Dev/issue 8894 premis parsing

docs/examples.rst

@@ -0,0 +1,65 @@
+Example usage
+=============
+
+Parsing METS documents
+----------------------
+
+Example of listing the relative file paths of preservation files referenced in
+a METS file:::
+
+  import metsrw
+
+  mets = metsrw.METSDocument.fromfile('fixtures/complete_mets.xml')
+  for entry in mets.all_files():
+      if entry.use == 'preservation':
+          print entry.path
+
+Example of retrieving a file by UUID:::
+
+  import metsrw
+
+  mets = metsrw.METSDocument.fromfile('fixtures/complete_mets.xml')
+  entry = mets.get_file('46b7cb96-792c-4441-a5d6-67c83313501c')
+  print entry.path
+
+Creating/modifying METS documents
+---------------------------------
+
+Example creation of a METS document (without PREMIS or Dublin Core metadata):::
+
+  import metsrw
+  import uuid
+
+  mw = metsrw.METSDocument()
+
+  # Create object entries
+  file1 = metsrw.FSEntry('objects/cat.png', file_uuid=str(uuid.uuid4()))
+  file2 = metsrw.FSEntry('objects/dog.jpg', file_uuid=str(uuid.uuid4()))
+
+  # Create preservation derivative entries
+  file1p = metsrw.FSEntry('objects/cat-preservation.tiff', use='preservation', file_uuid=str(uuid.uuid4()), derived_from=file1)
+  file2p = metsrw.FSEntry('objects/dog-preservation.tiff', use='preservation', file_uuid=str(uuid.uuid4()), derived_from=file2)
+
+  # Create object directory entry
+  objects = metsrw.FSEntry('objects', type='Directory', children=[file1, file2, file1p, file2p])
+
+  # Create metadata subdirectories then metadata directory entry
+  children = [
+      metsrw.FSEntry('transfers', type='Directory', children=[]),
+      metsrw.FSEntry('metadata/metadata.csv', use='metadata', file_uuid=str(uuid.uuid4())),
+  ]
+  metadata = metsrw.FSEntry('metadata', type='Directory', children=children)
+
+  # Create submission METS entry and submission documentation parent directory entry
+  children = [
+      metsrw.FSEntry('submissionDocumentation/METS.xml', use='submissionDocumentation', file_uuid=str(uuid.uuid4())),
+  ]
+  sub_doc = metsrw.FSEntry('submissionDocumentation', type='Directory', children=children)
+
+  # Create SIP entry containing objects, metadata, and submission documentaton entries
+  children = [objects, metadata, sub_doc]
+  sip = metsrw.FSEntry('sipname-uuid', type='Directory', children=children)
+
+  # Add SIP entry to METS document and write to file
+  mw.append_file(sip)
+  mw.write('mets.xml', fully_qualified=True, pretty_print=True)

metsrw/__init__.py

@@ -9,3 +9,4 @@
 from .metadata import *
 from .mets import *
 from .utils import *
+from . import premis

metsrw/exceptions.py

@@ -3,11 +3,18 @@
 
 All exceptions generated by this library will descend from MetsError.
 """
+
+
 class MetsError(Exception):
     """ Base Exception for this module. """
     pass
 
 
+class ConstructError(Exception):
+    """ Base constructing an object. """
+    pass
+
+
 class ParseError(MetsError):
-    """ Error parsing a METS file. """
+    """ Error parsing a METS file or PREMIS element. """
     pass

metsrw/fsentry.py

@@ -90,6 +90,11 @@ def __init__(self, path, label=None, use='original', type=u'Item', children=None
         self.amdsecs = []
         self.dmdsecs = []
 
+        # Convenient access to metadata (without cycling through amdsecs)
+        self.techmds = []
+        self.digiprovmds = []
+        self.rightsmds = []
+
     def __str__(self):
         return '{s.type}: {s.path}'.format(s=self)
 

metsrw/metadata.py

@@ -3,6 +3,7 @@
 """
 from __future__ import absolute_import
 
+from collections import OrderedDict
 import logging
 from lxml import etree
 from random import randint
@@ -280,6 +281,67 @@ def serialize(self):
         return el
 
 
+class DublinCoreXmlData(object):
+    """
+    An object representing a METS xmlData element containing a Dublin Core element.
+
+    :raises exceptions.ParseError: If the root element tag is not xmlData.
+    """
+    DC_ELEMENTS = ['title', 'creator', 'subject', 'description', 'publisher', 'contributor', 'date', 'format', 'identifier', 'source', 'relation', 'language', 'coverage', 'rights']
+
+    def __init__(self, title=None, creator=None, subject=None, description=None, publisher=None, contributor=None, date=None, format=None, identifier=None, source=None, relation=None, language=None, coverage=None, rights=None):
+        for element in self.DC_ELEMENTS:
+            setattr(self, element, locals()[element])
+
+    @classmethod
+    def parse(cls, root):
+        """
+        Parse an xmlData element containing a Dublin Core dublincore element.
+
+        :param root: Element or ElementTree to be parsed into an object.
+        :raises exceptions.ParseError: If the root is not xmlData or doesn't contain a dublincore element.
+        """
+        if root.tag != utils.lxmlns('mets') + 'xmlData':
+            raise exceptions.ParseError('DublinCoreXmlData can only parse xmlData elements with mets namespace.')
+
+        dc_el = root.find('dcterms:dublincore', namespaces=utils.NAMESPACES)
+
+        if dc_el is None or dc_el.tag != utils.lxmlns('dcterms') + 'dublincore':
+            raise exceptions.ParseError('xmlData can only contain a dublincore element with the dcterms namespace.')
+
+        args = []
+
+        for element in DublinCoreXmlData.DC_ELEMENTS:
+            args.append(dc_el.findtext("dc:" + element, namespaces=utils.NAMESPACES))
+
+        return cls(*args)
+
+    def serialize(self):
+        nsmap = OrderedDict([
+            ('mets', utils.NAMESPACES['mets']),
+            ('xsi', utils.NAMESPACES['xsi']),
+            ('xlink', utils.NAMESPACES['xlink'])
+        ])
+        root = etree.Element(utils.lxmlns('mets') + 'xmlData', nsmap=nsmap)
+        root.append(self._serialize_dublincore())
+        return root
+
+    def _serialize_dublincore(self):
+        nsmap = OrderedDict([
+            ('dcterms', utils.NAMESPACES['dcterms']),
+            ('dc', utils.NAMESPACES['dc'])
+        ])
+        attrib = {'{}schemaLocation'.format(utils.lxmlns('xsi')): utils.DUBLINCORE_SCHEMA_LOCATIONS}
+        dc_root = etree.Element(utils.lxmlns('dcterms') + 'dublincore', nsmap=nsmap, attrib=attrib)
+
+        for element in DublinCoreXmlData.DC_ELEMENTS:
+            dc_el = etree.Element(utils.lxmlns('dc') + element)
+            dc_el.text = getattr(self, element)
+            dc_root.append(dc_el)
+
+        return dc_root
+
+
 class MDWrap(object):
     """
     An object representing an XML document enclosed in a METS document.
@@ -291,13 +353,16 @@ class MDWrap(object):
     :param str mdtype: The MDTYPE of XML document being enclosed. Examples
         include "PREMIS:OBJECT" and "PREMIS:EVENT".
     """
-    def __init__(self, document, mdtype):
+    MDTYPE_CLASSES = {'DC': DublinCoreXmlData}
+
+    def __init__(self, document, mdtype, data=None):
         parser = etree.XMLParser(remove_blank_text=True)
         if isinstance(document, six.string_types):
             self.document = etree.fromstring(document, parser=parser)
         elif isinstance(document, etree._Element):
             self.document = document
         self.mdtype = mdtype
+        self.data = data
 
     @classmethod
     def parse(cls, root):
@@ -313,11 +378,17 @@ def parse(cls, root):
         mdtype = root.get('MDTYPE')
         if not mdtype:
             raise exceptions.ParseError('mdWrap must have a MDTYPE')
+        if mdtype in MDWrap.MDTYPE_CLASSES.keys():
+            mdtype_class = MDWrap.MDTYPE_CLASSES[mdtype]()
+            data = mdtype_class.parse(root.find('mets:xmlData', namespaces=utils.NAMESPACES)).__dict__
+        else:
+            data = None
+
         document = root.xpath('mets:xmlData/*', namespaces=utils.NAMESPACES)
         if len(document) != 1:
             raise exceptions.ParseError('mdWrap and xmlData can only have one child')
         document = document[0]
-        return cls(document, mdtype)
+        return cls(document, mdtype, data)
 
     def serialize(self):
         el = etree.Element(utils.lxmlns('mets') + 'mdWrap', MDTYPE=self.mdtype)

metsrw/mets.py

@@ -1,5 +1,6 @@
 from __future__ import absolute_import
 
+from collections import OrderedDict
 from datetime import datetime
 import logging
 from lxml import etree
@@ -108,17 +109,17 @@ def _document_root(self, fully_qualified=True):
         """
         Return the mets Element for the document root.
         """
-        nsmap = {
-            'xsi': utils.NAMESPACES['xsi'],
-            'xlink': utils.NAMESPACES['xlink']
-        }
+        nsmap = OrderedDict([
+            ('xsi', utils.NAMESPACES['xsi']),
+            ('xlink', utils.NAMESPACES['xlink'])
+        ])
         if fully_qualified:
             nsmap['mets'] = utils.NAMESPACES['mets']
         else:
             nsmap[None] = utils.NAMESPACES['mets']
         attrib = {
             '{}schemaLocation'.format(utils.lxmlns('xsi')):
-            utils.SCHEMA_LOCATIONS
+            utils.METS_SCHEMA_LOCATIONS
         }
         return etree.Element(utils.lxmlns('mets') + 'mets', nsmap=nsmap, attrib=attrib)
 
@@ -285,6 +286,10 @@ def _parse_tree_structmap(self, tree, parent_elem):
                     amdsec = metadata.AMDSec.parse(amdsec_elem)
                     fs_entry.amdsecs.append(amdsec)
 
+                    # Add subsections to convience properties
+                    for subsection in amdsec.subsections:
+                        getattr(fs_entry, subsection.subsection.lower() + 's').append(subsection)
+
             siblings.append(fs_entry)
         return siblings
 

metsrw/premis/__init__.py

@@ -0,0 +1,8 @@
+from __future__ import absolute_import
+
+import logging
+LOGGER = logging.getLogger(__name__)
+LOGGER.addHandler(logging.NullHandler())
+
+from .object import *
+from .event import *