Raise ValueError exception if bucket name is invalid.

[daspecster]

Fixes#2956.

[dhermes]

@daspecster I am 👎 on this change, not really sure what the right approach is but adding this check in the constructor makes EVERY constructed bucket pay the price

[daspecster]

@dhermes you mean every existing bucket? I don't believe there could be any existing buckets that violate this rule? I think the API would have complained.

#2956 (comment)

[dhermes]

@daspecster No I mean the local Bucket objects created by users of this library

[tseaver]

Quoting my follow up to #2956:

If the API allows for / in the bucket name, then we shouldn't be prohibiting such names: instead, we should be URL-escaping the bucket name when constructing the API request URL.

[daspecster]

@lukesneeringer@tseaver any alternative ideas?

ISTM that object creation would be the place to check this? Maybe there's a more efficient way to do it though?

[dhermes]

I think we can "fix" by just more clearly documenting the acceptable values.

[tseaver]

@dhermes If the API allows embedding / in a bucket name, then we have to honor that, which means that we need to be URL-escaping the bucket name when creating API URLs

[daspecster]

@tseaver a user might be confused when then see their bucket names listed from some other library that may/maynot handle the urlencoding/urldecoding right?

@dhermes sure, but the error that is returned is a 404 which doesn't explain what the allowed values are.
I'll update the docs to be more clear about the allowable name values since that should be in here anyway.

[lukesneeringer]

@tseaver, you are making an orthogonal point; in actuality, slashes are not permissible at the start or end of a bucket name (by the API), but the client library allows them to pass through.

[tseaver]

@lukesneeringer the OP in #2956 reported having created a bucket with a trailing slash.

[lukesneeringer]

@tseaver Yes, in our library, and then the API dropped the slash for the bucket name.

[lukesneeringer]

@daspecster I am 👎 on this change, not really sure what the right approach is but adding this check in the constructor makes EVERY constructed bucket pay the price

That seems fine to me. It is not terribly expensive.

[lukesneeringer]

Approved, pending @dhermes explaining his problem with it and resolving that.

[dhermes]

My issue is just that RegEx is expensive. In terms of this specific approach, we don't even need a regex:

name[0].isalnum()
name[-1].isalnum()

Though this is a little strange with unicode (2) / str (3)

>>>u'\xff'.isalnum()
True

[lukesneeringer]

Sold. Change to that, and add len(name) <= 222.

EDIT: We should also add all([len(i) <= 63 for i in name.split('.')]).

That is good enough. The API can complain about the remaining errors (e.g. "not containing google or a common misspelling thereof").

[lukesneeringer]

@daspecster Suggestion:

class_PropertyMixin(object):
def__init__(self, name=None):
ifname:
self._validate_name(name)
[...]
def_validate_name(self, name):
# Names must start and end with a letter or number.ifnotname[0].isalnum() ornotname[-1].isalnum():
raiseValueError('Bucket names must start and end with an alphanumeric character.')
# Names must be between 3 and 222 total characters in length.iflen(name) <3:
raiseValueError('Bucket names must be at least 3 characters.')
iflen(name) >222:
raiseValueError('Bucket names can not exceed 222 characters.')
# Each bucket name component can not exceed 63 characters.ifany([len(i) >63foriinname.split('.')]):
raiseValueError('Each dot-separated component in a bucket name ''can not exceed 63 characters.')

[lukesneeringer]

Updated based on previous discussion.

[daspecster]

@lukesneeringer where are you seeing the 63 char limit in the docs?

[lukesneeringer]

Bucket names must contain 3 to 63 characters. Names containing dots can contain up to 222 characters, but each dot-separated component can be no longer than 63 characters.

[daspecster]

@lukesneeringer found the link. Thanks!

[dhermes]

There is just so much complexity here. Why can't we just punt to the server validation and document the rules in our docstring?

[daspecster]

To echo @dhermes, there are more requirements for the names as well and I assume they're subject to change(with or without bumping the API version).

Your bucket names must meet the following requirements:
Bucket names must contain only lowercase letters, numbers, dashes (-), underscores (_), and dots (.). Names containing dots require verification.
Bucket names must start and end with a number or letter.
Bucket names must contain 3 to 63 characters. Names containing dots can contain up to 222 characters, but each dot-separated component can be no longer than 63 characters.
Bucket names cannot be represented as an IP address in dotted-decimal notation (for example, 192.168.5.4).
Bucket names cannot begin with the "goog" prefix.
Bucket names cannot contain "google" or close misspellings of "google".
Also, for DNS compliance and future compatibility, you should not use underscores (_) or have a period adjacent to another period or dash. For example, ".." or "-." or ".-" are not valid in DNS names.

[lukesneeringer]

I guess what I was thinking was, "catch the easy ones, let the API fail on the rest".
The issue with the forward slash in particular is that it gets stripped by the API, which leads to non-obvious errors later. So, at minimum, we should catch the alphanumeric 0 and -1 indices.

I could go either way on the rest.

[daspecster]

@lukesneeringer did you have anything else for this?