CWR V2.2 support: parser, encoder, and config updates

.gitignore

@@ -62,4 +62,7 @@ target/
 # Idea PyCharm
 .idea/
 .idea_modules/
-out/
+out/
+
+**/.DS_Store
+*__pycache__

README.rst

@@ -1,7 +1,7 @@
 CWR Data Model API
 ==================
 
-This projects offers a domain model for the CISAC CWR standard v2.1 to be
+This projects offers a domain model for the CISAC CWR standard v2.2 to be
 used on Python applications, along a series of parsing which allow
 transformations between the model and various data structures.
 
@@ -13,6 +13,9 @@ works data.
 While the CWR standard has been created by `CISAC`_ this library has been
 developed by `WESO`_ independently, with help from `BMAT`_.
 
+This is a fork of the `official CWR-DataApi repository`_ by WESO, updated to
+support CWR standard v2.2 by `max1millions`_.
+
 .. image:: https://badge.fury.io/py/cwr-api.svg
     :target: https://pypi.python.org/pypi/cwr-api
     :alt: CWR-API Pypi package page
@@ -51,9 +54,10 @@ Prerequisites
 
 The project has been tested in the following versions of the interpreter:
 
-- Python 3.4
-- Python 3.5
-- Python 3.6
+- Python 3.9
+- Python 3.10
+- Python 3.11
+- Python 3.12
 
 All other dependencies are indicated on the requirements.txt file.
 The included makefile can install them with the command:
@@ -146,9 +150,11 @@ The project has been released under the `MIT License`_.
 .. _BMAT: http://www.bmat.com/
 .. _WESO: http://www.weso.es/
 .. _project issues page: https://github.com/weso/CWR-DataApi/issues
-.. _Pyparsing: https://pyparsing.wikispaces.com/
+.. _Pyparsing: https://pyparsing-docs.readthedocs.io/
 .. _Pypi package: https://pypi.python.org/pypi/CWR-API
 .. _Sphinx: http://sphinx-doc.org/
 .. _latest docs: http://cwr-dataapi.readthedocs.org
 .. _GitHub project page: https://github.com/weso/CWR-DataApi
+.. _official CWR-DataApi repository: https://github.com/weso/CWR-DataApi
+.. _max1millions: https://github.com/max1millions
 .. _MIT License: http://www.opensource.org/licenses/mit-license.php

config_cwr/accessor.py

@@ -43,7 +43,7 @@ def read_yaml_file(self, file_name):
         """
         with open(os.path.join(self.__path(), os.path.basename(file_name)),
                   'rt') as yamlfile:
-            return yaml.load(yamlfile)
+            return yaml.load(yamlfile, Loader=yaml.FullLoader)
 
     def read_config_file(self, file_name):
         """

config_cwr/default_values.yml

@@ -1 +1 @@
-default_version: 2.1
+default_version: 2.2

config_cwr/field_config_common.yml

@@ -105,6 +105,16 @@ character_set:
   size: 15
   name: Character Set
 
+creation_date:
+  type: date
+  size: 8
+  name: Creation Date
+
+creation_time:
+  type: time
+  size: 6
+  name: Creation Time
+
 creation_date_time:
   type: date_time
   size: 16
@@ -137,7 +147,7 @@ date_publication_printed_edition:
 
 duration:
   type: time
-  size: 8
+  size: 6
   name: Duration
 
 ean13:
@@ -153,6 +163,12 @@ edi_standard:
   values:
     - '01.10'
 
+hdr_version_type:
+  type: alphanum
+  size: 3
+  name: Header Version Type
+  results_name: version_type
+
 entire_work_title:
   type: alphanum
   size: 60
@@ -357,6 +373,11 @@ original_transaction_sequence_n:
   size: 8
   name: Original Transaction Sequence Number
 
+organisation_code:
+  type: alphanum
+  size: 3
+  name: Organisation Code
+
 original_record_sequence_n:
   type: numeric
   size: 8
@@ -599,7 +620,7 @@ submitter_work_n:
   name: Submitter Work Number
 
 tax_id:
-  type: numeric
+  type: alphanum
   size: 9
   name: Tax ID Number
 
@@ -765,4 +786,50 @@ writer_unknown:
 year_production:
   type: numeric
   size: 4
-  name: Year of Production
+  name: Year of Production
+
+recording_title:
+  type: alphanum_ext
+  size: 60
+  name: Recording Title
+
+version_title:
+  type: alphanum_ext
+  size: 60
+  name: Version Title
+
+display_artist:
+  type: alphanum_ext
+  size: 60
+  name: Display Artist
+
+record_label:
+  type: alphanum
+  size: 60
+  name: Record Label
+
+isrc_validity:
+  type: alphanum
+  size: 20
+  name: ISRC Validity
+
+submitter_recording_identifier:
+  type: alphanum_end
+  size: 14
+  name: Submitter Recording Identifier
+
+identifier_type:
+  type: alphanum
+  size: 1
+  name: Identifier Type
+
+validity:
+  type: alphanum
+  size: 1
+  name: Validity
+
+xrf_identifier:
+  type: alphanum
+  size: 14
+  name: Identifier
+  results_name: identifier

config_cwr/record_config_common.cml

@@ -340,7 +340,7 @@ transaction_record:
 
 transaction_record:
     id: publisher_territory
-    head: SPT
+    head: SPT,OPT
     rules:
       [
         field: ip_n
@@ -396,6 +396,14 @@ transaction_record:
         optional [
             field: media_type
         ]
+        optional [
+            field: recording_title
+            field: version_title
+            field: display_artist
+            field: record_label
+            field: isrc_validity
+            field: submitter_recording_identifier
+        ]
       ]
 
 transaction_record:
@@ -422,6 +430,7 @@ record:
         optional
         [
           field: character_set
+          field: hdr_version_type
         ]
       ]
 
@@ -470,6 +479,17 @@ transaction_record:
         ]
       ]
 
+transaction_record:
+    id: work_id_cross_reference
+    head: XRF
+    rules:
+      [
+        field: organisation_code
+        field: xrf_identifier
+        field: identifier_type
+        field: validity
+      ]
+
 transaction_record:
     id: work_alternate_title
     head: ALT
@@ -573,6 +593,7 @@ transaction_record:
         field: submitter_agreement_n
         field: society_assigned_agreement_n
         field: writer_ip_n
+        field: publisher_sequence_n
       ]
 
 transaction_record:

cwr/cross_reference.py

@@ -0,0 +1,43 @@
+# -*- coding: utf-8 -*-
+
+from cwr.record import TransactionRecord
+
+"""
+CWR 2.2 § 5.32 Work ID Cross Reference (XRF) record.
+
+This record links a work transaction to identifiers issued by any organisation
+including the intended recipient, other PROs, ISWC (org_code='ISW'), or ISRC
+(org_code='ISR').
+
+The XRF record is an optional trailing record inside NWR, REV, and ISW
+transactions (see CWR 2.2 §6.1 BNF).
+"""
+
+__author__ = 'Rightstune'
+__license__ = 'MIT'
+__status__ = 'Development'
+
+
+class XrfRecord(TransactionRecord):
+    """CWR 2.2 § 5.32 Work ID Cross Reference (XRF).
+
+    Field layout after the 19-char Record Prefix:
+      organisation_code  3  alphanum  M
+      identifier        14  alphanum  M
+      identifier_type    1  alphanum  M  W=Work, R=Recording, P=Product, V=Video
+      validity           1  alphanum  M  Y=valid, U=link invalid, N=identifier invalid
+    """
+
+    def __init__(self, transaction_sequence_n, record_sequence_n,
+                 organisation_code, identifier,
+                 identifier_type='W', validity='Y'):
+        super(XrfRecord, self).__init__(
+            'XRF', transaction_sequence_n, record_sequence_n
+        )
+        org = str(organisation_code)
+        if org.isdigit():
+            org = org.zfill(3)
+        self.organisation_code = org[:3]
+        self.identifier = str(identifier)[:14]
+        self.identifier_type = identifier_type
+        self.validity = validity

cwr/grammar/factory/config.py

@@ -15,7 +15,7 @@
     pp.Word(pp.nums).setResultsName('count')
 rule_at_least.setParseAction(lambda v: int(v[0]))
 
-_rule_config_string = pp.Regex('[^()\[\]\\n,:]*')
+_rule_config_string = pp.Regex(r'[^()\[\]\n,:]*')
 _rule_config_string.setParseAction(lambda v: v[0].strip())
 
 _rule_identifier = _rule_config_string + pp.Literal(':').suppress()

cwr/grammar/field/basic.py

@@ -3,7 +3,6 @@
 import datetime
 
 import pyparsing as pp
-from pyparsing import ParseResults
 
 """
 CWR fields grammar.

cwr/group.py

@@ -44,7 +44,7 @@ def __init__(self,
                  record_type='',
                  group_id=0,
                  transaction_type='',
-                 version_number='02.10',
+                 version_number='02.20',
                  batch_request_id=0
                  ):
         super(GroupHeader, self).__init__(

cwr/interested_party.py

@@ -115,7 +115,7 @@ def __init__(self,
                  record_type='',
                  transaction_sequence_n=0,
                  record_sequence_n=0,
-                 first_recording_refusal='U',
+                 first_recording_refusal='N',
                  usa_license='',
                  pr_society=None,
                  pr_ownership_share=0,
@@ -292,7 +292,7 @@ def usa_license(self, value):
 
 class IPTerritoryOfControlRecord(TransactionRecord):
     """
-    Represents a CWR Publisher or Writer Territory of Control (SPT/SWT).
+    Represents a CWR Publisher or Writer Territory of Control (SPT/OPT/SWT).
 
     This indicates if a Publisher or Writer has control or not over a
     Territory, and the shares it has on it.
@@ -551,7 +551,7 @@ def __init__(self,
                  sr_society=None,
                  sr_ownership_share=0,
                  special_agreements=None,
-                 first_recording_refusal='U',
+                 first_recording_refusal='N',
                  usa_license=''
                  ):
         """
@@ -995,7 +995,8 @@ def __init__(self,
                  publisher_name='',
                  writer_ip_n='',
                  submitter_agreement_n=None,
-                 society_assigned_agreement_n=None
+                 society_assigned_agreement_n=None,
+                 publisher_sequence_n=0
                  ):
         super(PublisherForWriterRecord, self).__init__(
             record_type,
@@ -1011,6 +1012,9 @@ def __init__(self,
         self._submitter_agreement_n = submitter_agreement_n
         self._society_assigned_agreement_n = society_assigned_agreement_n
 
+        # Publisher chain link
+        self._publisher_sequence_n = publisher_sequence_n
+
     @property
     def publisher_ip_n(self):
         """
@@ -1082,6 +1086,21 @@ def publisher_name(self):
     def publisher_name(self, value):
         self._publisher_name = value
 
+    @property
+    def publisher_sequence_n(self):
+        """
+        Publisher Sequence # field. Numeric.
+
+        Must match the Publisher Sequence # of the relating SPU/OPU record.
+
+        :return: the publisher sequence number in the chain
+        """
+        return self._publisher_sequence_n
+
+    @publisher_sequence_n.setter
+    def publisher_sequence_n(self, value):
+        self._publisher_sequence_n = value
+
 
 class WriterRecord(InterestedPartyRecord):
     """
@@ -1100,9 +1119,9 @@ def __init__(self,
                  writer=None,
                  writer_designation=None,
                  work_for_hire=False,
-                 writer_unknown='F',
-                 reversionary='U',
-                 first_recording_refusal='U',
+                 writer_unknown='',
+                 reversionary='',
+                 first_recording_refusal='N',
                  usa_license='',
                  pr_society=None,
                  pr_ownership_share=0,