From de8d338b5a19073859706d725ea880ed4f779e8a Mon Sep 17 00:00:00 2001
From: Faris Ansari <netchamp.faris@gmail.com>
Date: Wed, 11 Sep 2019 14:04:09 +0530
Subject: [PATCH 1/2] fix: Check if duplicate file exists on disk

---
 frappe/core/doctype/file/file.py | 37 +++++++++++++++++++-------------
 1 file changed, 22 insertions(+), 15 deletions(-)

diff --git a/frappe/core/doctype/file/file.py b/frappe/core/doctype/file/file.py
index d334dc5346..390b091298 100755
--- a/frappe/core/doctype/file/file.py
+++ b/frappe/core/doctype/file/file.py
@@ -22,7 +22,7 @@ import requests
 import requests.exceptions
 import imghdr
 
-from frappe.utils import get_hook_method, get_files_path, random_string, encode, cstr, call_hook_method, cint
+from frappe.utils import get_hook_method, get_files_path, random_string, encode, cstr, call_hook_method, cint, get_site_path
 from frappe import _
 from frappe import conf
 from frappe.utils.nestedset import NestedSet
@@ -201,14 +201,16 @@ class File(NestedSet):
 			duplicate_file = frappe.db.get_value('File', filters, ['name', 'file_url'], as_dict=1)
 
 			if duplicate_file:
-				# if it is attached to a document then throw DuplicateEntryError
-				if self.attached_to_doctype and self.attached_to_name:
-					self.duplicate_entry = duplicate_file.name
-					frappe.throw(_("Same file has already been attached to the record"),
-						frappe.DuplicateEntryError)
-				# else just use the url, to avoid uploading a duplicate
-				else:
-					self.file_url = duplicate_file.file_url
+				duplicate_file_doc = frappe.get_cached_doc('File', duplicate_file.name)
+				if duplicate_file_doc.exists_on_disk():
+					# if it is attached to a document then throw DuplicateEntryError
+					if self.attached_to_doctype and self.attached_to_name:
+						self.duplicate_entry = duplicate_file.name
+						frappe.throw(_("Same file has already been attached to the record"),
+							frappe.DuplicateEntryError)
+					# else just use the url, to avoid uploading a duplicate
+					else:
+						self.file_url = duplicate_file.file_url
 
 	def validate_file_name(self):
 		if not self.file_name and self.file_url:
@@ -332,6 +334,9 @@ class File(NestedSet):
 		data = frappe.db.get_value("File", self.file_data_name, ["file_name", "file_url"], as_dict=True)
 		return data.file_url or data.file_name
 
+	def exists_on_disk(self):
+		exists = os.path.exists(self.get_full_path())
+		return exists
 
 	def upload(self):
 		# get record details
@@ -495,19 +500,21 @@ class File(NestedSet):
 		self.content_hash = get_content_hash(self.content)
 		self.content_type = mimetypes.guess_type(self.file_name)[0]
 
-		_file = False
+		duplicate_file = None
 
 		# check if a file exists with the same content hash and is also in the same folder (public or private)
 		if not ignore_existing_file_check:
-			_file = frappe.get_value("File", {
+			duplicate_file = frappe.get_value("File", {
 					"content_hash": self.content_hash,
 					"is_private": self.is_private
 				},
-				["file_url"])
+				["file_url", "name"], as_dict=True)
 
-		if _file:
-			self.file_url  = _file
-			file_exists = True
+		if duplicate_file:
+			file_doc = frappe.get_cached_doc('File', duplicate_file.name)
+			if file_doc.exists_on_disk():
+				self.file_url  = duplicate_file.file_url
+				file_exists = True
 
 		if os.path.exists(encode(get_files_path(self.file_name, is_private=self.is_private))):
 			self.file_name = get_file_name(self.file_name, self.content_hash[-6:])

From ba96f26f2481faa3480cf1c28a6e03150746a1c6 Mon Sep 17 00:00:00 2001
From: Faris Ansari <netchamp.faris@gmail.com>
Date: Wed, 11 Sep 2019 15:46:34 +0530
Subject: [PATCH 2/2] fix: Remove unused import

---
 frappe/core/doctype/file/file.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/frappe/core/doctype/file/file.py b/frappe/core/doctype/file/file.py
index 390b091298..3c89519faf 100755
--- a/frappe/core/doctype/file/file.py
+++ b/frappe/core/doctype/file/file.py
@@ -22,7 +22,7 @@ import requests
 import requests.exceptions
 import imghdr
 
-from frappe.utils import get_hook_method, get_files_path, random_string, encode, cstr, call_hook_method, cint, get_site_path
+from frappe.utils import get_hook_method, get_files_path, random_string, encode, cstr, call_hook_method, cint
 from frappe import _
 from frappe import conf
 from frappe.utils.nestedset import NestedSet