finprint

package finprint

finprint: record how a dataset was made in pandas, save the recipe beside the output, replay it to check, apply it to new data of the same shape.

%load_ext finprint                 # notebooks: also records cell history
import finprint as fp
rec = fp.session("northwind weekly prep")
raw = rec.read_csv("northwind_raw.csv")
clean = raw.drop(index=[12, 13])     # recorded
rec.save(clean, "northwind_weekly_clean.csv")   # writes the .csv and northwind_weekly_clean.recipe.json

Blueprint: docs/poc/TRANSFORM-RECIPES.md. Works without IPython.

Read the finprint guide first for the main entry points.

Modules

Module	What it holds
finprint.apply	Apply: run a saved recipe on new inputs of the same shape (more weeks, an extra column), not to check it.
finprint.catalogue	Thin HTTP client for apps/data (port 8108): read a catalogued dataset, publish an output with its recipe.
finprint.cli	finprint: show, verify, apply, export and schema.
finprint.codes	Note codes a recipe can carry (blueprint docs/poc/TRANSFORM-RECIPES.md sections 3 to 6).
finprint.describe	Plain-language summary of a recipe (finprint show): “Dropped 2 rows: 156 to 154”, “Added spend_net”.
finprint.encoding	Tagged JSON for step arguments and series expressions, and the inverse used by replay.
finprint.export	Export a recipe in existing wire formats (load-bearing decision 9): an OpenLineage RunEvent with schema and columnLineage facets, and W3C PROV-JSON with entities, activities, used and wasGeneratedBy.
finprint.fingerprint	Canonical-rows fingerprint, file hashes and the write-and-read-back helpers.
finprint.notebook	IPython extension: %load_ext finprint.
finprint.recording	The recording session (rec): readers, named steps, save, publish.
finprint.redact	Privacy rules for what goes into a recipe (blueprint section 4).
finprint.replay	Replay check: rerun a recipe from its inputs and compare fingerprints step by step.
finprint.schema	The recipe file, <output>.recipe.json (schema ump.transform_recipe/0.1).
finprint.stepcode	What a named step’s code needs besides its arguments, checked when the step runs.
finprint.tracked	Proxies that record what is done to a frame: TrackedFrame, TrackedSeries and their helpers.
finprint.trust	Which named-step code this machine recorded, so replay runs only that code without asking.

Re-exports

Documented on the page of the module that defines them.

Name	Kind	Summary
APPLY_CODES	constant
CODES	constant
FINGERPRINT_BASIS	constant
NOTE_CODES	constant
SCHEMA_VERSION	constant
ApplyCode	class
ApplyResult	class
CatalogueError	class
NoteCode	class
RecipeError	class
RecipePrivacyWarning	class	Cell source or step code that looks like it holds an email address or a key.
RecipeWarning	class	A recipe was saved with a note the analyst should see (partial capture, transient data).
ReplayResult	class
SaveResult	class
Session	class	Records reads, verbs and named steps on tracked frames. Create one with finprint.session(title).
TrackedBase	class	Base of the proxies, so the encoder can recognise them without importing tracked.
TrackedDeferred	class	A tracked groupby (or rolling, resample, expanding, ewm): the next aggregation is recorded as one step.
TrackedFrame	class	A DataFrame proxy bound to a session. Every frame version has an id (in1, f7…).
TrackedSeries	class	A Series proxy that remembers its expression over frame versions (never recorded as a step by itself).
TransformRecipe	class	<output>.recipe.json: inputs by hash, steps, notebook cells, the output’s fingerprint and the analyst’s own replay check.
TransientDataError	class
apply_verification	function
column_origins	function	For each output column, the input columns it came from, and a note for columns of unknown origin.
current	function
json_schema	function	JSON Schema 2020-12 for TransformRecipe (the form the published file holds).
json_schema_text	function
load_recipe	function
round_trip_fingerprint	function	Fingerprint of the frame as it would be written and read back, plus the bytes and the index flag.
row_bound_steps	function	Steps whose recorded arguments address rows by literal position or label, so they cannot carry over to new data ([{"step_id", "op", "reason"}]). Pure: reads only the recorded op and arguments.
data_derived_steps	function	Steps that use a literal the analyst took out of a value computed from the data (float(s.median()), int(s.max()), f"{s.mean():.1f}", np.float64(x), math.log(x)) or that follow an if on such a value. The recorder marks them (args.data_derived: how the value was taken out); on new data they would reuse the recorded value, so apply refuses them. Values kept as they come from pandas are recorded as expressions ({"__scalar__": ...}) and recomputed instead.
session	function	Start a recording session and make it current. In a notebook where %load_ext finprint started an empty session, that one is renamed and returned so the cells recorded so far stay with it.
sha256_file	function
step	function	@finprint.step: like @rec.step but bound to whichever session is current when the function runs.
summary	function	Multi-line text for finprint show.
to_openlineage	function	An OpenLineage 2-0-2 RunEvent (eventType COMPLETE) for the recipe.
to_prov	function	W3C PROV-JSON: inputs, intermediate frames and the output as entities, steps as activities.
untrack	function	The plain pandas object behind a tracked one (the same object, not a copy). Edits made to it are not recorded; the session notices them as unexplained changes and marks the recipe partial.
verify	function	Replay the recipe at path and, unless dry_run, write the verification back into the file.
write_recipe	function

Functions

load_ipython_extension

function

def load_ipython_extension(ip: Any) -> None

%load_ext finprint.

Source: packages/finprint/src/finprint/__init__.py, line 54

unload_ipython_extension

function

def unload_ipython_extension(ip: Any) -> None

Source: packages/finprint/src/finprint/__init__.py, line 59

Constants

Name	Type	Value	Description
__version__		'0.2.0'