python表中的keyerror

Question

I'm confused about the code below:

class Table(object):
    def __init__(self):
        self.d = collections.defaultdict(dict)
    def add(self, row, col, val):
        self.d[row][col] = val
    def get(self, row, col, default=None):
        return self.d[row].get(col, default)
    def incol(self, col, val):
        for x,y in self.d.iteritems():
            if y[col]==val:
                wholerow=y
        return wholerow.values()

After I created table, I cannot use the "incol" function to get all the values. It always give me key error. However, when I just use the same code without using function, I can get what I want. What happened to my code....

Answer 1

The above code does not "always give me key error" ("always" being the word I am disputing here). For example, consider the following code:

foo = Table()
foo.add("row_1", "col_1", "col_1_value")
foo.add("row_1", "col_2", "col_2_value")
print foo.get("row_1", "col_1")
print foo.incol("col_1", "col_1_value")

This works without any errors and produces the following output:

col_1_value
['col_2_value', 'col_1_value']

However, if you try to pass the name of a non-existent column to incol , you will get a KeyError . For example, if you append the following line to the above code:

print foo.incol("i do not exist", "my imaginary value")

You will end up with an error message resembling the following:

Traceback (most recent call last):
File "./stack_overflow-2013-03-14.py", line 21, in <module>
    print foo.incol("i do not exist", "my imaginary value")
File "./stack_overflow-2013-03-14.py", line 12, in incol
    if y[col]==val:
KeyError: 'i do not exist'

An example implementation of incol that can handle such situations is:

def incol(self, col, val):
    result = None
    for x,y in self.d.iteritems():
        if col in y and y[col]==val:
            result = y.values()
            break
    return result

Now, if you run the following lines:

foo = Table()
foo.add("row_1", "col_1", "col_1_value")
foo.add("row_1", "col_2", "col_2_value")
print foo.get("row_1", "col_1")
print foo.incol("col_1", "col_1_value")
print foo.incol("i do not exist", "my imaginary value")
print foo.incol("i do not exist", None)

You get:

col_1_value
['col_2_value', 'col_1_value']
None
None

You can also use the following implementation of incol as suggested by Martijn Pieters:

def incol(self, col, val):
    result = None
    for x,y in self.d.iteritems():
        if y.get(col)==val:
            result = y.values()
            break
    return result

But in this case if you pass a non-existent col name and val = None , then it will match any row. The corresponding output will be:

col_1_value
['col_2_value', 'col_1_value']
None
['col_2_value', 'col_1_value']

Maybe this is the behavior you want for your application. That is for you to decide.