Loading Issue in Local LEAN CLI Backtest (Global M2 CSV)

Here’s a focused diagnostic checklist and code example to help you debug why your custom GlobalM2Data class is not being loaded by QuantConnect LEAN CLI. These steps show exactly where things typically go wrong and how to catch issues early:

1. Class Location & Import

Issue: If GlobalM2Data isn’t defined in the same file—or isn’t properly imported—Python can’t find it.

• Solution: Make sure GlobalM2Data and your QCAlgorithm subclass live in the same file (main.py), or that you are importing it:

  from GlobalM2Data import GlobalM2Data  # if in another file
  

2. Path Formatting in get_source

Issue: LEAN requires forward slashes (/) and the path should be relative to project root.

• Code/Check:

  path = "data/custom/global_m2-weekly.csv"
  

3. Is get_source Triggering?

Issue: If get_source is never called, AddData is wrong, or LEAN can’t find the class.

• Diagnostic Insert:

  print("[GlobalM2Data.get_source] called")
  if not os.path.exists(path):
      print(f"[GlobalM2Data.get_source] FILE NOT FOUND: {path}")
  else:
      print(f"[GlobalM2Data.get_source] File found: {path}")
  

• And in your reader method:

  print(f"[GlobalM2Data.reader] {line}")
  

4. AddData Usage & Typo Checks

Issue: Wrong class reference (e.g., typo) or placed outside initialize.

• Should be, exactly:

  self.add_data(GlobalM2Data, "GLOBALM2", Resolution.Daily)
  

• And GlobalM2Data must be the exact class name (including case).

5. File Placement & File Check

Issue: File is in the wrong folder or not visible to Docker/LEAN.

• Always place in: [LEAN Root]/YourProject/data/custom/global_m2-weekly.csv
• Check existence inside get_source:

  if not os.path.exists("data/custom/global_m2-weekly.csv"):
      print("File not found!")
  

6. CSV File Format/Encoding

Check that:

• No extra header rows or BOM/hidden chars in your file.
• First line is exactly as expected and all lines are valid data.
• Open in Notepad++, Save As UTF-8 (without BOM).

7. lean.json Conflicts

• Check that data-folder is not set to an unexpected path. It should be ./ for local data if using relative paths.

  {
    "data-folder": "./"
  }
  

8. Use Both print and self.Debug

• In get_source and reader, use both for maximum visibility in CLI and Docker logs.

Recommended Diagnostic Code

from AlgorithmImports import *
import os
class GlobalM2Data(PythonData):
    def get_source(self, config: SubscriptionDataConfig,
                  date: datetime, is_live_mode: bool) -> SubscriptionDataSource:
        # ---- diagnostics ---------------------------------------------------
        print("[GlobalM2Data.get_source] called")
        path = "data/custom/global_m2-weekly.csv"       # forward slashes only
        if not os.path.exists(path):
            print(f"[GlobalM2Data.get_source] FILE NOT FOUND: {path}")
        else:
            print(f"[GlobalM2Data.get_source] File found: {path}")
        # --------------------------------------------------------------------
        return SubscriptionDataSource(path, SubscriptionTransportMedium.LOCAL_FILE)

    def reader(self, config: SubscriptionDataConfig,
               line: str, date: datetime, is_live_mode: bool) -> BaseData:
        # Raw-line echo for debugging ----------------------------------------
        print(f"[GlobalM2Data.reader] {line}")
        # --------------------------------------------------------------------
        if line.startswith('\ufeff'):          # strip UTF-8 BOM
            line = line.lstrip('\ufeff')
            print("[GlobalM2Data.reader] BOM removed")

        parts = line.split(',')
        if len(parts) < 2:
            return None                       # skip malformed rows

        self.symbol = config.symbol
        self.time   = date
        try:
            self.value = float(parts[1])
        except ValueError:
            print(f"[GlobalM2Data.reader] Bad float: {parts[1]}")
            return None
        return self
class GlobalM2DataAlgorithm(QCAlgorithm):
    def initialize(self) -> None:
        self.set_start_date(2021, 1, 1)
        self.set_end_date(2022, 1, 1)
        self.set_cash(100000)

        # Add the custom data stream
        self._symbol_m_2 = self.add_data(GlobalM2Data, "GLOBALM2").symbol
        self.debug(f"[Initialize] Added GlobalM2Data: {self._symbol_m_2}")

    def on_data(self, slice: Slice) -> None:
        if self._symbol_m_2 in slice:
            data = slice[self._symbol_m_2]
            if data:
                self.debug(f"[OnData] {data.time}  m_2={data.value}")

lean.json (if present, in project root):

{
  "data-folder": "./"
}

Summary of Common Pitfalls:

• Mistyped class name or file name.
• Path uses \ not /.
• Data file is not project-root relative or not visible to Docker.
• Missing or broken AddData.
• Syntax errors or header/encoding issues in CSV.
• Code structure/entrypoint issues (algorithm isn't run from the right file).

Double-check each section above! If you follow this pattern and still see no print/Debug messages from your custom data class, the root cause is almost always an import/scope, AddData, or path issue. If you do see those messages, but nothing in on_data, then it’s a parsing/timing/data issue.

Let me know which debug statements you see (or not) after trying this!

Sources: Custom Data - QuantConnect.com, Key Concepts - QuantConnect.com, Key Concepts - QuantConnect.com, CSV Format Example - QuantConnect.com, docs writing algorithms securities requesting data