Add file type parameter

This commit adds a file_type parameter to the new Data and API
endpoints and explicitly adds original and preservation versions of
each new report to the report selection UI.

AIPscan/API/namespace_data.py

@@ -12,6 +12,11 @@
 from flask_restx import Namespace, Resource
 from AIPscan.Data import data
 
+
+FILE_FORMAT_FIELD = "file_format"
+FILE_TYPE_FIELD = "file_type"
+PUID_FIELD = "puid"
+
 api = Namespace("data", description="Retrieve data from AIPscan to shape as you desire")
 
 """
@@ -91,7 +96,7 @@ class LargestFileList(Resource):
     @api.doc(
         "list_formats",
         params={
-            "file_type": {
+            FILE_TYPE_FIELD: {
                 "description": "Optional file type filter (original or preservation)",
                 "in": "query",
                 "type": "str",
@@ -105,7 +110,7 @@ class LargestFileList(Resource):
     )
     def get(self, storage_service_id, file_type=None, limit=20):
         """Largest files"""
-        file_type = request.args.get("file_type", None)
+        file_type = request.args.get(FILE_TYPE_FIELD)
         try:
             limit = int(request.args.get("limit", 20))
         except ValueError:
@@ -118,43 +123,55 @@ def get(self, storage_service_id, file_type=None, limit=20):
 
 @api.route("/storage_service/<storage_service_id>/file-format")
 class AIPsByFormatList(Resource):
-    FILE_FORMAT_FIELD = "file_format"
-
     @api.doc(
         "list_aips_by_format",
         params={
             FILE_FORMAT_FIELD: {
                 "description": "File format name (must be exact match)",
                 "in": "query",
                 "type": "str",
-            }
+            },
+            FILE_TYPE_FIELD: {
+                "description": "Optional file type filter (original or preservation)",
+                "in": "query",
+                "type": "str",
+            },
         },
     )
     def get(self, storage_service_id):
         """AIPs containing file format"""
-        file_format = request.args.get(self.FILE_FORMAT_FIELD, "")
+        file_format = request.args.get(FILE_FORMAT_FIELD, "")
+        file_type = request.args.get(FILE_TYPE_FIELD)
         aip_data = data.aips_by_file_format(
-            storage_service_id=storage_service_id, file_format=file_format
+            storage_service_id=storage_service_id,
+            file_format=file_format,
+            file_type=file_type,
         )
         return aip_data
 
 
 @api.route("/storage_service/<storage_service_id>/puid")
 class AIPsByPUIDList(Resource):
-    PUID_FIELD = "puid"
-
     @api.doc(
         "list_aips_by_puid",
         params={
             PUID_FIELD: {
                 "description": "PRONOM ID (PUID)",
                 "in": "query",
                 "type": "str",
-            }
+            },
+            FILE_TYPE_FIELD: {
+                "description": "Optional file type filter (original or preservation)",
+                "in": "query",
+                "type": "str",
+            },
         },
     )
     def get(self, storage_service_id):
         """AIPs containing PUID"""
-        puid = request.args.get(self.PUID_FIELD, "")
-        aip_data = data.aips_by_puid(storage_service_id=storage_service_id, puid=puid)
+        puid = request.args.get(PUID_FIELD, "")
+        file_type = request.args.get(FILE_TYPE_FIELD)
+        aip_data = data.aips_by_puid(
+            storage_service_id=storage_service_id, puid=puid, file_type=file_type
+        )
         return aip_data

AIPscan/Data/data.py

@@ -5,6 +5,7 @@
 from AIPscan import db
 from AIPscan.models import AIP, File, FileType, StorageService
 
+VALID_FILE_TYPES = set(item.value for item in FileType)
 
 FIELD_AIP = "AIP"
 FIELD_AIP_ID = "AIPID"
@@ -228,7 +229,6 @@ def _largest_files_query(storage_service_id, file_type, limit):
 
     This is separated into its own helper function to aid in testing.
     """
-    VALID_FILE_TYPES = set(item.value for item in FileType)
     if file_type is not None and file_type in VALID_FILE_TYPES:
         files = (
             File.query.join(AIP)
@@ -302,14 +302,16 @@ def largest_files(storage_service_id, file_type=None, limit=20):
 
 
 def _query_aips_by_file_format_or_puid(
-    storage_service_id, search_string, file_format=True
+    storage_service_id, search_string, file_format=True, file_type=None
 ):
     """Fetch information on all AIPs with given format or PUID from db.
 
     :param storage_service_id: Storage Service ID (int)
     :param search_string: File format or PUID (str)
     :param file_format: Boolean indicating whether to filter on file
-    format or PUID.
+    format (default) or PUID
+    :param file_type: Optional filter for type of file to return (str,
+    used only if value matches FileType enum options)
 
     :returns: SQLAlchemy query results
     """
@@ -327,18 +329,23 @@ def _query_aips_by_file_format_or_puid(
         .group_by(AIP.id)
         .order_by(db.func.count(File.id).desc(), db.func.sum(File.size).desc())
     )
+    if file_type is not None and file_type in VALID_FILE_TYPES:
+        aips = aips.filter(File.file_type == file_type)
     if file_format:
         return aips.filter(File.file_format == search_string)
     return aips.filter(File.puid == search_string)
 
 
-def _aips_by_file_format_or_puid(storage_service_id, search_string, file_format=True):
+def _aips_by_file_format_or_puid(
+    storage_service_id, search_string, file_type, file_format=True
+):
     """Return overview of all AIPs containing original files in format
 
     :param storage_service_id: Storage Service ID (int)
     :param search_string: File format name or PUID (str)
+    :param file_type: Optional filter for type of file to return (str)
     :param file_format: Boolean indicating whether to filter on file
-    format or PUID.
+    format or PUID (defaults to True)
 
     :returns: "report" dict containing following fields:
         report["StorageName"]: Name of Storage Service queried
@@ -360,7 +367,7 @@ def _aips_by_file_format_or_puid(storage_service_id, search_string, file_format=
 
     report[FIELD_AIPS] = []
     results = _query_aips_by_file_format_or_puid(
-        storage_service_id, search_string, file_format
+        storage_service_id, search_string, file_format, file_type
     )
     for result in results:
         aip_info = {}
@@ -376,23 +383,34 @@ def _aips_by_file_format_or_puid(storage_service_id, search_string, file_format=
     return report
 
 
-def aips_by_file_format(storage_service_id, file_format):
+def aips_by_file_format(storage_service_id, file_format, file_type=None):
     """Return overview of AIPs containing original files in format.
 
     :param storage_service_id: Storage Service ID (int)
     :param file_format: File format name (str)
+    :param file_type: Optional filter for type of file to return (str)
 
     :returns: Report dict provided by _aips_by_file_format_or_puid
     """
-    return _aips_by_file_format_or_puid(storage_service_id, file_format)
+    return _aips_by_file_format_or_puid(
+        storage_service_id=storage_service_id,
+        search_string=file_format,
+        file_type=file_type,
+    )
 
 
-def aips_by_puid(storage_service_id, puid):
+def aips_by_puid(storage_service_id, puid, file_type=None):
     """Return overview of AIPs containing original files in format.
 
     :param storage_service_id: Storage Service ID (int)
     :param puid: PUID (str)
+    :param file_type: Optional filter for type of file to return (str)
 
     :returns: Report dict provided by _aips_by_file_format_or_puid
     """
-    return _aips_by_file_format_or_puid(storage_service_id, puid, False)
+    return _aips_by_file_format_or_puid(
+        storage_service_id=storage_service_id,
+        search_string=puid,
+        file_type=file_type,
+        file_format=False,
+    )

AIPscan/Reporter/report_aips_by_format.py

@@ -2,6 +2,7 @@
 
 from flask import render_template, request
 
+from AIPscan.models import FileType
 from AIPscan.Data import data
 from AIPscan.Reporter import reporter, translate_headers, request_params
 
@@ -11,15 +12,19 @@ def aips_by_format():
     """Return AIPs containing file format, sorted by count and total size."""
     storage_service_id = request.args.get(request_params["storage_service_id"])
     file_format = request.args.get(request_params["file_format"])
+    file_type = request.args.get(request_params["file_type"], FileType.original.value)
     aip_data = data.aips_by_file_format(
-        storage_service_id=storage_service_id, file_format=file_format
+        storage_service_id=storage_service_id,
+        file_format=file_format,
+        file_type=file_type,
     )
     headers = [data.FIELD_AIP_NAME, data.FIELD_UUID, data.FIELD_COUNT, data.FIELD_SIZE]
     return render_template(
         "report_aips_by_format.html",
         storage_service_id=storage_service_id,
         storage_service_name=aip_data.get(data.FIELD_STORAGE_NAME),
         file_format=file_format,
+        file_type=file_type,
         columns=translate_headers(headers),
-        aips=aip_data[data.FIELD_AIPS],
+        aips=aip_data.get(data.FIELD_AIPS),
     )

AIPscan/Reporter/report_aips_by_puid.py

@@ -2,9 +2,9 @@
 
 from flask import render_template, request
 
+from AIPscan.models import File, FileType
 from AIPscan.Data import data
 from AIPscan.Reporter import reporter, translate_headers, request_params
-from AIPscan.models import File
 
 
 def get_format_string_from_puid(puid):
@@ -35,14 +35,18 @@ def aips_by_puid():
     """Return AIPs containing PUID, sorted by count and total size."""
     storage_service_id = request.args.get(request_params["storage_service_id"])
     puid = request.args.get(request_params["puid"])
-    aip_data = data.aips_by_puid(storage_service_id=storage_service_id, puid=puid)
+    file_type = request.args.get(request_params["file_type"], FileType.original.value)
+    aip_data = data.aips_by_puid(
+        storage_service_id=storage_service_id, puid=puid, file_type=file_type
+    )
     headers = [data.FIELD_AIP_NAME, data.FIELD_UUID, data.FIELD_COUNT, data.FIELD_SIZE]
     return render_template(
         "report_aips_by_puid.html",
         storage_service_id=storage_service_id,
         storage_service_name=aip_data.get(data.FIELD_STORAGE_NAME),
         puid=puid,
         file_format=get_format_string_from_puid(puid),
+        file_type=file_type,
         columns=translate_headers(headers),
-        aips=aip_data[data.FIELD_AIPS],
+        aips=aip_data.get(data.FIELD_AIPS),
     )

AIPscan/Reporter/templates/report_aips_by_format.html

@@ -2,7 +2,7 @@
 
 {% block content %}
 
-  <div class="alert alert-secondary" style="height: 100px;">
+  <div class="alert alert-secondary">
     <a class="noprint" onClick="window.print();return false">
     	<button type="button" class="btn btn-info" style="float:right;">Print</button>
     </a>
@@ -11,6 +11,8 @@
     <strong>Storage Service:</strong> {{ storage_service_name }}</strong>
     <br>
     <strong>File format:</strong> {{ file_format }}
+    <br>
+    <strong>File type:</strong> {{ file_type }} files
   </div>
 
   {% if aips %}

AIPscan/Reporter/templates/report_aips_by_puid.html

@@ -2,7 +2,7 @@
 
 {% block content %}
 
-  <div class="alert alert-secondary" style="height: 125px;">
+  <div class="alert alert-secondary">
     <a class="noprint" onClick="window.print();return false">
     	<button type="button" class="btn btn-info" style="float:right;">Print</button>
     </a>
@@ -13,8 +13,10 @@
     <strong>PUID:</strong> {{ puid }}
     {% if file_format %}
       <br>
-      <strong>File format/version:</strong> {{ file_format }}
+      <strong>File format and version:</strong> {{ file_format }}
     {% endif %}
+    <br>
+    <strong>File type:</strong> {{ file_type }} files
   </div>
 
   {% if aips %}