11 KiB
SOURCE_REGISTRY_V2
Supersedes SOURCE_REGISTRY_V1.md (kept for history). Every dataset here
follows the same lifecycle: acquire → hash → parse → normalize → validate →
persist → smoke → ACTIVE. A failed reimport never silently replaces a
valid ACTIVE version (see central/scripts/import_ba_st_mva.py's >10%
record-loss gate for a concrete example of that guard).
Update: the 9 domains this file originally listed as FAILED (external
acquisition blocker) were re-attempted after the operator manually placed
the official files that were blocked by the Portal NF-e/SPED session gate
into sources/raw/bootstrap_v1/. 6 of those 9 are now ACTIVE. The
remaining 3 (ICMS Monofásico, RTC cClassTrib, RTC Alíquotas CBS) plus all 8
PIS/COFINS Tabela 4.3.10–4.3.17 files are MISSING_BOOTSTRAP_FILE — no
file for them was provided, which is a different, more precise status than
"FAILED": it means acquisition is still blocked, not that something broke.
Every new file's domain was confirmed by its actual sheet names and header row, never by filename alone (see "Identification evidence" below).
Domains ACTIVE (preserved from before this session)
| DOMAIN | AUTHORITY | SOURCE | RAW_SHA256 (prefix) | PARSER_VERSION | RECORD_COUNT | STATUS |
|---|---|---|---|---|---|---|
| NCM | Receita Federal / Siscomex | Previously-fetched Siscomex JSON | 3abc0d2eb2501... |
ncm-v1 | 15156 | ACTIVE (preserved, no internet to re-verify) |
| TIPI_IPI | Receita Federal | 1 TIPI 2022 - Atualizada ADE 001-2026.xlsx |
d155f1baafb40... |
tipi-xlsx-v1 | 11103 | ACTIVE (hash-verified unchanged) |
| CEST | CONFAZ Convênio ICMS 142/18 | Previously-fetched CONFAZ HTML | d7fb7e8da0ace... |
cest-html-v1 | 1051 | ACTIVE (preserved) |
| BA_ST_MVA | SEFAZ-BA / RICMS-BA Anexo 1 | decreto_2012_13780_ricms_anexo_1_vigente_2026.pdf |
e07e908595... |
ba-st-mva-pdf-v1 | 303 | ACTIVE (re-imported, content identical) |
| ORIGEM_MERCADORIA / CST_ICMS / CSOSN / CST_PIS / CST_COFINS | Receita Federal / ENCAT (MOC 7.0 Anexo I) | moc7-anexo-i-leiaute-e-rv.pdf |
5eb4cf2010b1... |
moc7-tables-v1 | 9 / 11 / 10 / 33 / 33 | ACTIVE |
Domains ACTIVE this session (bootstrap_v1 operator-supplied files)
| DOMAIN | AUTHORITY | FILE | RAW_SHA256 (prefix) | PARSER_VERSION | RECORD_COUNT | IDENTIFICATION EVIDENCE |
|---|---|---|---|---|---|---|
| CFOP | Portal Nacional NF-e, Informe Técnico 2023.002 v2.10 | IT_2023_002_v2_10_Tabela_CFOP_Descrição.xlsx (an identical PDF of the same IT is also present, kept as a cross-check copy, not separately imported) |
c95c91ed1644... |
cfop-xlsx-v1 | 619 | Sheet literally named CFOP; header row is CFOP, Início de vigência, Fim de vigência, indNFe, indComunica, indTransp, indDevol, indRetor, indAnula, indRemes, indComb, indExcIBSCBS, Título do CFOP, Descrição — the exact official field set, including the newer indExcIBSCBS (Reforma Tributária) column |
| FCP | Portal Nacional NF-e | Tabela_Alíquotas_FCP_por_UF_2025-02_07.xlsx |
3079d08816... |
fcp-xlsx-v1 | 27 (26 states + DF) | Sheet FCP; header cUF, UF, Nome UF, Aliq_1, Aliq_2, Aliq_3, ., Observação; every Brazilian UF present exactly once |
| ANP | Portal Nacional NF-e | Tabela_Codigo_Produto_ANP-20241118.xlsx |
8271a80007... |
anp-xlsx-v1 | 1031 | Sheet cProdANP; single-column header cProdANP — confirmed this really is the NF-e portal's validation code list (no description column exists in the official source; not invented) |
| NCM_UTRIB | Portal Nacional NF-e, Informe Técnico 2024.001 v2.40 | IT2024.001v2.40 - Tabela NCM 2022 com Utrib_Comércio Exterior_vigência 01.10.26.xlsx |
afc8bfc94c... |
ncm-utrib-xlsx-v1 | 10516 | Sheet NCM X uTrib_Vig 1-10-2026; header NCM, Início de vigência da NCM, Fim de vigência da NCM, uTrib..., Descrição da uTrib...; effective from 2026-10-01 (only version present — no history to preserve) |
| RTC_BIOFUEL_INDEX | Portal Nacional NF-e, Informe Técnico 2025.004 v1.20 | IT 2025.004 v.1.20 - Tabela de Índice.xlsx |
c49a01150... |
rtc-biofuel-index-xlsx-v1 | 23 (14 current + 9 historical) | Two sheets, pBioObrigatorio and pBioObrigatorio_Historico; header cProdANP, Descrição, pBio (% Anidro Obrigatorio), Início de vigência, Fim de vigência; both current and historical rows preserved (is_historical flag), forward-filling only the repeated cProdANP/Descrição a historical block shares with its preceding row — never inventing new values |
| RTC_CCREDPRES | Portal Nacional NF-e / LC 214/2025 | cCredPres_2026-06-22.xlsx |
65f246139... |
rtc-cred-pres-xlsx-v1 | 13 | Sheet cCredPres; header has 18 real official columns (LC 214/2025, Apropria via NF?, ind_gCBSCredPres, Alíquota CBS, pAliqCredPresCBS, dIniVigCBS, etc.) — all 18 preserved as distinct typed columns, not reduced to code+description |
Bug found and fixed mid-import: rtc_cred_pres_entries.cclass_nota_referenciada
was originally VARCHAR(16) (assumed always a numeric class code); several
official rows carry free text there instead (e.g. "cClass de alíquota zero"), which Postgres rejected with a truncation error on insert.
Migration 0009 widens it to Text; no data was lost or silently truncated
— the import failed closed and was retried after the fix (see Dataset
history: the first failed attempt left no orphan row, confirmed via
ROLLBACK).
New read-only API endpoints for these domains (used by the coverage script,
reusable for future classifier work): GET /v1/reference/cfop/{code},
/v1/reference/ncm_utrib/{ncm}, /v1/reference/anp/{cprod_anp},
/v1/reference/fcp/{uf}.
Domains MISSING_BOOTSTRAP_FILE (still blocked — no file provided)
Not FAILED (which implied a broken acquisition attempt) — these simply
have no corresponding file in sources/raw/bootstrap_v1/ yet. The Portal
Nacional NF-e (nfe.fazenda.gov.br) remains the authority for all of them;
this session's Bash tool has zero DNS/internet resolution and the built-in
browser was denied navigation into the portal's document sections on every
attempt after one initial homepage load (see prior evidence, unchanged this
session — this mission's instructions explicitly said not to attempt
internet access again).
| DOMAIN | AUTHORITY | STATUS |
|---|---|---|
| ICMS Monofásico — Combustíveis | Portal Nacional NF-e | MISSING_BOOTSTRAP_FILE |
| RTC — cClassTrib (IBS/CBS) | Portal Nacional NF-e | MISSING_BOOTSTRAP_FILE |
| RTC — Alíquotas CBS | Portal Nacional NF-e | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.10 (alíquotas diferenciadas) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.11 (alíquota por unidade) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.12 (substituição tributária) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.13 (alíquota zero) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.14 (isenção) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.15 (sem incidência) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.16 (suspensão) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
| PIS/COFINS Tabela 4.3.17 (demais alíquotas diferenciadas) | Receita Federal / SPED | MISSING_BOOTSTRAP_FILE |
BLOCKER_EXACT=No file for ICMS Monofásico, RTC cClassTrib, RTC Alíquotas CBS, or any of the 8 PIS/COFINS Tabela 4.3.10-17 documents exists in sources/raw/bootstrap_v1/; the Portal Nacional NF-e/SPED remains unreachable from this environment (unchanged from the prior mission's evidence; not re-tested per this mission's explicit "não tente acessar internet" instruction).
EVIDENCE=find sources/raw -type f lists 11 files; none match these domains by content (only CFOP/FCP/ANP/NCM_UTRIB/RTC_BIOFUEL_INDEX/RTC_CCREDPRES files were provided).
OPERATOR_ACTION_REQUIRED=Place the corresponding official files into sources/raw/bootstrap_v1/ (or a new bootstrap_v3/) for a future mission to identify, hash, and import the same way this session did for the other 6.
Scope prioritization (business scope, 2026-09-19)
GTI clients: restaurants, markets, clothing/retail, workshops, general commerce. No fuel stations; no EFD-Contribuições/SPED in product scope. "Every Brazilian fiscal table" is NOT an MVP requirement. Nothing below is a technical error and no existing support was removed.
CORE_MVP (required): NCM, TIPI_IPI, CEST, BA_ST_MVA, CFOP, ORIGIN, CST_ICMS,
CSOSN, CST_PIS, CST_COFINS, FCP, RTC_CCLASSTRIB, RTC_CCREDPRES, RTC_CBS_RATES.
ACTIVE: all except RTC_CCLASSTRIB and RTC_CBS_RATES (MISSING_BOOTSTRAP_FILE).
| DOMAIN | SCOPE STATUS |
|---|---|
| ICMS_MONOPHASIC | OUT_OF_SCOPE_MVP (reason: fuel/station segment is not the target audience) |
| ANP | ACTIVE_REFERENCE_ONLY (no further rules for MVP) |
| RTC_BIOFUEL_INDEX | ACTIVE_REFERENCE_ONLY (no further rules for MVP) |
| PIS_COFINS_4_3_10..4_3_17 | DEFERRED_BY_BUSINESS_SCOPE (not required for deploy; measure need later from client data) |
CST_PIS / CST_COFINS basic code domains stay ACTIVE.
Operator action — the two remaining CORE files
Place in sources/raw/bootstrap_v1/ (Portal Nacional NF-e, Reforma Tributária / RTC area):
- RTC_CCLASSTRIB — official cClassTrib table (IBS/CBS classificação tributária, xlsx) with CST/cClassTrib codes and their descriptions/indicators.
- RTC_CBS_RATES — official CBS rates table (alíquotas CBS, xlsx) with rate and validity dates.
Status until then:
MISSING_BOOTSTRAP_FILE(verified: no matching file locally).
Update 2026-09-19 — CORE MVP closed (RTC_CCLASSTRIB + RTC_CBS_RATES ACTIVE)
Both files identified by sheet names/headers, not filename.
| DOMAIN | FILE | SHA256 (prefix) | PARSER | RECORDS | EVIDENCE |
|---|---|---|---|---|---|
| RTC_CCLASSTRIB | cClassTrib 2026-06-22.xlsx |
1448cb63a41b... |
rtc-cclasstrib-xlsx-v1 | 164 cClassTrib + 18 CST-IBS/CBS | Sheets CST 2026-06-01 Pub (CST, description, ind_* flags) and cClass 2026-06-01 Pub (CST, cClassTrib, name, description, LC/regulamento refs, tipo de alíquota, pRedIBS/pRedCBS, dIniVig/dFimVig, ind_* per document model, ANEXO, Link). All official columns kept (key fields typed, ind_* in JSON). Every cClass CST exists in the CST sheet; cClassTrib unique; dIniVig 2026-01-01; 3 rows have dFimVig |
| RTC_CBS_RATES | IT 2026.002 v.1.00 - Aliquotas da CBS.xlsx |
33f17aa332... |
rtc-cbs-rates-xlsx-v1 | 2 | Title "Aliquota da CBS", header Ano/Alíquota: 2026 = 0.009; 2027 = "aguarda legislação" (kept as text, rate NULL — not invented) |
Migration 0010 (3 tables). Reimport idempotent (same dataset ids 49/50, no new rows). Both kinds are in dataset_versions() (cache fingerprint changes naturally). New endpoints: GET /v1/reference/rtc_cclasstrib/{code}, /v1/reference/rtc_cbs_rate/{year}. CORE MVP domains: all 14 ACTIVE. Coverage: docs/FISCAL_COVERAGE_FULL_V3.json.