From b72f110c7d090a32bf356a0c29ca0aaeac0b3400 Mon Sep 17 00:00:00 2001
From: Anastasia Beglova <abeglova@mit.edu>
Date: Mon, 6 Oct 2025 14:36:48 -0400
Subject: [PATCH] truncate csv files in tutor problems api

---
 learning_resources/views.py      | 27 +++++++++++++++++++--------
 learning_resources/views_test.py | 21 ++++++++++++++++-----
 2 files changed, 35 insertions(+), 13 deletions(-)

diff --git a/learning_resources/views.py b/learning_resources/views.py
index a6a6597c04..899c3914bc 100644
--- a/learning_resources/views.py
+++ b/learning_resources/views.py
@@ -1471,18 +1471,29 @@ def retrieve_problem(self, request, run_readable_id, problem_title):  # noqa: AR
         return Response(
             {
                 "problem_set_files": [
-                    {
-                        "file_name": problem_file.file_name,
-                        "content": problem_file.content,
-                    }
+                    problem_set_file_output(problem_file)
                     for problem_file in problem_files
                 ],
                 "solution_set_files": [
-                    {
-                        "file_name": solution_file.file_name,
-                        "content": solution_file.content,
-                    }
+                    problem_set_file_output(solution_file)
                     for solution_file in solution_files
                 ],
             }
         )
+
+
+def problem_set_file_output(problem_set_file):
+    if problem_set_file.file_extension == ".csv":
+        csv_lines = problem_set_file.content.splitlines()
+        return {
+            "file_name": problem_set_file.file_name,
+            "file_extension": ".csv",
+            "truncated_content": "\n".join(csv_lines[0:5]),
+            "number_of_records": len(csv_lines) - 1,
+            "note": "The content of the data file has been truncated to the column headers and first 4 rows.",  # noqa: E501
+        }
+    return {
+        "file_name": problem_set_file.file_name,
+        "content": problem_set_file.content,
+        "file_extension": problem_set_file.file_extension,
+    }
diff --git a/learning_resources/views_test.py b/learning_resources/views_test.py
index 87ef7560d2..5ba2409108 100644
--- a/learning_resources/views_test.py
+++ b/learning_resources/views_test.py
@@ -1456,14 +1456,16 @@ def test_course_run_problems_endpoint(client, user_role, django_user_model):
         type="problem",
         content="Content for Problem Set 1",
         file_name="problem1.txt",
+        file_extension=".txt",
     )
 
     TutorProblemFileFactory.create(
         run=course_run,
         problem_title="Problem Set 1",
         type="problem",
-        content="Content for Problem Set 1 Part 2",
-        file_name="problem1-b.txt",
+        content="a,b\n1,1\n2,2\n3,3\n4,4\n5,5\n6,6",
+        file_name="problem1-data.csv",
+        file_extension=".csv",
     )
 
     TutorProblemFileFactory.create(
@@ -1472,6 +1474,7 @@ def test_course_run_problems_endpoint(client, user_role, django_user_model):
         type="solution",
         content="Content for Problem Set 1 Solution",
         file_name="solution1.txt",
+        file_extension=".txt",
     )
     TutorProblemFileFactory.create(
         run=course_run, problem_title="Problem Set 2", type="problem"
@@ -1496,16 +1499,24 @@ def test_course_run_problems_endpoint(client, user_role, django_user_model):
     if user_role in ["admin", "group_tutor_problem_viewer"]:
         assert detail_resp.json() == {
             "problem_set_files": [
-                {"file_name": "problem1.txt", "content": "Content for Problem Set 1"},
                 {
-                    "file_name": "problem1-b.txt",
-                    "content": "Content for Problem Set 1 Part 2",
+                    "file_name": "problem1.txt",
+                    "content": "Content for Problem Set 1",
+                    "file_extension": ".txt",
+                },
+                {
+                    "file_name": "problem1-data.csv",
+                    "truncated_content": "a,b\n1,1\n2,2\n3,3\n4,4",
+                    "file_extension": ".csv",
+                    "note": "The content of the data file has been truncated to the column headers and first 4 rows.",
+                    "number_of_records": 6,
                 },
             ],
             "solution_set_files": [
                 {
                     "file_name": "solution1.txt",
                     "content": "Content for Problem Set 1 Solution",
+                    "file_extension": ".txt",
                 },
             ],
         }