Merge pull request #44 from Bernardo-MG/dict_parsers

Dict parsers

cwr/file.py

@@ -118,10 +118,11 @@ def __str__(self):
             self._version)
 
     def __repr__(self):
-        return '<class %s>(year=%s, sequence_n=%r, sender=%r, receiver=%r, version=%r)' % (self.__class__.__name__, self._year,
-                                                                                           self._sequence_n,
-                                                                                           self._sender, self._receiver,
-                                                                                           self._version)
+        return '<class %s>(year=%s, sequence_n=%r, sender=%r, receiver=%r, version=%r)' % (
+            self.__class__.__name__, self._year,
+            self._sequence_n,
+            self._sender, self._receiver,
+            self._version)
 
     @property
     def year(self):

cwr/parser/decoder/cwrjson.py

@@ -1,6 +1,8 @@
 # -*- coding: utf-8 -*-
 
 from cwr.parser.decoder.common import Decoder
+from cwr.parser.decoder.dictionary import FileDictionaryDecoder
+import json
 
 """
 Classes for decoding CWR classes from JSON dictionaries.
@@ -14,6 +16,9 @@
 class JSONDecoder(Decoder):
     def __init__(self):
         super(JSONDecoder, self).__init__()
+        self._dict_decoder = FileDictionaryDecoder()
 
     def decode(self, data):
-        pass
+        decoded = json.loads(data)
+
+        return self._dict_decoder.decode(decoded)

cwr/parser/decoder/dictionary.py

@@ -31,6 +31,52 @@
 __status__ = 'Development'
 
 
+class TransactionRecordDictionaryDecoder(Decoder):
+    def __init__(self):
+        super(TransactionRecordDictionaryDecoder, self).__init__()
+        self._decoders = {}
+
+        self._decoders['ACK'] = AcknowledgementDictionaryDecoder()
+        self._decoders['AGR'] = AgreementDictionaryDecoder()
+        self._decoders['TER'] = AgreementTerritoryDictionaryDecoder()
+        self._decoders['ARI'] = AdditionalRelatedInformationDictionaryDecoder()
+        self._decoders['ALT'] = AlternateTitleDictionaryDecoder()
+        self._decoders['EWT'] = AuthoredWorkDictionaryDecoder()
+        self._decoders['VER'] = AuthoredWorkDictionaryDecoder()
+        self._decoders['COM'] = ComponentDictionaryDecoder()
+        self._decoders['IPA'] = InterestedPartyForAgreementDictionaryDecoder()
+        self._decoders['SPT'] = IPTerritoryOfControlDictionaryDecoder()
+        self._decoders['SWT'] = IPTerritoryOfControlDictionaryDecoder()
+        self._decoders['IND'] = InstrumentationDetailDictionaryDecoder()
+        self._decoders['INS'] = InstrumentationSummaryDictionaryDecoder()
+        self._decoders['MSG'] = MessageDictionaryDecoder()
+        self._decoders['PER'] = PerformingArtistDictionaryDecoder()
+        self._decoders['PWR'] = PublisherForWriterDictionaryDecoder()
+        self._decoders['REC'] = RecordingDetailDictionaryDecoder()
+        self._decoders['EXC'] = WorkDictionaryDecoder()
+        self._decoders['ISW'] = WorkDictionaryDecoder()
+        self._decoders['NWR'] = WorkDictionaryDecoder()
+        self._decoders['REV'] = WorkDictionaryDecoder()
+        self._decoders['ORN'] = WorkOriginDictionaryDecoder()
+        self._decoders['SWR'] = WriterRecordDictionaryDecoder()
+        self._decoders['OWR'] = WriterRecordDictionaryDecoder()
+        self._decoders['OWR'] = WriterRecordDictionaryDecoder()
+        self._decoders['NPA'] = NonRomanAlphabetAgreementPartyDictionaryDecoder()
+        self._decoders['NOW'] = NonRomanAlphabetOtherWriterDictionaryDecoder()
+        self._decoders['NPR'] = NonRomanAlphabetPerformanceDataDictionaryDecoder()
+        self._decoders['NPN'] = NonRomanAlphabetPublisherNameDictionaryDecoder()
+        self._decoders['NAT'] = NonRomanAlphabetTitleDictionaryDecoder()
+        self._decoders['NET'] = NonRomanAlphabetWorkDictionaryDecoder()
+        self._decoders['NCT'] = NonRomanAlphabetWorkDictionaryDecoder()
+        self._decoders['NVT'] = NonRomanAlphabetWorkDictionaryDecoder()
+        self._decoders['NWN'] = NonRomanAlphabetWriterNameDictionaryDecoder()
+        self._decoders['SPU'] = PublisherRecordDictionaryDecoder()
+        self._decoders['OPU'] = PublisherRecordDictionaryDecoder()
+
+    def decode(self, data):
+        return self._decoders[data['record_type']].decode(data)
+
+
 class AcknowledgementDictionaryDecoder(Decoder):
     def __init__(self):
         super(AcknowledgementDictionaryDecoder, self).__init__()
@@ -308,20 +354,82 @@ def decode(self, data):
                                      media_type=data['media_type'])
 
 
+class FileDictionaryDecoder(Decoder):
+    def __init__(self):
+        super(FileDictionaryDecoder, self).__init__()
+
+        self._tag_decoder = FileTagDictionaryDecoder()
+        self._transmission_decoder = TransmissionDictionaryDecoder()
+
+    def decode(self, data):
+        tag = data['tag']
+        if isinstance(tag, dict):
+            tag = self._tag_decoder.decode(tag)
+
+        transmission = data['transmission']
+        if isinstance(transmission, dict):
+            transmission = self._transmission_decoder.decode(transmission)
+
+        return CWRFile(tag, transmission)
+
+
 class TransmissionDictionaryDecoder(Decoder):
     def __init__(self):
         super(TransmissionDictionaryDecoder, self).__init__()
 
+        self._header_decoder = TransmissionHeaderDictionaryDecoder()
+        self._trailer_decoder = TransmissionTrailerDictionaryDecoder()
+        self._group_decoder = GroupDictionaryDecoder()
+
     def decode(self, data):
-        return Transmission(data['transmission_header'], data['transmission_trailer'], data['groups'])
+        header = data['header']
+        if isinstance(header, dict):
+            header = self._header_decoder.decode(header)
+
+        trailer = data['trailer']
+        if isinstance(trailer, dict):
+            trailer = self._trailer_decoder.decode(trailer)
+
+        groups = []
+        if len(data['groups']) > 0:
+            if isinstance(data['groups'][0], dict):
+                for group in data['groups']:
+                    groups.append(self._group_decoder.decode(group))
+            else:
+                groups = data['groups']
+
+        return Transmission(header, trailer, groups)
 
 
 class GroupDictionaryDecoder(Decoder):
     def __init__(self):
         super(GroupDictionaryDecoder, self).__init__()
 
+        self._header_decoder = GroupHeaderDictionaryDecoder()
+        self._trailer_decoder = GroupTrailerDictionaryDecoder()
+        self._transaction_decoder = TransactionRecordDictionaryDecoder()
+
     def decode(self, data):
-        return Group(data['group_header'], data['group_trailer'], data['transactions'])
+        header = data['group_header']
+        if isinstance(header, dict):
+            header = self._header_decoder.decode(header)
+
+        trailer = data['group_trailer']
+        if isinstance(trailer, dict):
+            trailer = self._trailer_decoder.decode(trailer)
+
+        transactions = []
+        if len(data['transactions']) > 0:
+            if isinstance(data['transactions'][0][0], dict):
+                for transaction in data['transactions']:
+                    transaction_records = []
+                    for record in transaction:
+                        transaction_records.append(self._transaction_decoder.decode(record))
+                    transactions.append(transaction_records)
+            else:
+                transactions = data['transactions']
+
+        return Group(header, trailer, transactions)
 
 
 class TransmissionHeaderDictionaryDecoder(Decoder):

data_commonworks/record_config_common.yml

@@ -630,6 +630,7 @@ transmission_header:
         id: transmission_date
       - rule_type: field
         id: character_set
+  results_name: header
 
 transmission_trailer:
   id: transmission_trailer
@@ -644,6 +645,7 @@ transmission_trailer:
         id: transaction_count
       - rule_type: field
         id: record_count
+  results_name: trailer
 
 work:
   id: work

tests/parser/dictionary/decoder/control/test_group.py

@@ -0,0 +1,67 @@
+# -*- coding: utf-8 -*-
+
+import unittest
+import datetime
+
+from cwr.parser.decoder.dictionary import GroupDictionaryDecoder
+
+
+"""
+Group from dictionary encoding tests.
+
+The following cases are tested:
+"""
+
+__author__ = 'Bernardo Martínez Garrido'
+__license__ = 'MIT'
+__version__ = '0.0.0'
+__status__ = 'Development'
+
+
+class TestGroupDictionaryEncoding(unittest.TestCase):
+    def setUp(self):
+        self._decoder = GroupDictionaryDecoder()
+
+    def test_encoded(self):
+        trailer = {}
+        trailer['record_count'] = 20
+        trailer['record_type'] = 'GRT'
+        trailer['group_id'] = 3
+        trailer['transaction_count'] = 15
+
+        header = {}
+        header['record_type'] = 'GRH'
+        header['version_number'] = '02.10'
+        header['group_id'] = 3
+        header['batch_request_id'] = 15
+        header['transaction_type'] = 'AGR'
+
+        transactions = [[{'original_group_id': 4, 'creation_date_time': datetime.date(2003, 2, 15),
+                          'original_transaction_type': 'AGR', 'record_sequence_n': 15, 'record_type': 'ACK',
+                          'creation_title': 'TITLE', 'original_transaction_sequence_n': 5, 'transaction_status': 'AS',
+                          'recipient_creation_n': 'B124', 'submitter_creation_n': 'A123', 'transaction_sequence_n': 3,
+                          'processing_date': datetime.date(2003, 2, 16)},
+                         {'original_record_sequence_n': 124, 'validation_n': 'AB3', 'message_record_type': 'AGR',
+                          'message_text': 'THE MESSAGE', 'record_sequence_n': 15, 'record_type': 'MSG',
+                          'message_level': 'F', 'message_type': 'G', 'transaction_sequence_n': 3},
+                         {'original_record_sequence_n': 124, 'validation_n': 'AB3', 'message_record_type': 'AGR',
+                          'message_text': 'THE MESSAGE', 'record_sequence_n': 15, 'record_type': 'MSG',
+                          'message_level': 'F', 'message_type': 'G', 'transaction_sequence_n': 3},
+                         {'sales_manufacture_clause': 'M', 'date_of_signature': datetime.date(2003, 2, 17),
+                          'prior_royalty_start_date': datetime.date(2003, 2, 19), 'advance_given': True,
+                          'retention_end_date': datetime.date(2003, 2, 18), 'international_standard_code': 'DFG135',
+                          'prior_royalty_status': 'D', 'agreement_end_date': datetime.date(2003, 2, 16),
+                          'record_type': 'AGR', 'shares_change': True, 'post_term_collection_status': 'D',
+                          'agreement_type': 'OS', 'submitter_agreement_n': 'AB12',
+                          'society_assigned_agreement_n': 'DF35', 'record_sequence_n': 15,
+                          'agreement_start_date': datetime.date(2003, 2, 15), 'transaction_sequence_n': 3,
+                          'post_term_collection_end_date': datetime.date(2003, 2, 20), 'number_of_works': 12}]]
+
+        dict = {'group_trailer': trailer, 'transactions': transactions, 'group_header': header}
+
+        record = self._decoder.decode(dict)
+
+        self.assertEqual('GRH', record.group_header.record_type)
+        self.assertEqual('GRT', record.group_trailer.record_type)
+
+        self.assertEqual('ACK', record.transactions[0][0].record_type)

tests/parser/dictionary/decoder/control/test_transmission.py

@@ -0,0 +1,92 @@
+# -*- coding: utf-8 -*-
+
+import unittest
+import datetime
+
+from cwr.parser.decoder.dictionary import TransmissionDictionaryDecoder
+
+
+"""
+Dictionary to Message decoding tests.
+
+The following cases are tested:
+"""
+
+__author__ = 'Bernardo Martínez Garrido'
+__license__ = 'MIT'
+__version__ = '0.0.0'
+__status__ = 'Development'
+
+
+class TestTransmissionDecoder(unittest.TestCase):
+    def setUp(self):
+        self._decoder = TransmissionDictionaryDecoder()
+
+    def test_encoded(self):
+        header = {}
+        header['record_type'] = 'HDR'
+        header['sender_id'] = 'SND123'
+        header['sender_name'] = 'THE SENDER'
+        header['sender_type'] = 'SO'
+        header['creation_date_time'] = datetime.datetime.strptime('20030216', '%Y%m%d').date()
+        header['transmission_date'] = datetime.datetime.strptime('20030217', '%Y%m%d').date()
+        header['edi_standard'] = '01.10'
+        header['character_set'] = 'ASCII'
+
+        trailer = {}
+        trailer['record_type'] = 'TRL'
+        trailer['group_count'] = 11
+        trailer['transaction_count'] = 22
+        trailer['record_count'] = 33
+
+        trailer_group = {}
+        trailer_group['record_count'] = 20
+        trailer_group['record_type'] = 'GRT'
+        trailer_group['group_id'] = 3
+        trailer_group['transaction_count'] = 15
+
+        header_group = {}
+        header_group['record_type'] = 'GRH'
+        header_group['version_number'] = '02.10'
+        header_group['group_id'] = 3
+        header_group['batch_request_id'] = 15
+        header_group['transaction_type'] = 'AGR'
+
+        transactions = [[{'original_group_id': 4, 'creation_date_time': datetime.date(2003, 2, 15),
+                          'original_transaction_type': 'AGR', 'record_sequence_n': 15, 'record_type': 'ACK',
+                          'creation_title': 'TITLE', 'original_transaction_sequence_n': 5, 'transaction_status': 'AS',
+                          'recipient_creation_n': 'B124', 'submitter_creation_n': 'A123', 'transaction_sequence_n': 3,
+                          'processing_date': datetime.date(2003, 2, 16)},
+                         {'original_record_sequence_n': 124, 'validation_n': 'AB3', 'message_record_type': 'AGR',
+                          'message_text': 'THE MESSAGE', 'record_sequence_n': 15, 'record_type': 'MSG',
+                          'message_level': 'F', 'message_type': 'G', 'transaction_sequence_n': 3},
+                         {'original_record_sequence_n': 124, 'validation_n': 'AB3', 'message_record_type': 'AGR',
+                          'message_text': 'THE MESSAGE', 'record_sequence_n': 15, 'record_type': 'MSG',
+                          'message_level': 'F', 'message_type': 'G', 'transaction_sequence_n': 3},
+                         {'sales_manufacture_clause': 'M', 'date_of_signature': datetime.date(2003, 2, 17),
+                          'prior_royalty_start_date': datetime.date(2003, 2, 19), 'advance_given': True,
+                          'retention_end_date': datetime.date(2003, 2, 18), 'international_standard_code': 'DFG135',
+                          'prior_royalty_status': 'D', 'agreement_end_date': datetime.date(2003, 2, 16),
+                          'record_type': 'AGR', 'shares_change': True, 'post_term_collection_status': 'D',
+                          'agreement_type': 'OS', 'submitter_agreement_n': 'AB12',
+                          'society_assigned_agreement_n': 'DF35', 'record_sequence_n': 15,
+                          'agreement_start_date': datetime.date(2003, 2, 15), 'transaction_sequence_n': 3,
+                          'post_term_collection_end_date': datetime.date(2003, 2, 20), 'number_of_works': 12}]]
+
+        groups = [{'group_header': header_group, 'transactions': transactions, 'group_trailer': trailer_group}]
+
+        transmission = {'header': header, 'groups': groups, 'trailer': trailer}
+
+        record = self._decoder.decode(transmission)
+
+        self.assertEqual('HDR', record.header.record_type)
+        self.assertEqual('TRL', record.trailer.record_type)
+
+        self.assertEqual(1, len(record.groups))
+
+        group = record.groups[0]
+
+        self.assertEqual('GRH', group.group_header.record_type)
+        self.assertEqual('GRT', group.group_trailer.record_type)
+
+        self.assertEqual('ACK', group.transactions[0][0].record_type)

tests/parser/dictionary/decoder/file/__init__.py

@@ -0,0 +1 @@
+__author__ = 'Bernardo'