ctags-lang-python

Random notes about tagging python source code with Universal-ctags

Version:5.9.0
Manual group:Universal-ctags
Manual section:7

SYNOPSIS

ctags … --languages=+Python …
ctags … --language-force=Python …
ctags … --map-Python=+.py …

DESCRIPTION

This man page gathers random notes about tagging python source code.

TAGGING import STATEMENTS

Summary

import X

name kind role other noticeable fields
X module imported N/A

import X as Y

name kind role other noticeable fields
X module indirectlyImported N/A
Y namespace definition nameref:module:X

from X import *

name kind role other noticeable fields
X module namespace N/A

from X import Y

name kind role other noticeable fields
X module namespace N/A
Y unknown imported scope:module:X

from X import Y as Z

name kind role other noticeable fields
X module namespace N/A
Y unknown indirectlyImported scope:module:X
Z unknown definition nameref:unknown:X

Examples

“input.py” .. code-block:: Python

import X0

“output.tags” with “--options=NONE -o - --extras=+r --fields=+rzK input.py” .. code-block:: tags

X0 input.py /^import X0$/;” kind:module roles:imported

A tag for an imported module has module kind with imported role. The module is not defined here; it is defined in another file. So the tag for the imported module is a reference tag; specify --extras=+r (or --extras=+{reference}) option for tagging it. “roles:” field enabled with --fields=+r is for recording the module is “imported” to the tag file.

“input.py” .. code-block:: Python

import X1 as Y1

“output.tags” with “--options=NONE -o - --extras=+r --fields=+rzK --fields-Python=’+{nameref}’ input.py”

X1      input.py        /^import X1 as Y1$/;"   kind:module     roles:indirectlyImported
Y1      input.py        /^import X1 as Y1$/;"   kind:namespace  roles:def       nameref:module:X1

“Y1” introduces a new name and is defined here. So “Y1” is tagged as a definition tag. “X1” is imported in a way that its name cannot be used in this source file. For letting client tools know that the name cannot be used, indirectlyImported role is assigned for “X1”. “Y1” is the name for accessing objects defined in the module imported via “X1”. For recording this relationship, nameref: field is attached to the tag of “Y1”. Instead of module kind, namespace kind is assigned to “Y1” because “Y1” itself isn’t a module.

“input.py” .. code-block:: Python

from X2 import *

“output.tags” with “--options=NONE -o - --extras=+r --fields=+rzK input.py”

X2      input.py        /^from X2 import \*$/;" kind:module     roles:namespace

The module is not defined here; it is defined in another file. So the tag for the imported module is a reference tag. Unlike “X0” in “import X0”, “X2” may not be used because the names defined in “X2” can be used in this source file. To represent the difference namespace role is attached to “X2” instead of imported.

“input.py” .. code-block:: Python

from X3 import Y3

“output.tags” with “--options=NONE -o - --extras=+r --fields=+rzKZ input.py”

X3      input.py        /^from X3 import Y3$/;" kind:module     roles:namespace
Y3      input.py        /^from X3 import Y3$/;" kind:unknown    scope:module:X3 roles:imported

“Y3” is a name for a language object defined in “X3” module. “scope:module:X3” attached to “Y3” represents this relation between “Y3” and “X3”. ctags assigns unknown kind to “Y3” because ctags cannot know whether “Y3” is a class, a variable, or a function from the input file.

“input.py” .. code-block:: Python

from X4 import Y4 as Z4

“output.tags” with “--options=NONE -o - --extras=+r --fields=+rzKZ input.py”

X4      input.py        /^from X4 import Y4 as Z4$/;"   kind:module     roles:namespace
Y4      input.py        /^from X4 import Y4 as Z4$/;"   kind:unknown    scope:module:X4 roles:indirectlyImported
Z4      input.py        /^from X4 import Y4 as Z4$/;"   kind:unknown    roles:def       nameref:unknown:Y4

“Y4” is similar to “Y3” of “from X3 import Y3” but the name cannot be used here. indirectlyImported role assigned to “Y4” representing this. “Z4” is the name for accessing the language object named in “Y4” in “X4” module. “nameref:unknown:Y4” attached to “Z4” and “scope:module:X4” attached to “Y4” represent the relations.

LAMBDA EXPRESSION AND TYPE HINT

Summary

id = lambda var0: var0

name kind role other noticeable fields
id function definition signature:(var0)

id_t: Callable[[int], int] = lambda var1: var1

name kind role other noticeable fields
id_t variable definition typeref:typename:Callable[[int], int] nameref:function:anonFuncN
anonFuncN function definition signature:(var1)

Examples

“input.py” .. code-block:: Python

from typing import Callable id = lambda var0: var0 id_t: Callable[[int], int] = lambda var1: var1

“output.tags” with “--options=NONE -o - --sort=no --fields=+KS --fields-Python=’+{nameref}’ --extras=’+{anonymous}’ input.py”

id      input.py        /^id = lambda var0: var0$/;"    function        signature:(var0)
id_t    input.py        /^id_t: Callable[[int], int] = lambda var1: var1$/;"\
        variable        typeref:typename:Callable[[int], int]   nameref:function:anonFunc84011d2c0101
anonFunc84011d2c0101    input.py        /^id_t: Callable[[int], int] = lambda var1: var1$/;"\
        function        signature:(var1)

If a variable (“id”) with no type hint is initialized with a lambda expression, ctags assigns function kind for the tag of “id”.

If a variable (“id_t”) with a type hint is initialized with a lambda expression, ctags assigns variable kind for the tag of “id_t” with typeref: and nameref: fields. ctags fills typeref: field with the value of the type hint. The way of filling nameref: is a bit complicated.

For the lambda expression used in initializing the type-hint’ed variable, ctags creates anonymous extra tag (“anonFunc84011d2c0101”). ctags fills the nameref: field of “id_t” with the name of anonymous extra tag: “nameref:function:anonFunc84011d2c0101”.

You may think why ctags does so complicated, and why ctags doesn’t emit following tags output for the input:

id      input.py        /^id = \\$/;"   function        signature:(var0)
id_t    input.py        /^id_t: \\$/;"  function        typeref:typename:Callable[[int], int]   signature:(var1)

There is a reason. The other languages of ctags obey the following rule: ctags fills typeref: field for a tag of a callable object (like function) with the type of its return value. If we consider “id_t” is a function, its typeref: field should have “typename:int”. However, for filling typeref: with “typename:int”, ctags has to analyze “Callable[[int], int]” deeper. We don’t want to do so.