LabFREED for Python

This is a Python implementation of LabFREED building blocks.

LabFREED itself is an open, vendor-neutral community initiative for pragmatic lab digitalization, contributed to mainly by ApiniLabs, Buchi, and wega, with 40+ supporting vendors/adopters. See labfreed.org for the building-block specs themselves, or join the community Discord for questions and discussion. This repo is only the Python reference implementation.

Contents

• LabFREED for Python
  • Contents
  • Supported Building Blocks
  • Installation
  • Using with Claude Code
  • Package Structure
  • Design Philosophy
  • Usage Examples
    • Parse a simple PAC-ID
    • Show recommendations:
    • Save as QR Code
    • PAC-CAT
    • Parse a PAC-ID with extensions
      • Display Name
      • TREX
    • Create a PAC-ID with Extensions
      • Create PAC-ID
      • Create a TREX
      • Combine PAC-ID and TREX and serialize
  • PAC-ID Resolver
  • PAC-ID Attributes
  • Versioning
  • Change Log
    • v1.0.0
    • v0.2.12
    • v0.2.11
    • v0.2.10
    • v0.2.9
    • v0.2.8
    • v0.2.7
    • v0.2.6
    • v0.2.5
    • v0.2.4
    • v0.2.3
    • v0.2.2
    • v0.2.1
    • v0.2.0b2
    • v0.1.1
    • v0.1.0
    • v0.0.20
    • v0.0.19
  • Getting Help / FAQ
  • Contributing
• Attributions

Supported Building Blocks

• PAC-ID

  • Parsing
  • Serialization

• PAC-CAT

  • Interpretation of PAC-ID as categories

• T-REX

  • Parsing
  • Serialization

• Display Extension

  • base36 <> str conversions

• PAC-ID Resolver

  • support for resolver configuration v2 (improved version)
  • use of multiple resolver configuration

• PAC-ID Attributes

  • client and server code

• Generation of QR codes (PAC-ID with extensions)

• Validation (with Errors Recommendations)

Installation

You can install LabFREED from PyPI using pip:

pip install labfreed

Some parts of the package need extra dependencies that are not installed by default — see Package Structure below.

Using with Claude Code

If you're writing Python against this package with Claude Code, install the bundled skill so Claude already knows the building blocks and has working code examples for every core operation, without you having to explain the ecosystem or paste in examples first.

Prerequisites: the Claude Code CLI installed (claude --version to check).

Install:

/plugin marketplace add retothuerer/LabFREED
/plugin install labfreed@labfreed-plugins

retothuerer/LabFREED (no @ref) resolves to this repo's default branch. If you want the plugin content tied to a released version rather than whatever's currently in active development, pin it explicitly instead:

/plugin marketplace add retothuerer/LabFREED@main

Verify it's working: run /plugin and check labfreed shows up under Installed — or just ask Claude something like "what is a PAC-CAT?" or "write Python to parse this PAC-ID" in any project; the skill should kick in on its own, without you mentioning it by name.

Keeping it up to date: /plugin marketplace update refreshes the marketplace listing from this repo, then /plugin update labfreed updates the installed plugin itself. Plugin content is refreshed as part of every labfreed release (see Versioning).

Uninstall: /plugin uninstall labfreed.

See plugins/labfreed/README.md for exactly what the skill covers.

Package Structure

The labfreed package is organized into three parts, reflecting how far the code strays from being a plain implementation of the building block specifications:

• labfreed/ (core) — the building blocks themselves (PAC-ID, PAC-CAT, T-REX, PAC-ID Resolver, PAC-Attributes), plus the Python-specific convenience code that goes with them (e.g. converting between spec types and native Python types). Only needs the base dependencies installed by pip install labfreed. Units (Quantity, T-REX's UNECE-code mapping) are UCUM throughout — the UCUM unit validator is the quickest way to check whether a unit string is valid UCUM. Structural UCUM validation works with no extra dependency; automatic mapping for compound/non-SI units, full symbol-level validation, and pretty-printing additionally need the units extra:

  pip install labfreed[units]
  

• labfreed/labfreed_extended/ — reference implementations built on top of the library that go beyond representing the specs in Python, such as a Flask-based attribute server and the PAC issuer landing page (see Setting up a PAC-ID Landing Page). Requires the extended extra:

  pip install labfreed[extended]
  

• labfreed/labfreed_experimental/ — early-stage, unstable code with no compatibility guarantees (currently BLE-based PAC discovery). Requires the experimental extra:

  pip install labfreed[experimental]
  

  This is distinct from the @experimental decorator (labfreed.experimental, defined in labfreed_infrastructure.py): labfreed_experimental/ is for whole modules that are undefined/unstable from top to bottom, while @experimental flags a single function, method, or class as a newly-added, still-shifting addition within a module that is otherwise stable and covered by the normal versioning/deprecation policy below. It emits a FutureWarning on every call and adds a note to the docstring.

Design Philosophy

This library optimizes for ease of use over the cleanest possible architecture. Concretely:

• Data types carry their own behavior. Rather than keeping models as plain data holders and pushing parsing, serialization, and validation into separate classes, types like PAC_ID, Quantity, and DataTable expose that behavior directly as methods (PAC_ID.from_url()/.to_url(), Quantity.from_str_value(), DataTable.append()/.get_column()). One import, one object — no separate factory or validator to look up.
• Few factories. Construction from external representations (URLs, strings, payload dicts) goes through classmethods on the type itself (from_url, from_str_value, from_payload_attributes, ...) rather than dedicated *Factory classes. The handful of factories that do exist live in labfreed_extended, not in the core building blocks.
• Validation lives on the model. LabFREED_BaseModel gives every core type is_valid / validation_messages() / print_validation_messages() directly, instead of routing through an external validator.

This is a deliberate tradeoff: a stricter separation-of-concerns design would be easier to unit-test in isolation and would keep each class's responsibility narrower, but it would also mean more classes to import and more indirection to trace for what is, for most users, a small set of well-defined operations — parse an identifier, serialize it, check it's valid.

Usage Examples

⚠️ Note: These examples are building on each other. Imports and parsing are not repeated in each example.

# import built ins
import os

Parse a simple PAC-ID

# Parse the PAC-ID
from labfreed import PAC_ID, LabFREED_ValidationError  

pac_str = 'HTTPS://PAC.METTORIUS.COM/-MD/bal500/@1234'
try:
    pac = PAC_ID.from_url(pac_str)
except LabFREED_ValidationError:
    pass
# Check validity of this PAC-ID
is_valid = pac.is_valid
print(f'PAC-ID is valid: {is_valid}')

>> PAC-ID is valid: True

Show recommendations:

Note that the PAC-ID -- while valid -- uses characters which are not recommended (results in larger QR code). There is a nice function to highlight problems

pac.print_validation_messages()

>> Validation Results                                                              
>> ┌──────────────────────────────────────────────────────────────────────────────┐
>> │ **RECOMMENDATION** in id segment value bal500                                │
>> │ Characters 'a','l','b' should not be used., Characters SHOULD be limited to  │
>> │ upper case letters (A-Z), numbers (0-9), '-' and '+'                         │
>> │                                                                              │
>> │ HTTPS://PAC.METTORIUS.COM/-MD/👉bal👈500/@1234                               │
>> ├──────────────────────────────────────────────────────────────────────────────┤
>> │ **RECOMMENDATION** in id segment value @1234                                 │
>> │ Characters '@' should not be used., Characters SHOULD be limited to upper    │
>> │ case letters (A-Z), numbers (0-9), '-' and '+'                               │
>> │                                                                              │
>> │ HTTPS://PAC.METTORIUS.COM/-MD/bal500/👉@👈1234                               │
>> ├──────────────────────────────────────────────────────────────────────────────┤
>> │ **RECOMMENDATION** in id segment value bal500                                │
>> │ Characters 'a','l','b' should not be used., Characters SHOULD be limited to  │
>> │ upper case letters (A-Z), numbers (0-9), '-' and '+'                         │
>> │                                                                              │
>> │ HTTPS://PAC.METTORIUS.COM/-MD/👉bal👈500/@1234                               │
>> ├──────────────────────────────────────────────────────────────────────────────┤
>> │ **RECOMMENDATION** in id segment value @1234                                 │
>> │ Characters '@' should not be used., Characters SHOULD be limited to upper    │
>> │ case letters (A-Z), numbers (0-9), '-' and '+'                               │
>> │                                                                              │
>> │ HTTPS://PAC.METTORIUS.COM/-MD/bal500/👉@👈1234                               │
>> ├──────────────────────────────────────────────────────────────────────────────┤
>> │ **RECOMMENDATION** in id segment value bal500                                │
>> │ Characters 'a','l','b' should not be used., Characters SHOULD be limited to  │
>> │ upper case letters (A-Z), numbers (0-9), '-' and '+'                         │
>> │                                                                              │
>> │ HTTPS://PAC.METTORIUS.COM/-MD/👉bal👈500/@1234                               │
>> ├──────────────────────────────────────────────────────────────────────────────┤
>> │ **RECOMMENDATION** in id segment value @1234                                 │
>> │ Characters '@' should not be used., Characters SHOULD be limited to upper    │
>> │ case letters (A-Z), numbers (0-9), '-' and '+'                               │
>> │                                                                              │
>> │ HTTPS://PAC.METTORIUS.COM/-MD/bal500/👉@👈1234                               │
>> └──────────────────────────────────────────────────────────────────────────────┘

Save as QR Code

from labfreed.qr import save_qr_with_markers  

save_qr_with_markers(pac_str, fmt='png')

>> Large QR: Provided URL is not alphanumeric!
>> Size: 29
>> Version: 3
>> Error Level: M

PAC-CAT

PAC-CAT defines a (optional) way how the identifier is structured. PAC_ID.from_url() automatically converts to PAC-CAT if possible.

from labfreed import PAC_CAT  
pac_str = 'HTTPS://PAC.METTORIUS.COM/-DR/XQ908756/-MD/bal500/@1234'
pac = PAC_ID.from_url(pac_str)
if isinstance(pac, PAC_CAT):
    categories = pac.categories 
    pac.print_categories()

>> Categories in           
>> HTTPS://PAC.METTORIUS.COM/-DR/XQ90
>>       8756/-MD/bal500/@1234       
>> ┌────────────────────┬───────────┐
>> │ Main Category      │           │
>> │ key ()             │  -DR      │
>> │ id (21)            │  XQ908756 │
>> ├────────────────────┼───────────┤
>> │ Category           │           │
>> │ key ()             │  -MD      │
>> │ model_number (240) │  bal500   │
>> │ serial_number (21) │  @1234    │
>> └────────────────────┴───────────┘

Parse a PAC-ID with extensions

PAC-ID can have extensions. Here we parse a PAC-ID with attached display names and summary.

pac_str = 'HTTPS://PAC.METTORIUS.COM/-MD/BAL500/1234*N$TEXT/WM633OV3E5DGJW2BEG0PDM1EA7*SUM$TREX/WEIGHT$GRM:67.89'
pac = PAC_ID.from_url(pac_str)

Display Name

Note that the Extension is automatically converted to a DisplayNameExtension

display_name = pac.get_extension('N') # display name has name 'N'
print(display_name) 

>> Text: My Balance ❤️

TREX

trexes = pac.get_extension_of_type('TREX')
trex_extension = trexes[0] # there could be multiple trexes. In this example there is only one, though
trex = trex_extension.trex
v = trex.get_segment('WEIGHT')
print(f'WEIGHT = {v.value}')

>> WEIGHT = 67.89

Create a PAC-ID with Extensions

Create PAC-ID

from labfreed import PAC_ID, IDSegment  
from labfreed.well_known_keys.labfreed.well_known_keys import WellKnownKeys  

pac = PAC_ID(issuer='METTORIUS.COM', identifier=[IDSegment(key=WellKnownKeys.SERIAL, value='1234')])
pac_str = pac.to_url()
print(pac_str)

>> HTTPS://PAC.METTORIUS.COM/21:1234

Create a TREX

TREX can conveniently be created from a python dictionary. Note that utility types for Quantity (number with unit) and table are needed

from datetime import datetime  
from labfreed.trex.facade import T_REX  
from labfreed.trex.facade import DataTable  
from labfreed.trex.facade import Quantity  
from labfreed.utilities.quantity import CommonQuantityUnit  

# Value segments of different type
# unit can be set from the CommonQuantityUnit enum (TEMP) or as a plain UCUM string (DURATION below)
segments = {
                'STOP': datetime(year=2024,month=5,day=5,hour=13,minute=6),
                'TEMP': Quantity(value=10.15, unit=CommonQuantityUnit.TEMPERATURE_KELVIN),
                'OK':False,
                'COMMENT': 'FOO',
                'COMMENT2':'£'
            }
mydata = T_REX(segments)

# Create a table
# A bare int/float is accepted for convenience (it's silently wrapped in a unitless
# Quantity, with a warning) - but LabFREED has no unitless numbers as a matter of
# principle, so always prefer an explicit Quantity with a real unit, as done here.
table = DataTable(col_names=['DURATION', 'Date', 'OK', 'COMMENT'])
table.append([Quantity(value=1, unit='h'), datetime.now(), True, 'FOO'])
table.append([Quantity(value=1.1, unit='h'), datetime.now(), True, 'BAR'])
table.append([Quantity(value=1.3, unit='h'), datetime.now(), False, 'BLUBB'])
#add the table to the pytrex
mydata.update({'TABLE': table})

# Create TREX
trex = mydata.to_trex()


# Validation also works the same way for TREX
trex.print_validation_messages()

>> Validation Results                                            
>> ┌────────────────────────────────────────────────────────────┐
>> │ **ERROR** in TREX table column Date                        │
>> │ Column header key contains invalid characters: 't','a','e' │
>> │                                                            │
>> │ STOP$T.D:20240505T1306                                     │
>> │ +TEMP$KEL:10.15                                            │
>> │ +OK$T.B:F                                                  │
>> │ +COMMENT$T.A:FOO                                           │
>> │ +COMMENT2$T.T:12G3                                         │
>> │ +TABLE$$DURATION$HUR:D👉ate👈$T.D:OK$T.B:COMMENT$T.A::     │
>> │  1:20260814T021153.586:T:FOO::                             │
>> │  1.1:20260814T021153.588:T:BAR::                           │
>> │  1.3:20260814T021153.590:F:BLUBB                           │
>> └────────────────────────────────────────────────────────────┘

Combine PAC-ID and TREX and serialize

from labfreed.well_known_extensions import TREX_Extension  
pac.extensions = [TREX_Extension(name='MYTREX', trex=trex)]
pac_str = pac.to_url()
print(pac_str)

>> HTTPS://PAC.METTORIUS.COM/21:1234*MYTREX$TREX/STOP$T.D:20240505T1306+TEMP$KEL:10.15+OK$T.B:F+COMMENT$T.A:FOO+COMMENT2$T.T:12G3+TABLE$$DURATION$HUR:Date$T.D:OK$T.B:COMMENT$T.A::1:20260814T021153.586:T:FOO::1.1:20260814T021153.588:T:BAR::1.3:20260814T021153.590:F:BLUBB

PAC-ID Resolver

from labfreed import PAC_ID_Resolver, load_cit  
from labfreed.pac_id_resolver.service_availability import check_service_group  
import requests_cache

# Get a CIT
dir = os.path.join(os.getcwd(), 'examples')
p = os.path.join(dir, 'cit_mine.yaml')       
cit = load_cit(p)

# validate the CIT
cit.is_valid
cit.print_validation_messages()

# get a second cit
p = os.path.join(dir, 'coupling-information-table')
cit2 = load_cit(p)
cit2.origin = 'MY_COMPANY'

# resolve a pac id
pac_str = 'HTTPS://PAC.METTORIUS.COM/-MS/X3511/CAS:7732-18-5'
service_groups = PAC_ID_Resolver(resolver_configs=[cit, cit2]).resolve(pac_str, check_service_status=False)
cached_session = requests_cache.CachedSession(backend='memory', expire_after=60)
for sg in service_groups:
    check_service_group(sg, cached_session)
    sg.print()

>> Services from origin 'PERSONAL                         
>> ┏━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
>> ┃ Service Name ┃ URL                        ┃ Service Type         ┃ Reachable ┃
>> ┡━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
>> │ CAS Search   │ https://pubchem.ncbi.nlm.… │ userhandover-generic │ UNKNOWN   │
>> └──────────────┴────────────────────────────┴──────────────────────┴───────────┘
>>                         Services from origin 'MY_COMPANY                        
>> ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
>> ┃ Service Name        ┃ URL                  ┃ Service Type        ┃ Reachable ┃
>> ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
>> │ Chemical Management │ https://chem-manage… │ userhandover-gener… │ UNKNOWN   │
>> └─────────────────────┴──────────────────────┴─────────────────────┴───────────┘

PAC-ID Attributes

Attributes attach lightweight metadata -- e.g. a display name, an image, a calibration due date -- to a PAC-ID, without baking it into the identifier itself.

This shows the core data model only: an in-memory data source, served in-process with no Flask/network involved. For an actual deployable server and a PAC-ID landing page built on the same classes, see Setting up a PAC-ID Landing Page.

from labfreed.pac_attributes.facade.attributes import Attribute, Attributes, Resource  
from labfreed.pac_attributes.facade.dict_data_source import Dict_DataSource  
from labfreed.pac_attributes.well_known_attribute_keys import MetaAttributeKeys  
from labfreed.pac_attributes.server.translation_data_sources import DictTranslationDataSource  
from labfreed.pac_attributes.server.server import AttributeServerRequestHandler  
from labfreed.pac_attributes.client.client import AttributeClient, local_attribute_request_callback_factory  
from labfreed.utilities.translations import Terms, Term  

# Attributes for one PAC-ID. A data source could just as well read this from a database, an Excel sheet, or anywhere else.
pac_str = 'HTTPS://PAC.METTORIUS.COM/-MD/BAL500/000001'
data_source = Dict_DataSource(
    attribute_group_key=MetaAttributeKeys.GROUPKEY,
    data={
        pac_str: Attributes([
            Attribute(key=MetaAttributeKeys.DISPLAYNAME, value="My Balance"),
            Attribute(key=MetaAttributeKeys.IMAGE, value=Resource("https://picsum.photos/id/82/200")),
        ])
    }
)

# Attribute keys need a translation, so the server can label them for a UI
translations = DictTranslationDataSource(
    supported_languages={'en'},
    data=Terms(terms=[
        Term.create(MetaAttributeKeys.GROUPKEY, [('en', 'Meta Data')]),
        Term.create(MetaAttributeKeys.DISPLAYNAME, [('en', 'Display Name')]),
        Term.create(MetaAttributeKeys.IMAGE, [('en', 'Image')]),
    ])
)

# The request handler is the framework-agnostic core of an attribute server
handler = AttributeServerRequestHandler(data_sources=[data_source], translation_data_sources=[translations], default_language='en')

Querying works the same whether the handler above is embedded in a Flask app or, as here, called in-process.

client = AttributeClient(http_post_callback=local_attribute_request_callback_factory(handler))
attribute_groups = client.get_attributes(server_url='', subject_id=pac_str)
for group in attribute_groups:
    for attr in Attributes.from_payload_attributes(group.attributes):
        values = ', '.join(str(v) for v in attr.value_list)
        print(f'{attr.label}: {values}')

>> Display Name: My Balance
>> Image: https://picsum.photos/id/82/200

Versioning

This package follows Semantic Versioning: MAJOR.MINOR.PATCH.

• MAJOR -- may introduce breaking changes. A major bump doesn't have to break anything, but it's where we reserve the right to.
• MINOR -- adds functionality. We aim for backward compatibility here too, but reserve the right to break edge cases where keeping compatibility isn't practical (e.g. tightening validation on input that was already spec-invalid). Any such case is called out explicitly in the changelog as BREAKING.
• PATCH -- bugfixes only, no intentional API changes.

Deprecation policy: when a public API needs to change or go away, we deprecate it first -- it keeps working, but raises a DeprecationWarning pointing at its replacement -- and keep that deprecated path working for at least one more major version before actually removing it. This policy doesn't apply to anything marked @experimental (see Package Structure) -- those can change or be removed in a minor/patch release without going through deprecation first.

Pre-releases: in-progress work toward the next version is published straight to PyPI as an alpha/beta pre-release (e.g. 1.0.0b44). A plain pip install labfreed always resolves to the latest stable release and skips these automatically -- opt in explicitly with pip install --pre labfreed, or pin an exact pre-release (pip install labfreed==1.0.0b44). Since the LabFREED building-block specs and this implementation are co-developed, a pre-release can be ahead of the currently published spec -- it may reflect a spec that's still being drafted. Treat pre-release behavior as experimental and subject to change before the real release ships.

Change Log

v1.0.1

PAC-ID

• BREAKING: PAC_ID.__eq__/__hash__ now scoped to (issuer, identifier) only, excluding extensions - two PAC_IDs differing only in an extension previously compared unequal and hashed differently, contradicting get_non_derived_pac_id()/get_parent_pac_id()/derive()'s own issuer+identifier identity model. Deliberately cross-type: a PAC_CAT and a PAC_ID with identical issuer+identifier now compare equal (accepted as a patch given ~zero adoption of v1.0.0, released hours earlier)
• new PAC_ID.values_for_key(key) (also on PAC_CAT, extended by extensions incl. T-REX table columns): joint lookup across identifier segments and extensions, tagging each match with its Origin (SegmentOrigin, ExtensionOrigin, TrexTableOrigin)
• bugfix: Extension subclasses' type-mismatch warning interpolated the wrong variable (DisplayNameExtension/TextBase36Extension logged the extension name instead of the mismatched type)

PAC-CAT

• new PAC_CAT.from_roles(issuer, main, processor=None) / Category.from_key(key, **fields) (plus .of aliases) for named-role construction; PAC_CAT.from_categories() is now deprecated in favor of it (kept working, scheduled for removal at v2.0)

PAC-ID Resolver

• bugfix: ResolverConfigEvaluator deep-copies its input before each jsonpath lookup - jsonpath_ng's recursive-descendant wildcard mutated the shared per-PAC-ID dict in place, silently corrupting every block evaluated after the first wildcard-shaped template_url/applicable_if (e.g. cit.yaml's Manual/CoA macros)

T-REX

• new explicit-type wrapper classes Alphanumeric, Text, Numeric, Bool, Date (labfreed.trex.facade) so a dict entry's wire type can be stated explicitly instead of always inferred from the Python value's type
• to_trex()/from_trex() renamed to to_trex_spec()/from_trex_spec(); T_REX gains its own serialize()/deserialize(s) as the everyday entry point (old names kept as deprecated aliases)
• bugfix: from_trex/from_trex_spec now actually return a T_REX instance instead of a plain dict
• bugfix: T_REX's dict-value union (and DataTable's cell type) had no bare int/float/None member - a plain int/float could be silently coerced into a datetime via Unix-timestamp interpretation, and constructing with None (a valid, undefined T-REX value) raised

General

• Quantity(value=..., unit=None) now emits a warning - LabFREED treats unitless numbers as a matter of principle, so this flags a bare int/float that got silently wrapped for convenience, or any other genuinely unitless construction, rather than passing silently

v1.0.0

PAC-ID

• supporting PAC.LI issuer
• BREAKING: WellKnownKeys and GS1ApplicationIdentifier now derive from StrEnum instead of Enum (same treatment as the PAC-ID Attributes key enums below); id_segment.py's well-known-key check simplified accordingly, no .value needed

PAC-CAT

• added new categories
• BREAKING: Renamed category MM to MX

PAC-ID Resolver

• Transition to improved resolver configuration ( replaces coupling information table )
• BREAKING: ServiceType now derives from StrEnum instead of Enum; _validate_service_type in both resolver_config_common.py and resolver_config.py dropped their manual .value/isinstance unwrapping now that members compare directly against plain strings
• BREAKING: ServiceStatus (Service.status) now derives from StrEnum with explicit string values ("active"/"inactive"/"unknown") instead of plain Enum with auto()-generated int values; any code reading .value directly (none found in this codebase) would now get a string instead of an int
• new optional key field on ResolverConfigEntry/Service: an absolute IRI, drawn from the same shared vocabularies PAC-ID Attributes already sources its own key from, identifying what an entry semantically is (e.g. "this is a Material Safety Data Sheet"), orthogonal to application_intents (which identifies which use case selects it). PacInfo gained matching get_user_handover(s)_by_key()/get_action(s)_by_key() lookups

PAC-ID Attributes

• new building block

General

• bugfix: Werkzeug moved from being pulled in only transitively via the extended/experimental extras' Flask dependency to a real base dependency - pac_attributes/server/server.py and api_data_models/request.py (both part of the always-imported base package, no extra required) use werkzeug's Accept-Language parsing (parse_accept_header, LanguageAccept) directly, so a bare pip install labfreed (no extras) could not even import labfreed once those modules landed - Flask itself stays extended/experimental-only, since nothing in the base import chain actually needs the web framework, only this one HTTP-header-parsing utility it depends on
• Minor Bugfixes
• labfreed_experimental.pac_disco's ServiceUUID/PAC_Characteristics now derive from StrEnum instead of Enum (no compatibility guarantee on this module, not tagged BREAKING)
• qr.generate_qr's Direction modernized from class Direction(str, Enum) to class Direction(StrEnum) - purely cosmetic, identical runtime behavior
• BREAKING: ValidationMsgLevel (used across every building block's validation messages) now derives from StrEnum with explicit string values ("error"/"warning"/"recommendation"/"info") instead of plain Enum with auto()-generated int values; ValidationMessage.level would now serialize as a string via model_dump()/model_dump_json() instead of an int if ever dumped directly (it's stored in a private attribute and excluded from the parent model's own serialization by default, but is a public field on ValidationMessage itself) - no in-repo call site read .value on it before this change
• BREAKING: reorganization of module structure > some import paths have changed
• BREAKING: labfreed.trex.pythonic and labfreed.pac_attributes.pythonic renamed to labfreed.trex.facade and labfreed.pac_attributes.facade; every py-prefixed convenience class (pyTREX, pyAttribute, pyAttributes, pyResource, pyReference, pyDict_DataSource) renamed to drop the py prefix (T_REX, Attribute, Attributes, Resource, Reference, Dict_DataSource) - old names kept as deprecated aliases for one more major version, but the old labfreed.*.pythonic submodule path itself is gone entirely, not just the symbol names
• BREAKING: renamed Quantity.float property to Quantity.as_float (the name collided with the float type used in Quantity's own annotations, breaking model construction on Python 3.14)
• moved Quantity from labfreed.trex.pythonic.quantity to labfreed.utilities.quantity (it's used by PAC-ID Attributes too, not only T-REX); also reachable via labfreed.trex.facade alongside T_REX/DataTable - unece_unit_code_from_quantity (T-REX-specific) moved into pyTREX.py
• AttributeClient.get_attributes()'s pac_id parameter renamed to subject_id, matching the rest of the IRI migration - old pac_id= keyword still works via a deprecated shim
• new optional units extra (pip install labfreed[units], adds pint+ucumvert): Quantity<->T-REX UNECE unit-code mapping is now automatic for compound/non-SI units (mol/L, kg/m3, Cel, ...) instead of only working when a unit's UNECE symbol happened to equal its UCUM string
• BREAKING (edge case): when a unit can't be resolved to a UNECE code and the units extra isn't installed, Quantity/pyTREX.to_trex() now raise UcumSupportError (an ImportError subclass) instead of ValueError - only observable if calling code specifically caught ValueError from this path, which previously fired for every non-exact-match unit
• BREAKING: Quantity now validates that unit is a valid UCUM unit at construction time (structure only without the units extra, full symbol-level check with it) and raises ValueError otherwise - previously any string was accepted. Pass dont_enforce_ucum_units=True to the constructor to bypass this (discouraged)
• Quantity.__str__ pretty-prints its unit (e.g. kg/m3 -> kg/m³) when the units extra is installed, instead of the old naive .->· substitution

v0.2.12

• bugfix:no warning message if PAC-CAT has same segment key in two segments

v0.2.11

• bugfix:added missing well known segment key '250'

v0.2.10

• bugfix:added missing well known segment key '20'

v0.2.9

• bugfix in serialization of PAC-CAT with multiple categories

v0.2.8

• option to pass cache to resolver for speedier check of service availability

v0.2.7

• Improved README. No functional changes

v0.2.6

• PAC_ID.to_url() preserves the identifier as is by default but allows to force short or long notation.
• PAC-ID Resolver does not try to resolve PAC-CAT with CIT v1.

v0.2.5

• resolvers checks service states by default
• improvements and bugfixes in conversion from python types to TREX
• follow better naming conventions in CIT v1

v0.2.4

• improvements in formatting of validation messages
• bugfix in DataTable

v0.2.3

• improvements in formatting of validation messages
• bugfix in DisplayNameExtension

v0.2.2

• minor changes for better access of subfunctions. No change in existing API

v0.2.1

• improved docu. no code changes

v0.2.0b2

• improvements in api consistency and ease of use
• restructured code for better separation of concerns
• support for coupling information table v1

v0.1.1

• minor internal improvements and bugfixes

v0.1.0

• DRAFT Support for PAC-ID Resolver

v0.0.20

• bugfix in TREX table to dict conversion
• markdown compatible validation printing

v0.0.19

• supports PAC-ID, PAC-CAT, TREX and DisplayName
• QR generation
• ok-ish test coverage

Getting Help / FAQ

• Questions about the LabFREED building blocks themselves (PAC-ID, PAC-CAT, T-REX, PAC-ID Resolver, PAC-ID Attributes) — join the community Discord or see labfreed.org.
• Bugs or feature requests for this Python package — open a GitHub issue.
• Found a security issue? See SECURITY.md instead of opening a public issue.
• Looking for the full API reference (all classes/functions, generated from docstrings via pdoc)? It's published at retothuerer.github.io/LabFREED.

Contributing

See CONTRIBUTING.md for setting up a dev environment, running the test suite, and what's expected of a pull request.

Attributions

The following tools were used:

• pdoc was a great help with generating documentation
• Pydantic
• json with UNECE units from (https://github.com/quadient/unece-units/blob/main/python/src/unece_excel_parser/parsedUneceUnits.json)
• json with GS1 codes from (https://ref.gs1.org/ai/GS1_Application_Identifiers.jsonld)