Add PREMIS Object XML to files

Adds display of PREMIS object XML, if populated, when viewing files.

AIPscan/Aggregator/database_helpers.py

@@ -5,6 +5,7 @@
 """
 
 from celery.utils.log import get_task_logger
+from lxml import etree
 
 from AIPscan import db
 from AIPscan.Aggregator import tasks
@@ -57,8 +58,7 @@ def _extract_event_detail(premis_event, file_id):
 
 
 def _create_agent_type_id(identifier_type, identifier_value):
-    """Create a key-pair string for the linking_type_value in the db.
-    """
+    """Create a key-pair string for the linking_type_value in the db."""
     return "{}-{}".format(identifier_type, identifier_value)
 
 
@@ -306,6 +306,19 @@ def _add_normalization_date(file_id):
         db.session.commit()
 
 
+def _add_premis_object_xml(fs_entry, file_id):
+    """Add string representation of PREMIS Object to File object."""
+    file_ = File.query.get(file_id)
+
+    if hasattr(fs_entry, "amdsecs"):
+        for ss in fs_entry.amdsecs[0].subsections:
+            if ss.contents.mdtype == fs_entry.PREMIS_OBJECT:
+                file_.premis_object = etree.tostring(
+                    ss.contents.serialize(), pretty_print=True
+                )
+        db.session.commit()
+
+
 def _get_original_file(related_uuid):
     """Get original file related to preservation derivative."""
     return File.query.filter_by(uuid=related_uuid, file_type=FileType.original).first()
@@ -352,6 +365,8 @@ def create_file_object(file_type, fs_entry, aip_id):
     if file_type == FileType.preservation:
         _add_normalization_date(new_file.id)
 
+    _add_premis_object_xml(fs_entry, new_file.id)
+
 
 def collect_mets_agents(mets):
     """Collect all of the unique agents in the METS file to write to the

AIPscan/Reporter/templates/file.html

@@ -66,6 +66,17 @@ <h2 class="h3">File: {{ file_.name }}</h2>
       {{ file_.checksum_value }}
     </td>
   </tr>
+  {% if file_.premis_object %}
+  <tr>
+    <td width=20%>
+      <strong>PREMIS Object XML</strong>
+    </td>
+    <td>
+      <pre>{% for line in premisxml: %}{{ line }}
+{% endfor %}</pre>
+    </td>
+  </tr>
+  {% endif %}
   {% if preservation_file %}
     <tr>
       <td width=20%>

AIPscan/Reporter/views.py

@@ -163,6 +163,7 @@ def view_file(file_id):
     return render_template(
         "file.html",
         file_=file_,
+        premisxml=file_.premis_object.decode("utf-8").split("\\n"),
         aip=aip,
         events=events,
         preservation_file=preservation_file,

AIPscan/models.py

@@ -384,6 +384,7 @@ class File(db.Model):
     format_version = db.Column(db.String(255))
     checksum_type = db.Column(db.String(255))
     checksum_value = db.Column(db.String(255), index=True)
+    premis_object = db.Column(db.Text())
 
     original_file_id = db.Column(db.Integer(), db.ForeignKey("file.id"))
     original_file = db.relationship(
@@ -405,6 +406,7 @@ def __init__(
         checksum_value,
         aip_id,
         file_type=FileType.original,
+        premis_object=None,
         format_version=None,
         puid=None,
         original_file_id=None,
@@ -420,6 +422,7 @@ def __init__(
         self.format_version = format_version
         self.checksum_type = checksum_type
         self.checksum_value = checksum_value
+        self.premis_object = premis_object
         self.original_file_id = original_file_id
         self.aip_id = aip_id
 

requirements/base.txt

@@ -18,7 +18,7 @@ Jinja2==2.11.3
 kombu==4.6.10
 lxml==4.6.5
 MarkupSafe==1.1.1
-metsrw==0.3.15
+metsrw==0.3.22
 natsort==7.0.1
 pandas==1.1.4
 plotly==5.7.0