From e55d0e6c75437af79f3338aca1ec12c7ea16e3bd Mon Sep 17 00:00:00 2001
From: Andrew Chen <48723787+chuenchen309@users.noreply.github.com>
Date: Sat, 18 Jul 2026 20:54:50 +0800
Subject: [PATCH] fix(dataprep): skip .jsonl lines that are valid JSON but not
 objects (#7195)

* fix(dataprep): skip .jsonl lines that are valid JSON but not objects

`_read_file_by_format` json.loads each line and hands the result to
`_extract_text_from_json`, which assumes a dict:

    for field in self._TEXT_FIELDS:
        if field in data and isinstance(data[field], str):

A JSON line does not have to be an object -- `"context"`, `["text"]` and
`42` are all valid JSON. For those, `field in data` stops being a key
lookup and becomes a substring/membership test, so `data[field]` raises:

    "context"        -> "text" in "context" is True (substring!)
                     -> TypeError: string indices must be integers
    ["text", "foo"]  -> TypeError: list indices must be integers
    42               -> TypeError: argument of type 'int' is not iterable

The TypeError escapes past `except json.JSONDecodeError: continue`, so the
whole load dies on one odd line.

That except clause is also the tell: a *malformed* line is already skipped
gracefully. A *well-formed* line that happens not to be an object should be
too -- it carries no text either way. This makes the two agree.

Reachable from `unsloth-cli.py:253` (`--dataset foo.jsonl` auto-detect) and
`RawTextDataLoader` is exported from `unsloth/__init__.py`.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* Slim the non-object jsonl regression test and shorten the guard comment

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: Daniel Han <danielhanchen@gmail.com>
---
 tests/test_raw_text.py       | 18 ++++++++++++++++++
 unsloth/dataprep/raw_text.py |  4 ++++
 2 files changed, 22 insertions(+)

diff --git a/tests/test_raw_text.py b/tests/test_raw_text.py
index ba16e0cfc4..18549adfe8 100644
--- a/tests/test_raw_text.py
+++ b/tests/test_raw_text.py
@@ -295,7 +295,25 @@ def test_smart_chunk_text_single_chunk_no_eos_returns_plain_list():
     return True
 
 
+def test_load_from_file_skips_non_object_json_lines():
+    """Non-object .jsonl lines (valid JSON, not dicts) are skipped, not fatal."""
+    # "context" contains "text", ["text"] holds it, 42 isn't iterable -- each
+    # would reach data[field] and raise TypeError without the isinstance guard.
+    with tempfile.NamedTemporaryFile("w", suffix = ".jsonl", delete = False) as f:
+        f.write('"context"\n["text", "x"]\n42\n{"text": "keep this"}\n')
+        path = f.name
+    try:
+        text = RawTextDataLoader(None)._read_file_by_format(path, "json_lines")
+        assert text == "keep this", text
+    finally:
+        os.unlink(path)
+
+    print("test_load_from_file_skips_non_object_json_lines passed")
+    return True
+
+
 if __name__ == "__main__":
     success = test_raw_text_loader()
     success = test_smart_chunk_text_single_chunk_no_eos_returns_plain_list() and success
+    success = test_load_from_file_skips_non_object_json_lines() and success
     sys.exit(0 if success else 1)
diff --git a/unsloth/dataprep/raw_text.py b/unsloth/dataprep/raw_text.py
index 128d966ecd..8623285a25 100644
--- a/unsloth/dataprep/raw_text.py
+++ b/unsloth/dataprep/raw_text.py
@@ -236,6 +236,10 @@ class RawTextDataLoader:
 
     def _extract_text_from_json(self, data):
         """Extract text from JSON object using common field names."""
+        # Skip non-object lines (str/list/number): `field in data` would be a
+        # substring/membership test, not a key lookup, and `data[field]` raises.
+        if not isinstance(data, dict):
+            return ""
         for field in self._TEXT_FIELDS:
             if field in data and isinstance(data[field], str):
                 return data[field]