Python：跨数千个Dictionary/XMLs/JSON比较和计算Dictionary结构

Question

我将成千上万的XML文件解析成字典，并将其结构存储在JSON中。你知道吗

它们的结构基本相同，但不同的标记命名方案的数目不详。在这数千个文件中，有各种不同的缩写用于命名标记。你知道吗

我需要找出有多少不同的标签来描述每一条信息，正确地解析它们。你知道吗

为此，我想创建一个XMLs/字典的主字典，其中包括标记名的所有变体，最好是它们在数千个XMLs/字典中的计数。你知道吗

以下是其中一本词典的小样本：

{
    "Header": {
        "Ts": {},
        "PeriodEndDt": {},
        "PreparedBy": {
            "PreparerID": {},
            "PreparerFirmName": {
                "BusinessNameLine1Txt": {}
            },
            "PreparerAddress": {
                "AddLn1Txt": {},
                "CityName": {},
                "StateAbbreviationCd": {},
                "ZIPCd": {}
            }
        },
        "FormTypeCd": {},
        "PeriodBeginDt": {},
        "Filer": {
            "UniqueID": {},
            "BusinessName": {
                "BusinessNameLine1Txt": {}
            },
            "BusinessNameControlTxt": {},
            "PhoneNum": {},
            "USAddress": {
                "AddressLine1Txt": {},
                "CityNm": {},
                "StateAbbreviationCd": {},
                "ZIPCd": {}
            }
        },

        "FormData": {
            "FormCodeType": {
                "BizType": {},
                "AssetsAtEOY": {},
                "AccountingMethod": {},
                "RevenueAndExpenses": {
                    "ScheduleBNotReqd": {},
                    "DivsRevAndExpenses": {},
                    "DivsNetInvstIncomeAmt": {},
                    "NetGainSaleAstRevAndExpnssAmt": {},
                    "RevsOvrExpenses": {},
                    "NetInvestmentIncomeAmt": {}
                },
                "BalanceSheetGroup": {
                    "CashInvstBOYAmt": {},
                    "CashInvstEOYAmt": {},
                    "CashInvstEOYFMVAmt": {},
                    "OtherInvestmentsBOYAmt": {},
                    "OtherInvestmentsEOYAmt": {},
                    "CapitalStockEOYAmt": {},
                    "TotalLiabilitiesNetAstEOYAmt": {}
                },
                "ChangeNetAssetsFundGroup": {
                    "NetAssettFundBalancesBOYAmt": {},
                    "ExcessRevExpensesAmt": {},
                    "OtherIncreasesAmt": {},
                    "SubtotalAmt": {},
                    "OtherDecreasesAmt": {},
                    "TotNetAstOrFundBalancesEOYAmt": {}
                },
                "CapGainsLossTxInvstIncmDetail": {
                    "CapGainsLossTxInvstIncmGrp": {
                        "PropertyDesc": {},
                        "HowAcquiredCd": {},
                        "GrossSalesPriceAmt": {},
                        "GainOrLossAmt": {},
                        "GainsMinusExcessOrLossesAmt": {}
                    },
                    "StatementsRegardingActyGrp": {
                        "LegislativePoliticalActyInd": {},
                        "MoreThan100SpentInd": {}
                    },
                    "PhoneNum": {},
                    "LocationOfBooksUSAddress": {
                        "AddressLine1Txt": {},
                        "CityNm": {},
                        "StateAbbreviationCd": {},
                        "ZIPCd": {}
                    },
                    "CorporateDirectorsGrp": {
                        "DirectorsGrp": {
                            "PersonNm": {},
                            "USAddress": {
                                "AddressLine1Txt": {},
                                "CityNm": {},
                                "StateAbbreviationCd": {},
                                "ZIPCd": {}
                            },
                            "EmpPrograms": {
                                "EmployeeBenefitGroupNum": {},
                                "GroupType": {
                                    "GroupElement": {},
                                    "GroupCharacter": {
                                        "GroupNames": {}
                                    }
                                }

                            },
                            "EmpOffice1": {},
                            "EmpOffice2": {},
                            "EmpOffice3": {},
                            "EmpOffice4": {}
                        }


                    }
                }
            }
        }
    }
}

首先，我用来创建dictionaries/JSON的代码如下：

import xml.etree.ElementTree as ET

strip_ns = lambda xx: str(xx).split('}', 1)[1]
tree = ET.parse('xmlpath.xml')
root = tree.getroot()


tierdict = {}
for tier1 in root:
    tier1var = strip_ns(tier1.tag)
    tierdict[tier1var] = {}
    for tier2 in tier1:
        tier2var = strip_ns(tier2.tag)
        tierdict[tier1var][tier2var] = {}
        for tier3 in tier2:
            tier3var = strip_ns(tier3.tag)
            tierdict[tier1var][tier2var][tier3var] = {}
            for tier4 in tier3:
                tier4var = strip_ns(tier4.tag)
                tierdict[tier1var][tier2var][tier3var][tier4var] = {}

我想看到的输出是：

{
    "Header": {
        "Header.Count": 5672,
        "Ts": {
            "Ts.Count": 3365
            },
        "Ss": {
            "Ss.Count": 2328
            },

Answer 1

我可能会对您想要的元素进行递归搜索，定义如下：

def get_elements(json_entry, child_elements=[]):

     if not child_elements:
         return json_entry

     el, other_children = child_elements[0], child_elements[1:]

     children = el.getchildren()
     rec = json_entry.get(el.tag)
     if not children:
         json_entry[el.tag] = {"Count": rec.get("Count",0)+1 if rec else 1}

     else:
         json_entry[el.tag] = {"Count": rec.get("Count",0) if rec else 1,
                                    **get_elements({}, children)}

     return get_elements(json_entry, other_children)

这样，您只需传递xml的根元素：

from lxml import etree

with open("myxml.xml", "r") as fh:
    tree = etree.parse(fh)

root = tree.getroot()

root_children = root.getchildren()

child_recs = get_elements({}, root_children)

{'tagOne': {'Count': 1}, 'tagTwo': {'Count': 1, 'tagThree': {'Count': 1}, 'tagFour': {'Count': 1, 'tagFive': {'Count': 1}}}}

如果要将根元素环绕在它周围，请按如下方式操作：

master_lookup = {root.tag: {"Count": 1, **child_recs}}

这可以很容易地扩展到许多文件的for循环

master_lookup = {}

for file in os.walk(path):
    with open(file) as fh:
        tree = etree.parse(fh)

    root = tree.getroot()
    root_entry = master_lookup.get(root.tag, {"Count": 0})
    root_children = root.getchildren()

    root_count = root_entry.pop("Count")

    master_lookup[root.tag] = {"Count": root_count, **get_elements({**root_entry}, root_children)}

有这样的意思吗