Google Cloud Vision Operators¶
CloudVisionAddProductToProductSetOperator¶
Creates a new ReferenceImage resource.
For parameter definition, take a look at
CloudVisionAddProductToProductSetOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_ID = os.environ.get('GCP_VISION_PRODUCT_ID', 'product_explicit_id')
GCP_VISION_PRODUCT_SET_ID = os.environ.get('GCP_VISION_PRODUCT_SET_ID', 'product_set_explicit_id')
Using the operator¶
We are using the Product,
ProductSet and Retry objects from
Google libraries:
from google.api_core.retry import Retry
from google.cloud.vision_v1.types import ProductSet
from google.cloud.vision_v1.types import Product
If product_set_id and product_id was generated by the API it can be extracted from XCOM:
add_product_to_product_set = CloudVisionAddProductToProductSetOperator(
location=GCP_VISION_LOCATION,
product_set_id="{{ task_instance.xcom_pull('product_set_create') }}",
product_id="{{ task_instance.xcom_pull('product_create') }}",
retry=Retry(maximum=10.0),
timeout=5,
task_id='add_product_to_product_set',
)
Otherwise it can be specified explicitly:
add_product_to_product_set_2 = CloudVisionAddProductToProductSetOperator(
location=GCP_VISION_LOCATION,
product_set_id=GCP_VISION_PRODUCT_SET_ID,
product_id=GCP_VISION_PRODUCT_ID,
retry=Retry(maximum=10.0),
timeout=5,
task_id='add_product_to_product_set_2',
)
Templating¶
template_fields = ("location", "product_set_id", "product_id", "project_id", "gcp_conn_id")
More information¶
See Google Cloud Vision Add Product To Product Set documentation.
CloudVisionAnnotateImageOperator¶
Run image detection and annotation for an image.
For parameter definition, take a look at
CloudVisionAnnotateImageOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_ANNOTATE_IMAGE_URL = os.environ.get('GCP_VISION_ANNOTATE_IMAGE_URL', 'gs://bucket/image2.jpg')
Using the operator¶
We are using the enums and Retry objects from
Google libraries:
from google.api_core.retry import Retry
from google.cloud.vision import enums
annotate_image = CloudVisionAnnotateImageOperator(
request=annotate_image_request, retry=Retry(maximum=10.0), timeout=5, task_id='annotate_image'
)
The result can be extracted from XCOM:
annotate_image_result = BashOperator(
bash_command="echo {{ task_instance.xcom_pull('annotate_image')"
"['logoAnnotations'][0]['description'] }}",
task_id='annotate_image_result',
)
Templating¶
template_fields = ('request', 'gcp_conn_id')
More information¶
CloudVisionProductCreateOperator¶
Creates and returns a new product resource.
Possible errors regarding the Product object provided:
Returns INVALID_ARGUMENT if display_name is missing or longer than 4096 characters.
Returns INVALID_ARGUMENT if description is longer than 4096 characters.
Returns INVALID_ARGUMENT if product_category is missing or invalid.
For parameter definition, take a look at
CloudVisionProductCreateOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_ID = os.environ.get('GCP_VISION_PRODUCT_ID', 'product_explicit_id')
Using the operator¶
We are using the Product and Retry objects from Google libraries:
from google.cloud.vision_v1.types import Product
from google.api_core.retry import Retry
product = Product(display_name='My Product 1', product_category='toys')
The product_id argument can be omitted (it will be generated by the API):
product_create = CloudVisionProductCreateOperator(
location=GCP_VISION_LOCATION,
product=product,
retry=Retry(maximum=10.0),
timeout=5,
task_id='product_create',
)
Or it can be specified explicitly:
product_create_2 = CloudVisionProductCreateOperator(
product_id=GCP_VISION_PRODUCT_ID,
location=GCP_VISION_LOCATION,
product=product,
retry=Retry(maximum=10.0),
timeout=5,
task_id='product_create_2',
)
Templating¶
template_fields = ('location', 'project_id', 'product_id', 'gcp_conn_id')
More information¶
CloudVisionProductDeleteOperator¶
Permanently deletes a product and its reference images.
Metadata of the product and all its images will be deleted right away, but search queries
against ProductSets containing the product may still work until all related
caches are refreshed.
Possible errors:
Returns NOT_FOUND if the product does not exist.
For parameter definition, take a look at
CloudVisionProductDeleteOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_ID = os.environ.get('GCP_VISION_PRODUCT_ID', 'product_explicit_id')
Using the operator¶
If product_id was generated by the API it can be extracted from XCOM:
product_delete = CloudVisionProductDeleteOperator(
location=GCP_VISION_LOCATION,
product_id="{{ task_instance.xcom_pull('product_create') }}",
task_id='product_delete',
)
Otherwise it can be specified explicitly:
product_delete_2 = CloudVisionProductDeleteOperator(
location=GCP_VISION_LOCATION, product_id=GCP_VISION_PRODUCT_ID, task_id='product_delete_2'
)
Templating¶
template_fields = ('location', 'project_id', 'product_id', 'gcp_conn_id')
More information¶
CloudVisionProductGetOperator¶
Gets information associated with a Product.
Possible errors:
Returns NOT_FOUND if the Product does not exist.
For parameter definition, take a look at
CloudVisionProductGetOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_ID = os.environ.get('GCP_VISION_PRODUCT_ID', 'product_explicit_id')
Using the operator¶
If product_id was generated by the API it can be extracted from XCOM:
product_get = CloudVisionProductGetOperator(
location=GCP_VISION_LOCATION,
product_id="{{ task_instance.xcom_pull('product_create') }}",
task_id='product_get',
)
Otherwise it can be specified explicitly:
product_get_2 = CloudVisionProductGetOperator(
location=GCP_VISION_LOCATION, product_id=GCP_VISION_PRODUCT_ID, task_id='product_get_2'
)
Templating¶
template_fields = ('location', 'project_id', 'product_id', 'gcp_conn_id')
More information¶
CloudVisionProductSetCreateOperator¶
Creates a new ProductSet resource.
For parameter definition, take a look at
CloudVisionProductSetCreateOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_SET_ID = os.environ.get('GCP_VISION_PRODUCT_SET_ID', 'product_set_explicit_id')
Using the operator¶
We are using the ProductSet and Retry objects from Google libraries:
from google.cloud.vision_v1.types import ProductSet
from google.api_core.retry import Retry
product_set = ProductSet(display_name='My Product Set')
The product_set_id argument can be omitted (it will be generated by the API):
product_set_create = CloudVisionProductSetCreateOperator(
location=GCP_VISION_LOCATION,
product_set=product_set,
retry=Retry(maximum=10.0),
timeout=5,
task_id='product_set_create',
)
Or it can be specified explicitly:
product_set_create_2 = CloudVisionProductSetCreateOperator(
product_set_id=GCP_VISION_PRODUCT_SET_ID,
location=GCP_VISION_LOCATION,
product_set=product_set,
retry=Retry(maximum=10.0),
timeout=5,
task_id='product_set_create_2',
)
Templating¶
template_fields = ("location", "project_id", "product_set_id", "gcp_conn_id")
More information¶
CloudVisionProductSetDeleteOperator¶
Permanently deletes a ProductSet. Products and ReferenceImages in
the ProductSet are not deleted. The actual image files are not deleted from
Google Cloud Storage.
For parameter definition, take a look at
CloudVisionProductSetDeleteOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_SET_ID = os.environ.get('GCP_VISION_PRODUCT_SET_ID', 'product_set_explicit_id')
Using the operator¶
If product_set_id was generated by the API it can be extracted from XCOM:
product_set_delete = CloudVisionProductSetDeleteOperator(
location=GCP_VISION_LOCATION,
product_set_id="{{ task_instance.xcom_pull('product_set_create') }}",
task_id='product_set_delete',
)
Otherwise it can be specified explicitly:
product_set_delete_2 = CloudVisionProductSetDeleteOperator(
location=GCP_VISION_LOCATION, product_set_id=GCP_VISION_PRODUCT_SET_ID, task_id='product_set_delete_2'
)
Templating¶
template_fields = ('location', 'project_id', 'product_set_id', 'gcp_conn_id')
More information¶
CloudVisionProductSetGetOperator¶
Gets information associated with a ProductSet.
For parameter definition, take a look at
CloudVisionProductSetGetOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_SET_ID = os.environ.get('GCP_VISION_PRODUCT_SET_ID', 'product_set_explicit_id')
Using the operator¶
If product_set_id was generated by the API it can be extracted from XCOM:
product_set_get = CloudVisionProductSetGetOperator(
location=GCP_VISION_LOCATION,
product_set_id="{{ task_instance.xcom_pull('product_set_create') }}",
task_id='product_set_get',
)
Otherwise it can be specified explicitly:
product_set_get_2 = CloudVisionProductSetGetOperator(
location=GCP_VISION_LOCATION, product_set_id=GCP_VISION_PRODUCT_SET_ID, task_id='product_set_get_2'
)
Templating¶
template_fields = ('location', 'project_id', 'product_set_id', 'gcp_conn_id')
More information¶
CloudVisionProductSetUpdateOperator¶
Makes changes to a ProductSet resource. Only display_name can be updated
currently.
Note
To locate the ProductSet resource, its name in the form
projects/PROJECT_ID/locations/LOC_ID/productSets/PRODUCT_SET_ID is necessary.
You can provide the name directly as an attribute of the product_set object. However, you can leave it blank and provide location and product_set_id instead (and optionally project_id - if not present, the connection default will be used) and the name will be created by the operator itself.
This mechanism exists for your convenience, to allow leaving the project_id empty and having Airflow use the connection default project_id.
For parameter definition, take a look at
CloudVisionProductSetUpdateOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_SET_ID = os.environ.get('GCP_VISION_PRODUCT_SET_ID', 'product_set_explicit_id')
Using the operator¶
We are using the ProductSet object from the Google Cloud Vision library:
from google.cloud.vision_v1.types import ProductSet
product_set = ProductSet(display_name='My Product Set')
Initialization of the task:
If product_set_id was generated by the API it can be extracted from XCOM:
product_set_update = CloudVisionProductSetUpdateOperator(
location=GCP_VISION_LOCATION,
product_set_id="{{ task_instance.xcom_pull('product_set_create') }}",
product_set=ProductSet(display_name='My Product Set 2'),
task_id='product_set_update',
)
Otherwise it can be specified explicitly:
product_set_update_2 = CloudVisionProductSetUpdateOperator(
location=GCP_VISION_LOCATION,
product_set_id=GCP_VISION_PRODUCT_SET_ID,
product_set=ProductSet(display_name='My Product Set 2'),
task_id='product_set_update_2',
)
Templating¶
template_fields = ('location', 'project_id', 'product_set_id', 'gcp_conn_id')
More information¶
CloudVisionProductUpdateOperator¶
Makes changes to a Product resource. Only the display_name,
description, and labels fields can be updated right now.
If labels are updated, the change will not be reflected in queries until the next index
time.
Note
To locate the Product resource, its name in the form
projects/PROJECT_ID/locations/LOC_ID/products/PRODUCT_ID is necessary.
You can provide the name directly as an attribute of the product object. However, you can leave it blank and provide location and product_id instead (and optionally project_id - if not present, the connection default will be used) and the name will be created by the operator itself.
This mechanism exists for your convenience, to allow leaving the project_id empty and having Airflow use the connection default project_id.
Possible errors:
Returns NOT_FOUND if the Product does not exist.
Returns INVALID_ARGUMENT if display_name is present in update_mask but is missing from the request or longer than 4096 characters.
Returns INVALID_ARGUMENT if description is present in update_mask but is longer than 4096 characters.
Returns INVALID_ARGUMENT if product_category is present in update_mask.
For parameter definition, take a look at
CloudVisionProductUpdateOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_ID = os.environ.get('GCP_VISION_PRODUCT_ID', 'product_explicit_id')
Using the operator¶
We are using the Product object from the Google Cloud Vision library:
from google.cloud.vision_v1.types import Product
product = Product(display_name='My Product 1', product_category='toys')
If product_id was generated by the API it can be extracted from XCOM:
product_update = CloudVisionProductUpdateOperator(
location=GCP_VISION_LOCATION,
product_id="{{ task_instance.xcom_pull('product_create') }}",
product=Product(display_name='My Product 2', description='My updated description'),
task_id='product_update',
)
Otherwise it can be specified explicitly:
product_update_2 = CloudVisionProductUpdateOperator(
location=GCP_VISION_LOCATION,
product_id=GCP_VISION_PRODUCT_ID,
product=Product(display_name='My Product 2', description='My updated description'),
task_id='product_update_2',
)
Templating¶
template_fields = ('location', 'project_id', 'product_id', 'gcp_conn_id')
More information¶
CloudVisionReferenceImageCreateOperator¶
Creates a new ReferenceImage resource.
For parameter definition, take a look at
CloudVisionReferenceImageCreateOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_REFERENCE_IMAGE_ID = os.environ.get('GCP_VISION_REFERENCE_IMAGE_ID', 'reference_image_explicit_id')
GCP_VISION_REFERENCE_IMAGE_URL = os.environ.get('GCP_VISION_REFERENCE_IMAGE_URL', 'gs://bucket/image1.jpg')
Using the operator¶
We are using the ReferenceImage and Retry objects from Google libraries:
from google.cloud.vision_v1.types import ReferenceImage
from google.api_core.retry import Retry
reference_image = ReferenceImage(uri=GCP_VISION_REFERENCE_IMAGE_URL)
The product_set_id argument can be omitted (it will be generated by the API):
reference_image_create = CloudVisionReferenceImageCreateOperator(
location=GCP_VISION_LOCATION,
reference_image=reference_image,
product_id="{{ task_instance.xcom_pull('product_create') }}",
reference_image_id=GCP_VISION_REFERENCE_IMAGE_ID,
retry=Retry(maximum=10.0),
timeout=5,
task_id='reference_image_create',
)
Or it can be specified explicitly:
reference_image_create_2 = CloudVisionReferenceImageCreateOperator(
location=GCP_VISION_LOCATION,
reference_image=reference_image,
product_id=GCP_VISION_PRODUCT_ID,
reference_image_id=GCP_VISION_REFERENCE_IMAGE_ID,
retry=Retry(maximum=10.0),
timeout=5,
task_id='reference_image_create_2',
)
Templating¶
template_fields = (
"location",
"reference_image",
"product_id",
"reference_image_id",
"project_id",
"gcp_conn_id",
)
More information¶
See Google Cloud Vision ReferenceImage create documentation.
CloudVisionRemoveProductFromProductSetOperator¶
Creates a new ReferenceImage resource.
For parameter definition, take a look at
CloudVisionRemoveProductFromProductSetOperator
Arguments¶
Some arguments in the example DAG are taken from the OS environment variables:
GCP_VISION_LOCATION = os.environ.get('GCP_VISION_LOCATION', 'europe-west1')
GCP_VISION_PRODUCT_ID = os.environ.get('GCP_VISION_PRODUCT_ID', 'product_explicit_id')
GCP_VISION_PRODUCT_SET_ID = os.environ.get('GCP_VISION_PRODUCT_SET_ID', 'product_set_explicit_id')
Using the operator¶
We are using the Product,
ProductSet and Retry objects from
Google libraries:
from google.api_core.retry import Retry
from google.cloud.vision_v1.types import ProductSet
from google.cloud.vision_v1.types import Product
If product_set_id and product_id was generated by the API it can be extracted from XCOM:
remove_product_from_product_set = CloudVisionRemoveProductFromProductSetOperator(
location=GCP_VISION_LOCATION,
product_set_id="{{ task_instance.xcom_pull('product_set_create') }}",
product_id="{{ task_instance.xcom_pull('product_create') }}",
retry=Retry(maximum=10.0),
timeout=5,
task_id='remove_product_from_product_set',
)
Otherwise it can be specified explicitly:
remove_product_from_product_set_2 = CloudVisionRemoveProductFromProductSetOperator(
location=GCP_VISION_LOCATION,
product_set_id=GCP_VISION_PRODUCT_SET_ID,
product_id=GCP_VISION_PRODUCT_ID,
retry=Retry(maximum=10.0),
timeout=5,
task_id='remove_product_from_product_set_2',
)
Templating¶
template_fields = ("location", "product_set_id", "product_id", "project_id", "gcp_conn_id")
More information¶
See Google Cloud Vision Remove Product From Product Set documentation.