add more logic to handle uploading prelabeled data

RTIInternational · Oct 1, 2024 · ecb104e · ecb104e
1 parent fab4c4f
commit ecb104e
Show file tree

Hide file tree

Showing 4 changed files with 32 additions and 7 deletions.
diff --git a/backend/django/core/utils/util.py b/backend/django/core/utils/util.py
@@ -202,6 +202,16 @@ def create_labels_from_csv(df, project):
     stream = StringIO()
 
     labels = {label.name: label.pk for label in project.labels.all()}
+
+    df["Label"] = df["Label"].apply(
+        lambda s: s.replace("\\n", "\n").replace("\\t", "\t").replace("\\r", "\r")
+    )
+
+    existing_labels = set(labels.keys())
+    df_labels = set(df["Label"].tolist())
+
+    quote_labels = df_labels - existing_labels
+    df["Label"] = df["Label"].apply(lambda s: f'"{s}"' if s in quote_labels else s)
     df["data_id"] = df["hash"].apply(
         lambda x: Data.objects.get(hash=x, project=project).pk
     )
@@ -841,10 +851,18 @@ def create_label_metadata(project, label_data):
     df_label_ids = set(label_data["Label"].tolist())
 
     need_quotes = df_label_ids - existing_label_ids
-    label_data["Label"] = label_data["Label"].apply(lambda s: f'"{s}"'.replace('\\n', '\n').replace('\\t', '\t').replace('\\r', '\r') if s in need_quotes else s)
+    label_data["Label"] = label_data["Label"].apply(
+        lambda s: (
+            f'"{s}"'.replace("\\n", "\n").replace("\\t", "\t").replace("\\r", "\r")
+            if s in need_quotes
+            else s
+        )
+    )
     df_label_ids = set(label_data["Label"].tolist())
     if len(df_label_ids - existing_label_ids) > 0:
-        raise ValidationError("ERROR loading in label metadata. Something is going wrong with the label file.")
+        raise ValidationError(
+            "ERROR loading in label metadata. Something is going wrong with the label file."
+        )
 
     label_data = label_data.merge(label_objects, on="Label", how="inner")
 

diff --git a/backend/django/core/utils/utils_form.py b/backend/django/core/utils/utils_form.py
@@ -79,9 +79,16 @@ def clean_data_helper(
         and len(set(labels_in_data) - set(supplied_labels)) > 0
     ):
         just_in_data = set(labels_in_data) - set(supplied_labels)
-        raise ValidationError(
-            f"There are extra labels in the file which were not created in step 2: {just_in_data}"
-        )
+        # add a correction for label descriptions with weird characters
+        labels_in_data_fixed = [
+            f'"{s}"'.replace("\\n", "\n").replace("\\t", "\t").replace("\\r", "\r")
+            for s in just_in_data
+        ] + list(set(labels_in_data) - just_in_data)
+
+        if len(set(labels_in_data_fixed) - set(supplied_labels)) > 0:
+            raise ValidationError(
+                f"There are extra labels in the file which were not in step 2 of project creation: {just_in_data}"
+            )
 
     if "ID" in data.columns:
         # there should be no null values

diff --git a/frontend/src/actions/skew.js b/frontend/src/actions/skew.js
@@ -90,7 +90,7 @@ export const skewLabel = (dataID, labelID, projectID) => {
                     dispatch(setMessage(response.error));
                 } else {
                     dispatch(getUnlabeled(projectID));
-                    dispatch(getLabelCounts(projectID));
+                    //dispatch(getLabelCounts(projectID));
                 }
             });
     };

diff --git a/frontend/src/components/Skew/index.jsx b/frontend/src/components/Skew/index.jsx
@@ -38,7 +38,7 @@ class Skew extends React.Component {
     componentDidMount() {
         this.props.setFilterStr("");
         this.props.getUnlabeled();
-        this.props.getLabelCounts();
+        //this.props.getLabelCounts();
     }
 
     getText(row) {