Seventh Arrow posted:

Baronash posted:

Is there a good way to have the extra series added as columns, rather than additional rows?

Thermopyle posted:

Here's a neat thing if you're one of the cool people like me who can mostly move most of their projects to keep up with the latest python versions. Coming in Python 3.7 are data classes!

Thermopyle posted:

I can't think of anything specific to PyCharm that would make that easier other than maybe some fancy regex search/replace.

You can write a class and override the appropriate dunder methods for attribute access...or you can use a library like Box which allows you to access your dict via attribute or item access. That would allow you to keep your functions as is and slowly convert them over to dotted access.

That being said, I'm not convinced it's a good idea. On the one hand, I also find dotted access smoother and less irritating to use, but on the other hand the way you access dicts and the way you access objects is an established thing in Python-land.

Whatever you do, I don't think there's an easy way to refactor this.

I've done the method where I implement the appropriate dunder methods on a class to let me access a dict via attribute and the bad part when it comes to PyCharm is that you don't get any help on autocompletion since the attributes are dynamic in nature.

wasey posted:

but I can't seem to properly assign the info. Thanks for helping out

Wallet posted:

If I have a list of numbers, like so:
[0, 14, 18, 20, 36, 41, 62, 70, 72]

And I want to extract all longest possible lists from it where no two consecutive items are more than, say, 5 apart:
[[14, 18, 20], [36, 41], [70, 72]]

It's not hard to construct a loop to do this with a few conditionals, but is there an elegant python solution?

Cingulate posted:

baka is, in fact, literally talking about map:

Cingulate posted:

You mean like this?

code:

df1 = pd.DataFrame({'A': ['a', 'b', 'c', 'd'],
                    'B': ['.']*4},
                    index=[0, 1, 2, 3])
df2 = pd.DataFrame({'A': ['a', 'c', 'd'],
                    'B': ['one', 'two', 'three']},
                    index=[0, 1, 2])
df1b = df1.set_index("A")
df1b["B"] = df2.set_index("A")["B"]

print(df1b)
       B
A       
a    one
b    NaN
c    two
d  three

(The assignment aligns on index.)

vikingstrike posted:

That's a one liner with pd.merge().

baka kaba posted:

While we're on this, this is the best one of those visualisation videos I've seen

https://www.youtube.com/watch?v=sYd_-pAfbBw

The hue is where the dot should be in the array, and the closeness to the centre is how out of position it is. As each one gets sorted it flies out to its position on the circle edge. Some wild stuff happens in there

SnatchRabbit posted:

Right, but what I need is message to be this portion, in JSON, so I can use individual values (strings) like resourceID, and put that in a statement later. Not sure that makes sense, like I said I'm terrible with python.

code:

{"awsAccountId":"#######","configRuleName":"restricted-sshv2","configRuleARN":"arn:aws:config:us-west-2:########:config-rule/config-rule-htirkv","resourceType":"AWS::EC2::SecurityGroup","resourceId":"sg-######","awsRegion":"us-west-2","newEvaluationResult":{"evaluationResultIdentifier":{"evaluationResultQualifier":{"configRuleName":"restricted-sshv2","resourceType":"AWS::EC2::SecurityGroup","resourceId":"sg-#######"},"orderingTimestamp":"2018-03-29T02:03:20.664Z"},"complianceType":"NON_COMPLIANT","resultRecordedTime":"2018-03-29T02:47:53.647Z","configRuleInvokedTime":"2018-03-29T02:47:53.207Z","annotation":null,"resultToken":null},"oldEvaluationResult":null,"notificationCreationTime":"2018-03-29T02:47:54.468Z","messageType":"ComplianceChangeNotification","recordVersion":"1.0"}

Cingulate posted:

Thread title

Seventh Arrow posted:

Yikes, really? So if it generated a hundred files, I'd be pretty hosed. I guess for now I'll just be thankful that the exercise only has five files.

So should I just do five "for img in img_url:" passes, each with a different filename?

breaks posted:

I don't comprehend at all what you're trying to accomplish. What you've got there at the moment seem to be one hell of a way to write self.some_list.append(some_stuff). On the other hand I've probably never manged to successfully understand a post in this thread when reading it at 2AM (or maybe ever) so I'll just answer these two specific questions:

Yes, if you're wrapping a function and want to make your wrapper appear to be what it's wrapping, use wraps.

The decorator syntax is only a shortcut for something that's otherwise ugly:

code:

def f(*args):
    print(*args)

f = decorator(f)  # If there was no @decorator syntax you would just write this

There is nothing magical about @decorator(a, b). Writing it out by hand results in a very literal translation:

code:

f = decorator(a, b)(f)
# or to really spell it out
actual_decorator = decorator_factory(a, b)
f = actual_decorator(f)

Disregarding multiple decorators and within the rather restrictive limits of what Python's grammar allows after the @, @X before f is the same as f = X(f) after it.

Does this help you understand what is going on? A decorator always gets called with one argument, which is the thing it's decorating. A "decorator with arguments" is a callable that returns a decorator. I think it's more clear to think of that as a decorator factory, but for whatever reasons the common terminology is just to lump it all together as "decorator". In what you wrote, __init__ is called as part of the process of instantiating the class. That's the factory part. Once the instance is created it's then called, which is possible since you defined __call__, and that's the decorator part.

vikingstrike posted:

Sure. Glad it’s working!

Query can be easier to use when method chaining and I believe on larger frames it gets processed using pandas’ eval() which speeds things up. I haven’t timed it recently though so I could be misremembering. I think this is actually discussed in the docs. I normally just use it when chaining methods. I would only worry about the input attack vector if this is public facing code. For personal use I don’t believe it matters. I’m also unsure what editor features you’re referring too but that could be an issue if you rely on something particular.

Zero Gravitas posted:

I'm trying to get an example of exactly how long it takes for the youtube algorithm to start pitching far right wing videos at a new user who views soft right or entryist political material.

baka kaba posted:

Another thing you could try is having a list containing the last 10 seconds' of IPs (filtered to only include the request types you're looking for). So when you get a new IP, add it to the list, then pop off the entries from the front of the list that have a timestamp > 10s earlier than your new entry. Then you can scan your list for matches on that new IP and see if you get 3+

Like a moving window, basically