Skip to content

Transforms

Module: rs_rich.ext.transform (Rust: rich_ext::transform). A Pipeline runs named stages in order. A stage is any object with an apply(value) method or any callable; its result feeds the next stage.

from rs_rich.console import Console
from rs_rich.ext.transform import HighlightMatches, KeepLines, Pipeline
from rs_rich.text import Text

console = Console(width=60)
log = Text("ok: started\nerror: disk full\nok: retry\nerror: gave up")
pipeline = (
    Pipeline()
    .then("errors", KeepLines("^error"))
    .then("mark", HighlightMatches("disk", "bold"))
    .then("upper", lambda text: Text(text.plain.upper()))
)
console.print(pipeline.apply(log))
print(KeepLines("^ok", invert=True)(log).plain)

ERROR: DISK FULL
ERROR: GAVE UP
error: disk full
error: gave up

KeepLines(pattern, invert=False) keeps the lines of a Text matching a regular expression; HighlightMatches(pattern, style) styles every match. Other modules provide stages for their own values: data.Select, data.Filter, data.Highlight, data.Redact for a data.Document, table.TableSort and table.TableGroup for a TableData, and diff.KeepFiles for a Patch.

A stage that raises TransformError stops the pipeline with a PipelineError naming the stage:

from rs_rich.ext.transform import PipelineError, TransformError

def reject(text):
    raise TransformError("empty input")

try:
    Pipeline().then("check", reject).apply(Text(""))
except PipelineError as error:
    print(error, "|", error.stage, "|", error.reason)

check: empty input | check | empty input

Pipeline or TextPipeline?

Both exist because Rust has both, and they answer different questions. Pipeline (this page, rich_ext::transform) is built in code from stage objects: any value, not only Text, with TransformError and PipelineError naming the stage. TextPipeline (from ExtensionRegistry.text_pipeline(names), rich_plugin_api) is assembled from transforms registered by name by plugins, works on Text only, and fails with PluginError(kind="pipeline"). A KeepLines or HighlightMatches stage can be registered as a plugin transform, so the same stages serve both; see Using what is registered.