Add CSV max filesize check (#190)

invenio_previewer/config.py

@@ -21,14 +21,17 @@
 """Confidence threshold for character encoding detection by `cchardet`."""
 
 PREVIEWER_MAX_FILE_SIZE_BYTES = 1 * 1024 * 1024
-"""Maximum file size in bytes for JSON/XML/CSV files."""
+"""Maximum file size in bytes for JSON/XML files."""
 
 PREVIEWER_MAX_IMAGE_SIZE_BYTES = 0.5 * 1024 * 1024
 """Maximum file size in bytes for image files."""
 
 PREVIEWER_TXT_MAX_BYTES = 1 * 1024 * 1024
 """Maximum number of .txt file bytes to preview before truncated."""
 
+PREVIEWER_CSV_MAX_BYTES = 100 * 1024 * 1024
+"""Maximum file size in bytes for CSV files."""
+
 PREVIEWER_ZIP_MAX_FILES = 1000
 """Max number of files showed in the ZIP previewer."""
 

invenio_previewer/extensions/csv_papaparsejs.py

@@ -14,10 +14,18 @@
 
 previewable_extensions = ["csv", "dsv"]
 
+def validate_csv(file):
+    """Validate a CSV file."""
+    max_file_size = current_app.config.get(
+        "PREVIEWER_CSV_MAX_BYTES", 100 * 1024 * 1024
+    )
+    if file.size > max_file_size:
+        return False
+    return True
 
 def can_preview(file):
     """Determine if the given file can be previewed."""
-    return file.is_local() and file.has_extensions(".csv", ".dsv")
+    return file.is_local() and file.has_extensions(".csv", ".dsv") and validate_csv(file)
 
 
 def preview(file):