Skip to contents

Reads what runPipeline() wrote: a table for each analysis step of the pipeline, named by the step's path through it, as windowing(60, time.unit = "seconds") / maad_temporal_entropy(), in the order of its outline. Each table holds one measurement at one grain, so values from windows of different lengths, or from raw and filtered audio, are never in the same table.

Usage

pipelineResults(store, p, step = NULL)

Arguments

store

The folder runPipeline() wrote to.

p

The Pipeline, or its id. The id changes with the code of the packages a pipeline uses, so results written before a package was updated are read with the id they were written under, which is the name of their folder in the store.

step

NULL for every table, or the name of one, or its number in the outline of the pipeline, as "2.1", to read that table alone.

Value

A named list of data frames, one for each analysis step, or a single data frame where step is given. The list has a "files" attribute, a data frame of every file the store holds: its file hash, the uri of its first recording, its location, its status, "ok" or "error", its error, the time it was run and the batch it was run in.

Details

Every table has these columns, in this order:

  • file: the SHA-256 hash of the file the row was measured on, which joins to the hash of the set that runPipeline() gives back.

  • pipeline: the id of the pipeline.

  • cut_start, cut_end, window_start, window_end, window_length: where in the recording the row was measured, in seconds from its start, for a step inside a cut or a windowing, as pipeline() describes.

  • window_time: when the window began, where all the recordings of the file give the same start.

  • error: why the file failed, or NA.

  • value and the columns beginning value.: the result, as flattenOutput() gives it.

A file that failed has one row in every table, holding only its file, pipeline and error. A file in which a step found nothing has no rows in its table, as flattenOutput() gives none for NULL, and the "files" attribute tells such a file apart from one that was never run. A table is listed once a file has run without failing.

A file that was run more than once, by retry = TRUE, gives the rows of its last run that did not fail, or its last error where every run failed. A file at a URL that failed before its hash was known is matched to its hash once a later run finds it.

See also