Crate openqbw 

QuickBooks .qbw file parser.

Provides parsing of invoice line-item records out of an SA17 page-store that was produced by QuickBooks. Built on the opensqlany crate which handles the lower-level page-store layer (CRC validation, AP-fill deobfuscation, slotted-page directories).

This release (v0.1) covers the invoice line-item record format only. Invoice headers, bills, checks, journal entries, and the system catalog are deferred to later releases.

Status

Prototype quality. See OpenQBW/re/NOTES.md (entries C.40–C.43) for the reverse-engineered record layout and remaining gaps.

Structs

AttributionAgreement

Statistics from comparing position-based and content-based attribution over a set of pages.

ContentAttribution

Content-signature attribution map.

CrossValidation

Per-table-id agreement counts.

FkEdge

One inferred foreign-key edge.

FkGraphStats

Summary of edge resolution rates.

LineItem

Parsed invoice line-item record.

NullsFlagBucket

One bucket in the histogram output.

PageAttribution

Resolves page numbers to the SYSTABLE entry that most likely owns them.

RowSignature

A row-0 prefix used as a per-table fingerprint.

SchemaAttribution

Schema-aware width-validator built from SYSCOLUMN + SYSTABLE.

SysColumn

One parsed SYSCOLUMN row.

SysIndexEntry

One parsed SYSINDEX row.

SysTableEntry

A single parsed SYSTABLE row.

TransactionHeader

A minimal transaction-header record.

ValidationStats

Summary of SchemaAttribution::validate_corpus over a page set.

WidthBand

A row-width band derived from a table’s SYSCOLUMN schema.

Enums

AmountType

Classification of the amount-type byte in a line item.

AuditOutcome

Outcome of auditing one SYSINDEX (root_page, table_id) pair against a position-heuristic PageAttribution.

LineItemError

Errors from line-item parsing.

Constants

APAGE_MAGIC

C.37 plaintext magic shared by every regular A-page (SA17 allocation/ free-space map B-tree). The 8-byte sequence appears at a varying offset (typically 0xC0..0xF7) inside the page body and is part of the SA17 page-level metadata block.

DATE_EPOCH_DAYS_BEFORE_UNIX

SA-day -> Unix-day offset: unix_day = sa_day - DATE_EPOCH_DAYS_BEFORE_UNIX. Equivalently, SA-day 0 = Unix-day -4017 = 1956-12-?? – but we use the “SA-day 4017 = Unix-day 0” framing throughout the codebase.

DISAGREE_SAMPLE_LIMIT

Maximum disagreement samples retained in CrossValidation::disagree_samples.

MIN_ROW_BODY_BYTES

Minimum row body length any plausible row must clear, regardless of schema. SA17 rows include a small row header that is not modelled per-column.

OPAQUE_ENTROPY_THRESHOLD

Threshold (bits/byte) above which the page body is treated as uniformly-random. Empirically every page in the four-file Phase 5 corpus that survives all four bv-recovery tiers scores in [7.5, 8.0); pages that do decode have body entropy well below 7.5 once the AP cipher is removed.

SA_DAY_MAX_PLAUSIBLE

Upper plausibility bound (~year 2200) for SA-day values. SA-day 80000 is roughly 2199-12-26 in the Unix calendar.

SA_DAY_MIN_PLAUSIBLE

Lower plausibility bound for SA-day values: any sa_day < 1 is rejected as a missing/zero placeholder.

SIG_LEN

Length in bytes of the row-0 prefix used as a table signature.

SYSCOLUMN_TAG

Fixed 8-byte anchor that precedes the numeric portion of every SYSCOLUMN row body.

SYSINDEX_CREATOR

Two-byte creator id appearing at row offset +2 of every SYSINDEX row. Empirically constant across all QBW files inspected so far (C.30).

SYSOBJECT_NAME_OFFSET

Byte distance between the object_id u32 and the <name_len> byte in a SYSOBJECT row. Determined empirically on RC page 550.

VARIABLE_COLUMN_UPPER_ALLOWANCE

Per-column variable-width allowance (bytes) added to the upper bound when the column’s domain is variable-length (Y, V, C, A).

Functions

bridge_owners_to_tables

Scan all Extent pages for SYSOBJECT-style rows and return a map from SysColumn::owner_object_id to SysTableEntry::name.

build_fk_graph

Build a heuristic FK edge list from the parsed SYSTABLE + SYSCOLUMN catalogs. The list is sorted by (source_table, source_column_id).

collect_unique

Collect a deduplicated catalog keyed by (table_id, name), choosing the first occurrence found.

collect_unique_syscolumns

Deduplicate SYSCOLUMN rows by (owner_object_id, column_id, name) and return them ordered by (owner_object_id, column_id).

collect_unique_sysindex

Collect a deduplicated set of SYSINDEX entries keyed by (table_id, root_page, name), preferring the first sighting.

deobfuscate_with_bv

Decode an E-page with an explicit bv (the inverse of the AP stream cipher).

fk_graph_stats

Compute simple counts over an edge set.

is_opaque_high_entropy

Returns true if the raw page bytes are indistinguishable from random data and therefore unlikely to be standard AP ciphertext.

iter_lineitems

Yields every line item found in store by scanning each E-type page.

iter_lineitems_with_attribution

Like iter_lineitems, but tags each emitted item with the SysTableEntry::name of the table that the line item’s source page is attributed to (via PageAttribution).

iter_syscolumns

Iterate every SYSCOLUMN row recovered from store.

iter_sysindex

Iterate every SYSINDEX row recovered from store.

iter_systable_entries

Iterate every SYSTABLE row recovered from store.

iter_transaction_headers

Iterate transaction headers across every E-page that PageAttribution attributes to a *_header table.

nulls_flag_histogram

Build a histogram of the nulls_flag byte across every SYSCOLUMN row in store, sorted ascending by byte value.

oracle_bv_e_page

Compute the E-page oracle bv assuming plain[0] == 0x00.

recover_bv_any

Cascade of bv-recovery oracles in order of decreasing confidence.

recover_bv_apage

Recover the page bv for an A-page (allocation/free-space map) using the C.37 magic anchor.

recover_bv_brute

Recover the page bv by exhaustive search over all 256 candidates.

recover_bv_qb_data

Try to recover the correct bv for E-page pn (zero-based) using the QB page-trailer anchor. Returns Some(bv) on success, None if no candidate bv decodes the anchor (page is not a QB user-data page).

sa_day_to_unix_day

Convert an SA-day to days since the Unix epoch (1970-01-01).

sa_day_to_unix_seconds

SA-day -> Unix seconds at midnight UTC. Convenience wrapper for downstream code that uses chrono or time for formatting.

scan_syscolumn_page

Scan every slotted-page row body on a single decoded page for SYSCOLUMN rows.

scan_sysindex_page

Scan a single decoded page body for SYSINDEX rows.

scan_systable_page

Scan a decoded page body for SYSTABLE row tags and append parsed entries to out. Matches every occurrence of the 16-byte framed tag followed by a plausible name-length byte; the file-specific 4-byte magic is accepted as wildcard.

schema_for

Return all columns for the table named table_name, ordered by column_id. The bridge is via the SYSOBJECT catalog (see crate::sysobject::bridge_owners_to_tables).

unix_day_to_sa_day

Inverse of sa_day_to_unix_day.