As discussed in https://mail.python.org/archives/list/python-dev@python.org/thread/JAQDBMC23HW2PQ27HQNJ7G244T423IDD/ I propose to expose the unparse.py tool as part of the standard library in the ast module.

The exposed function will have the interface:

ast.unparse(ast_obj, *, option1, option2,...)

and will return a string with the unparsed version of ast_obj.

The unparse tool will need some cosmetic improvements that will be tracked separately in this issue.

After PR17302 is merged we need to fix the following cosmetic issues indicated by Victor:

(*) unparse adds many useless parentheses. The algorithm seems naive.
For example, it adds "()" to "class _AddedDllDirectory():". It also
adds parenthesis around yield, like "(yield (top, dirs, nondirs))",
whereas the AST node was at "top level": it isn't a sub-expression.
Maybe this algortihm should be made smarter.

(*) newlines in docstring are rendered as two characters: "\" + "n"
(escaped newline: \n), rather than a newline character. I would expect
a newline, it's more common that \n... But it may "break" inline
doctests rendering... Maybe it should be an option (render newlines as
newline character or escape them?), or maybe even let the user choose
how to render them using a callback (that's related to the "pluggable
formatter" question).

(*) Indentation is hardcoded to 4 spaces. What if I want 2 spaces or
something different? Should it become an option?

(*) Float infinity is replaces with 1e309. Again, maybe someone wants
to render this differently? It sounds like an arbitrary choice (which
"works" as expected).

After PR 17302 is accepted, I'll work on refactorings including a precedence algorithm to find when to parentheses.

@gvanrossum are you OK with adding type comments support? Current version loses type comment information so if typed_ast parses this, they wont be the same in AST representation.

> @gvanrossum are you OK with adding type comments support? Current version loses type comment information so if typed_ast parses this, they wont be the same in AST representation.

Good catch, definitely do that!

New changeset 27fc3b6f3fc49a36d3f962caac5c5495696d12ed by Pablo Galindo in branch 'master':
bpo-38870: Expose a function to unparse an ast object in the ast module (GH-17302)
https://github.com/python/cpython/commit/27fc3b6f3fc49a36d3f962caac5c5495696d12ed

New changeset ded8888fbc33011dd39b7b1c86a5adfacc4943f3 by Miss Islington (bot) (Pablo Galindo) in branch 'master':
bpo-38870: Remove dependency on contextlib to avoid performance regression on import (GH-17376)
https://github.com/python/cpython/commit/ded8888fbc33011dd39b7b1c86a5adfacc4943f3

New changeset a322f50c369e2e4138266c88e32ef83af95b2da6 by Pablo Galindo (Batuhan Taşkaya) in branch 'master':
bpo-38870: Remove dead code related with argument unparsing (GH-17613)
https://github.com/python/cpython/commit/a322f50c369e2e4138266c88e32ef83af95b2da6

I created bpo-39069: "Move ast.unparse() function to a different module".

New changeset 4b3b1226e86df6cd45e921c8f2ad23c3639c43b2 by Pablo Galindo (Batuhan Taşkaya) in branch 'master':
bpo-38870: Refactor delimiting with context managers in ast.unparse (GH-17612)
https://github.com/python/cpython/commit/4b3b1226e86df6cd45e921c8f2ad23c3639c43b2

New changeset d69cbeb99d5fd0d5464e937202cca6a2024d1bcf by Pablo Galindo in branch 'master':
Revert "bpo-38870: Remove dependency on contextlib to avoid performance regression on import (GH-17376)" (GH-17687)
https://github.com/python/cpython/commit/d69cbeb99d5fd0d5464e937202cca6a2024d1bcf

New changeset 23a226bf3ae7b462084e899d007d12d9fe398ac5 by Pablo Galindo in branch 'master':
bpo-38870: Run always tests that heavily use grammar features in test_unparse (GH-17738)
https://github.com/python/cpython/commit/23a226bf3ae7b462084e899d007d12d9fe398ac5

New changeset 7b35bef9787cd80ed1e12124f759b4be03c849db by Pablo Galindo (Batuhan Taşkaya) in branch 'master':
bpo-38870: Throw ValueError on invalid yield from usage (GH-17798)
https://github.com/python/cpython/commit/7b35bef9787cd80ed1e12124f759b4be03c849db

ExtSlice nodes without second value doesn't roundtrip properly

source: x[1:2,]
         Expr(
             value=Subscript(
                 value=Name(id='x', ctx=Load()),
-                slice=ExtSlice(
-                    dims=[
-                        Slice(
-                            lower=Constant(value=1, kind=None),
-                            upper=Constant(value=2, kind=None),
-                            step=None)]),
+                slice=Slice(
+                    lower=Constant(value=1, kind=None),
+                    upper=Constant(value=2, kind=None),
+                    step=None),
                 ctx=Load()))],
     type_ignores=[])

(I have a fix for unifying both tuple, constant tuple and extslice dims traversing)

We might need to tweak the documentation @pablogsal, 

> Unparse an ast.AST object and generate a string with code that would produce an equivalent ast.AST object if parsed back with ast.parse().

If I interpret `equivalent` correctly, this explanation is false under cases like constant nodes has an immutable container value (which they can). 

>>> def wrap(expr):
...     return ast.Module(body=[ast.Expr(expr)], type_ignores=[])
... 
>>> constant_tuple = wrap(ast.Constant(value=(1, 2, 3), kind=None))
>>> normalpy_tuple = ast.parse("(1, 2, 3)")
>>> ast.unparse(constant_tuple) == ast.unparse(normalpy_tuple)
True
>>> ast.dump(ast.parse(ast.unparse(constant_tuple))) != ast.dump(constant_tuple)
True
>>> ast.dump(ast.parse(ast.unparse(constant_tuple))) == ast.dump(normalpy_tuple)
True

This isn't a bug in the code because there is no way we can generate a string that can produce a constant tuple. But this makes the docstring false, at least in certain cases.

>  We might need to tweak the documentation @pablogsal, 

Maybe we can say that is 'as close as possible' if you pass an arbitrary AST object (if only happens with Constants we could documment exactly that). Let me think about this and I will make PR updating the docs.

New changeset 397b96f6d7a89f778ebc0591e32216a8183fe667 by Batuhan Taşkaya in branch 'master':
bpo-38870: Implement a precedence algorithm in ast.unparse (GH-17377)
https://github.com/python/cpython/commit/397b96f6d7a89f778ebc0591e32216a8183fe667

New changeset 89aa4694fc8c6d190325ef8ed6ce6a6b8efb3e50 by Batuhan Taşkaya in branch 'master':
bpo-38870: Add docstring support to ast.unparse (GH-17760)
https://github.com/python/cpython/commit/89aa4694fc8c6d190325ef8ed6ce6a6b8efb3e50

New changeset e7cab7f780ac253999512ee86374fc3454342811 by Batuhan Taşkaya in branch 'master':
bpo-38870: Simplify sequence interleaves in ast.unparse (GH-17892)
https://github.com/python/cpython/commit/e7cab7f780ac253999512ee86374fc3454342811

New changeset 5b66ec166b81c8a77286da2c0d17be3579c3069a by Batuhan Taşkaya in branch 'master':
bpo-38870: Implement support for ast.FunctionType in ast.unparse (GH-19016)
https://github.com/python/cpython/commit/5b66ec166b81c8a77286da2c0d17be3579c3069a

New changeset 493bf1cc316b0b5bd90779ecd1132878c881669e by Batuhan Taskaya in branch 'master':
bpo-38870: Don't start generated output with newlines in ast.unparse (GH-19636)
https://github.com/python/cpython/commit/493bf1cc316b0b5bd90779ecd1132878c881669e

New changeset ce4a753dcb3eef3d68e892a6515490b1aa219651 by Batuhan Taskaya in branch 'master':
bpo-38870: Do not separate factor prefixes in ast.unparse (GH-20133)
https://github.com/python/cpython/commit/ce4a753dcb3eef3d68e892a6515490b1aa219651

New changeset 25160cdc4775a1ddb4e37c8bf5a6e31ad9c146ed by Batuhan Taskaya in branch 'master':
bpo-38870: Don't put unnecessary parentheses on class declarations in ast.parse (GH-20134)
https://github.com/python/cpython/commit/25160cdc4775a1ddb4e37c8bf5a6e31ad9c146ed

New changeset e966af7cff78e14e1d289db587433504b4b53533 by Batuhan Taskaya in branch 'master':
bpo-38870: Correctly handle empty docstrings in ast.unparse (GH-18768)
https://github.com/python/cpython/commit/e966af7cff78e14e1d289db587433504b4b53533

New changeset dff92bb31f7db1a80ac431811f8108bd0ef9be43 by Batuhan Taskaya in branch 'master':
bpo-38870: Implement round tripping support for typed AST in ast.unparse (GH-17797)
https://github.com/python/cpython/commit/dff92bb31f7db1a80ac431811f8108bd0ef9be43

New changeset 6341fc7257d89d798675ad6e425f7eb0b6f2b4bb by Pablo Galindo in branch 'master':
bpo-38870: Use subTest in test_unparse for better error reporting (GH-20141)
https://github.com/python/cpython/commit/6341fc7257d89d798675ad6e425f7eb0b6f2b4bb

New changeset d71a6492dbd5434dfa6a0ad95e3ad98aa690887a by CyberSaxosTiGER in branch 'master':
bpo-38870: correctly escape unprintable characters on ast.unparse (GH-20166)
https://github.com/python/cpython/commit/d71a6492dbd5434dfa6a0ad95e3ad98aa690887a

New changeset c102a148256b00b7d48c51a1a97df19042e603de by Batuhan Taskaya in branch 'master':
bpo-38870: Don't omit parenthesis when unparsing a slice in ast.unparse
https://github.com/python/cpython/commit/c102a148256b00b7d48c51a1a97df19042e603de

New changeset dd74b6fde31aff9aa46c4fc2a830c569764e1b63 by Batuhan Taskaya in branch 'master':
bpo-38870: invalid escape sequence (GH-20240)
https://github.com/python/cpython/commit/dd74b6fde31aff9aa46c4fc2a830c569764e1b63

New changeset 059279d8706074489144f6ba8ccc8723e0b85007 by Miss Islington (bot) in branch '3.9':
bpo-38870: invalid escape sequence (GH-20240)
https://github.com/python/cpython/commit/059279d8706074489144f6ba8ccc8723e0b85007

New changeset 8df1016e2ef8c0a9f4d15bf7894c284295c99d9f by Batuhan Taskaya in branch 'master':
bpo-38870: Extend subject of ast.unparse warnings (GH-21053)
https://github.com/python/cpython/commit/8df1016e2ef8c0a9f4d15bf7894c284295c99d9f

New changeset 6e39999a246a019a356e1befcdfd3d54a52cb7b5 by Pablo Galindo in branch '3.9':
[3.9] bpo-38870: Extend subject of ast.unparse warnings (GH-21053) (GH-21191)
https://github.com/python/cpython/commit/6e39999a246a019a356e1befcdfd3d54a52cb7b5

Well, I'm happy to say that 3.9.0 is finally released with the ast.unparse interface. After tens of PRs, I think it is time for us to move on. For all future bugs, please create a new issue and nosy me. Thanks everyone who has participated this journey!