When using alembic revision --autogenerate is it possible to preview the automatically generated file without saving it?

[ghost]

Hello,

I came across #936 but I believe my use-case to be different and is similar to how Django does it.

I'd like to know if it's possible to run alembic revision --autogenerate with a specific flag such that the Python file (not the raw SQL) that would normally be saved to the disk is redirected to stdout instead? Essentially, I'm looking for sort of a --dry-run flag, if one exists.

Here are the reasons for my request:

1. Sometimes empty migrations files can be produced
   • Running the command with no change in the database
   • Running the command with one or more of the changes that fall under "Autogenerate can not detect"
2. During prototyping/testing, it's common for me to make small changes to see how the file gets autogenerated.

While the empty migration issue can be prevented and one can argue that the unnecessary files can be deleted, I believe this feature would greatly improve the quality of life for users. 🙂

I am unsure if this feature exists, which is why I'm posing a question. If it doesn't, I'd like for this to be a considered a feature request.

[ghost]

@zzzeek @CaselIT apologies for the ping, just wanted to know if you have given this a look

To clarify, this comment was not meant to ask for any sort of commitment / deadline in case this turns out to be a feature request. As long as this issue is under the radar of the maintainers, I am fine with that.

[CaselIT]

(sorry was on holiday last week and missed this)

[ghost]

@CaselIT No problem at all, hope you had a great holiday. I very much appreciate the response. 🙂

[zzzeek]

this is not built in and would not be that simple because we have things like post write hooks that look for a file.

so the best we could do here would be a recipe that extends on the existing recipe for dont generate empty migrations which adds a render step inside of the custom hook. we have a utility function for this which is render_python_code

this is not tested but it makes use of the x argument to add alembic revision --autogenerate -x print:

from alembic.autogenerate import render_python_code
from mako.template import Template
from alembic import util
from pathlib import Path

def process_revision_directives(context, revision, directives):
    if config.cmd_opts.autogenerate and config.cmd_opts.x == ["print"]:
        script = directives[0]
        template_vars = {
            "upgrades": render_python_code(script.upgrade_ops),
            "downgrades": render_python_code(script.downgrade_ops),
            "imports": "\n".join(sorted(script.imports)),
            "message": "none",
            "up_revision": revision,
            "down_revision": None,
            "create_date": None,
            "branch_labels": None,
            "depends_on": None,
            "comma":util.format_as_comma
        }
        template = Template(
            filename=(Path(context.script.dir) / "script.py.mako").as_posix(),
        )
        print(template.render(**template_vars))
        # erase directives which prevents file from being written
        directives[:] = []

def run_migrations_online():

    # ...

    # connectable = ...

    with connectable.connect() as connection:
        context.configure(
            connection=connection,
            target_metadata=target_metadata,
            process_revision_directives=process_revision_directives
        )

        with context.begin_transaction():
            context.run_migrations()

[zzzeek]

this can of course be refined and added to https://alembic.sqlalchemy.org/en/latest/cookbook.html

[ghost]

Thanks for the prompt response @zzzeek. This should do exactly what I want.

A tangential question just to sate my curiosity. Why haven't you considered adding "Don’t Generate Empty Migrations with Autogenerate" to the base template(s)? Is there any reason one would want an empty migration?

[zzzeek]

sure, if they have constructs that were not detected by autogenerate that they wish to add

[CaselIT]

@iamrinshibuya once you have looked into it if you want to create a cookbook example a PR would be appreciated!

[ghost]

@CaselIT Sure, I'll see what I can come up with over the weekend

[me21]

The proposed answer doesn't work for me, or maybe it doesn't work with the new version of alembic:

>alembic revision --autogenerate -x print
usage: alembic [-h] [--version] [-c CONFIG] [-n NAME] [-x X] [--raiseerr] [-q]
               {branches,check,current,downgrade,edit,ensure_version,heads,history,init,list_templates,merge,revision,show,stamp,upgrade}
               ...
alembic: error: unrecognized arguments: -x print

If I move -x print to the front, I get an exception:

File "C:\Users\azarubkin\Documents\Python Projects\edoc2\alembic\env.py", line 66, in process_revision_directives
    print(render_python_code(script))
          ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "venv\Lib\site-packages\alembic\autogenerate\api.py", line 251, in render_python_code
    render._render_cmd_body(up_or_down_op, autogen_context)
  File "venv\Lib\site-packages\alembic\autogenerate\render.py", line 108, in _render_cmd_body
    for op in op_container.ops:
              ^^^^^^^^^^^^^^^^
AttributeError: 'MigrationScript' object has no attribute 'ops'

[odravison]

I'm having the same problem as you:

root@0d60e60431e4:/app# alembic current -x tables=subscription
usage: alembic [-h] [--version] [-c CONFIG] [-n NAME] [-x X] [--raiseerr] [-q] {branches,check,current,downgrade,edit,ensure_version,heads,history,init,list_templates,merge,revision,show,stamp,upgrade} ...
alembic: error: unrecognized arguments: -x tables=subscription

My env.py:

import os
from logging.config import fileConfig

from sqlalchemy import create_engine, pool
from alembic import context

# Carrega .env (opcional)
try:
    from dotenv import load_dotenv
    load_dotenv()
except Exception:
    pass

# this is the Alembic Config object, which provides
# access to the values within the .ini file in use.
config = context.config

# Interpret the config file for Python logging.
# This line sets up loggers basically.
if config.config_file_name is not None:
    fileConfig(config.config_file_name)

# add your model's MetaData object here
# for 'autogenerate' support
# from myapp import mymodel
# target_metadata = mymodel.Base.metadata
from app.db.db_models.base_class import Base
target_metadata = Base.metadata

DATABASE_URL = os.getenv("DATABASE_URL")
if not DATABASE_URL:
    raise RuntimeError("DATABASE_URL is not defined in environment (.env)")

# Read table names from -x tables=...
_x = context.get_x_argument(as_dictionary=True)
ONLY_TABLES = {t.strip() for t in _x.get("tables", "").split(",") if t.strip()}

def include_object(obj, name, type_, reflected, compare_to):
    # If no filter was passed, include everything
    if not ONLY_TABLES:
        return True

    # Tables
    if type_ == "table":
        return name in ONLY_TABLES

    # Columns and indexes: filter by parent table
    parent = getattr(obj, "table", None)
    if parent is not None:
        return parent.name in ONLY_TABLES

    # FKs/constraints: try to resolve by the parent table
    t = getattr(getattr(obj, "table", None), "name", None)
    if t:
        return t in ONLY_TABLES

    return True

print("x-args:", context.get_x_argument(as_dictionary=True))

def run_migrations_offline():
    context.configure(
        url=DATABASE_URL,
        target_metadata=target_metadata,
        literal_binds=True,
        compare_type=True,
        compare_server_default=True,
        include_object=include_object
    )
    with context.begin_transaction():
        context.run_migrations()

def run_migrations_online():
    engine = create_engine(
        DATABASE_URL,
        poolclass=pool.NullPool,
        future=True,  # SQLAlchemy 2.x style, still sync
    )
    with engine.connect() as connection:
        context.configure(
            connection=connection,
            target_metadata=target_metadata,
            compare_type=True,
            compare_server_default=True,
            version_table="alembic_version",
            include_object=include_object,
        )
        with context.begin_transaction():
            context.run_migrations()

if context.is_offline_mode():
    run_migrations_offline()
else:
    run_migrations_online()

[zzzeek]

here is a full template render fully tested with the included script.py.mako template

from alembic.autogenerate import render_python_code
from mako.template import Template
from alembic import util
from pathlib import Path

def process_revision_directives(context, revision, directives):
    if config.cmd_opts.autogenerate and config.cmd_opts.x == ["print"]:
        script = directives[0]
        template_vars = {
            "upgrades": render_python_code(script.upgrade_ops),
            "downgrades": render_python_code(script.downgrade_ops),
            "imports": "\n".join(sorted(script.imports)),
            "message": "none",
            "up_revision": revision,
            "down_revision": None,
            "create_date": None,
            "branch_labels": None,
            "depends_on": None,
            "comma":util.format_as_comma
        }
        template = Template(
            filename=(Path(context.script.dir) / "script.py.mako").as_posix(),
        )
        print(template.render(**template_vars))
        # erase directives which prevents file from being written
        directives[:] = []