Make graphviz dependency optional (#36647)

The `graphviz` dependency has been problematic as Airflow required
dependency - especially for ARM-based installations. Graphviz
packages require binary graphviz libraries - which is already a
limitation, but they also require to install graphviz Python
bindings to be build and installed. This does not work for older
Linux installation but - more importantly - when you try
to install Graphviz libraries for Python 3.8, 3.9 for ARM M1
MacBooks, the packages fail to install because Python bindings
compilation for M1 can only work for Python 3.10+.

There is not an easy solution for that except commenting out
graphviz dependency from setup.py, when you want to install Airflow
for Python 3.8, 3.9 for MacBook M1.

However Graphviz is really used in two places:

* when you want to render DAGs wia airflow CLI - either to an image
  or directly to terminal (for terminals/systems supporting imgcat)

* when you want to render ER diagram after you modified Airflow
  models

The latter is a development-only feature, the former is production
feature, however it is a very niche one.

This PR turns rendering of the images in Airflow in optional feature
(only working when graphviz python bindings are installed) and
effectively turns graphviz into an optional extra (and removes it
from requirements).

This is not a breaking change technically - the CLIs to render the
DAGs is still there and IF you already have graphviz installed, it
will continue working as it did before. The only problem when it
does not work is where you do not have graphviz installed for
fresh installation and it will raise an error and inform that you need it.

Graphviz will remain to be installed for most users:

* the Airflow Image will still contain graphviz library, because
  it is added there as extra
* when previous version of Airflow has been installed already, then
  graphviz library is already installed there and Airflow will
  continue working as it did

The only change will be a new installation of new version of Airflow
from the scratch, where graphviz will need to be specified as extra
or installed separately in order to enable DAG rendering option.

Taking into account this behaviour (which only requires to install
a graphviz package), this should not be considered as a breaking
change.

Extracted from: #36537

CONTRIBUTING.rst

@@ -856,13 +856,13 @@ arangodb, asana, async, atlas, atlassian.jira, aws, azure, cassandra, celery, cg
 cncf.kubernetes, cohere, common.io, common.sql, crypto, databricks, datadog, dbt.cloud,
 deprecated_api, devel, devel_all, devel_ci, devel_hadoop, dingding, discord, doc, doc_gen, docker,
 druid, elasticsearch, exasol, fab, facebook, ftp, gcp, gcp_api, github, github_enterprise, google,
-google_auth, grpc, hashicorp, hdfs, hive, http, imap, influxdb, jdbc, jenkins, kerberos, kubernetes,
-ldap, leveldb, microsoft.azure, microsoft.mssql, microsoft.psrp, microsoft.winrm, mongo, mssql,
-mysql, neo4j, odbc, openai, openfaas, openlineage, opensearch, opsgenie, oracle, otel, pagerduty,
-pandas, papermill, password, pgvector, pinecone, pinot, postgres, presto, rabbitmq, redis, s3, s3fs,
-salesforce, samba, saml, segment, sendgrid, sentry, sftp, singularity, slack, smtp, snowflake,
-spark, sqlite, ssh, statsd, tableau, tabular, telegram, trino, vertica, virtualenv, weaviate,
-webhdfs, winrm, yandex, zendesk
+google_auth, graphviz, grpc, hashicorp, hdfs, hive, http, imap, influxdb, jdbc, jenkins, kerberos,
+kubernetes, ldap, leveldb, microsoft.azure, microsoft.mssql, microsoft.psrp, microsoft.winrm, mongo,
+mssql, mysql, neo4j, odbc, openai, openfaas, openlineage, opensearch, opsgenie, oracle, otel,
+pagerduty, pandas, papermill, password, pgvector, pinecone, pinot, postgres, presto, rabbitmq,
+redis, s3, s3fs, salesforce, samba, saml, segment, sendgrid, sentry, sftp, singularity, slack, smtp,
+snowflake, spark, sqlite, ssh, statsd, tableau, tabular, telegram, trino, vertica, virtualenv,
+weaviate, webhdfs, winrm, yandex, zendesk
   .. END EXTRAS HERE
 
 Provider packages

Dockerfile

@@ -35,7 +35,7 @@
 #                        much smaller.
 #
 # Use the same builder frontend version for everyone
-ARG AIRFLOW_EXTRAS="aiobotocore,amazon,async,celery,cncf.kubernetes,common.io,docker,elasticsearch,ftp,google,google_auth,grpc,hashicorp,http,ldap,microsoft.azure,mysql,odbc,openlineage,pandas,postgres,redis,sendgrid,sftp,slack,snowflake,ssh,statsd,virtualenv"
+ARG AIRFLOW_EXTRAS="aiobotocore,amazon,async,celery,cncf.kubernetes,common.io,docker,elasticsearch,ftp,google,google_auth,graphviz,grpc,hashicorp,http,ldap,microsoft.azure,mysql,odbc,openlineage,pandas,postgres,redis,sendgrid,sftp,slack,snowflake,ssh,statsd,virtualenv"
 ARG ADDITIONAL_AIRFLOW_EXTRAS=""
 ARG ADDITIONAL_PYTHON_DEPS=""
 

INSTALL

@@ -101,13 +101,13 @@ arangodb, asana, async, atlas, atlassian.jira, aws, azure, cassandra, celery, cg
 cncf.kubernetes, cohere, common.io, common.sql, crypto, databricks, datadog, dbt.cloud,
 deprecated_api, devel, devel_all, devel_ci, devel_hadoop, dingding, discord, doc, doc_gen, docker,
 druid, elasticsearch, exasol, fab, facebook, ftp, gcp, gcp_api, github, github_enterprise, google,
-google_auth, grpc, hashicorp, hdfs, hive, http, imap, influxdb, jdbc, jenkins, kerberos, kubernetes,
-ldap, leveldb, microsoft.azure, microsoft.mssql, microsoft.psrp, microsoft.winrm, mongo, mssql,
-mysql, neo4j, odbc, openai, openfaas, openlineage, opensearch, opsgenie, oracle, otel, pagerduty,
-pandas, papermill, password, pgvector, pinecone, pinot, postgres, presto, rabbitmq, redis, s3, s3fs,
-salesforce, samba, saml, segment, sendgrid, sentry, sftp, singularity, slack, smtp, snowflake,
-spark, sqlite, ssh, statsd, tableau, tabular, telegram, trino, vertica, virtualenv, weaviate,
-webhdfs, winrm, yandex, zendesk
+google_auth, graphviz, grpc, hashicorp, hdfs, hive, http, imap, influxdb, jdbc, jenkins, kerberos,
+kubernetes, ldap, leveldb, microsoft.azure, microsoft.mssql, microsoft.psrp, microsoft.winrm, mongo,
+mssql, mysql, neo4j, odbc, openai, openfaas, openlineage, opensearch, opsgenie, oracle, otel,
+pagerduty, pandas, papermill, password, pgvector, pinecone, pinot, postgres, presto, rabbitmq,
+redis, s3, s3fs, salesforce, samba, saml, segment, sendgrid, sentry, sftp, singularity, slack, smtp,
+snowflake, spark, sqlite, ssh, statsd, tableau, tabular, telegram, trino, vertica, virtualenv,
+weaviate, webhdfs, winrm, yandex, zendesk
 # END EXTRAS HERE
 
 # For installing Airflow in development environments - see CONTRIBUTING.rst

airflow/utils/dot_renderer.py

@@ -19,9 +19,14 @@
 """Renderer DAG (tasks and dependencies) to the graphviz object."""
 from __future__ import annotations
 
+import warnings
 from typing import TYPE_CHECKING, Any
 
-import graphviz
+try:
+    import graphviz
+except ImportError:
+    warnings.warn("Could not import graphviz. Rendering graph to the graphical format will not be possible.")
+    graphviz = None
 
 from airflow.exceptions import AirflowException
 from airflow.models.baseoperator import BaseOperator
@@ -151,6 +156,10 @@ def render_dag_dependencies(deps: dict[str, list[DagDependency]]) -> graphviz.Di
     :param deps: List of DAG dependencies
     :return: Graphviz object
     """
+    if not graphviz:
+        raise AirflowException(
+            "Could not import graphviz. Install the graphviz python package to fix this error."
+        )
     dot = graphviz.Digraph(graph_attr={"rankdir": "LR"})
 
     for dag, dependencies in deps.items():
@@ -179,6 +188,10 @@ def render_dag(dag: DAG, tis: list[TaskInstance] | None = None) -> graphviz.Digr
     :param tis: List of task instances
     :return: Graphviz object
     """
+    if not graphviz:
+        raise AirflowException(
+            "Could not import graphviz. Install the graphviz python package to fix this error."
+        )
     dot = graphviz.Digraph(
         dag.dag_id,
         graph_attr={

dev/breeze/src/airflow_breeze/global_constants.py

@@ -432,6 +432,7 @@ def get_airflow_extras():
     "ftp",
     "google",
     "google_auth",
+    "graphviz",
     "grpc",
     "hashicorp",
     "http",

docs/apache-airflow/extra-packages-ref.rst

@@ -52,6 +52,8 @@ python dependencies for the provided package.
 +---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
 | google_auth         | ``pip install 'apache-airflow[google_auth]'``       | Google auth backend                                                        |
 +---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
+| graphviz            | ``pip install 'apache-airflow[graphviz]'``          | Graphviz renderer for converting DAG to graphical output                   |
++---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
 | kerberos            | ``pip install 'apache-airflow[kerberos]'``          | Kerberos integration for Kerberized services (Hadoop, Presto, Trino)       |
 +---------------------+-----------------------------------------------------+----------------------------------------------------------------------------+
 | ldap                | ``pip install 'apache-airflow[ldap]'``              | LDAP authentication for users                                              |

docs/docker-stack/build-arg-ref.rst

@@ -91,6 +91,7 @@ List of default extras in the production Dockerfile:
 * ftp
 * google
 * google_auth
+* graphviz
 * grpc
 * hashicorp
 * http

docs/spelling_wordlist.txt

@@ -687,6 +687,7 @@ googleapiclient
 GoogleDisplayVideo
 gpu
 gpus
+graphviz
 greenlet
 Groupalia
 groupId

images/breeze/output_prod-image_build.txt

@@ -1 +1 @@
-aa383b195f2991035d5333a6f37bebaa
+f7a753d66923772bfb6250d5b87d1f51

newsfragments/36647.significant.rst

@@ -0,0 +1,23 @@
+Graphviz dependency is now an optional one, not required one.
+
+The ``graphviz`` dependency has been problematic as Airflow required dependency - especially for
+ARM-based installations. Graphviz packages require binary graphviz libraries - which is already a
+limitation, but they also require to install graphviz Python bindings to be build and installed.
+This does not work for older Linux installation but - more importantly - when you try to install
+Graphviz libraries for Python 3.8, 3.9 for ARM M1 MacBooks, the packages fail to install because
+Python bindings compilation for M1 can only work for Python 3.10+.
+
+This is not a breaking change technically - the CLIs to render the DAGs is still there and IF you
+already have graphviz installed, it will continue working as it did before. The only problem when it
+does not work is where you do not have graphviz installed it will raise an error and inform that you need it.
+
+Graphviz will remain to be installed for most users:
+
+* the Airflow Image will still contain graphviz library, because
+  it is added there as extra
+* when previous version of Airflow has been installed already, then
+  graphviz library is already installed there and Airflow will
+  continue working as it did
+
+The only change will be a new installation of new version of Airflow from the scratch, where graphviz will
+need to be specified as extra or installed separately in order to enable DAG rendering option.

setup.cfg

@@ -107,7 +107,6 @@ install_requires =
     flask-wtf>=0.15
     fsspec>=2023.10.0
     google-re2>=1.0
-    graphviz>=0.12
     gunicorn>=20.1.0
     httpx
     importlib_metadata>=1.7;python_version<"3.9"

setup.py

@@ -318,12 +318,14 @@ def write_version(filename: str = str(AIRFLOW_SOURCES_ROOT / "airflow" / "git_ve
 ]
 doc_gen = [
     "eralchemy2",
+    "graphviz>=0.12",
 ]
 flask_appbuilder_oauth = [
     "authlib>=1.0.0",
     # The version here should be upgraded at the same time as flask-appbuilder in setup.cfg
     "flask-appbuilder[oauth]==4.3.10",
 ]
+graphviz = ["graphviz>=0.12"]
 kerberos = [
     "pykerberos>=1.1.13",
     "requests_kerberos>=0.10.0",
@@ -593,6 +595,7 @@ def get_unique_dependency_list(req_list_iterable: Iterable[list[str]]):
     "deprecated_api": deprecated_api,
     "github_enterprise": flask_appbuilder_oauth,
     "google_auth": flask_appbuilder_oauth,
+    "graphviz": graphviz,
     "kerberos": kerberos,
     "ldap": ldap,
     "leveldb": leveldb,