CroBank Manual Parser

Small, pure‑Python helper for inspecting and exporting Cronos/CroBank datasets when you want to hand‑craft schemas instead of relying on cronodump. You describe fixed‑width fields, preview raw records, then export to CSV.

Contents

• Prerequisites
• Core Ideas
• Workflow Overview
• Command Reference
• Building the Schema
• Advanced Tweaks
• Troubleshooting

Prerequisites

• Python 3.9+ (standard library only).
• Access to a CroBank dataset directory containing CroBank.dat, CroBank.tad, CroStru.dat, etc.
• Basic familiarity with fixed‑width records and CP1251 text (default encoding).

Core Ideas

• Manual schema: Define every field with name, byte offset, and byte length in SCHEMA.
• Fixed-size records: Assumes constant record width; variable/complex records require custom parsing.
• Two modes:
  • inspect: dump helper views so you can reverse‑engineer offsets.
  • export: apply your schema to CroBank.dat and write a CSV.

Workflow Overview

1. Inspect raw bytes to guess the layout.
2. Populate SCHEMA (and optionally RECORD_SIZE) in tad_parser.py.
3. Iterate until the decoded fields look correct.
4. Export to CSV.

Command Reference

Inspect

python tad_parser.py inspect "D:\Cronos\610.РФ Авиаперелеты 05.2021 (travellers)" ^
    --max-records 5 ^
    --guess-record-size 256

Outputs (saved beside the dataset):

• CroStru_dump.txt: first 1024 bytes of CroStru.dat (hex + cp1251 text).
• CroBank_sample_records.txt: first N records sliced by guess-record-size (or computed size if SCHEMA exists).

Use these dumps to decide field offsets/lengths.

Export

python tad_parser.py export "D:\Cronos\610.РФ Авиаперелеты 05.2021 (travellers)" ^
    --output travellers.csv

Prints the record size and field list, then writes a UTF‑8 CSV.

Building the Schema

Edit the top of tad_parser.py:

SCHEMA = [
    FieldSpec("ID",         0, 10),
    FieldSpec("LASTNAME",  10, 40),
    FieldSpec("FIRSTNAME", 50, 40),
    # ...
]
RECORD_SIZE = 256      # optional; otherwise sum of field lengths
ENCODING = "cp1251"    # change if your data uses another code page

Guidelines:

• Offsets are zero-based. Skip headers by starting the first field later.
• Lengths must cover the whole field; under-sizing truncates data.
• Set RECORD_SIZE if there is padding beyond your fields; otherwise it is computed.

Advanced Tweaks

• Parse non-text fields: extend decode_field() to cast to int/date/float.
• Separator-based records: modify parse_record() to read up to a delimiter (e.g., 0x1E) instead of fixed slices.
• Filtering: add logic in cmd_export() if you need to skip specific records.
• Large files: the script streams records, but you can flush the writer periodically for very large datasets.

Troubleshooting

Symptom	Likely Cause	Fix
RECORD_SIZE is None and SCHEMA is empty	You ran export without defining a schema.	Fill SCHEMA or set RECORD_SIZE.
Garbled CSV fields	Wrong offsets/lengths or encoding.	Re-run inspect, adjust SCHEMA, verify ENCODING.
Warning: file size not multiple of record size	Record size guess is off or there is footer data.	Correct RECORD_SIZE or handle trailing bytes.
Lots of � characters	Wrong encoding.	Set ENCODING to the correct code page.

Use inspect to explore, adjust SCHEMA until fields align, then export to dump the dataset. Remember: this tool assumes fixed-width records and plain text unless you extend it.