This product bundles software and data that originate from other projects.
The Apache License, Version 2.0 applies to this product as a whole; the
notices below identify what is third-party and under what terms it may be
redistributed.

--------------------------------------------------------------------------------
1. Upstream R packages (code, data and documentation)
--------------------------------------------------------------------------------

Derived from:

  * pharmaverseadam - "ADaM Test Data for the 'Pharmaverse' Family of
    Packages", taken from an upstream development version
    https://github.com/pharmaverse/pharmaverseadam
    shipped as the pharmadata.pharmaverseadam collection
  * pharmaversesdtm - "SDTM Test Data for the 'Pharmaverse' Family of
    Packages", taken from an upstream development version
    https://github.com/pharmaverse/pharmaversesdtm
    shipped as the pharmadata.pharmaversesdtm collection

Both upstream packages are also published on CRAN. This product has its own
version (0.1.0) that does not identify an upstream release; data-raw/export
scripts read the upstream sources, so running them against your own checkout
reproduces these inputs without the shipped parquet files.

Both are licensed under the Apache License, Version 2.0 or later (their
DESCRIPTION files declare `License: Apache License (>= 2)` and
`License: Apache License (>= 2.0)` respectively), and their copyright
holders - the entries with role "cph" in the Authors@R field of their
DESCRIPTION files - are:

  Copyright Cytel Inc.
  Copyright F. Hoffmann-La Roche AG
  Copyright GlaxoSmithKline LLC

The ADaM datasets come from the admiral template scripts
(https://github.com/pharmaverse/admiral, Apache-2.0) applied to the SDTM
datasets above. Their labels, types, mandatory flags and roles come from
`pharmaverseadam/inst/extdata/adams-specs.json`, drawn from the CDISC ADaM
Implementation Guide.

Changes made by this product (per Apache-2.0 Section4(b)):

  * `.rda` -> Parquet conversion, plus per-dataset metadata tables
    exported alongside (the pharmaverseadam and pharmaversesdtm collections);
  * bytes that are invalid UTF-8 in the upstream `.rda` (for example the
    0x92 right single quotation mark in the `ts` dataset) are re-decoded as
    CP1252, and an empty character is exported as a Parquet null - the missing
    value it stands for. No other value is edited;
  * column labels and roles are additionally embedded in the Parquet files as
    Arrow field metadata, and a merged `_meta.json` plus generated IDE stubs and
    MkDocs documentation are added.

--------------------------------------------------------------------------------
2. The CDISC pilot submission packages, redistributed directly
--------------------------------------------------------------------------------

The pharmadata.cdiscpilotadam and pharmadata.cdiscpilotsdtm collections are
built from two archives published by the PHUSE Scripts repository:

  * data/adam/cdiscpilot_update1.zip
  * data/sdtm/cdiscpilot_update2.zip
    https://github.com/phuse-org/phuse-scripts

They are the CDISC SDTM/ADaM Pilot Project's own submission packages as updated
by the PHUSE Test Data Factory working group (the repository's
`data/sdtm/README.md` names those updates). The PHUSE Scripts repository is
licensed under the MIT (Expat) licence, names no copyright holder, and is in its
`LICENSE.md`. The data inside the archives is CDISC's; section 3 records the
terms it carries.

Changes made by this product (per Apache-2.0 Section4(b)):

  * SAS transport files (`.xpt`) -> Parquet conversion, and the per-dataset
    metadata tables are exported alongside (the cdiscpilotadam and cdiscpilotsdtm
    collections, sharing a common helper);
  * a blank character is exported as a Parquet null, the missing value it stands
    for: the SAS transport format has no missing character value, so a text
    variable a record does not use is written there as an empty string;
  * the dataset label, the structure, the mandatory flags and the roles are read
    from the Define-XML 2.0 each archive ships, and the column labels from the
    transport files' own variable labels with the Define-XML as fallback;
  * those labels and roles are additionally embedded in the Parquet files as
    Arrow field metadata, and a merged `_meta.json` plus generated IDE stubs and
    MkDocs documentation are added.

No other data value is edited: the conversion is a container change, and the
metadata published beside the data is the one the submission package itself
records.

--------------------------------------------------------------------------------
3. The CDISC pilot project data at the root of the chain
--------------------------------------------------------------------------------

Both routes above lead back to the CDISC SDTM/ADaM Pilot Project, whose files are
published at

  https://github.com/cdisc-org/sdtm-adam-pilot-project

and that repository states a Terms of Use for the data, of which the following
conditions travel with any redistribution:

  * attribution back to CDISC is required, and users must not be misled about
    the origin of the data;
  * the data may not be sublicensed or distributed for a fee;
  * the data may not be used to compete with CDISC, or in any manner that
    violates a law or regulation;
  * the data is provided "AS IS", without warranties of any kind, and CDISC
    disclaims the currency, accuracy, relevance and completeness of it.

This product is distributed free of charge, attributes CDISC as the source of the
underlying data, and makes no claim that the data is fit for any regulatory or
analytical purpose: it is synthetic test data for exercising software.

Note that the pilot project's Terms of Use also ask that the data not be
"disassembled, reverse engineered, decompiled, modified, or altered". The
upstream pharmaverse packages already rework it, and this product's Parquet
conversion is a format change that reads a blank character as the missing value
it stands for. If a stricter reading of that clause matters to your use case, the
regeneration programs in `data-raw/` rebuild the same datasets from the upstream
sources without shipping Parquet files.

--------------------------------------------------------------------------------
4. CDISC standards documentation
--------------------------------------------------------------------------------

CDISC standards and Implementation Guides are copyrighted by CDISC, which grants
open public use of its published standards (https://www.cdisc.org/). This product
reproduces the variable names, labels, mandatory flags and roles recorded in the
upstream specs and Define-XML files; it does not reproduce the Implementation
Guide documents themselves.

--------------------------------------------------------------------------------
5. No endorsement
--------------------------------------------------------------------------------

Cytel, F. Hoffmann-La Roche AG, GlaxoSmithKline LLC, CDISC, PHUSE and the
pharmaverse initiative are not affiliated with this product and do not endorse it.
The datasets carry no real subject data: they are simulated test data.
