10):\n        """\n\n        :param root:        \u6839\u76ee\u5f55\n        :param recursion:   \u662f\u5426\u9012\u5f52\u5904\u7406\u6240\u6709\u5b50\u6587\u4ef6\u5939\n        """\n        self.root = root\n        self.ops = []\n        self.xmls = []\n        self.jpgs = []\n\n        self.pool = Pool(processes=num_processes)\n\n        self.find_xml(self.root, recursion)\n\n    def register(self, op:BaseOp):\n        self.ops.append(op)\n\n    def find_xml(self, root, recursion):\n        fs = os.listdir(root)\n        for f in fs:\n            f = os.path.join(root, f)\n            if os.path.isdir(f) and recursion:\n                self.find_xml(f, recursion)\n            if os.path.isfile(f):\n                if f.endswith('.xml'):\n                    self.xmls.append(f)\n                else:\n                    pass\n\n    def run(self):\n        for xml in self.xmls:\n            for op in self.ops:\n                self.pool.apply_async(op.run, (xml,))\n\n        self.pool.close()\n        self.pool.join()\n\n\nif __name__ == '__main__':\n    import argparse\n\n    parse = argparse.ArgumentParser(description='VOC\u683c\u5f0f\u6570\u636e\u96c6,xml\u6587\u4ef6\u64cd\u4f5c.')\n    parse.add_argument('--root', required=True, help='\u6587\u4ef6\u6839\u76ee\u5f55.')\n    parse.add_argument('--recursion', action='store_true', help='\u5426\u5b9a\u9012\u5f52\u5904\u7406\u6240\u6709\u5b50\u6587\u4ef6\u5939\u4e2d\u7684xml\u6587\u4ef6.')\n    parse.add_argument('--num_processes', default=10, type=int, help='\u8fdb\u7a0b\u6570.')\n\n    subparser = parse.add_subparsers(dest='subparser_name', description='\u5b50\u547d\u4ee4')\n    invaild = subparser.add_parser('invaild', help='\u67e5\u627e\u65e0\u6548\u6216\u65e0\u76ee\u6807\u7684xml\u6587\u4ef6', description='\u67e5\u627e\u65e0\u6548\u6216\u65e0\u76ee\u6807\u7684xml\u6587\u4ef6')\n    invaild.add_argument('--remove', action='store_true', help='\u79fb\u9664\u65e0\u6548\u6587\u4ef6')\n\n    replace = subparser.add_parser('replace', help='\u66ff\u6362xml\u7c7b\u522b\u540d', description='\u66ff\u6362xml\u7c7b\u522b\u540d')\n    replace.add_argument('--old_name', required=True, help='\u65e7\u7c7b\u522b\u540d')\n    replace.add_argument('--new_name', required=True, help='\u65b0\u7c7b\u522b\u540d')\n\n    delobjects = subparser.add_parser('delobjects', help='\u5220\u9664xml\u7c7b\u522b.\u53ef\u540c\u65f6\u5220\u9664\u591a\u4e2a\u7c7b\u522b\uff0c\u591a\u4e2a\u7c7b\u522b\u95f4\u7528\u7a7a\u683c\u9694\u5f00', description='\u5220\u9664xml\u7c7b\u522b')\n    delobjects.add_argument('--del_objects', nargs='+', help='\u5220\u9664\u7684\u7c7b\u522b\u540d')\n\n    move = subparser.add_parser('move', help='\u79fb\u52a8xml\u6587\u4ef6', description='\u79fb\u52a8xml\u6587\u4ef6')\n    move.add_argument('--to_root', required=True, help='\u76ee\u6807\u6587\u4ef6\u5939.')\n\n    copy = subparser.add_parser('copy', help='\u590d\u5236xml\u6587\u4ef6', description='\u590d\u5236xml\u6587\u4ef6')\n    copy.add_argument('--to_root', required=True, help='\u76ee\u6807\u6587\u4ef6\u5939.')\n\n    args = parse.parse_args()\n\n    vocparser = VOCParser(args.root, args.recursion, args.num_processes)\n\n    if args.subparser_name == 'invaild':\n        vocparser.register(IsInvaild(args.remove))\n\n    elif args.subparser_name == 'replace':\n        vocparser.register(ReplaceNameOp(args.old_name, args.new_name))\n\n    elif args.subparser_name == 'move':\n        vocparser.register(MoveOp(args.to_root))\n\n    elif args.subparser_name == 'copy':\n        vocparser.register(CopyOp(args.to_root))\n\n    elif args.subparser_name == 'delobjects':\n        vocparser.register(DelObject(args.del_objects))\n\n    vocparser.run()<\/code><\/pre><\/div>\n\n\n\n
<\/p>\n","protected":false},"excerpt":{"rendered":"
VOC\u683c\u5f0f\u6570\u636e\uff0cxml\u6587\u4ef6\u64cd\u4f5c\u96c6\u5408\u3002<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":[],"categories":[8,5],"tags":[],"_links":{"self":[{"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=\/wp\/v2\/posts\/1588"}],"collection":[{"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=1588"}],"version-history":[{"count":47,"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=\/wp\/v2\/posts\/1588\/revisions"}],"predecessor-version":[{"id":1642,"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=\/wp\/v2\/posts\/1588\/revisions\/1642"}],"wp:attachment":[{"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=1588"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=1588"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.yatenglg.cn\/blog\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=1588"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}