The missing step is a decision, not another parser
Our Markdown example includes a second table with opening cash, operating, investing, financing, FX and closing balances. Exporting that table yields strings. The cash-flow checker expects a named JSON object with period, currency, unit and cash basis. No automatic adapter connects the two: a row called “cash” would not reliably tell a program which balance it represents.
Make the handoff reproducible
Place the three repositories side by side in a working directory. From md-table-to-csv, export the second synthetic table. Then inspect it before preparing a separate JSON file in the cash-flow project.
python md_table_csv.py examples/table-sample.md --table 2 --output out/cash-extract.csv
# Inspect Item/Amount pairs: Opening 100; Operating 35;
# Investing -20; Financing -10; FX -2; Closing 103Compare the six values with cashflow-bridge-check/examples/reconciled.json. Confirm USD millions and the same cash-and-cash-equivalents definition. Keep the negative signs. The first revenue table's decimal-comma strings do not belong in these fields: explicitly normalize a locale-specific number only after establishing its convention.
python cashflow_bridge.py examples/reconciled.json --tolerance 0 --json
# calculated_closing 103; difference 0; matches trueKeep the transformation separate from the conclusion
Record table number 2, the mapping of row names to JSON fields and the source unit in a short delivery note. If you alter a value, preserve the original CSV and explain the change. Avoid the spreadsheet-safe CSV option for a numeric handoff unless you consciously remove and document its protective apostrophes.
A successful calculation establishes neither that a table was extracted correctly nor that the reporting definition is appropriate. The useful result is that another person can repeat the export, review the mapping, and obtain the same arithmetic. Tool syntax lives in Table Workshop and Cashflow Fieldnotes; this handoff is the part neither tool automates.
If the starting document is a PDF, the related KolmoPDF cloud converter is an optional upstream way to obtain Markdown; it uploads the document and is not called by these local scripts. After checking the cash bridge, Eqvero company research is the related place to continue investigating public disclosures with source references. Neither step removes the need to review the manual row-to-field mapping.
← Field notes