The most basic way to configure the flow of isort within a single file is action comments. These comments are picked up and interpreted by the isort parser during parsing.
Tells isort to skip the entire file.
# !/bin/python3 # isort: skip_file import os import sys ...
This should be placed as high in the file as reasonably possible. Since isort uses a streaming architecture, it may have already completed some work before it reaches the comment. Usually, this is okay - but can be confusing if --diff or any interactive options are used from the command line.
If placed on the same line as (or within the continuation of a) an import statement, isort will not sort this import. More specifically, it prevents the import statement from being recognized by isort as an import. In consequence, this line will be treated as code and be pushed down to below the import section of the file.
import b import a # isort: skip <- this will now stay below b
It is recommended to where possible use
# isort: off and
# isort: on or
# isort: split instead as the behavior is more explicit and predictable.
Turns isort parsing off. Every line after an
# isort: off statement will be passed along unchanged until an
# isort: on comment or the end of the file.
import e import f # isort: off import b import a
Turns isort parsing back on. This only makes sense if an
# isort: off comment exists higher in the file! This allows you to have blocks of unsorted imports, around otherwise sorted ones.
import e import f # isort: off import b import a # isort: on import c import d
Tells isort the current sort section is finished, and all future imports belong to a new sort grouping.
import e import f # isort: split import a import b import c import d
You can also use it inline to keep an import from having imports above or below it swap position:
import c import b # isort: split import a
isort split is exactly the same as placing an
# isort: on immediately below an
# isort: off