Add validtion to prevent duplicate files for multi-invoice receiving

by comparing sha256 hash values for each file

tailbone/forms/core.py

@@ -24,6 +24,7 @@
 Forms Core
 """
 
+import hashlib
 import json
 import logging
 import warnings
@@ -659,11 +660,25 @@ class Form(object):
                   'widget': MultiFileUploadWidget(tmpstore)}
             # if 'required' in kwargs and not kwargs['required']:
             #     kw['missing'] = colander.null
+            if kwargs.get('validate_unique'):
+                kw['validator'] = self.validate_multiple_files_unique
             files_node = colander.SequenceSchema(file_node, **kw)
             self.set_node(key, files_node)
         else:
             raise ValueError("unknown type for '{}' field: {}".format(key, type_))
 
+    def validate_multiple_files_unique(self, node, value):
+
+        # get SHA256 hash for each file; error if duplicates encountered
+        hashes = {}
+        for fileinfo in value:
+            fp = fileinfo['fp']
+            fp.seek(0)
+            filehash = hashlib.sha256(fp.read()).hexdigest()
+            if filehash in hashes:
+                node.raise_invalid(f"Duplicate file detected: {fileinfo['filename']}")
+            hashes[filehash] = fileinfo
+
     def set_enum(self, key, enum, empty=None):
         if enum:
             self.enums[key] = enum
@@ -906,6 +921,11 @@ class Form(object):
                 return json.dumps({'name': value['filename']})
             return 'null'
 
+        elif isinstance(value, list) and all([isinstance(f, dfwidget.filedict)
+                                              for f in value]):
+            return json.dumps([{'name': f['filename']}
+                               for f in value])
+
         app = self.request.rattail_config.get_app()
         value = app.json_friendly(value)
         return json.dumps(value)

tailbone/forms/widgets.py

@@ -323,6 +323,21 @@ class MultiFileUploadWidget(dfwidget.FileUploadWidget):
     template = 'multi_file_upload'
     requirements = ()
 
+    def serialize(self, field, cstruct, **kw):
+        if cstruct in (colander.null, None):
+            cstruct = []
+
+        if cstruct:
+            for fileinfo in cstruct:
+                uid = fileinfo['uid']
+                if uid not in self.tmpstore:
+                    self.tmpstore[uid] = fileinfo
+
+        readonly = kw.get("readonly", self.readonly)
+        template = readonly and self.readonly_template or self.template
+        values = self.get_template_values(field, cstruct, kw)
+        return field.renderer(template, **values)
+
     def deserialize(self, field, pstruct):
         if pstruct is colander.null:
             return colander.null

tailbone/views/purchasing/receiving.py

@@ -570,7 +570,7 @@ class ReceivingBatchView(PurchasingBatchView):
             elif workflow == 'from_multi_invoice':
                 if 'invoice_files' not in f:
                     f.insert_before('invoice_file', 'invoice_files')
-                f.set_type('invoice_files', 'multi_file')
+                f.set_type('invoice_files', 'multi_file', validate_unique=True)
                 f.set_required('invoice_parser_key')
                 f.remove('truck_dump_batch_uuid',
                          'po_number',