From f57c718877226959ab3cde76cb05b45d80012b52 Mon Sep 17 00:00:00 2001
From: BryanFauble <17128019+BryanFauble@users.noreply.github.com>
Date: Thu, 7 Nov 2024 13:25:21 -0700
Subject: [PATCH 1/3] Update file annotation store process to require filename
 be present in order to annotate file

---
 schematic/store/synapse.py | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/schematic/store/synapse.py b/schematic/store/synapse.py
index 4c1b900a8..973bdabcc 100644
--- a/schematic/store/synapse.py
+++ b/schematic/store/synapse.py
@@ -2266,8 +2266,19 @@ async def add_annotations_to_entities_files(
 
         """
 
-        # Expected behavior is to annotate files if `Filename` is present and if file_annotations_upload is set to True regardless of `-mrt` setting
-        if "filename" in [col.lower() for col in manifest.columns]:
+        filename_column_present = False
+        filename_column = None
+
+        # Find the exact column name for 'Filename'
+        for col in manifest.columns:
+            if col.lower() == "filename":
+                filename_column_present = True
+                filename_column = col
+                break
+
+        # Expected behavior is to annotate files if `Filename` is present and if
+        # file_annotations_upload is set to True regardless of `-mrt` setting
+        if filename_column_present:
             # get current list of files and store as dataframe
             dataset_files = self.getFilesInStorageDataset(datasetId)
             files_and_entityIds = self._get_file_entityIds(
@@ -2295,6 +2306,9 @@ async def add_annotations_to_entities_files(
                 row["entityId"] = manifest_synapse_table_id
                 manifest.loc[idx, "entityId"] = manifest_synapse_table_id
                 entityId = ""
+            elif filename_column_present and not row[filename_column]:
+                # Skip any attempt to annotate a row which does not have a filename
+                entityId = ""
             else:
                 # get the file id of the file to annotate, collected in above step.
                 entityId = row["entityId"]

From e832d0266c748a1c3737d83a2f1cf9fdd141027b Mon Sep 17 00:00:00 2001
From: BryanFauble <17128019+BryanFauble@users.noreply.github.com>
Date: Thu, 7 Nov 2024 14:53:20 -0700
Subject: [PATCH 2/3] Revert "Update file annotation store process to require
 filename be present in order to annotate file"

This reverts commit f57c718877226959ab3cde76cb05b45d80012b52.
---
 schematic/store/synapse.py | 18 ++----------------
 1 file changed, 2 insertions(+), 16 deletions(-)

diff --git a/schematic/store/synapse.py b/schematic/store/synapse.py
index 973bdabcc..4c1b900a8 100644
--- a/schematic/store/synapse.py
+++ b/schematic/store/synapse.py
@@ -2266,19 +2266,8 @@ async def add_annotations_to_entities_files(
 
         """
 
-        filename_column_present = False
-        filename_column = None
-
-        # Find the exact column name for 'Filename'
-        for col in manifest.columns:
-            if col.lower() == "filename":
-                filename_column_present = True
-                filename_column = col
-                break
-
-        # Expected behavior is to annotate files if `Filename` is present and if
-        # file_annotations_upload is set to True regardless of `-mrt` setting
-        if filename_column_present:
+        # Expected behavior is to annotate files if `Filename` is present and if file_annotations_upload is set to True regardless of `-mrt` setting
+        if "filename" in [col.lower() for col in manifest.columns]:
             # get current list of files and store as dataframe
             dataset_files = self.getFilesInStorageDataset(datasetId)
             files_and_entityIds = self._get_file_entityIds(
@@ -2306,9 +2295,6 @@ async def add_annotations_to_entities_files(
                 row["entityId"] = manifest_synapse_table_id
                 manifest.loc[idx, "entityId"] = manifest_synapse_table_id
                 entityId = ""
-            elif filename_column_present and not row[filename_column]:
-                # Skip any attempt to annotate a row which does not have a filename
-                entityId = ""
             else:
                 # get the file id of the file to annotate, collected in above step.
                 entityId = row["entityId"]

From 11be01a755d0a32fc68a83f10bde3077222e5d11 Mon Sep 17 00:00:00 2001
From: BryanFauble <17128019+BryanFauble@users.noreply.github.com>
Date: Thu, 7 Nov 2024 14:55:06 -0700
Subject: [PATCH 3/3] Don't attempt to annotate the table

---
 schematic/store/synapse.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/schematic/store/synapse.py b/schematic/store/synapse.py
index 4c1b900a8..9caffca13 100644
--- a/schematic/store/synapse.py
+++ b/schematic/store/synapse.py
@@ -2250,7 +2250,10 @@ async def add_annotations_to_entities_files(
         manifest_synapse_table_id="",
         annotation_keys: str = "class_label",
     ):
-        """Depending on upload type add Ids to entityId row. Add anotations to connected files.
+        """
+        Depending on upload type add Ids to entityId row. Add anotations to connected
+        files and folders. Despite the name of this function, it also applies to folders.
+
         Args:
             dmge: DataModelGraphExplorer Object
             manifest (pd.DataFrame): loaded df containing user supplied data.
@@ -2295,6 +2298,9 @@ async def add_annotations_to_entities_files(
                 row["entityId"] = manifest_synapse_table_id
                 manifest.loc[idx, "entityId"] = manifest_synapse_table_id
                 entityId = ""
+                # If the row is the manifest table, do not add annotations
+            elif row["entityId"] == manifest_synapse_table_id:
+                entityId = ""
             else:
                 # get the file id of the file to annotate, collected in above step.
                 entityId = row["entityId"]