@mailwoman/corpus 9.4.0 → 10.1.0

package/out/tools/source-register/build.d.ts

@@ -0,0 +1,145 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Derive the committed address-source register from a research pass's two CSVs.
+ *
+ * The second input is mostly not input: the eight global discovery lookups repeat once per jurisdiction and are dropped, because they name a lookup rather than a national source.
+ */
+import type { PathBuilderLike } from "path-ts";
+import { JurisdictionResearchState, SourceStatus, type AddressSourceRecord, type ElectedLicense, type LicenseDecision, type RefusedLicense } from "#source-register";
+/**
+ * What the build produced, for a caller that wants to print it rather than re-read the file.
+ */
+export interface SourceRegisterBuildResult {
+    outPath: string;
+    jurisdictions: number;
+    sources: number;
+    licenses: number;
+    /**
+     * Rows dropped because their source identifies a global discovery lookup rather than a national source.
+     */
+    discoveryRailRowsDropped: number;
+    researchStates: Readonly<Record<JurisdictionResearchState, number>>;
+    statuses: Readonly<Record<SourceStatus, number>>;
+}
+export interface BuildSourceRegisterOptions {
+    /**
+     * The 250-row jurisdiction inventory CSV.
+     */
+    inventoryPath: PathBuilderLike;
+    /**
+     * The functional-authority CSV, discovery lookups included.
+     *
+     * This build drops them.
+     */
+    sourcesPath: PathBuilderLike;
+    outPath: PathBuilderLike;
+    /**
+     * License decisions somebody made by reading a publisher's terms, applied over the
+     * `unchecked` defaults this build derives from the research pass's access labels.
+     *
+     * An input rather than an edit of the output, because the register is generated
+     * and {@linkcode buildSourceRegister} rewrites it whole.
+     * A path that does not exist is read as no decisions recorded.
+     */
+    decisionsPath?: PathBuilderLike;
+    /**
+     * Source fields a review resolved, applied over the rows this build derives from the research CSV.
+     *
+     * An input rather than an edit of the output, for the same reason as {@linkcode
+     * BuildSourceRegisterOptions.decisionsPath}: this build rewrites the register whole.
+     * A path that does not exist is read as no resolutions recorded.
+     */
+    resolutionsPath?: PathBuilderLike;
+    version: string;
+    /**
+     * ISO 8601 calendar date the research pass was taken, `yyyy-MM-DD`.
+     */
+    authoredAt: string;
+    /**
+     * The research document the pass was written against, recorded in the register's provenance.
+     */
+    sourceVersion?: string;
+}
+/**
+ * Replace every retired word in one string.
+ */
+export declare function rewriteRetiredVocabulary(text: string): string;
+/**
+ * A column's value, or `undefined` when the research pass left it unresolved.
+ *
+ * Every value outside the placeholder set is returned, so a value somebody fills in
+ * later reaches the register or fails the build rather than being lost.
+ */
+export declare function readUnresolvedColumn(value: string | undefined, column: string, row: number): string | undefined;
+/**
+ * Build the register and write it, refusing to write one that fails the audit.
+ *
+ * The output is tab-indented JSON.
+ * `oxfmt` applies additional formatting.
+ *
+ * @throws When an input row contains a vocabulary this build has no mapping for, when a declared
+ * rewrite never fires, or when the finished register fails {@linkcode auditAddressSourceRegister}.
+ */
+/**
+ * The shape of `license-decisions.json`: license id to the decision minus its own id,
+ * keyed by id so each license has one decision.
+ */
+/**
+ * A decision body without its own id, as `license-decisions.json` records one.
+ */
+type RecordedDecision = Omit<ElectedLicense, "licenseID"> | Omit<RefusedLicense, "licenseID">;
+/**
+ * A decision that reuses a body declared once under `sharedReadings`.
+ *
+ * One publisher can hold a license over several jurisdictions, and the register
+ * scopes a decision to one publisher in one jurisdiction, so reading that publisher's
+ * terms once produces several decisions with the same body.
+ * Naming the shared body keeps each license id's own decision while the prose behind
+ * it has one home, so nine copies cannot drift apart under later editing.
+ */
+interface SharedDecisionReference {
+    sameAs: string;
+}
+/**
+ * The shape of `license-decisions.json`.
+ */
+export interface LicenseDecisionsFile {
+    decisions?: Record<string, RecordedDecision | SharedDecisionReference>;
+    /**
+     * Decision bodies that several license ids reuse, keyed by a name the ids refer to.
+     */
+    sharedReadings?: Record<string, RecordedDecision>;
+}
+/**
+ * The source fields a review resolves, which the research CSV records as placeholders.
+ *
+ * The CSV holds the research pass's findings and is a working document of that pass,
+ * so a later review records its own findings here rather than by editing the pass's record.
+ */
+type SourceResolution = Pick<AddressSourceRecord, "addressRoles" | "coverage" | "upstreamLineage" | "personalDataReview" | "adapterID">;
+/**
+ * Merges each recorded resolution onto the source whose id it holds.
+ *
+ * @throws When a resolution's source id is absent from the register.
+ * Such an entry is a typo or a source that has been removed.
+ * Keeping it silently would leave a review whose fields reach no row, which reads as work already done.
+ */
+export declare function applySourceResolutions(sources: readonly AddressSourceRecord[], recorded: ReadonlyMap<string, SourceResolution>): AddressSourceRecord[];
+/**
+ * Turns a decisions file into decisions keyed by license id, resolving every `sameAs` reference.
+ *
+ * The id comes from the key, so a record cannot disagree with the license it is filed under.
+ * `auditAddressSourceRegister` decides whether the resolved fields form a well-formed decision, so this
+ * function checks only that the file declares a reading for each reference.
+ *
+ * @throws When a `sameAs` reads a reading absent from `sharedReadings`.
+ * Such a reference would otherwise produce a decision carrying no terms, which the audit
+ * would report as an unknown review state rather than as the typo it is.
+ */
+export declare function resolveRecordedDecisions(file: LicenseDecisionsFile): Map<string, LicenseDecision>;
+export declare function buildSourceRegister(options: BuildSourceRegisterOptions): Promise<SourceRegisterBuildResult>;
+export {};
+//# sourceMappingURL=build.d.ts.map

package/out/tools/source-register/build.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"build.d.ts","sourceRoot":"","sources":["../../../lib/tools/source-register/build.ts"],"names":[],"mappings":"AAAA;;;;;;;;GAQG;AAKH,OAAO,KAAK,EAAE,eAAe,EAAE,MAAM,SAAS,CAAA;AAI9C,OAAO,EAKN,yBAAyB,EAIzB,YAAY,EAEZ,KAAK,mBAAmB,EAExB,KAAK,cAAc,EAEnB,KAAK,eAAe,EACpB,KAAK,cAAc,EAGnB,MAAM,kBAAkB,CAAA;AA4KzB;;GAEG;AACH,MAAM,WAAW,yBAAyB;IACzC,OAAO,EAAE,MAAM,CAAA;IACf,aAAa,EAAE,MAAM,CAAA;IACrB,OAAO,EAAE,MAAM,CAAA;IACf,QAAQ,EAAE,MAAM,CAAA;IAChB;;OAEG;IACH,wBAAwB,EAAE,MAAM,CAAA;IAChC,cAAc,EAAE,QAAQ,CAAC,MAAM,CAAC,yBAAyB,EAAE,MAAM,CAAC,CAAC,CAAA;IACnE,QAAQ,EAAE,QAAQ,CAAC,MAAM,CAAC,YAAY,EAAE,MAAM,CAAC,CAAC,CAAA;CAChD;AAED,MAAM,WAAW,0BAA0B;IAC1C;;OAEG;IACH,aAAa,EAAE,eAAe,CAAA;IAC9B;;;;OAIG;IACH,WAAW,EAAE,eAAe,CAAA;IAC5B,OAAO,EAAE,eAAe,CAAA;IACxB;;;;;;;OAOG;IACH,aAAa,CAAC,EAAE,eAAe,CAAA;IAC/B;;;;;;OAMG;IACH,eAAe,CAAC,EAAE,eAAe,CAAA;IACjC,OAAO,EAAE,MAAM,CAAA;IACf;;OAEG;IACH,UAAU,EAAE,MAAM,CAAA;IAClB;;OAEG;IACH,aAAa,CAAC,EAAE,MAAM,CAAA;CACtB;AAoBD;;GAEG;AACH,wBAAgB,wBAAwB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAQ7D;AA2PD;;;;;GAKG;AACH,wBAAgB,oBAAoB,CAAC,KAAK,EAAE,MAAM,GAAG,SAAS,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,MAAM,GAAG,SAAS,CAgB/G;AAED;;;;;;;;GAQG;AACH;;;GAGG;AACH;;GAEG;AACH,KAAK,gBAAgB,GAAG,IAAI,CAAC,cAAc,EAAE,WAAW,CAAC,GAAG,IAAI,CAAC,cAAc,EAAE,WAAW,CAAC,CAAA;AAE7F;;;;;;;;GAQG;AACH,UAAU,uBAAuB;IAChC,MAAM,EAAE,MAAM,CAAA;CACd;AAED;;GAEG;AACH,MAAM,WAAW,oBAAoB;IACpC,SAAS,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,gBAAgB,GAAG,uBAAuB,CAAC,CAAA;IACtE;;OAEG;IACH,cAAc,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,gBAAgB,CAAC,CAAA;CACjD;AAMD;;;;;GAKG;AACH,KAAK,gBAAgB,GAAG,IAAI,CAC3B,mBAAmB,EACnB,cAAc,GAAG,UAAU,GAAG,iBAAiB,GAAG,oBAAoB,GAAG,WAAW,CACpF,CAAA;AA0BD;;;;;;GAMG;AACH,wBAAgB,sBAAsB,CACrC,OAAO,EAAE,SAAS,mBAAmB,EAAE,EACvC,QAAQ,EAAE,WAAW,CAAC,MAAM,EAAE,gBAAgB,CAAC,GAC7C,mBAAmB,EAAE,CAiBvB;AAeD;;;;;;;;;;GAUG;AACH,wBAAgB,wBAAwB,CAAC,IAAI,EAAE,oBAAoB,GAAG,GAAG,CAAC,MAAM,EAAE,eAAe,CAAC,CAoBjG;AAED,wBAAsB,mBAAmB,CAAC,OAAO,EAAE,0BAA0B,GAAG,OAAO,CAAC,yBAAyB,CAAC,CAoFjH"}

package/out/tools/source-register/build.js

@@ -0,0 +1,515 @@
+/**
+ * @copyright Sister Software
+ * @license AGPL-3.0
+ * @author Teffen Ellis, et al.
+ *
+ * Derive the committed address-source register from a research pass's two CSVs.
+ *
+ * The second input is mostly not input: the eight global discovery lookups repeat once per jurisdiction and are dropped, because they name a lookup rather than a national source.
+ */
+import { pathExists, readLocalJSONFile } from "@mailwoman/core/fs/readers";
+import { writeLocalFile } from "@mailwoman/core/fs/writers";
+import { prettyJSON, stringifyJSON } from "@mailwoman/core/json";
+import { TextSpliterator } from "spliterator";
+import { readCSVRecords } from "#recipes/scaffold";
+import { applyLicenseDecisions, auditAddressSourceRegister, registerContentDigest, BackboneState, JurisdictionResearchState, LicenseReviewState, ResearchPass, SourceGeometry, SourceStatus, UNRESOLVED_FIELDS, } from "#source-register";
+import { AddressRole } from "#types";
+/**
+ * The `origin` value the research pass gives its eight repeated discovery lookups.
+ * Every row carrying it is dropped.
+ */
+const DISCOVERY_RAIL_ORIGIN = "global_research_rail";
+/**
+ * Joins the two parts of the per-sector counter's key.
+ *
+ * A solidus separates the fields because a NUL cannot survive a format pass.
+ * The solidus is safe because an ISO 3166-1 code has two letters and a sector is kebab-case.
+ * Neither field can contain a solidus.
+ */
+const KEY_SEPARATOR = "/";
+const BACKBONE_STATE_BY_LETTER = {
+    A: BackboneState.Verified,
+    "A~": BackboneState.VerifiedPartial,
+    B: BackboneState.Restricted,
+    C: BackboneState.Unverified,
+    D: BackboneState.Exceptional,
+};
+const RESEARCH_STATE_BY_NAME = {
+    SEEDED: JurisdictionResearchState.Seeded,
+    EXCEPTION: JurisdictionResearchState.Exception,
+    RAILS_ONLY: JurisdictionResearchState.Unexamined,
+};
+/**
+ * Every `AddressRole` by its wire value, derived from the enum so a role added to
+ * `AddressRole` is readable here without a second edit.
+ * A CSV value outside this role set fails the build.
+ */
+const ADDRESS_ROLE_BY_NAME = Object.fromEntries(Object.values(AddressRole).map((role) => [role, role]));
+const SOURCE_STATUS_BY_NAME = {
+    VERIFIED_AUTHORITY: SourceStatus.VerifiedAuthority,
+    RETAINED_ORIGINAL: SourceStatus.RetainedOriginal,
+    VERIFIED_CORPUS: SourceStatus.VerifiedCorpus,
+    VERIFIED_CORPUS_STALE: SourceStatus.VerifiedCorpusStale,
+};
+const RESEARCH_PASS_BY_ORIGIN = {
+    "2026-09-18_web_research": ResearchPass.WebResearch,
+    original_memo: ResearchPass.OriginalMemo,
+    "2026-09-30_rights_review": ResearchPass.RightsReview,
+    "2026-10-01_rights_review": ResearchPass.RightsReview,
+    "2026-10-02_rights_review": ResearchPass.RightsReview,
+    "2026-10-03_decoder_exposure": ResearchPass.DecoderExposureSurvey,
+};
+const GEOMETRY_BY_NAME = {
+    Y: SourceGeometry.Present,
+    N: SourceGeometry.Absent,
+    Partial: SourceGeometry.Partial,
+    varies: SourceGeometry.Unresolved,
+    "": SourceGeometry.Unresolved,
+};
+/**
+ * The research pass writes a source's propositions as one string.
+ *
+ * The table is closed, because a spelling it does not include is a vocabulary the
+ * register has not agreed to rather than a row to guess at.
+ */
+const ASSERTS_BY_ROLE = {
+    "IDENTITY + OBSERVATION": ["identity", "observation"],
+    OBSERVATION: ["observation"],
+};
+/**
+ * The access labels the research pass recorded, each with the id prefix
+ * and note a decision derived from it records.
+ *
+ * A decision is scoped to one publisher in one jurisdiction by {@link scopedLicenseID},
+ * so an election cannot reach past the grant it was made about.
+ * Every decision is `unchecked`, because the pass recorded what a register costs
+ * to reach and never opened anybody's terms.
+ */
+const LICENSE_DECISIONS = [
+    [
+        "CHECK NATIONAL / DATASET TERMS",
+        "unchecked-national-terms",
+        "The research pass recorded the register and did not open its terms.",
+    ],
+    [
+        "Free",
+        "unchecked-access-free",
+        "An access label from the original memo: the download costs nothing. It states no grant.",
+    ],
+    ["Free-reg", "unchecked-access-free-registration", "An access label: free after registration. It states no grant."],
+    [
+        "Free/varied",
+        "unchecked-access-free-varied",
+        "An access label: free in part, varying by dataset. It states no grant.",
+    ],
+    [
+        "Free/restricted",
+        "unchecked-access-free-or-restricted",
+        "An access label: free in part, restricted in part. It states no grant.",
+    ],
+    [
+        "Free/Free-reg",
+        "unchecked-access-free-or-registration",
+        "An access label: free, or free after registration. It states no grant.",
+    ],
+    [
+        "Free/Licensed",
+        "unchecked-access-free-or-licensed",
+        "An access label: free in part, licensed in part. It states no grant.",
+    ],
+    [
+        "Free-reg/Licensed",
+        "unchecked-access-registration-or-licensed",
+        "An access label: free after registration, or licensed. It states no grant.",
+    ],
+    ["Restricted", "unchecked-access-restricted", "An access label: access is restricted. It states no grant."],
+    ["Varied", "unchecked-access-varied", "An access label: varying by dataset. It states no grant."],
+    [
+        "Licensed",
+        "unchecked-access-licensed",
+        "An access label: a license is sold or granted on application. It states no terms.",
+    ],
+    [
+        "Licensed/query",
+        "unchecked-access-licensed-or-query",
+        "An access label: licensed, or answerable per query. It states no terms.",
+    ],
+];
+/**
+ * Rewrites for complete notes that use retired vocabulary.
+ * Each entry records the replacement note.
+ *
+ * Each rewrite must fire at least once, so a pair left behind after the upstream
+ * text changes fails the build.
+ */
+const RETIRED_NOTE_REWRITES = [
+    [
+        "Official CC BY dataset, but portal states only ~4% geographic coverage; not national LABEL.",
+        "Official CC BY dataset, but portal states only ~4% geographic coverage; not a national address backbone.",
+    ],
+    [
+        "High-value taxpayer address observations; no verified national open premise LABEL.",
+        "High-value taxpayer address observations; no verified national open premise address backbone.",
+    ],
+    [
+        "2026 system is new; validate operational bulk access before corpus promotion.",
+        "The national system is new; validate operational bulk access before corpus promotion.",
+    ],
+];
+/**
+ * Rewrites for words removed from the repository's vocabulary.
+ *
+ * Each entry gives the plain synonym used when the research pass emitted that word.
+ *
+ * The substitution runs over every string this build emits, because `repo-health`'s
+ * `bannedVocabulary` counter reads every tracked text file and a committed JSON artifact is one.
+ */
+const RETIRED_WORD_REWRITES = [
+    [/\bGated\b/g, "Restricted"],
+    [/\bgated\b/g, "restricted"],
+    [/-gated\b/g, "-restricted"],
+];
+function required(value, field, row) {
+    const trimmed = (value ?? "").trim();
+    if (!trimmed)
+        throw new Error(`row ${row}: ${field} is empty, and the register has nowhere to put an unnamed value`);
+    return trimmed;
+}
+function mapped(table, key, field, row) {
+    const value = table[key];
+    if (value === undefined) {
+        throw new Error(`row ${row}: ${field} is ${stringifyJSON(key)}, which this build has no mapping for`);
+    }
+    return value;
+}
+/**
+ * Replace every retired word in one string.
+ */
+export function rewriteRetiredVocabulary(text) {
+    let out = text;
+    for (const [pattern, to] of RETIRED_WORD_REWRITES) {
+        out = out.replaceAll(pattern, to);
+    }
+    return out;
+}
+/**
+ * Apply the whole-note rewrites, recording which fired, then the word-level ones.
+ */
+function rewriteNote(note, fired) {
+    for (const [from, to] of RETIRED_NOTE_REWRITES) {
+        if (note === from) {
+            fired.add(from);
+            return to;
+        }
+    }
+    return rewriteRetiredVocabulary(note);
+}
+async function readJurisdictions(inventoryPath, fired) {
+    const records = [];
+    let row = 1;
+    for await (const record of readCSVRecords(inventoryPath)) {
+        row++;
+        const researchState = mapped(RESEARCH_STATE_BY_NAME, required(record["functional_research_state"], "functional_research_state", row), "functional_research_state", row);
+        const jurisdiction = {
+            iso2: required(record["iso2"], "iso2", row),
+            name: required(record["jurisdiction"], "jurisdiction", row),
+            backboneState: mapped(BACKBONE_STATE_BY_LETTER, required(record["tier"], "tier", row), "tier", row),
+            researchState,
+            bestPath: rewriteRetiredVocabulary(required(record["source"], "source", row)),
+            assertionPlan: required(record["role"], "role", row),
+            note: rewriteNote(required(record["notes"], "notes", row), fired),
+        };
+        if (researchState !== JurisdictionResearchState.Seeded) {
+            jurisdiction.stateReason = rewriteRetiredVocabulary(required(record["functional_next_action"], "functional_next_action", row));
+        }
+        records.push(jurisdiction);
+    }
+    return records;
+}
+/**
+ * A party's name reduced to the form that decides whether two spellings name the same party:
+ * lower-cased with every run outside `a-z0-9` collapsed to one hyphen, so case
+ * and punctuation merge rather than issue two decisions over one grant.
+ */
+function partyKey(value) {
+    const slug = value
+        .toLowerCase()
+        .replaceAll(/[^a-z0-9]+/gu, "-")
+        .replaceAll(/^-+|-+$/gu, "");
+    return slug || "unnamed";
+}
+/**
+ * The id fragment for a party is its key trimmed to prevent a long ministry key from dominating
+ * the id. {@link scopedLicenseID} compares the untrimmed key before accepting a match.
+ */
+const ID_FRAGMENT_LENGTH = 48;
+function idFragment(partyKeyValue) {
+    return partyKeyValue.slice(0, ID_FRAGMENT_LENGTH).replaceAll(/-+$/gu, "") || "unnamed";
+}
+/**
+ * The license id for one source row, scoped to the party a reviewer would read.
+ *
+ * The id includes the publisher when that value exists.
+ * It uses the source id when the publisher value is empty.
+ *
+ * The jurisdiction also forms part of the scope because a publisher can use the same name in several states.
+ */
+function scopedLicenseID(statementID, publisher, sourceID, iso2, statement, scopes) {
+    const scopedTo = publisher || sourceID;
+    const key = `${iso2.toLowerCase()}-${partyKey(scopedTo)}`;
+    const licenseID = `${statementID}-${idFragment(key)}`;
+    const existing = scopes.get(licenseID);
+    if (existing && existing.partyKey !== key) {
+        throw new Error(`two license scopes collapse to ${stringifyJSON(licenseID)}: ${stringifyJSON(existing.scopedTo)} and ` +
+            `${stringifyJSON(scopedTo)}. They are different parties whose names agree over the first ` +
+            `${ID_FRAGMENT_LENGTH} characters of their id fragment. Merging them would let one reading grant both, ` +
+            "which is what scoping the id to a publisher exists to prevent.");
+    }
+    if (!existing) {
+        scopes.set(licenseID, { licenseID, statement, publisher: publisher || null, scopedTo, partyKey: key, iso2 });
+    }
+    return licenseID;
+}
+async function readSources(sourcesPath, licenseByStatement, fired, scopes) {
+    const records = [];
+    const ordinalByKey = new Map();
+    let discoveryRailRowsDropped = 0;
+    let row = 1;
+    for await (const record of readCSVRecords(sourcesPath)) {
+        row++;
+        const origin = required(record["origin"], "origin", row);
+        if (origin === DISCOVERY_RAIL_ORIGIN) {
+            discoveryRailRowsDropped++;
+            continue;
+        }
+        const iso2 = required(record["iso2"], "iso2", row);
+        const sector = required(record["sector"], "sector", row).replaceAll("_", "-");
+        const key = iso2 + KEY_SEPARATOR + sector;
+        const ordinal = (ordinalByKey.get(key) ?? 0) + 1;
+        ordinalByKey.set(key, ordinal);
+        const statement = rewriteRetiredVocabulary(required(record["license"], "license", row));
+        const statementID = licenseByStatement.get(statement);
+        if (!statementID) {
+            throw new Error(`row ${row}: license ${stringifyJSON(statement)} has no decision record in this build`);
+        }
+        const sourceID = `${iso2.toLowerCase()}-${sector}-${ordinal}`;
+        const publisherName = rewriteRetiredVocabulary((record["owner"] ?? "").trim());
+        const licenseID = scopedLicenseID(statementID, publisherName, sourceID, iso2, statement, scopes);
+        const source = {
+            sourceID,
+            iso2,
+            sector,
+            name: rewriteRetiredVocabulary(required(record["source"], "source", row)),
+            status: mapped(SOURCE_STATUS_BY_NAME, required(record["status"], "status", row), "status", row),
+            asserts: mapped(ASSERTS_BY_ROLE, required(record["role"], "role", row), "role", row),
+            authorityBasis: rewriteRetiredVocabulary(required(record["authority_basis"], "authority_basis", row)),
+            geometry: mapped(GEOMETRY_BY_NAME, (record["geometry"] ?? "").trim(), "geometry", row),
+            license: licenseID,
+            researchPass: mapped(RESEARCH_PASS_BY_ORIGIN, origin, "origin", row),
+        };
+        const access = (record["access"] ?? "").trim();
+        const sourceURL = (record["source_url"] ?? "").trim();
+        const note = (record["notes"] ?? "").trim();
+        if (access) {
+            source.access = rewriteRetiredVocabulary(access);
+        }
+        if (publisherName) {
+            source.publisher = publisherName;
+        }
+        if (sourceURL) {
+            source.sourceURL = sourceURL;
+        }
+        if (note) {
+            source.note = rewriteNote(note, fired);
+        }
+        const role = readUnresolvedColumn(record["address_role"], "address_role", row);
+        if (role) {
+            // The research pass records one role for a whole source and names no column,
+            // so the role is filed under the record's address field.
+            // A source carrying two roles on one record gets a column-by-column reading
+            // in `source-resolutions.json` instead.
+            source.addressRoles = { address: mapped(ADDRESS_ROLE_BY_NAME, role, "address_role", row) };
+        }
+        const coverage = readUnresolvedColumn(record["coverage"], "coverage", row);
+        if (coverage) {
+            source.coverage = coverage;
+        }
+        const upstream = readUnresolvedColumn(record["upstream"], "upstream", row);
+        if (upstream) {
+            source.upstreamLineage = TextSpliterator.from(upstream, { delimiter: ";" }).toArray();
+        }
+        records.push(source);
+    }
+    return { sources: records, discoveryRailRowsDropped };
+}
+/**
+ * Values the research pass wrote into a column it did not resolve per source.
+ *
+ * A copy onto a record would turn "uninspected" into a value a consumer reads as an answer.
+ */
+const UNRESOLVED_COLUMN_PLACEHOLDERS = new Set(["varies", "country-specific", "unknown", "n/a"]);
+/**
+ * A column's value, or `undefined` when the research pass left it unresolved.
+ *
+ * Every value outside the placeholder set is returned, so a value somebody fills in
+ * later reaches the register or fails the build rather than being lost.
+ */
+export function readUnresolvedColumn(value, column, row) {
+    const trimmed = (value ?? "").trim();
+    if (!trimmed)
+        return undefined;
+    if (UNRESOLVED_COLUMN_PLACEHOLDERS.has(trimmed.toLowerCase()))
+        return undefined;
+    if (column === "address_role" && !ADDRESS_ROLE_BY_NAME[trimmed.toLowerCase()]) {
+        throw new Error(`row ${row}: address_role ${stringifyJSON(trimmed)} is neither a declared placeholder nor an \`AddressRole\`. ` +
+            `Add it to \`AddressRole\` if it is a role, or to the placeholder set if it is the research pass declining ` +
+            "to answer.");
+    }
+    return trimmed;
+}
+function isSharedReference(value) {
+    return typeof value.sameAs === "string";
+}
+/**
+ * Source resolutions read from `resolutionsPath`, keyed by source id.
+ *
+ * An absent file answers an empty map.
+ * A file that exists and cannot be parsed raises, because reading it as empty would
+ * drop somebody's recorded review without saying so.
+ */
+async function readSourceResolutions(resolutionsPath) {
+    if (!resolutionsPath || !(await pathExists(resolutionsPath)))
+        return new Map();
+    const file = await readLocalJSONFile(resolutionsPath);
+    return new Map(Object.entries(file.resolutions ?? {}));
+}
+/**
+ * Merges each recorded resolution onto the source whose id it holds.
+ *
+ * @throws When a resolution's source id is absent from the register.
+ * Such an entry is a typo or a source that has been removed.
+ * Keeping it silently would leave a review whose fields reach no row, which reads as work already done.
+ */
+export function applySourceResolutions(sources, recorded) {
+    const known = new Set(sources.map((source) => source.sourceID));
+    for (const sourceID of recorded.keys()) {
+        if (!known.has(sourceID)) {
+            throw new Error(`a recorded source resolution names ${stringifyJSON(sourceID)}, which the register does not carry. ` +
+                "A resolution that resolves nothing is a typo or a source that has been removed.");
+        }
+    }
+    return sources.map((source) => {
+        const resolution = recorded.get(source.sourceID);
+        return resolution ? { ...source, ...resolution } : source;
+    });
+}
+/**
+ * License decisions read from `decisionsPath`, keyed by license id.
+ *
+ * An absent file answers an empty map, but a file that exists and cannot be parsed raises,
+ * because reading it as empty would silently drop somebody's recorded work.
+ * This function validates no decision it reads, because `auditAddressSourceRegister` already does.
+ */
+async function readLicenseDecisions(decisionsPath) {
+    if (!decisionsPath || !(await pathExists(decisionsPath)))
+        return new Map();
+    return resolveRecordedDecisions(await readLocalJSONFile(decisionsPath));
+}
+/**
+ * Turns a decisions file into decisions keyed by license id, resolving every `sameAs` reference.
+ *
+ * The id comes from the key, so a record cannot disagree with the license it is filed under.
+ * `auditAddressSourceRegister` decides whether the resolved fields form a well-formed decision, so this
+ * function checks only that the file declares a reading for each reference.
+ *
+ * @throws When a `sameAs` reads a reading absent from `sharedReadings`.
+ * Such a reference would otherwise produce a decision carrying no terms, which the audit
+ * would report as an unknown review state rather than as the typo it is.
+ */
+export function resolveRecordedDecisions(file) {
+    const shared = file.sharedReadings ?? {};
+    return new Map(Object.entries(file.decisions ?? {}).map(([licenseID, recorded]) => {
+        if (!isSharedReference(recorded))
+            return [licenseID, { ...recorded, licenseID }];
+        const body = shared[recorded.sameAs];
+        if (!body) {
+            throw new Error(`licence ${stringifyJSON(licenseID)} reads \`sameAs\` ${stringifyJSON(recorded.sameAs)}, ` +
+                "which `sharedReadings` does not declare. A reference to a reading that is absent would " +
+                "otherwise produce a decision with no terms.");
+        }
+        return [licenseID, { ...body, licenseID }];
+    }));
+}
+export async function buildSourceRegister(options) {
+    const fired = new Set();
+    const licenseByStatement = new Map(LICENSE_DECISIONS.map(([statement, licenseID]) => [statement, licenseID]));
+    const scopes = new Map();
+    const jurisdictions = await readJurisdictions(options.inventoryPath, fired);
+    const { sources: derived, discoveryRailRowsDropped } = await readSources(options.sourcesPath, licenseByStatement, fired, scopes);
+    const sources = applySourceResolutions(derived, await readSourceResolutions(options.resolutionsPath));
+    for (const [from] of RETIRED_NOTE_REWRITES) {
+        if (!fired.has(from)) {
+            throw new Error(`the rewrite of ${stringifyJSON(from)} never fired — the upstream text has moved on`);
+        }
+    }
+    const used = new Set(sources.map((source) => source.license));
+    const noteByStatement = new Map(LICENSE_DECISIONS.map(([statement, , note]) => [statement, note]));
+    const generated = [...scopes.values()]
+        .filter((scope) => used.has(scope.licenseID))
+        .map((scope) => {
+        const label = noteByStatement.get(scope.statement) ?? "";
+        const scopedNote = scope.publisher
+            ? `${label} Scoped to ${scope.publisher} in ${scope.iso2}, so electing it grants that publisher's sources in that jurisdiction alone.`
+            : `${label} The research pass recorded no publisher, so this decision is scoped to ${scope.scopedTo} alone.`;
+        return {
+            licenseID: scope.licenseID,
+            state: LicenseReviewState.Unchecked,
+            publisherStatement: scope.statement,
+            note: scopedNote,
+        };
+    })
+        .toSorted((left, right) => left.licenseID.localeCompare(right.licenseID));
+    const licenses = applyLicenseDecisions(generated, await readLicenseDecisions(options.decisionsPath));
+    const register = {
+        registerID: "address-source-register",
+        version: options.version,
+        // Filled below, once every other field is in place, because the digest covers the rest
+        // of the register and cannot be computed while the object is still being assembled.
+        contentDigest: "",
+        provenance: {
+            source: "mailwoman-research",
+            sourceVersion: options.sourceVersion,
+            authoredAt: options.authoredAt,
+            notes: "Derived from a research pass over all 249 ISO 3166-1 codes plus the operational code XK. " +
+                "The eight global discovery lookups the pass repeated once per jurisdiction are the package " +
+                "README's procedure, not rows here.",
+        },
+        unresolved: UNRESOLVED_FIELDS.filter((field) => !sources.some((source) => source[field] !== undefined)),
+        licenses,
+        jurisdictions,
+        sources,
+    };
+    register.contentDigest = registerContentDigest(register);
+    const problems = auditAddressSourceRegister(register);
+    if (problems.length) {
+        throw new Error(`the derived register fails its own audit:\n  ${problems.join("\n  ")}`);
+    }
+    await writeLocalFile(prettyJSON(register), options.outPath);
+    return {
+        outPath: options.outPath.toString(),
+        jurisdictions: jurisdictions.length,
+        sources: sources.length,
+        licenses: licenses.length,
+        discoveryRailRowsDropped,
+        researchStates: countBy(jurisdictions, (record) => record.researchState),
+        statuses: countBy(sources, (record) => record.status),
+    };
+}
+function countBy(rows, key) {
+    const counts = {};
+    for (const row of rows) {
+        counts[key(row)] = (counts[key(row)] ?? 0) + 1;
+    }
+    return counts;
+}
+//# sourceMappingURL=build.js.map