如何从csv构建嵌套的有序字典？

Question

How can I get a nested dictionary, where both the keys and the subkeys are precisely in the same order as in the csv file?

I tried

import csv
from collections import OrderedDict

filename = "test.csv"
aDict = OrderedDict()

with open(filename, 'r') as f:
    csvReader = csv.DictReader(f)
    for row in csvReader:
        key = row.pop("key")
        aDict[key] = row

where test.csv looks like

key,number,letter
eins,1,a
zwei,2,b
drei,3,c

But the sub-dictionaries are not ordered (rows letter and number are changed). So how can I populate aDict[key] in an ordered manner?

Answer 1

You have to build the dictionaries and sub-dictionaries yourself from rows returned from csv.reader which are sequences, instead of using csv.DictReader .

Fortunately that's fairly easy:

import csv
from collections import OrderedDict

filename = 'test.csv'
aDict = OrderedDict()

with open(filename, 'rb') as f:
    csvReader = csv.reader(f)
    fields = next(csvReader)
    for row in csvReader:
        temp = OrderedDict(zip(fields, row))
        key = temp.pop("key")
        aDict[key] = temp

import json  # just to create output
print(json.dumps(aDict, indent=4))

Output:

{
    "eins": {
        "number": "1",
        "letter": "a"
    },
    "zwei": {
        "number": "2",
        "letter": "b"
    },
    "drei": {
        "number": "3",
        "letter": "c"
    }
}

Answer 2

This is one way:

import csv
from collections import OrderedDict

filename = "test.csv"
aDict = OrderedDict()

with open(filename, 'r') as f:
    order = next(csv.reader(f))[1:]
    f.seek(0)

    csvReader = csv.DictReader(f)
    for row in csvReader:
        key = row.pop("key")
        aDict[key] = OrderedDict((k, row[k]) for k in order)

Answer 3

csv.DictReader loads the rows into a regular dict and not an ordered one. You'll have to read the csv manually into an OrderedDict to get the order you need:

from collections import OrderedDict

filename = "test.csv"
dictRows = []

with open(filename, 'r') as f:
    rows = (line.strip().split(',') for line in f)
    # read column names from first row
    columns = rows.next()
    for row in rows:
        dictRows.append(OrderedDict(zip(columns, row)))

Answer 4

You can take advantage of the existing csv.DictReader class, but alter the rows it returns. To do that, add the following class to the beginning of your script:

class OrderedDictReader(csv.DictReader):
    def next(self):
        # Get a row using csv.DictReader
        row = csv.DictReader.next(self)

        # Create a new row using OrderedDict
        new_row = OrderedDict(((k, row[k]) for k in self.fieldnames))
        return new_row

Then, use this class in place of csv.DictReader :

csvReader = OrderedDictReader(f)

The rest of your code remains the same.