This is an automated email from the ASF dual-hosted git repository. github-merge-queue[bot] pushed a commit to branch gh-readonly-queue/main/pr-7143-2b030973f3a2ab8356307e36a5cd2a5998b1b99d in repository https://gitbox.apache.org/repos/asf/texera.git
commit 8f1bef66e9fed51023a5e5d223ccd5040334aa1c Author: Prateek Ganigi <[email protected]> AuthorDate: Thu Jul 30 15:21:43 2026 -0700 refactor(workflow-operator): rename PythonCodegenBase to HuggingFaceCodegenBase for clarity (#7143) ### What changes were proposed in this PR? Pure internal rename: `PythonCodegenBase` -> `HuggingFaceCodegenBase` (object, file, and spec). Despite the generic name, this object is HF-specific, it lives in `.../huggingFace/codegen/`, is used only there, and is the shared base of the Hugging Face per-task `TaskCodegen`s. The old name led a reviewer to ask whether the HF logic belonged elsewhere (comment on #6972); the new name matches the sibling codegens and `HuggingFaceInferenceOpDesc`. Also updated the references in `HuggingFaceInferenceOpDesc.scala`, `ImageTaskCodegen.scala`, and `TaskCodegen.scala`. Behavior-neutral: no change to the generated Python, workflow JSON, or any public API; git tracks both files as renames. ### Any related issues, documentation, discussions? Closes #7140 Follow-up to the review discussion on #6972. ### How was this PR tested? No new tests as this is a rename, and existing coverage confirms it's behavior-neutral. sbt "WorkflowOperator/testOnly org.apache.texera.amber.operator.huggingFace.*" The full HF package passes (124 tests), including `PythonCodeRawInvalidTextSpec`, which py-compiles the generated Python of all 117 operators — confirming the generated output is unchanged. scalafmt clean. ### Was this PR authored or co-authored using generative AI tooling? Co-authored with Claude Opus 4.7 in compliance with ASF. --- .../huggingFace/HuggingFaceInferenceOpDesc.scala | 8 ++++---- ...CodegenBase.scala => HuggingFaceCodegenBase.scala} | 2 +- .../huggingFace/codegen/ImageTaskCodegen.scala | 4 ++-- .../operator/huggingFace/codegen/TaskCodegen.scala | 4 ++-- ...aseSpec.scala => HuggingFaceCodegenBaseSpec.scala} | 19 ++++++++++--------- .../hugging-face-image-upload.component.ts | 2 +- 6 files changed, 20 insertions(+), 19 deletions(-) diff --git a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDesc.scala b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDesc.scala index 0e1062c75a..c08084135b 100644 --- a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDesc.scala +++ b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceInferenceOpDesc.scala @@ -29,7 +29,7 @@ import org.apache.texera.amber.operator.huggingFace.codegen.{ CodegenContext, ImageTaskCodegen, MediaGenCodegen, - PythonCodegenBase, + HuggingFaceCodegenBase, QaRankingCodegen, TaskCodegen, TextGenCodegen @@ -47,12 +47,12 @@ import org.apache.texera.amber.pybuilder.PyStringTypes.EncodableString * `TaskCodegen` implementations registered in `registeredCodegens`. * * The Python script that runs at execution time is assembled by - * `PythonCodegenBase.render(ctx, codegen)`, which composes the shared + * `HuggingFaceCodegenBase.render(ctx, codegen)`, which composes the shared * provider-fallback / request-loop infrastructure with the per-task * payload + parse snippets supplied by the selected `TaskCodegen`. * * User-provided string fields are typed as [[EncodableString]] so the - * `pyb"..."` macro inside `PythonCodegenBase` emits them as + * `pyb"..."` macro inside `HuggingFaceCodegenBase` emits them as * base64-decoded expressions at runtime instead of raw Python literals — * this is what allows the operator to satisfy * `PythonCodeRawInvalidTextSpec`'s contract that arbitrary `@JsonProperty` @@ -246,7 +246,7 @@ class HuggingFaceInferenceOpDesc extends PythonOperatorDescriptor { sentencesColumn = safeSentencesColumn ) - PythonCodegenBase.render(ctx, codegenForTask(safeTask)) + HuggingFaceCodegenBase.render(ctx, codegenForTask(safeTask)) } override def operatorInfo: OperatorInfo = diff --git a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/PythonCodegenBase.scala b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala similarity index 99% rename from common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/PythonCodegenBase.scala rename to common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala index 36a2a329a0..77bf488d6c 100644 --- a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/PythonCodegenBase.scala +++ b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBase.scala @@ -42,7 +42,7 @@ import org.apache.texera.amber.pybuilder.PythonTemplateBuilder.PythonTemplateBui * loading, audio MIME inference, media-URL fetching, etc.) will be added * in subsequent PRs as the corresponding task families land. */ -object PythonCodegenBase { +object HuggingFaceCodegenBase { def render(ctx: CodegenContext, codegen: TaskCodegen): String = { val payload = codegen.payloadPython(ctx) diff --git a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/ImageTaskCodegen.scala b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/ImageTaskCodegen.scala index 5a5ee0a937..26abcf3215 100644 --- a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/ImageTaskCodegen.scala +++ b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/ImageTaskCodegen.scala @@ -31,11 +31,11 @@ package org.apache.texera.amber.operator.huggingFace.codegen * zero-shot-image-classification, image-text-to-text, image-to-image. * * Per-row `current_image_bytes` is resolved upstream in - * [[PythonCodegenBase]]'s `process_table` (either from the operator's + * [[HuggingFaceCodegenBase]]'s `process_table` (either from the operator's * uploaded image or from `INPUT_IMAGE_COLUMN`). The image helpers * (`_read_image_input`, `_compress_image_bytes`, `_image_input_as_base64`, * `_read_binary_value`, `_looks_like_html`, `_html_to_image_bytes`, - * `_extract_json_arg`) live in PythonCodegenBase alongside the per-task + * `_extract_json_arg`) live in HuggingFaceCodegenBase alongside the per-task * tuples (`image_only_tasks`, `image_prompt_tasks`, `image_tasks`). */ object ImageTaskCodegen extends TaskCodegen { diff --git a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/TaskCodegen.scala b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/TaskCodegen.scala index 8abcef721b..3e40a2085e 100644 --- a/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/TaskCodegen.scala +++ b/common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/codegen/TaskCodegen.scala @@ -25,7 +25,7 @@ import org.apache.texera.amber.pybuilder.PyStringTypes.EncodableString * Inputs the dispatcher passes through to each TaskCodegen. * * User-provided string fields are typed as [[EncodableString]] so the - * `pyb"..."` macro in [[PythonCodegenBase]] emits them as base64-decoded + * `pyb"..."` macro in [[HuggingFaceCodegenBase]] emits them as base64-decoded * runtime expressions rather than raw Python string literals — required to * pass `PythonCodeRawInvalidTextSpec`'s leakage check. */ @@ -57,7 +57,7 @@ final case class CodegenContext( * `*Codegen` objects and adding them to that map. * * Snippets returned by these methods are Python source spliced into the - * shared template assembled by [[PythonCodegenBase.render]]. Snippets must + * shared template assembled by [[HuggingFaceCodegenBase.render]]. Snippets must * NOT directly inline user-provided strings — reference the per-instance * attributes `self.HF_API_TOKEN`, `self.MODEL_ID`, `self.PROMPT_COLUMN`, * etc. that the base class initializes from `CodegenContext` via the diff --git a/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/codegen/PythonCodegenBaseSpec.scala b/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBaseSpec.scala similarity index 90% rename from common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/codegen/PythonCodegenBaseSpec.scala rename to common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBaseSpec.scala index 3705fad47a..0a5e96f330 100644 --- a/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/codegen/PythonCodegenBaseSpec.scala +++ b/common/workflow-operator/src/test/scala/org/apache/texera/amber/operator/huggingFace/codegen/HuggingFaceCodegenBaseSpec.scala @@ -23,7 +23,7 @@ import org.apache.texera.amber.pybuilder.PyStringTypes.EncodableString import org.scalatest.flatspec.AnyFlatSpec import org.scalatest.matchers.should.Matchers -class PythonCodegenBaseSpec extends AnyFlatSpec with Matchers { +class HuggingFaceCodegenBaseSpec extends AnyFlatSpec with Matchers { // A stand-in TaskCodegen whose payload/parse snippets are distinctive // sentinels. Because payloadPython/parsePython return plain Python source @@ -59,8 +59,8 @@ class PythonCodegenBaseSpec extends AnyFlatSpec with Matchers { safeTemp = safeTemp ) - "PythonCodegenBase.render" should "emit the ProcessTableOperator skeleton and shared helpers" in { - val out = PythonCodegenBase.render(makeCtx(), StubCodegen) + "HuggingFaceCodegenBase.render" should "emit the ProcessTableOperator skeleton and shared helpers" in { + val out = HuggingFaceCodegenBase.render(makeCtx(), StubCodegen) out should include("class ProcessTableOperator(UDFTableOperator):") out should include("def open(self):") out should include("def process_table(") @@ -73,7 +73,7 @@ class PythonCodegenBaseSpec extends AnyFlatSpec with Matchers { } it should "splice the per-task codegen's payload and parse snippets into the template" in { - val out = PythonCodegenBase.render(makeCtx(), StubCodegen) + val out = HuggingFaceCodegenBase.render(makeCtx(), StubCodegen) out should include("STUB_PAYLOAD_zX7q42") out should include("STUB_PARSE_zX7q42") } @@ -82,8 +82,8 @@ class PythonCodegenBaseSpec extends AnyFlatSpec with Matchers { // The same context routed through two different codegens must yield the // spliced fragments of whichever codegen is passed, proving the base is a // pure host for the codegen's snippets. - val stub = PythonCodegenBase.render(makeCtx(), StubCodegen) - val real = PythonCodegenBase.render(makeCtx(), TextGenCodegen) + val stub = HuggingFaceCodegenBase.render(makeCtx(), StubCodegen) + val real = HuggingFaceCodegenBase.render(makeCtx(), TextGenCodegen) stub should include("STUB_PAYLOAD_zX7q42") real should not include "STUB_PAYLOAD_zX7q42" // TextGenCodegen's real payload/parse markers appear only in the real run. @@ -96,13 +96,14 @@ class PythonCodegenBaseSpec extends AnyFlatSpec with Matchers { } it should "interpolate the numeric context fields as raw Python literals" in { - val out = PythonCodegenBase.render(makeCtx(safeMaxTokens = 512, safeTemp = 0.9), StubCodegen) + val out = + HuggingFaceCodegenBase.render(makeCtx(safeMaxTokens = 512, safeTemp = 0.9), StubCodegen) out should include("self.MAX_NEW_TOKENS = 512") out should include("self.TEMPERATURE = 0.9") } it should "assign user-provided strings via runtime base64 decode expressions, not raw literals" in { - val out = PythonCodegenBase.render(makeCtx(), StubCodegen) + val out = HuggingFaceCodegenBase.render(makeCtx(), StubCodegen) // Every user-supplied string field open() assigns is set through the safe // decode helper. This covers all EncodableString context fields, including // the result/task and per-task (image/audio/context/labels/sentences) @@ -125,7 +126,7 @@ class PythonCodegenBaseSpec extends AnyFlatSpec with Matchers { it should "never leak raw user-provided string values into the generated source" in { // Sentinels contain underscores, which base64 output cannot contain, so a // literal match here can only mean the raw value leaked past the encoder. - val out = PythonCodegenBase.render( + val out = HuggingFaceCodegenBase.render( makeCtx( hfApiToken = "MARKER_TOKEN_zXyq42", modelId = "MARKER_MODEL_zXyq42", diff --git a/frontend/src/app/workspace/component/hugging-face-image-upload/hugging-face-image-upload.component.ts b/frontend/src/app/workspace/component/hugging-face-image-upload/hugging-face-image-upload.component.ts index 4b72e14aa5..9556a72670 100644 --- a/frontend/src/app/workspace/component/hugging-face-image-upload/hugging-face-image-upload.component.ts +++ b/frontend/src/app/workspace/component/hugging-face-image-upload/hugging-face-image-upload.component.ts @@ -22,7 +22,7 @@ import { CommonModule } from "@angular/common"; import { FieldType, FieldTypeConfig } from "@ngx-formly/core"; import { NzButtonModule } from "ng-zorro-antd/button"; -// Keep in sync with PythonCodegenBase._compress_image_bytes(max_bytes) on the backend: +// Keep in sync with HuggingFaceCodegenBase._compress_image_bytes(max_bytes) on the backend: // the uploaded data URL must stay within the size the inference helpers expect. const MAX_DATA_URL_LENGTH = 45000; const INITIAL_MAX_DIMENSION = 512;
