Skip to content

Network Data Issues

Common network data problems and solutions when using legacy or filtered networks.

Transit Lines with Highway Modes

Symptom: Error during create_tod_scenarios: KeyError: 'x'

Full Error: 

File "tm2py/components/network/create_tod_scenarios.py", line 435
    line["@invehicle_factor"] = in_vehicle_factors[line.vehicle.mode.id]
KeyError: 'x'

Cause: - Transit lines in the EMME network have been assigned highway mode ‘x’ (MAZ-to-MAZ mode) - Mode ‘x’ is not defined in [transit.modes] config - Common in networks built before November 2025 (sprint-04 and earlier)

Root Cause: - Network build process contamination where MAZ centroid connectors or highway modes leaked into transit line definitions - Possible filtering artifacts when extracting subregional networks

Solution: Code now handles this gracefully (as of January 2026): 

# Checks if mode exists before accessing
if line.vehicle.mode.id in in_vehicle_factors:
    line["@invehicle_factor"] = in_vehicle_factors[line.vehicle.mode.id]
else:
    # Logs warning and skips

Warning Message: 

Warning: Transit line 12345 uses mode 'x' not defined in transit.modes config. 
This may indicate network build issues. Skipping perception factors for this line.

Prevention: - Use networks from sprint-05 or later (post-November 2025) - Verify transit line modes during network build QA - Check that all transit lines use only transit modes (not ‘x’, ‘c’, etc.)

“Illegal character ‘+’” in reportlexer.py

Symptom: SOLA traffic assignment crashes with lexer error: 

File "inro\emme\procedure\reportlexer.py", line 103
LexError: Illegal character '+' at position X

Full Error Context: Assignment runs for ~30 seconds then crashes during EMME’s internal report parsing.

Root Cause Chain: 1. @lanes = 0 for all links (attribute not copied from base network) 2. @capacity = 0 (calculated from @lanes in VDF/capacity formulas) 3. VDF divides by @capacity → produces infinity 4. EMME formats infinity as 0.135972+124 (scientific notation without ‘e’) 5. Report lexer crashes on unexpected + character

Why @lanes Was Zero:

Two potential causes:

  1. Case sensitivity in period attribute copying (fixed January 2026):
  2. Period-specific attributes like @lanes_am use lowercase suffixes
  3. Period names in config are uppercase ("AM")
  4. The endswith("AM") check failed for @lanes_am

  5. Fix: Use case-insensitive comparison and try lowercase suffix first

  6. get_attribute_values() API misuse: The API returns [id_array, value_array], not just values: 

    # WRONG - iterates over 2 arrays, not individual values
lanes_result = scenario.get_attribute_values("LINK", ["@lanes"])
if all(v == 0 for v in lanes_result):  # Always False!
    ...

# CORRECT - extract the value array (index 1)
lanes_result = scenario.get_attribute_values("LINK", ["@lanes"])
lanes_values = lanes_result[1]  # The actual values
if all(v == 0 for v in lanes_values):
    ...

Fix Applied: create_tod_scenarios.py now: - Uses case-insensitive matching for period attributes - Correctly extracts lanes_result[1] for value comparisons - Uses link.num_lanes (not link.lanes) for EMME standard attribute

Missing Node Crosswalk Files

See Node ID Crosswalk Documentation

See Also