diff --git a/.make.versions b/.make.versions index ed36fe8c8..e3a8e8239 100644 --- a/.make.versions +++ b/.make.versions @@ -19,7 +19,7 @@ DPK_MINOR_VERSION=2 DPK_MICRO_VERSION=2 # The suffix is generally always set in the main/development branch and only nulled out when creating release branches. # It can be manually incremented, for example, to allow publishing a new intermediate version wheel to pypi. -DPK_VERSION_SUFFIX=.dev2 +DPK_VERSION_SUFFIX= DPK_VERSION=$(DPK_MAJOR_VERSION).$(DPK_MINOR_VERSION).$(DPK_MICRO_VERSION)$(DPK_VERSION_SUFFIX) @@ -39,7 +39,7 @@ DPK_LIB_KFP_SHARED=$(DPK_VERSION) KFP_DOCKER_VERSION=$(DOCKER_IMAGE_VERSION) KFP_DOCKER_VERSION_v2=$(DOCKER_IMAGE_VERSION) -DPK_CONNECTOR_VERSION=0.2.3.dev0 +DPK_CONNECTOR_VERSION=0.2.3 ################## ################## ################## ################## ################## ################## # Begin versions that the repo depends on. @@ -59,3 +59,11 @@ else WORKFLOW_SUPPORT_LIB=kfp_v1_workflow_support endif +################################################################################ +# This defines the transforms' package version number as would be used +# when publishing the wheel. In general, only the micro version +# number should be advanced relative to the DPK_VERSION. +# +# 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=0.2.2 diff --git a/data-connector-lib/pyproject.toml b/data-connector-lib/pyproject.toml index 4fcc97ed9..d3d213946 100644 --- a/data-connector-lib/pyproject.toml +++ b/data-connector-lib/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_connector" -version = "0.2.3.dev1" +version = "0.2.3" requires-python = ">=3.10,<3.13" keywords = [ "data", diff --git a/data-processing-lib/pyproject.toml b/data-processing-lib/pyproject.toml index 2e827ea82..36e4e155f 100644 --- a/data-processing-lib/pyproject.toml +++ b/data-processing-lib/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_toolkit" -version = "0.2.2.dev2" +version = "0.2.2" keywords = ["data", "data preprocessing", "data preparation", "llm", "generative", "ai", "fine-tuning", "llmapps" ] requires-python = ">=3.10,<3.13" description = "Data Preparation Toolkit Library for Ray and Python" diff --git a/data-processing-lib/spark/pyproject.toml b/data-processing-lib/spark/pyproject.toml index 89b4d9bf8..c0be43920 100644 --- a/data-processing-lib/spark/pyproject.toml +++ b/data-processing-lib/spark/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_toolkit_spark" -version = "0.2.2.dev2" +version = "0.2.2" keywords = ["data", "data preprocessing", "data preparation", "llm", "generative", "ai", "fine-tuning", "llmapps" ] requires-python = ">=3.10,<3.13" description = "Data Preparation Toolkit Library for Spark" @@ -11,7 +11,7 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "data-prep-toolkit==0.2.2.dev2", + "data-prep-toolkit==0.2.2", "pyspark>=3.5.2", "psutil>=6.0.0", "PyYAML>=6.0.2" diff --git a/kfp/kfp_ray_components/createRayClusterComponent.yaml b/kfp/kfp_ray_components/createRayClusterComponent.yaml index 30b0b66d8..78976a97c 100644 --- a/kfp/kfp_ray_components/createRayClusterComponent.yaml +++ b/kfp/kfp_ray_components/createRayClusterComponent.yaml @@ -11,7 +11,7 @@ inputs: implementation: container: - image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" + image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # command is a list of strings (command-line arguments). # The YAML language has two syntaxes for lists and you can use either of them. # Here we use the "flow syntax" - comma-separated strings inside square brackets. diff --git a/kfp/kfp_ray_components/deleteRayClusterComponent.yaml b/kfp/kfp_ray_components/deleteRayClusterComponent.yaml index 44e199c47..c75554d5f 100644 --- a/kfp/kfp_ray_components/deleteRayClusterComponent.yaml +++ b/kfp/kfp_ray_components/deleteRayClusterComponent.yaml @@ -9,7 +9,7 @@ inputs: implementation: container: - image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" + image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # command is a list of strings (command-line arguments). # The YAML language has two syntaxes for lists and you can use either of them. # Here we use the "flow syntax" - comma-separated strings inside square brackets. diff --git a/kfp/kfp_ray_components/executeRayJobComponent.yaml b/kfp/kfp_ray_components/executeRayJobComponent.yaml index 7ab517bff..2e02c3adf 100644 --- a/kfp/kfp_ray_components/executeRayJobComponent.yaml +++ b/kfp/kfp_ray_components/executeRayJobComponent.yaml @@ -12,7 +12,7 @@ inputs: implementation: container: - image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" + image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # command is a list of strings (command-line arguments). # The YAML language has two syntaxes for lists and you can use either of them. # Here we use the "flow syntax" - comma-separated strings inside square brackets. diff --git a/kfp/kfp_ray_components/executeRayJobComponent_multi_s3.yaml b/kfp/kfp_ray_components/executeRayJobComponent_multi_s3.yaml index 9b98912f0..37c0198bf 100644 --- a/kfp/kfp_ray_components/executeRayJobComponent_multi_s3.yaml +++ b/kfp/kfp_ray_components/executeRayJobComponent_multi_s3.yaml @@ -13,7 +13,7 @@ inputs: implementation: container: - image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" + image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # command is a list of strings (command-line arguments). # The YAML language has two syntaxes for lists and you can use either of them. # Here we use the "flow syntax" - comma-separated strings inside square brackets. diff --git a/kfp/kfp_ray_components/executeSubWorkflowComponent.yaml b/kfp/kfp_ray_components/executeSubWorkflowComponent.yaml index 6b261a003..ec82e9484 100644 --- a/kfp/kfp_ray_components/executeSubWorkflowComponent.yaml +++ b/kfp/kfp_ray_components/executeSubWorkflowComponent.yaml @@ -27,7 +27,7 @@ outputs: implementation: container: - image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" + image: "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # command is a list of strings (command-line arguments). # The YAML language has two syntaxes for lists, and you can use either of them. # Here we use the "flow syntax" - comma-separated strings inside square brackets. diff --git a/kfp/kfp_support_lib/kfp_v1_workflow_support/pyproject.toml b/kfp/kfp_support_lib/kfp_v1_workflow_support/pyproject.toml index d7058f2ae..daa903aaf 100644 --- a/kfp/kfp_support_lib/kfp_v1_workflow_support/pyproject.toml +++ b/kfp/kfp_support_lib/kfp_v1_workflow_support/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_toolkit_kfp_v1" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Data Preparation Kit Library. KFP support" license = {text = "Apache-2.0"} @@ -13,7 +13,7 @@ authors = [ ] dependencies = [ "kfp==1.8.22", - "data-prep-toolkit-kfp-shared==0.2.2.dev2", + "data-prep-toolkit-kfp-shared==0.2.2", ] [build-system] diff --git a/kfp/kfp_support_lib/kfp_v2_workflow_support/pyproject.toml b/kfp/kfp_support_lib/kfp_v2_workflow_support/pyproject.toml index 04b6bc7a2..61f54663f 100644 --- a/kfp/kfp_support_lib/kfp_v2_workflow_support/pyproject.toml +++ b/kfp/kfp_support_lib/kfp_v2_workflow_support/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_toolkit_kfp_v2" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Data Preparation Kit Library. KFP support" license = {text = "Apache-2.0"} @@ -14,7 +14,7 @@ authors = [ dependencies = [ "kfp==2.8.0", "kfp-kubernetes==1.2.0", - "data-prep-toolkit-kfp-shared==0.2.2.dev2", + "data-prep-toolkit-kfp-shared==0.2.2", ] [build-system] diff --git a/kfp/kfp_support_lib/shared_workflow_support/pyproject.toml b/kfp/kfp_support_lib/shared_workflow_support/pyproject.toml index df27ad1cf..3ba7491bc 100644 --- a/kfp/kfp_support_lib/shared_workflow_support/pyproject.toml +++ b/kfp/kfp_support_lib/shared_workflow_support/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_toolkit_kfp_shared" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Data Preparation Kit Library. KFP support" license = {text = "Apache-2.0"} @@ -14,7 +14,7 @@ authors = [ dependencies = [ "requests", "kubernetes", - "data-prep-toolkit[ray]==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/release-notes.md b/release-notes.md index 15f23c542..4b7b8d553 100644 --- a/release-notes.md +++ b/release-notes.md @@ -1,5 +1,42 @@ # Data Prep Kit Release notes +## Release 0.2.2 - 11/25/2024 + +### General +1. Update RAG example to use granite model +1. Updated transforms with Docling 2 +1. Added single package for dpk with extra for \[spark\] and \[ray\] +1. Added single package for transforms with extra for \[all\] or \[individual-transform-name\] + + +### data-prep-toolkit libraries (python, ray, spark) + +1. Fix metadata logging even when actors crash +1. Add multilock for ray workers downloads/cleanup +1. Multiple updates to spark runtime +1. Added support for python 3.12 +1. refactoring of data access code + + +### KFP Workloads + +1. Modify superpipeline params type Str/json +1. Set kuberay apiserver version +1. Add Super pipeline for code transforms + + +### Transforms + +1. Enhance pdf2parquet with docling2 support for extracting HTML, DOCS, etc. +1. Added web2parquet transform +1. Added HAP transform + +### HTTP Connector 0.2.3 + +1. Enhanced parameter/configuration allows the user to customize crawler settings +1. implement subdomain focus feature in data-prep-connector + + ## Release 0.2.2- HTTP Connector Module - 10/23/2024 ### General diff --git a/transforms/code/code2parquet/kfp_ray/code2parquet_wf.py b/transforms/code/code2parquet/kfp_ray/code2parquet_wf.py index f3f491e4b..3e5f262b9 100644 --- a/transforms/code/code2parquet/kfp_ray/code2parquet_wf.py +++ b/transforms/code/code2parquet/kfp_ray/code2parquet_wf.py @@ -25,7 +25,7 @@ # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/code2parquet/python/pyproject.toml b/transforms/code/code2parquet/python/pyproject.toml index 5e6f41bb2..d4f8c11cf 100644 --- a/transforms/code/code2parquet/python/pyproject.toml +++ b/transforms/code/code2parquet/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_code2parquet_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "code2parquet Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/code/code2parquet/python/requirements.txt b/transforms/code/code2parquet/python/requirements.txt index bbb84b749..4a217ff8c 100644 --- a/transforms/code/code2parquet/python/requirements.txt +++ b/transforms/code/code2parquet/python/requirements.txt @@ -1,3 +1,3 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 parameterized pandas diff --git a/transforms/code/code2parquet/ray/pyproject.toml b/transforms/code/code2parquet/ray/pyproject.toml index 15a4be4c1..98b2e3a65 100644 --- a/transforms/code/code2parquet/ray/pyproject.toml +++ b/transforms/code/code2parquet/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_code2parquet_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "code2parquet Ray Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "data-prep-toolkit[ray]==0.2.2.dev2", - "dpk-code2parquet-transform-python==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", + "dpk-code2parquet-transform-python==0.2.2", "parameterized", "pandas", ] diff --git a/transforms/code/code_profiler/python/pyproject.toml b/transforms/code/code_profiler/python/pyproject.toml index 492603d54..d3c2c2196 100644 --- a/transforms/code/code_profiler/python/pyproject.toml +++ b/transforms/code/code_profiler/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_code_profiler_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Code Profiler Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/code/code_profiler/python/requirements.txt b/transforms/code/code_profiler/python/requirements.txt index 8608c6d6e..31509b291 100644 --- a/transforms/code/code_profiler/python/requirements.txt +++ b/transforms/code/code_profiler/python/requirements.txt @@ -1,4 +1,4 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 parameterized pandas aiolimiter==1.1.0 diff --git a/transforms/code/code_profiler/ray/pyproject.toml b/transforms/code/code_profiler/ray/pyproject.toml index 933152e3f..0c9457efc 100644 --- a/transforms/code/code_profiler/ray/pyproject.toml +++ b/transforms/code/code_profiler/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_code_profiler_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Code Profiler Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Pankaj Thorat", email = "pankaj.thorat@ibm.com" }, ] dependencies = [ - "dpk-code-profiler-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-code-profiler-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/code/code_quality/kfp_ray/code_quality_wf.py b/transforms/code/code_quality/kfp_ray/code_quality_wf.py index 6a4ccec1b..7f5aa9768 100644 --- a/transforms/code/code_quality/kfp_ray/code_quality_wf.py +++ b/transforms/code/code_quality/kfp_ray/code_quality_wf.py @@ -24,7 +24,7 @@ task_image = "quay.io/dataprep1/data-prep-kit/code_quality-ray:latest" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/code_quality/python/pyproject.toml b/transforms/code/code_quality/python/pyproject.toml index 5f201c8ae..d7b452d6b 100644 --- a/transforms/code/code_quality/python/pyproject.toml +++ b/transforms/code/code_quality/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_code_quality_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Code Quality Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/code/code_quality/python/requirements.txt b/transforms/code/code_quality/python/requirements.txt index 0bd936ef2..a50ddff5c 100644 --- a/transforms/code/code_quality/python/requirements.txt +++ b/transforms/code/code_quality/python/requirements.txt @@ -1,3 +1,3 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 bs4==0.0.2 transformers==4.38.2 diff --git a/transforms/code/code_quality/ray/pyproject.toml b/transforms/code/code_quality/ray/pyproject.toml index 290429f95..ea6aad8ae 100644 --- a/transforms/code/code_quality/ray/pyproject.toml +++ b/transforms/code/code_quality/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_code_quality_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Code Quality Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Shivdeep Singh", email = "shivdeep.singh@ibm.com" }, ] dependencies = [ - "dpk-code-quality-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-code-quality-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/code/header_cleanser/kfp_ray/header_cleanser_wf.py b/transforms/code/header_cleanser/kfp_ray/header_cleanser_wf.py index 9bb315569..5049a9c11 100644 --- a/transforms/code/header_cleanser/kfp_ray/header_cleanser_wf.py +++ b/transforms/code/header_cleanser/kfp_ray/header_cleanser_wf.py @@ -24,7 +24,7 @@ task_image = "quay.io/dataprep1/data-prep-kit/header_cleanser-ray:latest" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/header_cleanser/python/pyproject.toml b/transforms/code/header_cleanser/python/pyproject.toml index ecaf4d7bb..2dadeaf02 100644 --- a/transforms/code/header_cleanser/python/pyproject.toml +++ b/transforms/code/header_cleanser/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_header_cleanser_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "License and Copyright Removal Transform for Python" license = {text = "Apache-2.0"} diff --git a/transforms/code/header_cleanser/python/requirements.txt b/transforms/code/header_cleanser/python/requirements.txt index c2d0d8793..fd3fc0de4 100644 --- a/transforms/code/header_cleanser/python/requirements.txt +++ b/transforms/code/header_cleanser/python/requirements.txt @@ -1,3 +1,3 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 scancode-toolkit==32.1.0 ; platform_system != 'Darwin' diff --git a/transforms/code/header_cleanser/ray/pyproject.toml b/transforms/code/header_cleanser/ray/pyproject.toml index adff71cfc..471ce1d5e 100644 --- a/transforms/code/header_cleanser/ray/pyproject.toml +++ b/transforms/code/header_cleanser/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_header_cleanser_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "License and copyright removal Transform for Ray" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Yash kalathiya", email = "yashkalathiya164@gmail.com" }, ] dependencies = [ - "dpk-header-cleanser-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-header-cleanser-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", "scancode-toolkit==32.1.0", ] diff --git a/transforms/code/license_select/kfp_ray/license_select_wf.py b/transforms/code/license_select/kfp_ray/license_select_wf.py index 7dba0d9d1..9bdcc6e96 100644 --- a/transforms/code/license_select/kfp_ray/license_select_wf.py +++ b/transforms/code/license_select/kfp_ray/license_select_wf.py @@ -25,7 +25,7 @@ # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/license_select/python/pyproject.toml b/transforms/code/license_select/python/pyproject.toml index 30f2f001e..b445c6b09 100644 --- a/transforms/code/license_select/python/pyproject.toml +++ b/transforms/code/license_select/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_license_select_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "License Select Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/code/license_select/python/requirements.txt b/transforms/code/license_select/python/requirements.txt index 368287e5d..880c7c2c7 100644 --- a/transforms/code/license_select/python/requirements.txt +++ b/transforms/code/license_select/python/requirements.txt @@ -1 +1 @@ -data-prep-toolkit==0.2.2.dev2 \ No newline at end of file +data-prep-toolkit==0.2.2 \ No newline at end of file diff --git a/transforms/code/license_select/ray/pyproject.toml b/transforms/code/license_select/ray/pyproject.toml index 815121787..b2c56e940 100644 --- a/transforms/code/license_select/ray/pyproject.toml +++ b/transforms/code/license_select/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_license_select_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "License Select Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Mark Lewis", email = "mark_lewis@uk.ibm.com" }, ] dependencies = [ - "dpk-license-select-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-license-select-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/code/malware/kfp_ray/malware_wf.py b/transforms/code/malware/kfp_ray/malware_wf.py index bede80b88..89eb9d730 100644 --- a/transforms/code/malware/kfp_ray/malware_wf.py +++ b/transforms/code/malware/kfp_ray/malware_wf.py @@ -24,7 +24,7 @@ task_image = "quay.io/dataprep1/data-prep-kit/malware-ray:latest" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/malware/python/pyproject.toml b/transforms/code/malware/python/pyproject.toml index 22d92fd8c..2a7d1a5b9 100644 --- a/transforms/code/malware/python/pyproject.toml +++ b/transforms/code/malware/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_malware_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Malware Python Transform" license = {text = "Apache-2.0"} @@ -9,7 +9,7 @@ authors = [ { name = "Takuya Goto", email = "tkyg@jp.ibm.com" }, ] dependencies = [ - "data-prep-toolkit==0.2.2.dev2", + "data-prep-toolkit==0.2.2", "clamd==1.0.2", ] diff --git a/transforms/code/malware/ray/pyproject.toml b/transforms/code/malware/ray/pyproject.toml index 791b8d253..36901b88c 100644 --- a/transforms/code/malware/ray/pyproject.toml +++ b/transforms/code/malware/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_malware_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Malware Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Takuya Goto", email = "tkyg@jp.ibm.com" }, ] dependencies = [ - "dpk-malware-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-malware-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/code/proglang_select/kfp_ray/proglang_select_wf.py b/transforms/code/proglang_select/kfp_ray/proglang_select_wf.py index 11f001bfa..bb114e3d6 100644 --- a/transforms/code/proglang_select/kfp_ray/proglang_select_wf.py +++ b/transforms/code/proglang_select/kfp_ray/proglang_select_wf.py @@ -24,7 +24,7 @@ task_image = "quay.io/dataprep1/data-prep-kit/proglang_select-ray:latest" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/proglang_select/python/pyproject.toml b/transforms/code/proglang_select/python/pyproject.toml index 186198d83..e20a62f7c 100644 --- a/transforms/code/proglang_select/python/pyproject.toml +++ b/transforms/code/proglang_select/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_proglang_select_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Programming Language Selection Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/code/proglang_select/python/requirements.txt b/transforms/code/proglang_select/python/requirements.txt index 368287e5d..880c7c2c7 100644 --- a/transforms/code/proglang_select/python/requirements.txt +++ b/transforms/code/proglang_select/python/requirements.txt @@ -1 +1 @@ -data-prep-toolkit==0.2.2.dev2 \ No newline at end of file +data-prep-toolkit==0.2.2 \ No newline at end of file diff --git a/transforms/code/proglang_select/ray/pyproject.toml b/transforms/code/proglang_select/ray/pyproject.toml index bf3e5f9f4..d2e820d99 100644 --- a/transforms/code/proglang_select/ray/pyproject.toml +++ b/transforms/code/proglang_select/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_proglang_select_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Programming Language Selection Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Shivdeep Singh", email = "shivdeep.singh@ibm.com" }, ] dependencies = [ - "dpk-proglang-select-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-proglang-select-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/code/repo_level_ordering/kfp_ray/repo_level_order_wf.py b/transforms/code/repo_level_ordering/kfp_ray/repo_level_order_wf.py index 38a829fab..fa739bfd0 100644 --- a/transforms/code/repo_level_ordering/kfp_ray/repo_level_order_wf.py +++ b/transforms/code/repo_level_ordering/kfp_ray/repo_level_order_wf.py @@ -24,7 +24,7 @@ EXEC_SCRIPT_NAME: str = "repo_level_order_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/code/repo_level_ordering/ray/pyproject.toml b/transforms/code/repo_level_ordering/ray/pyproject.toml index 80440a362..5fb561d67 100644 --- a/transforms/code/repo_level_ordering/ray/pyproject.toml +++ b/transforms/code/repo_level_ordering/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_repo_level_order_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "repo_level_order Ray Transform" license = {text = "Apache-2.0"} @@ -11,7 +11,7 @@ authors = [ { name = "Shanmukha Guttula", email = "shagutt1@in.ibm.com" }, ] dependencies = [ - "data-prep-toolkit[ray]==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", "networkx==3.3", "colorlog==6.8.2", "func-timeout==4.3.5", diff --git a/transforms/language/doc_chunk/kfp_ray/doc_chunk_multiple_wf.py b/transforms/language/doc_chunk/kfp_ray/doc_chunk_multiple_wf.py index 7e30ee8b8..1fd927356 100644 --- a/transforms/language/doc_chunk/kfp_ray/doc_chunk_multiple_wf.py +++ b/transforms/language/doc_chunk/kfp_ray/doc_chunk_multiple_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "doc_chunk_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/doc_chunk/kfp_ray/doc_chunk_wf.py b/transforms/language/doc_chunk/kfp_ray/doc_chunk_wf.py index 387c3bda7..e128df8b0 100644 --- a/transforms/language/doc_chunk/kfp_ray/doc_chunk_wf.py +++ b/transforms/language/doc_chunk/kfp_ray/doc_chunk_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "doc_chunk_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/doc_chunk/python/requirements.txt b/transforms/language/doc_chunk/python/requirements.txt index c24d0c3e2..144688f63 100644 --- a/transforms/language/doc_chunk/python/requirements.txt +++ b/transforms/language/doc_chunk/python/requirements.txt @@ -1,4 +1,4 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 docling-core==2.3.0 pydantic>=2.0.0,<2.10.0 llama-index-core>=0.11.22,<0.12.0 diff --git a/transforms/language/doc_chunk/ray/pyproject.toml b/transforms/language/doc_chunk/ray/pyproject.toml index 29b594fac..ed8f5d60b 100644 --- a/transforms/language/doc_chunk/ray/pyproject.toml +++ b/transforms/language/doc_chunk/ray/pyproject.toml @@ -12,7 +12,7 @@ authors = [ ] dependencies = [ "dpk-doc-chunk-transform-python==0.3.0", - "data-prep-toolkit[ray]==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/language/doc_quality/kfp_ray/doc_quality_multiple_wf.py b/transforms/language/doc_quality/kfp_ray/doc_quality_multiple_wf.py index 436d93ff3..f103b7269 100644 --- a/transforms/language/doc_quality/kfp_ray/doc_quality_multiple_wf.py +++ b/transforms/language/doc_quality/kfp_ray/doc_quality_multiple_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "doc_quality_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/doc_quality/kfp_ray/doc_quality_wf.py b/transforms/language/doc_quality/kfp_ray/doc_quality_wf.py index f39fd7e39..0ca4fb865 100644 --- a/transforms/language/doc_quality/kfp_ray/doc_quality_wf.py +++ b/transforms/language/doc_quality/kfp_ray/doc_quality_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "doc_quality_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/doc_quality/python/pyproject.toml b/transforms/language/doc_quality/python/pyproject.toml index 72406e945..f3abe0337 100644 --- a/transforms/language/doc_quality/python/pyproject.toml +++ b/transforms/language/doc_quality/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_doc_quality_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Document Quality Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/language/doc_quality/python/requirements.txt b/transforms/language/doc_quality/python/requirements.txt index 2993d6b12..de76cb006 100644 --- a/transforms/language/doc_quality/python/requirements.txt +++ b/transforms/language/doc_quality/python/requirements.txt @@ -1,2 +1,2 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 diff --git a/transforms/language/doc_quality/ray/pyproject.toml b/transforms/language/doc_quality/ray/pyproject.toml index dc13d5f94..c1433d29b 100644 --- a/transforms/language/doc_quality/ray/pyproject.toml +++ b/transforms/language/doc_quality/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_doc_quality_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Document Quality Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Daiki Tsuzuku", email = "dtsuzuku@jp.ibm.com" } ] dependencies = [ - "dpk-doc_quality-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-doc_quality-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/language/html2parquet/kfp_ray/html2parquet_wf.py b/transforms/language/html2parquet/kfp_ray/html2parquet_wf.py index 4eb8b9de1..4eaef2fea 100644 --- a/transforms/language/html2parquet/kfp_ray/html2parquet_wf.py +++ b/transforms/language/html2parquet/kfp_ray/html2parquet_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "html2parquet_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/html2parquet/python/pyproject.toml b/transforms/language/html2parquet/python/pyproject.toml index dfd0c3928..af6b64763 100644 --- a/transforms/language/html2parquet/python/pyproject.toml +++ b/transforms/language/html2parquet/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_html2parquet_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "HTML2PARQUET Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/language/html2parquet/python/requirements.txt b/transforms/language/html2parquet/python/requirements.txt index af6ffe1e5..432362451 100644 --- a/transforms/language/html2parquet/python/requirements.txt +++ b/transforms/language/html2parquet/python/requirements.txt @@ -1,2 +1,2 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 trafilatura==1.12.0 diff --git a/transforms/language/html2parquet/ray/pyproject.toml b/transforms/language/html2parquet/ray/pyproject.toml index 873883e49..859706621 100644 --- a/transforms/language/html2parquet/ray/pyproject.toml +++ b/transforms/language/html2parquet/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_html2parquet_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "HTML2PARQUET Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/language/html2parquet/ray/requirements.txt b/transforms/language/html2parquet/ray/requirements.txt index 151d05a3e..7e543b153 100644 --- a/transforms/language/html2parquet/ray/requirements.txt +++ b/transforms/language/html2parquet/ray/requirements.txt @@ -1,3 +1,3 @@ -dpk-html2parquet-transform-python==0.2.2.dev2 -data-prep-toolkit[ray]==0.2.2.dev2 +dpk-html2parquet-transform-python==0.2.2 +data-prep-toolkit[ray]==0.2.2 trafilatura==1.12.0 \ No newline at end of file diff --git a/transforms/language/lang_id/kfp_ray/lang_id_multiple_wf.py b/transforms/language/lang_id/kfp_ray/lang_id_multiple_wf.py index a89c54ab3..e853c2328 100644 --- a/transforms/language/lang_id/kfp_ray/lang_id_multiple_wf.py +++ b/transforms/language/lang_id/kfp_ray/lang_id_multiple_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "lang_id_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/lang_id/kfp_ray/lang_id_wf.py b/transforms/language/lang_id/kfp_ray/lang_id_wf.py index 2ac84645d..5aed719c5 100644 --- a/transforms/language/lang_id/kfp_ray/lang_id_wf.py +++ b/transforms/language/lang_id/kfp_ray/lang_id_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "lang_id_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/lang_id/python/pyproject.toml b/transforms/language/lang_id/python/pyproject.toml index c5de6826a..43650a50a 100644 --- a/transforms/language/lang_id/python/pyproject.toml +++ b/transforms/language/lang_id/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_lang_id_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Language Identification Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/language/lang_id/python/requirements.txt b/transforms/language/lang_id/python/requirements.txt index a405f7afc..2cd053cfb 100644 --- a/transforms/language/lang_id/python/requirements.txt +++ b/transforms/language/lang_id/python/requirements.txt @@ -1,4 +1,4 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 fasttext==0.9.2 langcodes==3.3.0 huggingface-hub >= 0.21.4, <1.0.0 diff --git a/transforms/language/lang_id/ray/pyproject.toml b/transforms/language/lang_id/ray/pyproject.toml index ac45a167e..6347bda71 100644 --- a/transforms/language/lang_id/ray/pyproject.toml +++ b/transforms/language/lang_id/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_lang_id_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Language Identification Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Daiki Tsuzuku", email = "dtsuzuku@jp.ibm.com" } ] dependencies = [ - "dpk-lang_id-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-lang_id-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_multiple_wf.py b/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_multiple_wf.py index 8992f1145..56e881b5e 100644 --- a/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_multiple_wf.py +++ b/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_multiple_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "pdf2parquet_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_wf.py b/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_wf.py index c9cdbf652..395918ac3 100644 --- a/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_wf.py +++ b/transforms/language/pdf2parquet/kfp_ray/pdf2parquet_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "pdf2parquet_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/pdf2parquet/python/requirements.txt b/transforms/language/pdf2parquet/python/requirements.txt index 2912af252..4d09ff394 100644 --- a/transforms/language/pdf2parquet/python/requirements.txt +++ b/transforms/language/pdf2parquet/python/requirements.txt @@ -1,4 +1,4 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 docling-core==2.3.0 docling-ibm-models==2.0.3 deepsearch-glm==0.26.1 diff --git a/transforms/language/pdf2parquet/ray/requirements.txt b/transforms/language/pdf2parquet/ray/requirements.txt index 2b414c59e..abec5044d 100644 --- a/transforms/language/pdf2parquet/ray/requirements.txt +++ b/transforms/language/pdf2parquet/ray/requirements.txt @@ -1,5 +1,5 @@ dpk-pdf2parquet-transform-python==0.3.0 -data-prep-toolkit[ray]==0.2.2.dev2 +data-prep-toolkit[ray]==0.2.2 # docling-core==1.7.2 # docling-ibm-models==2.0.0 # deepsearch-glm==0.22.0 diff --git a/transforms/language/pii_redactor/python/requirements.txt b/transforms/language/pii_redactor/python/requirements.txt index 958210865..1fb9c95b9 100644 --- a/transforms/language/pii_redactor/python/requirements.txt +++ b/transforms/language/pii_redactor/python/requirements.txt @@ -1,4 +1,4 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 presidio-analyzer>=2.2.355 presidio-anonymizer>=2.2.355 flair>=0.14.0 diff --git a/transforms/language/pii_redactor/ray/pyproject.toml b/transforms/language/pii_redactor/ray/pyproject.toml index b96f16615..b98b2c9af 100644 --- a/transforms/language/pii_redactor/ray/pyproject.toml +++ b/transforms/language/pii_redactor/ray/pyproject.toml @@ -11,7 +11,7 @@ authors = [ ] dependencies = [ "dpk_pii_redactor_transform_python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", "presidio-analyzer>=2.2.355", "presidio-anonymizer>=2.2.355", "flair>=0.14.0", diff --git a/transforms/language/text_encoder/kfp_ray/text_encoder_multiple_wf.py b/transforms/language/text_encoder/kfp_ray/text_encoder_multiple_wf.py index e522737a1..bad5e24cd 100644 --- a/transforms/language/text_encoder/kfp_ray/text_encoder_multiple_wf.py +++ b/transforms/language/text_encoder/kfp_ray/text_encoder_multiple_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "text_encoder_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/text_encoder/kfp_ray/text_encoder_wf.py b/transforms/language/text_encoder/kfp_ray/text_encoder_wf.py index f88fe9eef..5c762c2a1 100644 --- a/transforms/language/text_encoder/kfp_ray/text_encoder_wf.py +++ b/transforms/language/text_encoder/kfp_ray/text_encoder_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "text_encoder_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/language/text_encoder/python/pyproject.toml b/transforms/language/text_encoder/python/pyproject.toml index 87dad3c1c..62182b27b 100644 --- a/transforms/language/text_encoder/python/pyproject.toml +++ b/transforms/language/text_encoder/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_text_encoder_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Text Encoder Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/language/text_encoder/python/requirements.txt b/transforms/language/text_encoder/python/requirements.txt index 2eb79e69b..32bf83692 100644 --- a/transforms/language/text_encoder/python/requirements.txt +++ b/transforms/language/text_encoder/python/requirements.txt @@ -1,2 +1,2 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 sentence-transformers==3.0.1 diff --git a/transforms/language/text_encoder/ray/pyproject.toml b/transforms/language/text_encoder/ray/pyproject.toml index ef08f697a..c6d49701b 100644 --- a/transforms/language/text_encoder/ray/pyproject.toml +++ b/transforms/language/text_encoder/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_text_encoder_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Text Encoder Ray Transform" license = {text = "Apache-2.0"} @@ -11,8 +11,8 @@ authors = [ { name = "Peter Staar", email = "taa@zurich.ibm.com" }, ] dependencies = [ - "dpk-text_encoder-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-text_encoder-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/pyproject.toml b/transforms/pyproject.toml index 2357553e4..3b853cbe7 100644 --- a/transforms/pyproject.toml +++ b/transforms/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "data_prep_toolkit_transforms" -version = "0.2.2.dev3" +version = "0.2.2" requires-python = ">=3.10,<3.13" keywords = ["transforms", "data preprocessing", "data preparation", "llm", "generative", "ai", "fine-tuning", "llmapps" ] description = "Data Preparation Toolkit Transforms using Ray" @@ -80,11 +80,11 @@ web2parquet = { file = ["universal/web2parquet/requirements.txt"]} # Does not seem to work for our custom layout # copy all files to a single src and let automatic discovery find them -[tool.setuptools.package-data] -"*" = ["*.txt"] +#[tool.setuptools.package-data] +#"*" = ["*.txt"] -[tool.setuptools.packages.find] -where = ["src"] +#[tool.setuptools.packages.find] +#where = ["src"] #[tool.setuptools.package-dir] #dpk_web2parquet = "universal/web2parquet/dpk_web2parquet" diff --git a/transforms/requirements-ray.txt b/transforms/requirements-ray.txt index 9012f685b..11d0decf5 100644 --- a/transforms/requirements-ray.txt +++ b/transforms/requirements-ray.txt @@ -1,4 +1,4 @@ -data-prep-toolkit[ray]>=0.2.2.dev2 +data-prep-toolkit[ray]>=0.2.2 networkx==3.3 colorlog==6.8.2 func-timeout==4.3.5 diff --git a/transforms/requirements.txt b/transforms/requirements.txt index 8b48a970f..7317d33e3 100644 --- a/transforms/requirements.txt +++ b/transforms/requirements.txt @@ -1 +1 @@ -data-prep-toolkit>=0.2.2.dev2 +data-prep-toolkit>=0.2.2 diff --git a/transforms/transform.config b/transforms/transform.config index c226171c6..7bafba684 100644 --- a/transforms/transform.config +++ b/transforms/transform.config @@ -7,11 +7,3 @@ # expected files and is used to define the transform's image name. TRANSFORM_NAME=data-prep-kit-transforms -################################################################################ -# This defines the transforms' package version number as would be used -# when publishing the wheel. In general, only the micro version -# number should be advanced relative to the DPK_VERSION. -# -# 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=0.2.2.dev2 diff --git a/transforms/universal/doc_id/kfp_ray/doc_id_wf.py b/transforms/universal/doc_id/kfp_ray/doc_id_wf.py index f41231159..7e1bd0b8e 100644 --- a/transforms/universal/doc_id/kfp_ray/doc_id_wf.py +++ b/transforms/universal/doc_id/kfp_ray/doc_id_wf.py @@ -22,7 +22,7 @@ # the name of the job script EXEC_SCRIPT_NAME: str = "doc_id_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/doc_id/python/pyproject.toml b/transforms/universal/doc_id/python/pyproject.toml index 0e2658087..a9e69f0bf 100644 --- a/transforms/universal/doc_id/python/pyproject.toml +++ b/transforms/universal/doc_id/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_doc_id_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "ededup Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/universal/doc_id/python/requirements.txt b/transforms/universal/doc_id/python/requirements.txt index 368287e5d..880c7c2c7 100644 --- a/transforms/universal/doc_id/python/requirements.txt +++ b/transforms/universal/doc_id/python/requirements.txt @@ -1 +1 @@ -data-prep-toolkit==0.2.2.dev2 \ No newline at end of file +data-prep-toolkit==0.2.2 \ No newline at end of file diff --git a/transforms/universal/doc_id/ray/pyproject.toml b/transforms/universal/doc_id/ray/pyproject.toml index 5a5941155..fc6a37b19 100644 --- a/transforms/universal/doc_id/ray/pyproject.toml +++ b/transforms/universal/doc_id/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_doc_id_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "docid Ray Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "dpk_doc_id_transform_python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk_doc_id_transform_python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/universal/doc_id/spark/pyproject.toml b/transforms/universal/doc_id/spark/pyproject.toml index 36f345c09..f50d4f70d 100644 --- a/transforms/universal/doc_id/spark/pyproject.toml +++ b/transforms/universal/doc_id/spark/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_doc_id_transform_spark" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Doc ID Spark Transform" license = {text = "Apache-2.0"} @@ -10,7 +10,7 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "data-prep-toolkit[spark]==0.2.2.dev2", + "data-prep-toolkit[spark]==0.2.2", ] [build-system] diff --git a/transforms/universal/ededup/kfp_ray/ededup_wf.py b/transforms/universal/ededup/kfp_ray/ededup_wf.py index ab46daadb..d878bd3e2 100644 --- a/transforms/universal/ededup/kfp_ray/ededup_wf.py +++ b/transforms/universal/ededup/kfp_ray/ededup_wf.py @@ -24,7 +24,7 @@ EXEC_SCRIPT_NAME: str = "ededup_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/ededup/python/pyproject.toml b/transforms/universal/ededup/python/pyproject.toml index 735104f20..67fd0f758 100644 --- a/transforms/universal/ededup/python/pyproject.toml +++ b/transforms/universal/ededup/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_ededup_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "ededup Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/universal/ededup/python/requirements.txt b/transforms/universal/ededup/python/requirements.txt index 75baaef62..45b4cfd50 100644 --- a/transforms/universal/ededup/python/requirements.txt +++ b/transforms/universal/ededup/python/requirements.txt @@ -1,3 +1,3 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 mmh3>=4.1.0 xxhash==3.4.1 diff --git a/transforms/universal/ededup/ray/pyproject.toml b/transforms/universal/ededup/ray/pyproject.toml index 9e3885e50..d74fa0637 100644 --- a/transforms/universal/ededup/ray/pyproject.toml +++ b/transforms/universal/ededup/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_ededup_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "ededup Ray Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "data-prep-toolkit[ray]==0.2.2.dev2", - "dpk_ededup_transform_python==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", + "dpk_ededup_transform_python==0.2.2", "tqdm==4.66.3", ] diff --git a/transforms/universal/fdedup/kfp_ray/fdedup_wf.py b/transforms/universal/fdedup/kfp_ray/fdedup_wf.py index 3156ab6f1..da431d030 100644 --- a/transforms/universal/fdedup/kfp_ray/fdedup_wf.py +++ b/transforms/universal/fdedup/kfp_ray/fdedup_wf.py @@ -24,7 +24,7 @@ EXEC_SCRIPT_NAME: str = "fdedup_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/fdedup/ray/pyproject.toml b/transforms/universal/fdedup/ray/pyproject.toml index 923cbdf82..7c59dcff9 100644 --- a/transforms/universal/fdedup/ray/pyproject.toml +++ b/transforms/universal/fdedup/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_fdedup_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "fdedup Ray Transform" license = {text = "Apache-2.0"} @@ -10,7 +10,7 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "data-prep-toolkit[ray]==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", "mmh3>=4.1.0", "xxhash==3.4.1", "tqdm==4.66.3", diff --git a/transforms/universal/filter/kfp_ray/filter_wf.py b/transforms/universal/filter/kfp_ray/filter_wf.py index b856b1007..4b122d98f 100644 --- a/transforms/universal/filter/kfp_ray/filter_wf.py +++ b/transforms/universal/filter/kfp_ray/filter_wf.py @@ -24,7 +24,7 @@ task_image = "quay.io/dataprep1/data-prep-kit/filter-ray:latest" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/filter/python/pyproject.toml b/transforms/universal/filter/python/pyproject.toml index 64f148799..8e9bb2366 100644 --- a/transforms/universal/filter/python/pyproject.toml +++ b/transforms/universal/filter/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_filter_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Filter Transform for Python" license = {text = "Apache-2.0"} diff --git a/transforms/universal/filter/python/requirements.txt b/transforms/universal/filter/python/requirements.txt index 9f1feff29..5e3e783c8 100644 --- a/transforms/universal/filter/python/requirements.txt +++ b/transforms/universal/filter/python/requirements.txt @@ -1,3 +1,3 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 duckdb>=0.10.1 diff --git a/transforms/universal/filter/ray/pyproject.toml b/transforms/universal/filter/ray/pyproject.toml index a794a1a0b..a8ec7bb4d 100644 --- a/transforms/universal/filter/ray/pyproject.toml +++ b/transforms/universal/filter/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_filter_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Filter Transform for Ray" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Constantin Adam", email = "cmadam@us.ibm.com" }, ] dependencies = [ - "dpk-filter-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-filter-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/universal/filter/spark/pyproject.toml b/transforms/universal/filter/spark/pyproject.toml index 7b60dba46..85403487a 100644 --- a/transforms/universal/filter/spark/pyproject.toml +++ b/transforms/universal/filter/spark/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_filter_transform_spark" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Filter Spark Transform" license = {text = "Apache-2.0"} @@ -9,7 +9,7 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "data-prep-toolkit[spark]==0.2.2.dev2", + "data-prep-toolkit[spark]==0.2.2", ] [project.optional-dependencies] diff --git a/transforms/universal/hap/kfp_ray.disable/hap_wf.py b/transforms/universal/hap/kfp_ray.disable/hap_wf.py index 786011d4d..8069ec181 100644 --- a/transforms/universal/hap/kfp_ray.disable/hap_wf.py +++ b/transforms/universal/hap/kfp_ray.disable/hap_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "hap_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/hap/python/pyproject.toml b/transforms/universal/hap/python/pyproject.toml index 389788363..7b30dd72e 100644 --- a/transforms/universal/hap/python/pyproject.toml +++ b/transforms/universal/hap/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_hap_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "HAP Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/universal/hap/python/requirements.txt b/transforms/universal/hap/python/requirements.txt index 505dd9ceb..07c5f854a 100644 --- a/transforms/universal/hap/python/requirements.txt +++ b/transforms/universal/hap/python/requirements.txt @@ -1,5 +1,5 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 nltk==3.9.1 transformers==4.38.2 -torch==2.4.1 +torch>=2.2.2,<=2.4.1 pandas==2.2.2 diff --git a/transforms/universal/hap/ray/pyproject.toml b/transforms/universal/hap/ray/pyproject.toml index abbb1a30c..6518e5277 100644 --- a/transforms/universal/hap/ray/pyproject.toml +++ b/transforms/universal/hap/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_hap_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "HAP Ray Transform" license = {text = "Apache-2.0"} diff --git a/transforms/universal/hap/ray/requirements.txt b/transforms/universal/hap/ray/requirements.txt index 0ed65f625..119167ca2 100644 --- a/transforms/universal/hap/ray/requirements.txt +++ b/transforms/universal/hap/ray/requirements.txt @@ -1,6 +1,6 @@ -data-prep-toolkit[ray]==0.2.2.dev2 -dpk-hap-transform-python==0.2.2.dev2 +data-prep-toolkit[ray]==0.2.2 +dpk-hap-transform-python==0.2.2 nltk==3.9.1 transformers==4.38.2 -torch==2.4.1 +torch>=2.2.2,<=2.4.1 pandas==2.2.2 diff --git a/transforms/universal/noop/kfp_ray/noop_multiple_wf.py b/transforms/universal/noop/kfp_ray/noop_multiple_wf.py index 3b102d205..737b60121 100644 --- a/transforms/universal/noop/kfp_ray/noop_multiple_wf.py +++ b/transforms/universal/noop/kfp_ray/noop_multiple_wf.py @@ -23,7 +23,7 @@ EXEC_SCRIPT_NAME: str = "noop_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/noop/kfp_ray/noop_wf.py b/transforms/universal/noop/kfp_ray/noop_wf.py index e8125328b..9dbdaf3b0 100644 --- a/transforms/universal/noop/kfp_ray/noop_wf.py +++ b/transforms/universal/noop/kfp_ray/noop_wf.py @@ -24,7 +24,7 @@ EXEC_SCRIPT_NAME: str = "noop_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/noop/python/pyproject.toml b/transforms/universal/noop/python/pyproject.toml index 998161e31..e8c089ef0 100644 --- a/transforms/universal/noop/python/pyproject.toml +++ b/transforms/universal/noop/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_noop_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "NOOP Python Transform" license = {text = "Apache-2.0"} @@ -10,7 +10,7 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "data-prep-toolkit==0.2.2.dev2", + "data-prep-toolkit==0.2.2", ] [build-system] diff --git a/transforms/universal/noop/ray/pyproject.toml b/transforms/universal/noop/ray/pyproject.toml index 5d475fe12..19fe77560 100644 --- a/transforms/universal/noop/ray/pyproject.toml +++ b/transforms/universal/noop/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_noop_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "NOOP Ray Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "dpk-noop-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-noop-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/universal/noop/spark/pyproject.toml b/transforms/universal/noop/spark/pyproject.toml index f867fb070..495d827a0 100644 --- a/transforms/universal/noop/spark/pyproject.toml +++ b/transforms/universal/noop/spark/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_noop_transform_spark" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "NOOP Spark Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "dpk-noop-transform-python==0.2.2.dev2", - "data-prep-toolkit[spark]==0.2.2.dev2", + "dpk-noop-transform-python==0.2.2", + "data-prep-toolkit[spark]==0.2.2", ] [build-system] diff --git a/transforms/universal/profiler/kfp_ray/profiler_wf.py b/transforms/universal/profiler/kfp_ray/profiler_wf.py index 914637895..ee6323d74 100644 --- a/transforms/universal/profiler/kfp_ray/profiler_wf.py +++ b/transforms/universal/profiler/kfp_ray/profiler_wf.py @@ -24,7 +24,7 @@ EXEC_SCRIPT_NAME: str = "profiler_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/profiler/python/pyproject.toml b/transforms/universal/profiler/python/pyproject.toml index 95775e3a6..117be53c0 100644 --- a/transforms/universal/profiler/python/pyproject.toml +++ b/transforms/universal/profiler/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_profiler_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "profiler Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/universal/profiler/python/requirements.txt b/transforms/universal/profiler/python/requirements.txt index 89801e4ad..fee352d4a 100644 --- a/transforms/universal/profiler/python/requirements.txt +++ b/transforms/universal/profiler/python/requirements.txt @@ -1,5 +1,5 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 mmh3==4.1.0 xxhash==3.4.1 diff --git a/transforms/universal/profiler/ray/pyproject.toml b/transforms/universal/profiler/ray/pyproject.toml index 6060653fa..c9f1b1da3 100644 --- a/transforms/universal/profiler/ray/pyproject.toml +++ b/transforms/universal/profiler/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_profiler_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "profiler Ray Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "data-prep-toolkit[ray]==0.2.2.dev2", - "dpk_profiler_transform_python==0.2.2.dev2", + "data-prep-toolkit[ray]==0.2.2", + "dpk_profiler_transform_python==0.2.2", "tqdm==4.66.3", ] diff --git a/transforms/universal/profiler/spark/pyproject.toml b/transforms/universal/profiler/spark/pyproject.toml index 455684b4f..05602dc26 100644 --- a/transforms/universal/profiler/spark/pyproject.toml +++ b/transforms/universal/profiler/spark/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_profiler_transform_spark" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Profiler Spark Transform" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "dpk-profiler-transform-python==0.2.2.dev2", - "data-prep-toolkit[spark]==0.2.2.dev2", + "dpk-profiler-transform-python==0.2.2", + "data-prep-toolkit[spark]==0.2.2", ] [build-system] diff --git a/transforms/universal/resize/kfp_ray/resize_wf.py b/transforms/universal/resize/kfp_ray/resize_wf.py index 0724ed731..0a9be8e95 100644 --- a/transforms/universal/resize/kfp_ray/resize_wf.py +++ b/transforms/universal/resize/kfp_ray/resize_wf.py @@ -22,7 +22,7 @@ # the name of the job script EXEC_SCRIPT_NAME: str = "resize_transform_ray.py" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files component_spec_path = "../../../../kfp/kfp_ray_components/" diff --git a/transforms/universal/resize/python/pyproject.toml b/transforms/universal/resize/python/pyproject.toml index 082f37f0c..836388694 100644 --- a/transforms/universal/resize/python/pyproject.toml +++ b/transforms/universal/resize/python/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_resize_transform_python" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "resize Python Transform" license = {text = "Apache-2.0"} diff --git a/transforms/universal/resize/python/requirements.txt b/transforms/universal/resize/python/requirements.txt index 368287e5d..880c7c2c7 100644 --- a/transforms/universal/resize/python/requirements.txt +++ b/transforms/universal/resize/python/requirements.txt @@ -1 +1 @@ -data-prep-toolkit==0.2.2.dev2 \ No newline at end of file +data-prep-toolkit==0.2.2 \ No newline at end of file diff --git a/transforms/universal/resize/ray/pyproject.toml b/transforms/universal/resize/ray/pyproject.toml index 1490303bb..4f7603f6f 100644 --- a/transforms/universal/resize/ray/pyproject.toml +++ b/transforms/universal/resize/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_resize_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Resize Ray Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsky@ibm.com" }, ] dependencies = [ - "dpk-resize-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-resize-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/universal/resize/spark/pyproject.toml b/transforms/universal/resize/spark/pyproject.toml index 538c12d20..c8bb67111 100644 --- a/transforms/universal/resize/spark/pyproject.toml +++ b/transforms/universal/resize/spark/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_resize_transform_spark" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Resize Spark Transform" license = {text = "Apache-2.0"} @@ -10,8 +10,8 @@ authors = [ { name = "Boris Lublinsky", email = "blublinsk@ibm.com" }, ] dependencies = [ - "dpk-resize-transform-python==0.2.2.dev2", - "data-prep-toolkit[spark]==0.2.2.dev2", + "dpk-resize-transform-python==0.2.2", + "data-prep-toolkit[spark]==0.2.2", ] [build-system] diff --git a/transforms/universal/tokenization/kfp_ray/tokenization_wf.py b/transforms/universal/tokenization/kfp_ray/tokenization_wf.py index c131d11ea..243cac6be 100644 --- a/transforms/universal/tokenization/kfp_ray/tokenization_wf.py +++ b/transforms/universal/tokenization/kfp_ray/tokenization_wf.py @@ -23,7 +23,7 @@ task_image = "quay.io/dataprep1/data-prep-kit/tokenization-ray:latest" # components -base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:latest" +base_kfp_image = "quay.io/dataprep1/data-prep-kit/kfp-data-processing:0.2.2" # path to kfp component specifications files # path to kfp component specifications files diff --git a/transforms/universal/tokenization/python/pyproject.toml b/transforms/universal/tokenization/python/pyproject.toml index bc352f0fd..021a1427f 100644 --- a/transforms/universal/tokenization/python/pyproject.toml +++ b/transforms/universal/tokenization/python/pyproject.toml @@ -1,7 +1,7 @@ [project] name = "dpk_tokenization_transform_python" keywords = ["tokenizer", "data", "data preprocessing", "data preparation", "llm", "generative", "ai", "fine-tuning", "llmapps" ] -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Tokenization Transform for Python" license = {text = "Apache-2.0"} diff --git a/transforms/universal/tokenization/python/requirements.txt b/transforms/universal/tokenization/python/requirements.txt index 5e00dbaa1..afd567d8b 100644 --- a/transforms/universal/tokenization/python/requirements.txt +++ b/transforms/universal/tokenization/python/requirements.txt @@ -1,2 +1,2 @@ -data-prep-toolkit==0.2.2.dev2 +data-prep-toolkit==0.2.2 transformers==4.38.2 diff --git a/transforms/universal/tokenization/ray/pyproject.toml b/transforms/universal/tokenization/ray/pyproject.toml index 095cb63e0..3cc4bcf80 100644 --- a/transforms/universal/tokenization/ray/pyproject.toml +++ b/transforms/universal/tokenization/ray/pyproject.toml @@ -1,6 +1,6 @@ [project] name = "dpk_tokenization_transform_ray" -version = "0.2.2.dev2" +version = "0.2.2" requires-python = ">=3.10,<3.13" description = "Tokenization Transform for Ray" license = {text = "Apache-2.0"} @@ -9,8 +9,8 @@ authors = [ { name = "Xuan-Hong Dang", email = "xuan-hong.dang@ibm.com"}, ] dependencies = [ - "dpk-tokenization-transform-python==0.2.2.dev2", - "data-prep-toolkit[ray]==0.2.2.dev2", + "dpk-tokenization-transform-python==0.2.2", + "data-prep-toolkit[ray]==0.2.2", ] [build-system] diff --git a/transforms/universal/web2parquet/requirements.txt b/transforms/universal/web2parquet/requirements.txt index 5c989591d..dfb74a6ca 100644 --- a/transforms/universal/web2parquet/requirements.txt +++ b/transforms/universal/web2parquet/requirements.txt @@ -1,2 +1,2 @@ -data-prep-toolkit>=0.2.2.dev2 -data_prep_connector>=0.2.3.dev0 \ No newline at end of file +data-prep-toolkit>=0.2.2 +data_prep_connector>=0.2.3 \ No newline at end of file