python读取xml数据库中表内所有数据，获取数据库中所有表的字段名称

工作中需要读取指定xml数据库中的数据以及 表所需字段名，所以在已有例子中改进实现：

xml 数据库 xmldabase.xml：

<database>
	<manifest>
		<pair key="schema_major_vsn" value="5"/>
		<pair key="schema_minor_vsn" value="63"/>
		<pair key="generation_count" value="8541968"/>
	</manifest>
	<table name="PCI">
		<row ref="OpaqueRef:1e2c98ee-a6fe-b1d5-e91c-d27911deb21d" __ctime="8489729" __mtime="8489775" _ref="OpaqueRef:1e2c98ee-a6fe-b1d5-e91c-d27911deb21d" attached_VMs="()" class_id="03" class_name="Display controller" dependencies="()" device_id="2000" device_name="ASPEED Graphics Family" functions="1" host="OpaqueRef:d92d7211-482f-c68f-f89b-4c8b22faec1a" other_config="()" pci_id="0000:09:00.0" uuid="5907d30d-7d53-4682-4ff1-ed32974cb3b6" vendor_id="1a03" vendor_name="ASPEED Technology, Inc."/>
	</table>
<table name="xx_metrics">
		<row ref="OpaqueRef:5105a1b6-a689-15a9-708d-57417c69d4b4" __ctime="8489729" __mtime="8490755" _ref="OpaqueRef:5105a1b6-a689-15a9-708d-57417c69d4b4" carrier="true" device_id="10d3" device_name="82574L Gigabit Network Connection" duplex="true" io__read_kbs="0." io__write_kbs="0." last_updated="20130508T00:02:46Z" other_config="()" pci_bus_path="0000:06:00.0" speed="1000" uuid="a26a8b42-b4c1-1b3c-1236-ea62d2d05236" vendor_id="8086" vendor_name="Intel Corporation"/>
		<row ref="OpaqueRef:63f81794-55b1-af60-5225-900673b300d2" __ctime="8489729" __mtime="8490752" _ref="OpaqueRef:63f81794-55b1-af60-5225-900673b300d2" carrier="false" device_id="10d3" device_name="82574L Gigabit Network Connection" duplex="false" io__read_kbs="0." io__write_kbs="0." last_updated="20130508T00:02:46Z" other_config="()" pci_bus_path="0000:07:00.0" speed="65535" uuid="74bc3019-9c04-ca9d-6ea8-3be63840620d" vendor_id="8086" vendor_name="Intel Corporation"/>
		<row ref="OpaqueRef:a0e992d8-5293-581a-0ad1-2b9a00c53d2a" __ctime="8489729" __mtime="8490754" _ref="OpaqueRef:a0e992d8-5293-581a-0ad1-2b9a00c53d2a" carrier="true" device_id="10d3" device_name="82574L Gigabit Network Connection" duplex="true" io__read_kbs="0." io__write_kbs="0." last_updated="20130508T00:02:46Z" other_config="()" pci_bus_path="0000:06:00.0" speed="1000" uuid="fc9d0a76-eb60-dfc8-18b6-bfe26fe8be9d" vendor_id="8086" vendor_name="Intel Corporation"/>
		<row ref="OpaqueRef:f9374ed7-316b-0f53-92d1-f7f70823f6ce" __ctime="8489729" __mtime="8490753" _ref="OpaqueRef:f9374ed7-316b-0f53-92d1-f7f70823f6ce" carrier="true" device_id="10d3" device_name="82574L Gigabit Network Connection" duplex="true" io__read_kbs="0." io__write_kbs="0." last_updated="20130508T00:02:46Z" other_config="()" pci_bus_path="0000:06:00.0" speed="1000" uuid="e4f38332-1317-b9df-07f9-ab5681569c91" vendor_id="8086" vendor_name="Intel Corporation"/>
	</table>
</database>

获取指定表的数据：

from xml.etree import ElementTree


def get_table_records( xmlfile = None,text = None,table_name=None):
    root=None
    if xmlfile:
        root = ElementTree.parse(xmlfile)
    elif text:
        root = ElementTree.fromstring(text)
    
    node_findall = root.findall("table")
    for e in node_findall:
       
        for key,value in e.items():
            if value==table_name:
                print( "%s:%s" % (key, value)   )
                ar=e.findall('row')
                for row in ar:
                    for k,v in row.items():
                        
                        print( "%s:%s" % (k, v)   )

获取所有表的字段名称：

def get_table_field(text =None, xmlfile =None):
    root=None
    if xmlfile:
        root = ElementTree.parse(xmlfile)
    elif text:
        root = ElementTree.fromstring(text)
    
    node_findall = root.findall("table")
    for e in node_findall:
        list_field=[]    
        for key,value in e.items():
#                print( "%s:%s" % (key, value)   )
                first_row=e.find('row')
#                print(type(first_row))
                if first_row==None :
                    print('%s_need_fields=%s'%(value,list_field))
#                    continue
                else:
                    for k,v in first_row.items():
                        list_field.append(k)
    #                        print( "%s:%s" % (k, v)   )    
                    print('%s_need_fields=%s'%(value,list_field))

测试用例：

if __name__ == '__main__':
    

#    get_table_records(text=open("xmldabase.xml").read(),table_name='PCI')  #获取xml的方式不同，皆可用 
    get_table_records(xmlfile="xmldabase.xml",table_name='PCI') 
    get_table_field(xmlfile="xmldabase.xml") 

输出结果：

name:PCI
class_name:Display controller
vendor_id:1a03
__mtime:8489775
host:OpaqueRef:d92d7211-482f-c68f-f89b-4c8b22faec1a
dependencies:()
_ref:OpaqueRef:1e2c98ee-a6fe-b1d5-e91c-d27911deb21d
device_id:2000
functions:1
uuid:5907d30d-7d53-4682-4ff1-ed32974cb3b6
class_id:03
other_config:()
__ctime:8489729
pci_id:0000:09:00.0
device_name:ASPEED Graphics Family
attached_VMs:()
vendor_name:ASPEED Technology, Inc.
ref:OpaqueRef:1e2c98ee-a6fe-b1d5-e91c-d27911deb21d


PCI_need_fields=['class_name', 'vendor_id', '__mtime', 'host', 'dependencies', '_ref', 'device_id', 'functions', 'uuid', 'class_id', 'other_config', '__ctime', 'pci_id', 'device_name', 'attached_VMs', 'vendor_name', 'ref']
xx_metrics_need_fields=['last_updated', 'vendor_id', '__mtime', '_ref', 'device_name', 'speed', 'device_id', 'uuid', 'duplex', 'other_config', '__ctime', 'pci_bus_path', 'io__write_kbs', 'carrier', 'io__read_kbs', 'vendor_name', 'ref']