我有一些需要转换为json的csv文件。 csv中的某些float值是数字字符串（以保持尾随零）。转换为json时，所有键和值都用双引号引起来。我需要数字字符串浮点值不带引号，但保持尾随零。
以下是输入的csv文件的示例：
ACCOUNTNAMEDENORM,DELINQUENCYSTATUS,RETIRED,INVOICEDAYOFWEEK,ID,BEANVERSION,ACCOUNTTYPE,ORGANIZATIONTYPEDENORM,HIDDENTACCOUNTCONTAINERID,NEWPOLICYPAYMENTDISTRIBUTABLE,ACCOUNTNUMBER,PAYMENTMETHOD,INVOICEDELIVERYTYPE,DISTRIBUTIONLIMITTYPE,CLOSEDATE,FIRSTTWICEPERMTHINVOICEDOM,HELDFORINVOICESENDING,FEINDENORM,COLLECTING,ACCOUNTNUMBERDENORM,CHARGEHELD,PUBLICID
John Smith,2.0000000000,0.0000000000,5.0000000000,1234567.0000000000,69.0000000000,1.0000000000,,4321987.0000000000,1,000-000-000-00,10012.0000000000,10002.0000000000,3.0000000000,,1.0000000000,0,,0,000-000-000-00,0,bc:1234346
我得到的json输出是：
{"ACCOUNTNAMEDENORM":"John Smith","DELINQUENCYSTATUS":"2.0000000000","RETIRED":"0.0000000000","INVOICEDAYOFWEEK":"5.0000000000","ID":"1234567.0000000000","BEANVERSION":"69.0000000000","ACCOUNTTYPE":"1.0000000000","ORGANIZATIONTYPEDENORM":null,"HIDDENTACCOUNTCONTAINERID":"4321987.0000000000","NEWPOLICYPAYMENTDISTRIBUTABLE":"1","ACCOUNTNUMBER":"000-000-000-00","PAYMENTMETHOD":"12345.0000000000","INVOICEDELIVERYTYPE":"98765.0000000000","DISTRIBUTIONLIMITTYPE":"3.0000000000","CLOSEDATE":null,"FIRSTTWICEPERMTHINVOICEDOM":"1.0000000000","HELDFORINVOICESENDING":"0","FEINDENORM":null,"COLLECTING":"0","ACCOUNTNUMBERDENORM":"000-000-000-00","CHARGEHELD":"0","PUBLICID":"xx:1234346"}
这是我正在使用的代码：
import csv
import json
csvfile = open('output2.csv', 'r')
jsonfile = open('output2.json', 'w')
readHeaders = csv.reader(csvfile)
fieldnames = next(readHeaders)
reader = csv.DictReader(csvfile, fieldnames)
for row in reader:
json.dump(row, jsonfile, separators=(',', ':'))
jsonfile.write('\n')
我希望输出的浮点值不带引号，类似于以下内容：
{"ACCOUNTNAMEDENORM":"John Smith","DELINQUENCYSTATUS":2.0000000000,"RETIRED":0.0000000000,"INVOICEDAYOFWEEK":5.0000000000,"ID":1234567.0000000000,"BEANVERSION":69.0000000000,"ACCOUNTTYPE":1.0000000000,"ORGANIZATIONTYPEDENORM":null,"HIDDENTACCOUNTCONTAINERID":4321987.0000000000,"NEWPOLICYPAYMENTDISTRIBUTABLE":"1","ACCOUNTNUMBER":"000-000-000-00","PAYMENTMETHOD":12345.0000000000,"INVOICEDELIVERYTYPE":98765.0000000000,"DISTRIBUTIONLIMITTYPE":3.0000000000,"CLOSEDATE":null,"FIRSTTWICEPERMTHINVOICEDOM":1.0000000000,"HELDFORINVOICESENDING":"0","FEINDENORM":null,"COLLECTING":"0","ACCOUNTNUMBERDENORM":"000-000-000-00","CHARGEHELD":"0","PUBLICID":"xx:1234346"}
答案 0 :(得分：1)
现在，根据您的评论，我可以更好地理解您的问题，这是一个完全不同的答案。请注意，它不使用json模块，而只是“手动”执行所需的处理。尽管它可能可以使用模块来完成，但是与下面使用的相对简单的逻辑相比，默认情况下，它可以完全不同地格式化它可以识别的Python数据类型的格式：
另一个注意事项：与您的代码一样，这会将csv文件的每一行转换为有效的JSON对象，并将每一行写入文件中的单独一行。但是，结果文件的内容将不是有效的JSON，因为所有这些单独的对象都需要用逗号分隔并放在[]括号中（即有效的JSON数组对象）。
import csv
with open('output2.csv', 'r', newline='') as csvfile, \
open('output2.json', 'w') as jsonfile:
for row in csv.DictReader(csvfile):
newfmt = []
for field, value in row.items():
field = '"{}"'.format(field)
try:
float(value)
except ValueError:
value = 'null' if value == '' else '"{}"'.format(value)
else:
# Avoid changing integer values.
try:
int(value)
except ValueError:
pass
else:
value = '"{}"'.format(value)
newfmt.append((field, value))
my_json = '{' + ','.join(':'.join(pair) for pair in newfmt) + '}'
jsonfile.write(my_json + '\n')
这是写入文件的JSON：
{"ACCOUNTNAMEDENORM":"John Smith","DELINQUENCYSTATUS":2.0000000000,"RETIRED":0.0000000000,"INVOICEDAYOFWEEK":5.0000000000,"ID":1234567.0000000000,"BEANVERSION":69.0000000000,"ACCOUNTTYPE":1.0000000000,"ORGANIZATIONTYPEDENORM":null,"HIDDENTACCOUNTCONTAINERID":4321987.0000000000,"NEWPOLICYPAYMENTDISTRIBUTABLE":"1","ACCOUNTNUMBER":"000-000-000-00","PAYMENTMETHOD":12345.0000000000,"INVOICEDELIVERYTYPE":98765.0000000000,"DISTRIBUTIONLIMITTYPE":3.0000000000,"CLOSEDATE":null,"FIRSTTWICEPERMTHINVOICEDOM":1.0000000000,"HELDFORINVOICESENDING":"0","FEINDENORM":null,"COLLECTING":"0","ACCOUNTNUMBERDENORM":"000-000-000-00","CHARGEHELD":"0","PUBLICID":"bc:1234346"}
下面再次显示，并添加了空格：
{"ACCOUNTNAMEDENORM": "John Smith",
"DELINQUENCYSTATUS": 2.0000000000,
"RETIRED": 0.0000000000,
"INVOICEDAYOFWEEK": 5.0000000000,
"ID": 1234567.0000000000,
"BEANVERSION": 69.0000000000,
"ACCOUNTTYPE": 1.0000000000,
"ORGANIZATIONTYPEDENORM": null,
"HIDDENTACCOUNTCONTAINERID": 4321987.0000000000,
"NEWPOLICYPAYMENTDISTRIBUTABLE": "1",
"ACCOUNTNUMBER": "000-000-000-00",
"PAYMENTMETHOD": 12345.0000000000,
"INVOICEDELIVERYTYPE": 98765.0000000000,
"DISTRIBUTIONLIMITTYPE": 3.0000000000,
"CLOSEDATE": null,
"FIRSTTWICEPERMTHINVOICEDOM": 1.0000000000,
"HELDFORINVOICESENDING": "0",
"FEINDENORM": null,
"COLLECTING": "0",
"ACCOUNTNUMBERDENORM": "000-000-000-00",
"CHARGEHELD": "0",
"PUBLICID": "bc:1234346"}
答案 1 :(得分：0)
可能有些矫kill过正，但是使用pandas会很简单：
validate_username(...)答案 2 :(得分：0)
一种解决方案是使用正则表达式查看字符串值是否看起来像浮点数，然后将其转换为浮点数。
import re
null = None
j = {"ACCOUNTNAMEDENORM":"John Smith","DELINQUENCYSTATUS":"2.0000000000",
"RETIRED":"0.0000000000","INVOICEDAYOFWEEK":"5.0000000000",
"ID":"1234567.0000000000","BEANVERSION":"69.0000000000",
"ACCOUNTTYPE":"1.0000000000","ORGANIZATIONTYPEDENORM":null,
"HIDDENTACCOUNTCONTAINERID":"4321987.0000000000",
"NEWPOLICYPAYMENTDISTRIBUTABLE":"1","ACCOUNTNUMBER":"000-000-000-00",
"PAYMENTMETHOD":"12345.0000000000","INVOICEDELIVERYTYPE":"98765.0000000000",
"DISTRIBUTIONLIMITTYPE":"3.0000000000","CLOSEDATE":null,
"FIRSTTWICEPERMTHINVOICEDOM":"1.0000000000","HELDFORINVOICESENDING":"0",
"FEINDENORM":null,"COLLECTING":"0","ACCOUNTNUMBERDENORM":"000-000-000-00",
"CHARGEHELD":"0","PUBLICID":"xx:1234346"}
for key in j:
if j[key] is not None:
if re.match("^\d+?\.\d+?$", j[key]):
j[key] = float(j[key])
我在这里使用null = None处理JSON中显示的“ null”。但是您可以在此处用要读取的每个CSV行替换“ j”，然后使用此更新行，然后用浮点数替换字符串将其写回。
如果可以将任何数字字符串转换为浮点数，那么可以跳过正则表达式（re.match()命令），并用j[key].isnumeric()替换它（如果您的Python版本可用）。
编辑：我不认为Python中的float会以您可能认为的方式处理“精度”。似乎2.0000000000被“截断”为2.0，但是我认为这更多是格式化和显示问题，而不是丢失信息。考虑以下示例：
>>> float(2.0000000000)
2.0
>>> float(2.00000000001)
2.00000000001
>>> float(1.00) == float(1.000000000)
True
>>> float(3.141) == float(3.140999999)
False
>>> float(3.141) == float(3.1409999999999999)
True
>>> print('%.10f' % 3.14)
3.1400000000
虽然可以使JSON具有这些零，但是在这种情况下，它只能将数字视为字符串，即格式化的字符串。
答案 3 :(得分：0)
哈，真的很有趣，我想和你找到相反的答案，就是结果带引号。
其实很容易自动去掉，去掉参数"separators=(',', ':')"即可。
对我来说，只需添加这个参数就可以了。