Transforms 0.2.3.dev1

.make.defaults

@@ -209,7 +209,7 @@ __check_defined = \
 # We create both local and remote tags.  Local seems to be needed when using our spark
 # base image.  Remote seems to be needed by kfp.
 .PHONY: .defaults.image
-.defaults.image:: # Must be called with a DOCKER_IMAGE= settings.
+.defaults.image:: # Must be called with a DOCKER_IMAGE_NAME= settings.
 	@# Help: Create the docker image $(DOCKER_LOCAL_IMAGE) and a tag for $(DOCKER_REMOTE_IMAGE)
 	$(call check_defined, DOCKER_IMAGE_NAME)
 	# The following touch seems to be needed to work around a docker build problem in which
@@ -222,14 +222,15 @@ __check_defined = \
 	if [ -e pyproject.toml ]; then 	\
 		touch pyproject.toml;	\
 	fi
-	$(DOCKER) build -t $(DOCKER_LOCAL_IMAGE) $(DOCKER_BUILD_EXTRA_ARGS)	\
+	$(DOCKER) build -f $(DOCKER_FILE) -t $(DOCKER_IMAGE_NAME):$(DOCKER_IMAGE_VERSION) $(DOCKER_BUILD_EXTRA_ARGS)	\
 		--platform $(DOCKER_PLATFORM)  			\
 		--build-arg EXTRA_INDEX_URL=$(EXTRA_INDEX_URL)	\
 		--build-arg BASE_IMAGE=$(BASE_IMAGE)  		\
 		--build-arg DPK_WHEEL_FILE_NAME=$(DPK_WHEEL_FILE_NAME) \
 		--build-arg BUILD_DATE=$(shell date -u +'%Y-%m-%dT%H:%M:%SZ')  	\
 		--build-arg GIT_COMMIT=$(shell git log -1 --format=%h) . 
-	$(DOCKER) tag $(DOCKER_LOCAL_IMAGE)  $(DOCKER_REMOTE_IMAGE)
+	$(DOCKER) tag $(DOCKER_IMAGE_NAME):$(DOCKER_IMAGE_VERSION) $(DOCKER_REGISTRY_ENDPOINT)/$(DOCKER_IMAGE_NAME):$(DOCKER_IMAGE_VERSION)
+
 
 # Copy a source tree in LIB_PATH, including src, pyproject.toml to LIB_NAME
 # Generally used to copy source from within the repo into a local directory for use by a Dockerfile
@@ -244,24 +245,25 @@ __check_defined = \
 	    cp -p ${LIB_PATH}/requirements.txt ${LIB_NAME};	\
 	fi
 
-
-# Build and image using the local Dockerfile and make the data-processing-lib/python
-# available in the current directory for use by the Dockerfile (i.e. to install the library).
-#.PHONY: .defaults.python-lib-src-image
-#.defaults.python-lib-src-image:: # Must be called with a DOCKER_LOCAL_IMAGE= settings.
-#	@# Help: Build the Python $(DOCKER_LOCAL_IMAGE) using the $(DOCKER_FILE), requirements.txt and data-processing-lib/python source
-#ifeq ($(USE_REPO_LIB_SRC), 1)
-#	$(MAKE) LIB_PATH=$(DPK_PYTHON_LIB_DIR) LIB_NAME=data-processing-lib-python .defaults.copy-lib
-#endif
-#	$(MAKE) DOCKER_IMAGE=$(DOCKER_LOCAL_IMAGE) .defaults.image
-#	-rm -rf data-processing-lib-python
 
 .PHONY: .default.build-lib-wheel
 .default.build-lib-wheel:
-	make -C $(REPOROOT)/data-processing-lib build-pkg-dist
+	$(MAKE) -C $(REPOROOT)/data-processing-lib build-pkg-dist
 	rm -rf data-processing-dist && mkdir data-processing-dist
 	cp $(REPOROOT)/data-processing-lib/dist/*.whl data-processing-dist
 
+
+# Build and image using the local Dockerfile 
+# Assumes wheel has already been created
+.PHONY: .defaults.lib-whl-image 
+.defaults.lib-whl-image::
+	# Must be called with a DOCKER_LOCAL_IMAGE= settings.
+	@# Help: Build the Python $(DOCKER_LOCAL_IMAGE) using the the wheel file for the library
+	@$(eval LIB_WHEEL_FILE := $(shell find data-processing-dist/*.whl))
+	$(eval LIB_WHEEL_FILE := $(shell basename $(LIB_WHEEL_FILE)))
+	$(MAKE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
+
+
 # Build and image using the local Dockerfile and make the wheel for data-processing-lib
 # available in the current directory for use by the Dockerfile (i.e. to install the library).
 .PHONY: .defaults.python-lib-whl-image 
@@ -270,28 +272,9 @@ __check_defined = \
 	@# Help: Build the Python $(DOCKER_LOCAL_IMAGE) using the the wheel file for the library
 	@$(eval LIB_WHEEL_FILE := $(shell find data-processing-dist/*.whl))
 	$(eval LIB_WHEEL_FILE := $(shell basename $(LIB_WHEEL_FILE)))
-	$(MAKE) DOCKER_IMAGE=$(DOCKER_LOCAL_IMAGE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
+	$(MAKE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
 	-rm -rf data-processing-dist
 
-# Build an image using the local Dockerfile and make the data-processing-lib/ray
-# available in the current directory for use by the Dockerfile (i.e. to install the library). 
-# Note that this looks for the ../python directory, which is currently only used in the transform projects,
-# but we add it here as a convenience to avoid duplicating a lot of this in transforms/.make.transforms.
-#.PHONY: .defaults.ray-lib-src-image
-#.defaults.ray-lib-src-image:: # Must be called with a DOCKER_LOCAL_IMAGE= settings.
-#	@# Help: Build the Ray $(DOCKER_LOCAL_IMAGE) using the $(DOCKER_FILE), requirements.txt and data-processing-libs source
-#ifeq ($(USE_REPO_LIB_SRC), 1)
-#	$(MAKE) LIB_PATH=$(DPK_PYTHON_LIB_DIR) LIB_NAME=data-processing-lib-python .defaults.copy-lib
-#	$(MAKE) LIB_PATH=$(DPK_RAY_LIB_DIR) LIB_NAME=data-processing-lib-ray .defaults.copy-lib
-#endif
-#	if [ -e ../python ]; then								\
-#		$(MAKE) LIB_PATH=../python LIB_NAME=python-transform .defaults.copy-lib;	\
-#	fi
-#	$(MAKE) DOCKER_IMAGE=$(DOCKER_LOCAL_IMAGE) .defaults.image
-#	-rm -rf data-processing-lib-python
-#	-rm -rf data-processing-lib-ray
-#	-rm -rf python-transform 
-
 
 # Build an image using the local Dockerfile and make the data-processing wheel 
 # available in the current directory for use by the Dockerfile (i.e. to install the library). 
@@ -306,7 +289,7 @@ __check_defined = \
 	if [ -e ../python ]; then								\
 		$(MAKE) LIB_PATH=../python LIB_NAME=python-transform .defaults.copy-lib;	\
 	fi
-	$(MAKE) DOCKER_IMAGE=$(DOCKER_LOCAL_IMAGE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
+	$(MAKE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
 	-rm -rf python-transform 
 	-rm -rf data-processing-dist
 
@@ -316,24 +299,6 @@ __check_defined = \
 .defaults.spark-lib-base-image: 
 	$(MAKE) -C $(DPK_SPARK_LIB_DIR) image
 
-# Note that this looks for the ../python directory, which is currently only used in the transform projects,
-# but we add it here as a convenience to avoid duplicating a lot of this in transforms/.make.transforms.
-# Must be called with a DOCKER_LOCAL_IMAGE= settings.
-#.PHONY: .defaults.spark-lib-src-image
-#.defaults.spark-lib-src-image:: .defaults.spark-lib-base-image
-#	@# Help: Build the Spark $(DOCKER_LOCAL_IMAGE) using the $(DOCKER_FILE), requirements.txt and data-processing-libs source
-#	$(MAKE) IMAGE_NAME_TO_VERIFY=$(DOCKER_SPARK_BASE_IMAGE_NAME) .defaults.verify-image-availability
-#ifeq ($(USE_REPO_LIB_SRC), 1)
-#	$(MAKE) LIB_PATH=$(DPK_PYTHON_LIB_DIR) LIB_NAME=data-processing-lib-python .defaults.copy-lib
-#	$(MAKE) LIB_PATH=$(DPK_SPARK_LIB_DIR) LIB_NAME=data-processing-lib-spark .defaults.copy-lib
-#endif
-#	if [ -e ../python ]; then								\
-#		$(MAKE) LIB_PATH=../python LIB_NAME=python-transform .defaults.copy-lib;	\
-#	fi
-#	$(MAKE) DOCKER_IMAGE=$(DOCKER_LOCAL_IMAGE) BASE_IMAGE=$(DOCKER_SPARK_BASE_IMAGE) .defaults.image
-#	-rm -rf data-processing-lib-python
-#	-rm -rf data-processing-lib-spark
-#	-rm -rf python-transform 
 
 .PHONY: .defaults.spark-lib-whl-image
 .defaults.spark-lib-whl-image:: .default.build-lib-wheel .defaults.spark-lib-base-image
@@ -345,7 +310,7 @@ __check_defined = \
 	if [ -e ../python ]; then								\
 		$(MAKE) LIB_PATH=../python LIB_NAME=python-transform .defaults.copy-lib;	\
 	fi
-	$(MAKE) DOCKER_IMAGE=$(DOCKER_LOCAL_IMAGE) BASE_IMAGE=$(DOCKER_SPARK_BASE_IMAGE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
+	$(MAKE) BASE_IMAGE=$(DOCKER_SPARK_BASE_IMAGE) DPK_WHEEL_FILE_NAME=$(LIB_WHEEL_FILE) .defaults.image
 	-rm -rf python-transform 
 	-rm -rf data-processing-dist
 

.make.versions

@@ -66,4 +66,4 @@ endif
 #
 # If you change the versions numbers, be sure to run "make set-versions" to 
 # update version numbers across the transform (e.g., pyproject.toml).
-TRANSFORMS_PKG_VERSION=0.2.3.dev0
+TRANSFORMS_PKG_VERSION=0.2.3.dev1

kfp/kfp_support_lib/shared_workflow_support/pyproject.toml

@@ -14,7 +14,7 @@ authors = [
 dependencies = [
     "requests",
     "kubernetes",
-    "data-prep-toolkit[ray]==0.2.3.dev0",
+    "data-prep-toolkit[ray]>=0.2.3.dev0",
 ]
 
 [build-system]

transforms/.make.cicd.targets

@@ -7,10 +7,15 @@
 include $(REPOROOT)/transforms/.make.transforms
 
 ######################################################################
-## Default setting for TRANSFORM_RUNTIME uses folder name-- Old layout
-TRANSFORM_PYTHON_RUNTIME_SRC_FILE=-m dpk_$(TRANSFORM_NAME).transform
-TRANSFORM_RAY_RUNTIME_SRC_FILE=-m dpk_$(TRANSFORM_NAME).ray.transform
-TRANSFORM_PYTHON_RUNTIME_SRC_FILE=-m dpk_$(TRANSFORM_NAME).spark.transform
+## Default setting for TRANSFORM_RUNTIME entry point: 
+# python -m dpk_html2parquet.ray.transform --help
+# or
+# python -m dpk_html2parquet.transform_python --help
+#
+TRANSFORM_PYTHON_SRC?="-m dpk_$(TRANSFORM_NAME).transform_python"
+TRANSFORM_RAY_SRC?="-m dpk_$(TRANSFORM_NAME).ray.transform"
+TRANSFORM_SPARK_SRC?="-m dpk_$(TRANSFORM_NAME).spark.transform"
+
 
 venv::	.defaults.create-venv
 	source venv/bin/activate && $(PIP) install -e $(REPOROOT)/data-processing-lib[ray,spark]
@@ -19,7 +24,6 @@ venv::	.defaults.create-venv
 		source venv/bin/activate && $(PIP) install -r  requirements.txt; \
 	fi; 
 
-
 test:: .transforms.test-src test-image
 
 clean:: .transforms.clean
@@ -28,62 +32,113 @@ clean:: .transforms.clean
 set-versions::
 
 ## We need to think how we want to do this going forward
-build::
 
-image::
-	@if [ -e Dockerfile ]; then \
-		$(MAKE) image-default ; \
-	else \
-		echo "Skipping image for $(shell pwd) since no Dockerfile is present"; \
+build:: image
+
+publish:
+	@if [ -e Dockerfile.python ]; then \
+		$(MAKE) DOCKER_REMOTE_IMAGE=$(DOCKER_REGISTRY_ENDPOINT)/$(TRANSFORM_NAME)-python:$(DOCKER_IMAGE_VERSION) \
+				.defaults.publish-image ; \
+	fi
+	@if [ -e Dockerfile.ray ]; then \
+		$(MAKE) DOCKER_REMOTE_IMAGE=$(DOCKER_REGISTRY_ENDPOINT)/$(TRANSFORM_NAME)-ray:$(DOCKER_IMAGE_VERSION) \
+				.defaults.publish-image ; \
+	fi
+	@if [ -e Dockerfile.spark ]; then \
+		$(MAKE) DOCKER_REMOTE_IMAGE=$(DOCKER_REGISTRY_ENDPOINT)/$(TRANSFORM_NAME)-spark:$(DOCKER_IMAGE_VERSION) \
+				.defaults.publish-image ; \
+	fi
+
+test-image-sequence:: .defaults.lib-whl-image .transforms.test-image-help .transforms.clean
+
+test-image:: .default.build-lib-wheel
+	@if [ -e Dockerfile.python ]; then \
+		$(MAKE) DOCKER_FILE=Dockerfile.python \
+				TRANSFORM_RUNTIME_SRC_FILE=$(TRANSFORM_PYTHON_SRC) \
+				DOCKER_IMAGE_NAME=$(TRANSFORM_NAME)-python \
+				test-image-sequence ; \
+	fi
+	@if [ -e Dockerfile.ray ]; then \
+		$(MAKE) DOCKER_FILE=Dockerfile.ray \
+				TRANSFORM_RUNTIME_SRC_FILE=$(TRANSFORM_RAY_SRC) \
+				DOCKER_IMAGE_NAME=$(TRANSFORM_NAME)-ray \
+				BASE_IMAGE=$(RAY_BASE_IMAGE)  \
+				test-image-sequence ; \
 	fi
+	@if [ -e Dockerfile.spark ]; then \
+		$(MAKE) DOCKER_FILE=Dockerfile.spark \
+				TRANSFORM_RUNTIME_SRC_FILE=$(TRANSFORM_SPARK_SRC) \
+				DOCKER_IMAGE_NAME=$(TRANSFORM_NAME)-spark \
+				BASE_IMAGE=$(SPARK_BASE_IMAGE)  \
+				test-image-sequence ; \
+	fi
+	-rm -rf data-processing-dist
+
 
-publish::
-	@if [ -e Dockerfile ]; then \
-		$(MAKE) publish-default ; \
-	else \
-		echo "Skipping publish for $(shell pwd) since no Dockerfile is present"; \
+image-python:
+	@if [ -e Dockerfile.python ]; then \
+		$(MAKE) DOCKER_FILE=Dockerfile.python \
+				DOCKER_IMAGE_NAME=$(TRANSFORM_NAME)-python \
+				.defaults.lib-whl-image ; \
 	fi
 
-publish-image::
-	@if [ -e Dockerfile ]; then \
-		$(MAKE) publish-image-default ; \
-	else \
-		echo "Skipping publish-image for $(shell pwd) since no Dockerfile is present"; \
+image-ray:
+	@if [ -e Dockerfile.ray ]; then \
+		$(MAKE) DOCKER_FILE=Dockerfile.ray \
+				DOCKER_IMAGE_NAME=$(TRANSFORM_NAME)-ray \
+				BASE_IMAGE=$(RAY_BASE_IMAGE)  \
+				.defaults.lib-whl-image ; \
 	fi
 
-test-image::
-	@if [ -e Dockerfile ]; then \
-		$(MAKE) test-image-default ; \
-	else \
-		echo "Skipping test-image for $(shell pwd) since no Dockerfile is present"; \
+image-spark:
+	@if [ -e Dockerfile.spark ]; then \
+		$(MAKE) DOCKER_FILE=Dockerfile.spark \
+				DOCKER_IMAGE_NAME=$(TRANSFORM_NAME)-spark \
+				BASE_IMAGE=$(SPARK_BASE_IMAGE)  \
+				.defaults.lib-whl-image ; \
 	fi
 
+image:: .default.build-lib-wheel
+	## Build all possible images unless a specific runtime is specified
+	@if [ -z "$(BUILD_SPECIFIC_RUNTIME)" ] || [ "$(BUILD_SPECIFIC_RUNTIME)" == "python" ]; then \
+		$(MAKE) image-python ; \
+	fi
+	@if [ -z "$(BUILD_SPECIFIC_RUNTIME)" ] || [ "$(BUILD_SPECIFIC_RUNTIME)" == "ray" ]; then \
+		$(MAKE) image-ray ; \
+	fi
+	@if [ -z "$(BUILD_SPECIFIC_RUNTIME)" ] || [ "$(BUILD_SPECIFIC_RUNTIME)" == "spark" ]; then \
+		$(MAKE) image-spark ; \
+	fi
+	-rm -rf data-processing-dist
+
 test-src:: .transforms.test-src
 
 setup:: .transforms.setup
 
-publish-default:: publish-image
-
-publish-image-default:: .defaults.publish-image
-
-test-image-default:: image .transforms.test-image-help .defaults.test-image-pytest .transforms.clean
-
-build-lib-wheel:
-	make -C $(REPOROOT)/data-processing-lib build-pkg-dist
-
-image-default:: build-lib-wheel
-	@$(eval LIB_WHEEL_FILE := $(shell find $(REPOROOT)/data-processing-lib/dist/*.whl))
-	rm -fr dist && mv $(REPOROOT)/data-processing-lib/dist .
-	$(eval WHEEL_FILE_NAME := $(shell basename $(LIB_WHEEL_FILE)))
-	$(DOCKER) build -t $(DOCKER_IMAGE_NAME) $(DOCKER_BUILD_EXTRA_ARGS)	\
-		--platform $(DOCKER_PLATFORM)  			\
-		--build-arg EXTRA_INDEX_URL=$(EXTRA_INDEX_URL)	\
-		--build-arg BASE_IMAGE=$(RAY_BASE_IMAGE)  		\
-		--build-arg BUILD_DATE=$(shell date -u +'%Y-%m-%dT%H:%M:%SZ')  	\
-		--build-arg WHEEL_FILE_NAME=$(WHEEL_FILE_NAME) \
-		--build-arg TRANSFORM_NAME=$(TRANSFORM_NAME) \
-		--build-arg GIT_COMMIT=$(shell git log -1 --format=%h) .
-	$(DOCKER) tag $(DOCKER_LOCAL_IMAGE)  $(DOCKER_REMOTE_IMAGE)
-	rm -fr dist
+kind-load-image:: .transforms.kind-load-image
+
+.PHONY: workflow-vent
+workflow-venv:
+	if [ -e kfp_ray ]; then                 \
+	    $(MAKE) -C kfp_ray TRANSFORM_NAME=$(TRANSFORM_NAME) workflow-venv;   \
+	fi
+
+.PHONY: workflow-test
+workflow-test:
+	if [ -e kfp_ray ]; then                 \
+	    $(MAKE) -C kfp_ray TRANSFORM_NAME=$(TRANSFORM_NAME) workflow-test;   \
+	fi
+
+.PHONY: workflow-upload
+workflow-upload:
+	if [ -e kfp_ray ]; then                 \
+	    $(MAKE) -C kfp_ray TRANSFORM_NAME=$(TRANSFORM_NAME) workflow-upload; \
+	fi
+
+.PHONY: workflow-build
+workflow-build:
+	if [ -e kfp_ray ]; then                 \
+	    $(MAKE) -C  kfp_ray TRANSFORM_NAME=$(TRANSFORM_NAME) workflow-build; \
+	fi
 
 

transforms/Makefile

@@ -107,7 +107,7 @@ build-pkg-dist:
 	-rm -fr src
 	mkdir src
 	# Copy all the src folders recursively (not clear if they have subfolders)
-	for x in $(shell find . | grep '[ray| python]/src$$') ; do \
+	for x in $(shell find . | grep '[ray| python | spark]/src$$') ; do \
 		echo $$x ; \
 		if [ -d "$$x" ]; then \
 		   cp -r $$x/* src ; \

transforms/code/code2parquet/python/requirements.txt

@@ -1,3 +1,3 @@
-data-prep-toolkit==0.2.3.dev0
+data-prep-toolkit>=0.2.3.dev0
 parameterized
 pandas

transforms/code/code2parquet/ray/pyproject.toml

@@ -10,7 +10,7 @@ authors = [
     { name = "Boris Lublinsky", email = "[email protected]" },
 ]
 dependencies = [
-    "data-prep-toolkit[ray]==0.2.3.dev0",
+    "data-prep-toolkit[ray]>=0.2.3.dev0",
     "dpk-code2parquet-transform-python==0.2.3.dev0",
     "parameterized",
     "pandas",

transforms/code/code_profiler/python/requirements.txt

@@ -1,4 +1,4 @@
-data-prep-toolkit==0.2.3.dev0
+data-prep-toolkit>=0.2.3.dev0
 parameterized
 pandas
 aiolimiter==1.1.0

transforms/code/code_profiler/ray/pyproject.toml

@@ -10,7 +10,7 @@ authors = [
 ]
 dependencies = [
 	"dpk-code-profiler-transform-python==0.2.3.dev0",
-    "data-prep-toolkit[ray]==0.2.3.dev0",
+    "data-prep-toolkit[ray]>=0.2.3.dev0",
 	]
 
 [build-system]

transforms/code/code_quality/python/requirements.txt

@@ -1,3 +1,3 @@
-data-prep-toolkit==0.2.3.dev0
+data-prep-toolkit>=0.2.3.dev0
 bs4==0.0.2
 transformers==4.38.2

transforms/code/code_quality/ray/pyproject.toml

@@ -10,7 +10,7 @@ authors = [
 ]
 dependencies = [
     "dpk-code-quality-transform-python==0.2.3.dev0",
-    "data-prep-toolkit[ray]==0.2.3.dev0",
+    "data-prep-toolkit[ray]>=0.2.3.dev0",
 ]
 
 [build-system]

transforms/code/header_cleanser/python/requirements.txt

@@ -1,3 +1,3 @@
-data-prep-toolkit==0.2.3.dev0
+data-prep-toolkit>=0.2.3.dev0
 scancode-toolkit==32.1.0 ; platform_system != 'Darwin'