Operator catalog
| Operator | Versions | Executor | Trust | Downloads | Actions |
|---|---|---|---|---|---|
piflow_email_cleanPiFlow · Data cleaning Convert full-width characters, lowercase the address and return an empty string for malformed addresses. Replace the selected string columns while preserving other fields. | python12.6 KiB | 5 | |||
piflow_evaluate_json_pathPiFlow · JSON Extract one field from each JSON object. Dot-separated paths traverse nested objects; missing fields return null. | python19.9 KiB | 4 | |||
piflow_excel_parserPiFlow · Excel Read the first sheet of each XLS or XLSX Item. Normalize first-row headers and merge string columns by name. | python699.8 KiB | 4 | |||
piflow_excel_readPiFlow · Data conversion XLSX byte stream to string records (no inference), dates as ISO strings, blanks null, cached formula values only. Params optional header boolean(default true),dataAddress [Sheet!]A1[:F35]. Replaces filePath with source bytes. Limits compressed/expanded128MiB and one million selected cells; materializes workbook bytes, preserves worksheet selection. No legacy XLS. | python519.3 KiB | 9 | |||
piflow_excel_writePiFlow · Data conversion Records to XLSX byte stream; replaces filePath/saveMode with a new workbook output. Params optional header boolean(default true),dataAddress start cell e.g. Sheet1!A1 (no bounded write ranges). Dates become ISO text, formula-looking strings remain text. Bounded128MiB/one million rows; temporary XLSX serialization, not constant-memory streaming. Vendored openpyxl3.1.5/et_xmlfile2.0.0. | python518.5 KiB | 6 | |||
piflow_excel_write_multiple_sheetsPiFlow · Excel Write one worksheet for each of the left and right record inputs, and return one XLSX stream. | python538.5 KiB | 4 | |||
piflow_excel_write_sortPiFlow · Excel Sort records on the selected columns and emit an XLSX byte stream with data written at the requested cell. | python560.2 KiB | 3 | |||
piflow_execute_sql_stopData transformation · SQL Read-only SQLite SQL over a named input relation (not Spark SQL). Params: sql, viewName. Scalar inputs, max 128 MiB. Output types inferred; empty computed columns have null type. | jvm61.0 KiB | 6 | |||
piflow_field_type_conversionPiFlow · Data conversion Cast selected fields to double precision, 32-bit integer or fixed-precision decimal. Preserve other columns. | python35.7 KiB | 3 | |||
piflow_filterData transformation · SQL Filter Arrow batches by a SQLite SQL predicate (not Spark SQL). Params: condition, e.g. quality = 0 AND temperature >= 12. Scalar columns only. | jvm61.0 KiB | 9 | |||
piflow_flatten_xml_parserPiFlow · XML Flatten a selected XML parent and optionally emit one row per child of a nested collection. | python19.9 KiB | 4 | |||
piflow_folder_csv_parserPiFlow · CSV Read CSV items from one folder. Skip a first row matching the configured schema and preserve field text. | python30.5 KiB | 5 | |||
piflow_folder_json_parserPiFlow · JSON Read JSON objects and arrays from one folder and concatenate by field name. | python30.5 KiB | 4 | |||
piflow_forkPiFlow · Streams Duplicate each input batch to two fixed ports, left and right. No params. Adaptation: explicit ports replace PiFlow dynamic outports. | python8.5 KiB | 8 | |||
piflow_front_and_back_values_are_validPiFlow · Data processing EcoFlow same-hour interpolation using valid preceding and following days. Requires day_id,hour. Params fields and qualityControlMarkerValue integer string; fixes upstream ignored marker. | jvm190.8 KiB | 6 | |||
piflow_gaussian_mixture_predictionPiFlow · Machine learning Apply a trained Spark Gaussian mixture model to Arrow records and append prediction. | jvm22.6 KiB | 5 | |||
piflow_gaussian_mixture_trainingPiFlow · Machine learning Train a Spark Gaussian mixture model on Arrow records and emit the trained model as a ZIP byte stream. | jvm21.2 KiB | 4 | |||
piflow_gbt_predictionPiFlow · Machine learning Apply a trained Spark gradient-boosted trees model to Arrow records and append prediction. | jvm22.5 KiB | 5 | |||
piflow_gbt_trainingPiFlow · Machine learning Train a Spark gradient-boosted trees model on Arrow records and emit the trained model as a ZIP byte stream. | jvm21.1 KiB | 4 | |||
piflow_histogramPiFlow · Visualization Produce grouped histogram data as records, not an image. Params: abscissa, customizedProperties mapping columns to COUNT/SUM/AVG/MAX/MIN. SQLite aggregates, scalar columns, 128 MiB. | python10.5 KiB | 5 | |||
piflow_html_parserPiFlow · Documents Extract visible UTF-8 text in document order. Each logical text block becomes one record. | python19.9 KiB | 4 | |||
piflow_identity_number_cleanPiFlow · Data cleaning Normalize Chinese identity number text and check the final checksum character of an 18-digit number. Replace the selected string columns while preserving other fields. | python12.6 KiB | 5 | |||
piflow_joinData transformation · Relational Key-based inner/left/right/full_outer record join. Params: joinMode, correlationColumn. Colliding right non-key columns receive _right suffix. 128 MiB per input. | jvm52.0 KiB | 8 | |||
piflow_json_parserPiFlow · JSON Read one UTF-8 JSON object or a top-level array of objects as Arrow records, preserving nested fields. | python12.6 KiB | 4 | |||
piflow_json_savePiFlow · JSON Serialize records as UTF-8 JSON Lines. The result is a bytes stream and each line is one JSON object. | python12.6 KiB | 4 | |||
piflow_json_string_parserPiFlow · JSON Parse a literal JSON object or array of objects into a records stream. | python12.6 KiB | 5 | |||
piflow_kmeans_predictionPiFlow · Machine learning Apply a trained Spark k-means model to Arrow records and append prediction. | jvm22.5 KiB | 5 | |||
piflow_kmeans_trainingPiFlow · Machine learning Train a Spark k-means model on Arrow records and emit the trained model as a ZIP byte stream. | jvm21.2 KiB | 4 | |||
piflow_label_propagationPiFlow · Graphs Run synchronous label propagation using both directions of each edge. The lowest label wins a tie. | python30.6 KiB | 5 | |||
piflow_lda_predictionPiFlow · Machine learning Apply a trained Spark online latent Dirichlet allocation model to Arrow records and append topicDistribution. | jvm22.5 KiB | 5 | |||
piflow_lda_trainingPiFlow · Machine learning Train a Spark online latent Dirichlet allocation model on Arrow records and emit the trained model as a ZIP byte stream. | jvm21.1 KiB | 4 | |||
piflow_line_chartPiFlow · Visualization Produce grouped line-chart data as records, not an image. Params: abscissa, customizedProperties mapping columns to COUNT/SUM/AVG/MAX/MIN. SQLite aggregates, scalar columns, 128 MiB. | python10.5 KiB | 5 | |||
piflow_load_graphPiFlow · Graphs Read whitespace-separated integer edges and output edge and distinct vertex records. | python30.5 KiB | 6 | |||
piflow_logistic_regression_predictionPiFlow · Machine learning Apply a trained Spark logistic regression model to Arrow records and append prediction. | jvm22.6 KiB | 5 | |||
piflow_logistic_regression_trainingPiFlow · Machine learning Train a Spark logistic regression model on Arrow records and emit the trained model as a ZIP byte stream. | jvm21.3 KiB | 4 | |||
piflow_max_min_normalizationPiFlow · Statistics Scale the selected column to [0, 1]. The step materializes the input; a constant column produces nulls. | python12.7 KiB | 4 | |||
piflow_mdv_gPiFlow · Data processing EcoFlow MDV sliding-window interpolation at the same hour on neighboring days, excluding current day. Requires day_id,hour. Params fields,mdv_num (positive odd string),qualityControlMarkerValue (integer string). | jvm190.5 KiB | 5 | |||
piflow_mdv_iPiFlow · Data processing EcoFlow fixed-window MDV interpolation grouped by ceiling(day_id/mdv_num),hour. Preserves original zero contribution of flagged values to the group mean. Adds zone; fixes multiple-field replacement. Params fields,mdv_num positive string,qualityControlMarkerValue integer string. | jvm190.5 KiB | 5 | |||
piflow_mergeData transformation · Relational Stream left records followed by right records, retaining duplicates. Merge schemas by field name: preserve left field order, append right-only fields and fill missing values with null. Conflicting types are rejected. No params. Fixed left/right ports. | jvm52.0 KiB | 11 | |||
piflow_mock_dataPiFlow · Data processing Random Arrow records. source is a trigger stream consumed and discarded (matching upstream unused default input); output independent of input. Params schema e.g. name:String,age:Int,optional:Double:true; count string0..1000000. Types String/Int/Long/Float/Double/Boolean. Batches1024. | python10.3 KiB | 5 |