python.txt

Type reference a symbol before it's defined (like in a class definition)?;;Wrap the symbol in a string, it'll be eval'd later.<br>"Forward reference";

Display the differences between an environment and Django's default settings;;;manage.py diffsettings

Roll back a migration?;;;server/manage.py migrate seminar 0073

What are Django's five log levels?;;DEBUG, INFO, WARN(ING), ERROR, CRITICAL;

What are the four parts of Django's logging system?;;Loggers, Handlers, Filters, Formatters;

How do you create a timezone-aware datetime?;;;from django.utils import timezone<br>timezone.now()

Inspect Python's import search path?;;;import sys<br/>sys.path

Step one line in ipdb;;;s(tep)

IPython: list all variables + info;;;%whos

IPython: display environment vars;;;%env

What is the difference between a Python module and a package?;;Module: python file with functions (file) <br/>Package: directory with __init__.py, contians packages & modules (dir);

Read all of the functions implemented by module `url`?;;;dir(url)

How many layers of nesting does a decorator that takes arguments require?;;3;def outer_layer(takes_arguments):<br>....def second_layer(old_function):<br>........def inner_layer(*args, **kwds):<br>............new_function_work<br>........return inner_layer<br>....return second_layer

/ vs //?;;Float division (py3 only) vs integer floor division.;

np: Create a new array of 2*2 integers?;;;np.zeros([2,2], int)

np: Create a 3x3 matrix with ones on the diagonal and zeros elsewhere.;;;np.identity(3)

np: Create a new array of 2*5 uints, filled with 6.;;;np.full((2, 5), 6, dtype=np.uint)

np: What are the 4 _like methods, and what do they do?;;Copy shape and datatype. zeros, ones, full, empty;

np: create a 3x5 matrix of random data;;;np.random.random((3, 5))

np: Create an array of 2, 4, 6, 8, ..., 100.;;;np.arange(2, 101, 2)

np: Create a 1-D array of 50 linearly (or logarithmically) spaced elements between 3. and 10., inclusive.;;;np.linspace(3., 10, 50)<br>np.logspace

np: zero out the bottom triangle of an array?;;;np.tril(X, -1)  # -1 chooses the diagonal

np: reshape the second dimension to be 75, let the first dimension be inferred?;;;x.reshape((-1, 150))

np: flatten a 2D array by iterating through in column order?; [[1, 2, 3], [4, 5, 6]] => [1 4 2 5 3 6];Use "Fortran" order. Default is "C", for C-style order (row-major). Fortran used column-major ordering.;np.asarray([np.arange(1, 4), np.arange(4, 7)]).flatten(order='F')

np: get the 10th element in a matrix?;;;x.flat[10]

np: get transpose of matrix?;;;x.T

np: expand dimensionality of matrix? Make each element of a 3, 4 array sit inside its own array.;;Shape will go from (3, 4) to (3, 4, 1);np.expand_dims(axis=2)

np: remove single-dimensional axes?;;;ndarray.squeeze()

np: Let x be an array <br/>[[ 1 2 3]<br/>[ 4 5 6].<br/><br/>and y be an array <br/>[[ 7 8 9]<br/>[10 11 12]].<br/>Concatenate x and y so that a new array looks like <br/>[[1, 2, 3, 7, 8, 9], <br/>[4,5, 6, 10, 11, 12]].;;;np.concatenate((a1, a2), axis=1)

np: Let x be an array [1, 2, 3, ..., 9]. Split x into 3 arrays, each of which has 4, 2, and 3 elements in the original order.;;;np.split(x, [4, 6, 9])

np: separate or join arrays across various dimensions?;;vsplit (row-wise), hsplit (column-wise), dsplit (depth-wise)<br>vstack, hstack, dstack<br>split, concatenate, stack;

np: concatenate vs stack?;;Concatenate adds matrices together along existing axes, stack inserts new ones.;

np: Compare two arrays elementwise?;;;np.allclose(a, b)

np: Let x be an array [0, 1, 2]. Convert it to <br/>[[0, 1, 2, 0, 1, 2],<br/>[0, 1, 2, 0, 1, 2]].;;A, reps=times to repeat array along each axis;np.tile(x, [2, 2])

np: Let x be an array [0, 1, 2]. Convert it to <br/>[0, 0, 1.1, 1, 2, 2].;;;np.repeat([0, 1, 2], 2)

np: Let x be an array [2, 2, 1, 5, 4, 5, 1, 2, 3]. Get two arrays of unique elements and their counts.;;return_index=True: indices of input that give first unique val<br>return_inverse=False: return indices that you can use to reconstruct original<br>return_counts=False: return counts<br>axis=None: if None, flatten first, otherwise only get unique per axis.;np.unique(x, return_counts=True)

np: Let x be an array <br/>[[ 1 2]<br/>[ 3 4].<br/>Rotate x 90 degrees counter-clockwise.;;m, # times to rotate clockwise, rotate through plane defined by these axes=(0, 1);np.rot90(x, -1)

np: Let x be an array <br/>[[ 1 2 3 4]<br/>[ 5 6 7 8].<br/>Shift elements one step to right along the second axis.;;a, shift, axis;np.roll(x, 0, axis=1)

np: Count the lowest index of "l" in x, element-wise.;;All sorts of string ops live in np.char.;np.char.find(x, 'l')

np: keyword search on docstrings?;;;np.lookfor(keyword)

np: customize how NP objects are displayed?;;precision=8: digits for floating points<br>threshold=1000: points at which np summarizes<br>suppress=False: show small floats?;np.set_printoptions

Show binary representation of decimal number?;;digit, width;np.binary_repr(12)<br>np.base_rep(12, 2)

np: what happens when you matmul a vector and 2x2 matrix? Isn't 2x1 * 2x2 multiplication impossible?;@;It promotes the vector to a matrix by PREPENDING a 1 to the dimensions;

enumerate through an array and its indices?;;;for index, value in np.ndenumerate(x):

When you index by an array or a matrix, what happens?;;Each element of matrix is interpreted as a coordinate (for that dimension) and mapped to corresponding value.;

What are the three parts of an index for a numpy matrix?;;slice start, slice end, stride;

What happens when you index one dimension with tensors and not the others?;;The others are broadcast to match dimensions, if possible, then mapped.;

Indexing: select in full from unspecified dimensions?;;Use the Ellipsis object literal;y[..., :2, :2]

How are tuples and indexes treated differently when slicing?;;Arrays are mapped to entire dimensions of values, tuples are coordinates;y[(1, 1, 2)] => scalar, y[[1, 1, 2]] => the matrix at depth 1 twice, then the matrix at depth 2

Render a matrix as an image?;;;import matplotlib.pyplot as plt<br>plt.imshow(X, cmap='bone')<br>plt.show()

imshow: what dimensions does it accept?;;MxN: color mappable<br>MxNx3: RGB<br>MxNx4: RGBA;

What types of colormaps are there?;;Sequential: lightness increases<br>Diverging: neutral at 0<br>Qualitative: pretty<br>Miscellaneous: rainbow, terrain, etc;

get a copy of a slice?;;;B = A.copy()

Select matrix values above 10?;;;B = A > 10

matplotlib.pyplot.scatter: args?;;x, y: input data<br>s: size or sequence of sizes<br>c: color or sequence of colors (RGB);

compute e^x of a tensor?;;;np.exp(x)

Make a new virtualenv?;;;pyenv virtualenv 3.6.2 new-env

List virtualenvs?;;;pyenv virtualenvs

Automatically switch to a virtualenv?;;;pyenv activate new-env<br>cat 'new-env' >> .python-version

Uninstall a virtualenv?;;;pyenv uninstall new-env

Difference between list and collections.deque (double-ended queue)?;;lists are arrays (O(1) for random access and approx O(1) for appending, O(N) for prepending)<br>deques are doubly-linked lists (O(1) push and pop from both ends).<br>deque also has maxlen, which shifts items off the front when a push would exceed it;

What's the problem with mutable default arguments?;def uh_oh(arg={}):<br>    pass;They're instantiated once, not every time you invoke the function;

Constrain memory by Python;;;resource.setrlimit()

Why is Django delete() taking too much mem?;;It does SELECT, loads in Python, then sends signals around while deleting. Use _raw_delete to do it in SQL.;

Recover OOM EB instance?;;Make sure no connection to DB, reboot in EC2, deploy.;

Physical location of postgres record?;;;ctid

psql pretty print;;;\x auto

Postgres activity table?;;;pg_stat_activity

What are common reasons for a segfaults in Python?;;Binary module & library version mismatch<br>C extension (or Cython) is broke;

How does IPython's %debug magic work?;;IPython keeps a process hanging around for the last exception, and %debug can connect to it later.;

How are Python debuggers implemented?;;sys.settrace is used to set the system's trace function for the current thread, this function is then invoked with every line and exception. A debugger process runs in the background with breakpoints in mind, when those lines are hit, it opens a REPL.;

What is PyDevd?;;A high-performance debugger that handles settrace calls and decides what to do. Pydev is fast because it disables tracing entirely for contexts with no breakpoints!;

Why is Pandas chained indexing bad?;df[df.C <= 0].ix[:,'B':'E'] = 2;Because the get indexer will often (but not always!) return a view, and the set indexer will operate in place.;df[df.C <= 0, 'B':'E'] = 2

When does Pandas copy?;;(subsequent rules override).<br>1. Always.<br>2. Except if inplace=True.<br>3. Get/set indexers have their own rules;

When do Pandas indexers return copies, and when do they return views?;;Set indexers (.loc[], .iloc[], .at[]): views (in-place).<br>Get indexers on single-dtyped frames: views (mostly).<br>Get indexers on multi-dtyped frames: copies.;

GROUP BY multiple columns in Django?;;House.objects.values('price', 'rooms').annotate(Count('rooms'));