feat(workflow-operator): leave explanatory comments out of exported Python

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceIrisLogisticRegressionOpDesc.scala

@@ -137,6 +137,7 @@ class HuggingFaceIrisLogisticRegressionOpDesc
   override def generateStandaloneCode(): String = {
     val lengthLit = pyStringLiteral(petalLengthCmAttribute)
     val widthLit = pyStringLiteral(petalWidthCmAttribute)
+    // The operator's guard: pd.isna answers for a JSON frame's NaN and a Tuple's None.
     s"""import numpy as np
        |import torch
        |import torch.nn as nn
@@ -159,8 +160,6 @@ class HuggingFaceIrisLogisticRegressionOpDesc
        |_classes = []
        |_probs = []
        |for _length, _width in zip(out1df[$lengthLit], out1df[$widthLit]):
-       |    # The operator's guard: pandas answers for the NaN a frame read from
-       |    # JSON holds and for the None a Tuple hands over.
        |    if pd.isna(_length) or pd.isna(_width):
        |        _classes.append(None)
        |        _probs.append(None)

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceSentimentAnalysisOpDesc.scala

@@ -121,6 +121,7 @@ class HuggingFaceSentimentAnalysisOpDesc
     val positiveLit = pyStringLiteral(resultAttributePositive)
     val neutralLit = pyStringLiteral(resultAttributeNeutral)
     val negativeLit = pyStringLiteral(resultAttributeNegative)
+    // An empty cell, which the tokenizer rejects, keeps its row with the scores left empty.
     s"""from transformers import AutoModelForSequenceClassification
        |from transformers import AutoTokenizer, AutoConfig
        |import numpy as np
@@ -137,10 +138,6 @@ class HuggingFaceSentimentAnalysisOpDesc
        |for _col in ($positiveLit, $neutralLit, $negativeLit):
        |    out1df[_col] = 0.0
        |for _idx, _text in out1df[${pyStringLiteral(attribute)}].items():
-       |    # An empty cell reaches the frame as None, or as NaN once a column holds
-       |    # nothing else, and the tokenizer rejects both. Keep the row and leave the
-       |    # scores empty rather than ending the run over a value the model has
-       |    # nothing to say about.
        |    if pd.isna(_text) or (isinstance(_text, str) and not _text.strip()):
        |        for _col in ($positiveLit, $neutralLit, $negativeLit):
        |            out1df.at[_idx, _col] = None

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceSpamSMSDetectionOpDesc.scala

@@ -100,17 +100,14 @@ class HuggingFaceSpamSMSDetectionOpDesc
     val attributeLit = pyStringLiteral(attribute)
     val spamLit = pyStringLiteral(resultAttributeSpam)
     val probabilityLit = pyStringLiteral(resultAttributeProbability)
+    // An empty cell, which the pipeline rejects, keeps its row with the results left empty.
     s"""from transformers import pipeline
        |import pandas as pd
        |
        |_pipeline = pipeline("text-classification", model="mrm8488/bert-tiny-finetuned-sms-spam-detection")
        |out1df = in1df.copy()
        |
        |def _classify(_t):
-       |    # An empty cell reaches the frame as None, or as NaN once a column holds
-       |    # nothing else, and the pipeline rejects both. Keep the row and leave the
-       |    # results empty rather than ending the run over a value the model has
-       |    # nothing to say about.
        |    if pd.isna(_t) or (isinstance(_t, str) and not _t.strip()):
        |        return None
        |    return _pipeline(_t)[0]

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/huggingFace/HuggingFaceTextSummarizationOpDesc.scala

@@ -98,6 +98,7 @@ class HuggingFaceTextSummarizationOpDesc
   // once, generate a summary per row, and add the STRING result column to
   // produce out1df.
   override def generateStandaloneCode(): String = {
+    // An empty cell, which the tokenizer rejects, keeps its row with the summary left empty.
     s"""from transformers import BertTokenizerFast, EncoderDecoderModel
        |import pandas as pd
        |import torch
@@ -110,10 +111,6 @@ class HuggingFaceTextSummarizationOpDesc
        |out1df = in1df.copy()
        |_summaries = []
        |for _text in out1df[${pyStringLiteral(attribute)}]:
-       |    # An empty cell reaches the frame as None, or as NaN once a column holds
-       |    # nothing else, and the tokenizer rejects both. Keep the row and leave the
-       |    # summary empty rather than ending the run over a value the model has
-       |    # nothing to say about.
        |    if pd.isna(_text) or (isinstance(_text, str) and not _text.strip()):
        |        _summaries.append(None)
        |        continue

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/machineLearning/Scorer/MachineLearningScorerOpDesc.scala

@@ -300,6 +300,8 @@ class MachineLearningScorerOpDesc extends PythonOperatorDescriptor with Standalo
 
   override def generateStandaloneCode(): String = {
     val isRegressionStr = if (isRegression) "True" else "False"
+    // Rows come back Class first, then the metrics in the order picked, as the operator declares.
+    // The drop goes to its own name, and an empty result raises rather than scoring nothing.
     s"""import pandas as pd
        |from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score, mean_squared_error, root_mean_squared_error, mean_absolute_error, r2_score
        |
@@ -320,9 +322,6 @@ class MachineLearningScorerOpDesc extends PythonOperatorDescriptor with Standalo
        |
        |  labels = ['class_' + str(label) if type(label) != str else label for label in labels]
        |  result['Class'] = labels
-       |  # Class first and the metrics in the order they were picked, which is
-       |  # the column order this operator declares and so the order of the rows
-       |  # the workflow shows.
        |  return pd.DataFrame(result)[['Class'] + metric_list]
        |
        |def regression_metrics(y_true, y_pred, metric_list):
@@ -332,14 +331,9 @@ class MachineLearningScorerOpDesc extends PythonOperatorDescriptor with Standalo
        |    result[metric] = metrics_func[metric](y_true, y_pred)
        |  return pd.DataFrame(result, index=[0])
        |
-       |# Bound to a name of its own: the same frame can feed another branch of the
-       |# plan, which must still see every row.
        |scored_df = in1df.dropna(subset=[${pyStringLiteral(actualValueColumn)}, ${pyStringLiteral(
       predictValueColumn
     )}])
-       |# Nothing survived the drop. The metrics answer that badly and each in its own
-       |# way: the regression ones raise from inside scikit-learn, and the
-       |# classification ones return a NaN score, which reads as a result.
        |if scored_df.empty:
        |    raise ValueError("No rows left to score: every row is missing the actual value, the predicted value, or both.")
        |y_true = scored_df[${pyStringLiteral(actualValueColumn)}]

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/sklearn/SklearnClassifierOpDesc.scala

@@ -87,6 +87,7 @@ $reportMissingKept
     val narrowTrain = dropNonFeatureColumns("X_train", "")
     val narrowTest = dropNonFeatureColumns("X_test", "")
 
+    // Both frames drop the rows the operator drops, into local names, and each reports its skips.
     s"""${getImportStatements}
        |from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score
        |from sklearn.pipeline import make_pipeline
@@ -95,10 +96,6 @@ $reportMissingKept
        |import numpy as np
        |import pandas as pd
        |
-       |# The same rows the operator drops, and on both frames: the model is fitted
-       |# on one and scored on the other, so narrowing only one side would fit and
-       |# score on different data. A local name rather than a reassignment, since
-       |# the input variable belongs to whichever operator produced it.
        |_train = ${dropMissingRowsStandalone("in1df")}
        |if len(_train) < len(in1df):
        |    print("Skipped", len(in1df) - len(_train), "of", len(in1df), "rows with missing values")
@@ -108,8 +105,6 @@ $reportMissingKept
        |model = make_pipeline(${vectorizerStage(c => pyStringLiteral(c))}$tfidfPart$estimator()).fit(X_train, Y_train)
        |
        |_test = ${dropMissingRowsStandalone("in2df")}
-       |# The operator runs once per port, so it says this for the table it scores
-       |# as well as for the one it fits.
        |if len(_test) < len(in2df):
        |    print("Skipped", len(in2df) - len(_test), "of", len(in2df), "rows with missing values")
        |Y_test = _test[$targetLit]

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/sklearn/SklearnLinearRegressionOpDesc.scala

@@ -104,16 +104,13 @@ class SklearnLinearRegressionOpDesc
 
   override def generateStandaloneCode(): String = {
     val targetLit = pyStringLiteral(target)
+    // Both frames drop the rows the operator drops, into local names, and each reports its skips.
     s"""from sklearn.metrics import mean_absolute_error, r2_score
        |from sklearn.pipeline import make_pipeline
        |from sklearn.linear_model import LinearRegression
        |from sklearn.preprocessing import PolynomialFeatures
        |import pandas as pd
        |
-       |# The same rows the operator drops, and on both frames: the model is fitted
-       |# on one and scored on the other, so narrowing only one side would fit and
-       |# score on different data. Local names rather than reassignments, since the
-       |# input variables belong to whichever operators produced them.
        |_train = in1df.dropna()
        |if len(_train) < len(in1df):
        |    print("Skipped", len(in1df) - len(_train), "of", len(in1df), "rows with missing values")
@@ -127,8 +124,6 @@ class SklearnLinearRegressionOpDesc
        |model = pipeline.fit(X_train, Y_train)
        |
        |_test = in2df.dropna()
-       |# The operator runs once per port, so it says this for the table it scores
-       |# as well as for the one it fits.
        |if len(_test) < len(in2df):
        |    print("Skipped", len(in2df) - len(_test), "of", len(in2df), "rows with missing values")
        |Y_test = _test[$targetLit]

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/sklearn/testing/SklearnTestingOpDesc.scala

@@ -132,28 +132,18 @@ class SklearnTestingOpDesc
     val isRegressionStr = if (isRegression) "True" else "False"
     val modelLit = pyStringLiteral(model)
     val targetLit = pyStringLiteral(target)
+    // Drops on every column as the executor does, since a fitted model hides its estimator.
+    // Each row's model scores the frame itself, kept two-dimensional for predict.
     s"""from sklearn.metrics import accuracy_score, f1_score, precision_score, recall_score, root_mean_squared_error, mean_absolute_error, r2_score
        |
        |out1df = in1df.copy()
-       |# The same drop the executor makes before it scores: the model arrives
-       |# already fitted, so this operator cannot ask which estimator it holds and
-       |# drops on every column to be safe. Scoring the rows it skips would answer
-       |# with a number the run never reported, or fail inside scikit-learn.
        |rows_read = len(in2df)
        |scored_df = in2df.dropna()
        |if len(scored_df) < rows_read:
        |    print("Skipped", rows_read - len(scored_df), "of", rows_read, "rows with missing values")
        |Y = scored_df[$targetLit]
        |X = scored_df.drop($targetLit, axis=1)
        |${narrowToFittableColumns("X", "")}
-       |# predict wants (n_samples, n_features). Squeezing the frame first collapsed
-       |# a single feature, or a single surviving row, to one dimension, which
-       |# scikit-learn rejects.
-       |#
-       |# The model port carries one model per row and the executor scores each of
-       |# them on its own, so a row has to be given the score of the model it holds.
-       |# One score written down the column would report the first model's answer
-       |# for all of them.
        |predictions = [_model.predict(X) for _model in in1df[$modelLit]]
        |if $isRegressionStr:
        |    out1df["R2"] = [r2_score(Y, _p) for _p in predictions]

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/sklearn/training/SklearnTrainingOpDesc.scala

@@ -74,15 +74,13 @@ $reportMissingKept
     val modelNameLit = pyStringLiteral(getUserFriendlyModelName)
     val narrowX = dropNonFeatureColumns("X", "")
 
+    // The rows the operator drops, into a local name, since the input belongs to its producer.
     s"""${getImportStatements}
        |from sklearn.pipeline import make_pipeline
        |from sklearn.compose import ColumnTransformer
        |from sklearn.feature_extraction.text import CountVectorizer, TfidfTransformer
        |import pandas as pd
        |
-       |# The same rows the operator drops. A local name rather than a
-       |# reassignment, since the input variable belongs to whichever operator
-       |# produced it.
        |_train = ${dropMissingRowsStandalone("in1df")}
        |if len(_train) < len(in1df):
        |    print("Skipped", len(in1df) - len(_train), "of", len(in1df), "rows with missing values")

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/unneststring/UnnestStringOpDesc.scala

@@ -152,10 +152,8 @@ class UnnestStringOpDesc extends FlatMapOpDesc with StandaloneCodeGenerator {
       .flatMap(schema => scala.util.Try(schema.getAttribute(attribute)).toOption)
       .map(_.getType)
     val column = renderedAsText(s"out1df[$attributeLit]", declared)
-    s"""# Nothing in the column unnests to nothing, the way the operator answers a null
-       |# field with no rows at all. Dropped before the split rather than after: the
-       |# rendering would turn the empty cell into text and unnest that.
-       |out1df = in1df[in1df[$attributeLit].notna()].copy()
+    // A null unnests to no rows, so it is dropped before the rendering turns it into text.
+    s"""out1df = in1df[in1df[$attributeLit].notna()].copy()
        |out1df[$resultLit] = $column.str.split($delim, regex=True)
        |out1df = out1df.explode($resultLit, ignore_index=True)
        |out1df = out1df[(out1df[$resultLit].notna()) & (out1df[$resultLit] != "")].reset_index(drop=True)""".stripMargin

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/visualization/IcicleChart/IcicleChartOpDesc.scala

@@ -147,7 +147,7 @@ class IcicleChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCo
     // reason for "no chart" is where the reader looks for the chart — printing it
     // to the terminal alone left that file absent. render_error's continuation
     // line keeps the runtime path's indentation, since the HTML is triple-quoted
-    // and those spaces reach the browser.
+    // and those spaces reach the browser. chart_df is a copy: the input can feed another branch.
     s"""def render_error(error_msg):
        |    return '''<h1>Icicle chart is not available.</h1>
        |                  <p>Reason is: {} </p>
@@ -161,8 +161,6 @@ class IcicleChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCo
        |if in1df.empty:
        |    fail("input table is empty.")
        |else:
-       |    # On a copy: the same frame can feed another branch of the plan, and
-       |    # both the assignment and the drop below would otherwise reach it.
        |    chart_df = in1df.copy()
        |    chart_df[$valueLit] = chart_df[chart_df[$valueLit] > 0][$valueLit]
        |    chart_df = chart_df.dropna(subset=[$attributes])

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/visualization/barChart/BarChartOpDesc.scala

@@ -180,7 +180,8 @@ class BarChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCode
     // reason for "no chart" is where the reader looks for the chart — printing it
     // to the terminal alone left that file absent. render_error's continuation
     // line keeps the runtime path's indentation, since the HTML is triple-quoted
-    // and those spaces reach the browser.
+    // and those spaces reach the browser. Empty is checked before the dropna, since an empty
+    // table read back from JSONL has no columns to name, and chart_df keeps the input whole.
     s"""def render_error(error_msg):
        |    return '''<h1>Bar chart is not available.</h1>
        |                  <p>Reason is: {} </p>
@@ -194,13 +195,8 @@ class BarChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCode
        |if $fieldsLit == $valueLit:
        |    fail("Fields should not have the same value.")
        |elif in1df.empty:
-       |    # Checked before the dropna, unlike the runtime path: an empty table read
-       |    # back from JSONL carries no columns at all, so naming them in `subset`
-       |    # would raise a KeyError instead of reporting the empty table.
        |    fail("Table should not have any empty/null values or fields.")
        |else:
-       |    # Bound to a name of its own: the same frame can feed another branch
-       |    # of the plan, which must still see every row.
        |    chart_df = in1df.dropna(subset=[$valueLit, $fieldsLit])
        |    if chart_df.empty:
        |        fail("Table should not have any empty/null values or fields.")

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/visualization/boxViolinPlot/BoxViolinPlotOpDesc.scala

@@ -171,7 +171,7 @@ class BoxViolinPlotOpDesc extends PythonOperatorDescriptor with PlotlyStandalone
     // reason for "no chart" is where the reader looks for the chart — printing it
     // to the terminal alone left that file absent. render_error's continuation
     // line keeps the runtime path's indentation, since the HTML is triple-quoted
-    // and those spaces reach the browser.
+    // and those spaces reach the browser. chart_df keeps the input whole for other branches.
     s"""def render_error(error_msg):
        |    return '''<h1>Box/Violin Plot is not available.</h1>
        |                  <p>Reason is: {} </p>
@@ -185,8 +185,6 @@ class BoxViolinPlotOpDesc extends PythonOperatorDescriptor with PlotlyStandalone
        |if in1df.empty:
        |    fail("input table is empty.")
        |else:
-       |    # Bound to a name of its own: the same frame can feed another branch
-       |    # of the plan, which must still see every row.
        |    chart_df = in1df.dropna(subset=[$valueLit])
        |    if chart_df.empty:
        |        fail("value column contains only non-positive numbers or nulls.")

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/visualization/bubbleChart/BubbleChartOpDesc.scala

@@ -181,7 +181,8 @@ class BubbleChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCo
     // to the terminal alone left that file absent. The heading is the
     // runtime path's, TreeMap and all, so both paths say the same thing.
     // render_error's continuation line keeps the runtime path's indentation, since
-    // the HTML is triple-quoted and those spaces reach the browser.
+    // the HTML is triple-quoted and those spaces reach the browser. chart_df keeps the
+    // input whole for other branches.
     s"""def render_error(error_msg):
        |    return '''<h1>TreeMap is not available.</h1>
        |                  <p>Reasons are: {} </p>
@@ -195,8 +196,6 @@ class BubbleChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCo
        |if in1df.empty:
        |    fail("Input table is empty.")
        |else:
-       |    # Bound to a name of its own: the same frame can feed another branch
-       |    # of the plan, which must still see every row.
        |    chart_df = in1df.dropna(subset=[$xLit, $yLit, $zLit])
        |    if chart_df.empty:
        |        fail("No valid rows left (every row has at least 1 missing value).")

common/workflow-operator/src/main/scala/org/apache/texera/amber/operator/visualization/bulletChart/BulletChartOpDesc.scala

@@ -268,7 +268,8 @@ class BulletChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCo
         .mkString("[", ", ", "]")
 
     // render_error's continuation line keeps the runtime path's indentation — the
-    // HTML is triple-quoted, so those spaces reach the browser.
+    // HTML is triple-quoted, so those spaces reach the browser. The JSON holds one chart
+    // per row, and a lone figure is written as one object, as every one-chart operator does.
     s"""def render_error(error_msg):
        |    return '''<h1>Bullet chart is not available.</h1>
        |                  <p>Reason: {} </p>'''.format(error_msg)
@@ -353,9 +354,6 @@ class BulletChartOpDesc extends PythonOperatorDescriptor with PlotlyStandaloneCo
        |                        html_chunks.append(html_chunk)
        |                    except Exception as e:
        |                        html_chunks.append(render_error(f"Error generating bullet chart: {str(e)}"))
-       |                # One chart per row, so the file holds the whole sequence.
-       |                # A single figure still writes as a lone object, which is
-       |                # what every one-chart operator writes.
        |                if len(figs) == 1:
        |                    figs[0].write_json(outputJson)
        |                elif figs: