Data Structures

File: raw/{game}_raw_tables.json
Source: raw_table_extractor.py

Structure preserves PDF extraction exactly as read:

• Array of scenarios, each with confederate_tables and union_tables
• Each table has header_row, rows[][], and annotations{}
• Values are raw strings from PDF, may span multiple tokens
• Footnote symbols embedded in cell values (e.g., "2*")

Stage 2: Parsed Units

File: parsed/{game}_parsed.json
Source: parse_raw_tables.py

Structured unit objects with semantic fields:

• unit_leader, size, command, unit_type, manpower_value, hex_location
• Footnote symbols extracted to notes[] array
• Side assignment: "Confederate" or "Union"
• Optional: turn, reinforcement_set, table_name

Stage 3: Web JSON

File: web/public/data/{game}.json
Source: convert_to_web.py

Frontend-ready format:

• CamelCase field names (hexLocation, manpowerValue)
• Gunboats/special units separated into confederateGunboats / unionGunboats
• Matches TypeScript interfaces in web/src/types.ts

Inspection Tools

Three utility scripts help debug data at each stage:

# View raw table data
cd parser && uv run python utils/inspect_raw.py <game_id>
cd parser && uv run python utils/inspect_raw.py <game_id> --scenario 1 --table "Confederate Set-Up"

# View parsed units
cd parser && uv run python utils/inspect_parsed.py <game_id>
cd parser && uv run python utils/inspect_parsed.py <game_id> --scenario 1 --side Confederate

# Compare raw vs parsed (for debugging parsing issues)
cd parser && uv run python utils/compare_data.py <game_id> <scenario>
cd parser && uv run python utils/compare_data.py <game_id> <scenario> --side Union --table "Union Set-Up"

All scripts support --help for detailed usage.

Common Debugging Patterns

Missing Units in Parsed Data

1. Check raw extraction: uv run python inspect_raw.py <game> --scenario N
2. Look for the missing unit in raw table rows
3. If present in raw but missing in parsed, check column config in game_configs.json
4. If absent from raw, check PDF extraction with utils/diagnose_pdf.py

Wrong Unit Attributes

1. Compare raw vs parsed: uv run python compare_data.py <game> <scenario>
2. Check column configuration in game_configs.json
3. Verify column mappings match the raw header_row

Footnotes Not Matching

1. Check if symbol is in footnote_symbols list in config
2. Verify annotations dict in raw tables contains the symbol
3. Check parser's footnote extraction logic

Field Naming Across Stages

Detailed Documentation

For complete structure details, examples, and edge cases, see references/structures.md.

That file contains:

• Full JSON schemas for each stage
• Field-by-field descriptions
• TypeScript interface definitions
• Data flow examples
• Unit type explanations (Ldr, Inf, Cav, Art, Special)
• Size hierarchy (Army > Corps > Div > Brig > Regt)
• Hex location formats
• Footnote symbol reference
• Common patterns and edge cases

Related Files

• game_configs.json - Column mappings and parsing rules
• raw_table_extractor.py - Stage 1: PDF → raw
• parse_raw_tables.py - Stage 2: raw → parsed
• convert_to_web.py - Stage 3: parsed → web
• types.ts - TypeScript interfaces