@budiauktioner/buildi-primitives 0.1.0

package/dist/text-scanning/index.d.cts

@@ -0,0 +1,153 @@
+import { T as TextCandidate, a as TextCandidateCategory, b as TextMatchConfidence } from '../text-candidate-qg2bD9PY.cjs';
+import { I as Iban, B as Bic, d as SwedishBankgiroNumber, f as SwedishPostgiroNumber, b as SwedishBankAccount, e as SwedishOcrReferenceNumber, g as SwedishSwishNumber } from '../swedish-swish-number-CL6Gx_h3.cjs';
+import { c as AddressZipCode, A as Address, S as SwedishAddress } from '../swedish-address-DNC3ftRF.cjs';
+import { a as PhoneNumber } from '../phone-number-DZ-YxCQQ.cjs';
+import { I as Isin } from '../isin-ZlQLBVm8.cjs';
+import { a as Country } from '../country-BhD51LDL.cjs';
+import { S as SwedishCounty, a as SwedishMunicipality } from '../swedish-municipality-DlUS7g7h.cjs';
+import { S as SwedishOrganizationNumber, E as EuVatNumber, L as LeiCode, D as DunsNumber } from '../swedish-organization-number-Bcj7V_lh.cjs';
+import { b as SwedishPersonalIdentityNumber, S as SwedishCoordinationNumber } from '../swedish-personal-identity-number-Cp9Coz55.cjs';
+import { S as SwedishPropertyDesignation } from '../swedish-property-designation-BaHXufqV.cjs';
+import { EmailAddress, Url } from '../web/index.cjs';
+import '../validation/index.cjs';
+import '../swedish-organization-type-CZOdETwG.cjs';
+
+/** Registers every built-in `findCandidatesInText` implementation on the shared scanner. */
+declare function registerDefaultTextScanners(): void;
+
+/**
+ * The result of scanning unstructured text for structured value types.
+ * Contains all found candidates, a resolved non-overlapping subset, and
+ * methods for bulk replacement / masking.
+ *
+ * Text scanning is heuristic-based. Candidates may be false positives, and
+ * valid occurrences may be missed.
+ */
+declare class TextScanResult {
+    /** All candidates sorted by `startIndex`, including overlapping ones. */
+    readonly all: readonly TextCandidate[];
+    /**
+     * Non-overlapping subset after resolution. When candidates overlap or one
+     * contains another, the winner is chosen by: containment (longer enclosing
+     * match wins), then higher {@link TextMatchConfidence}, then longer span,
+     * then category priority.
+     */
+    readonly resolvedCandidates: readonly TextCandidate[];
+    constructor(candidates: readonly TextCandidate[]);
+    get totalCount(): number;
+    /** Returns the number of candidates in the given category. */
+    countByCategory(category: TextCandidateCategory): number;
+    /** Returns the number of candidates at or above the given confidence level. */
+    countByConfidence(minimum: TextMatchConfidence): number;
+    /** Returns all candidates whose `typeName` equals `typeName`. */
+    byTypeName<T = unknown>(typeName: string): TextCandidate<T>[];
+    get emails(): TextCandidate<EmailAddress>[];
+    get urls(): TextCandidate<Url>[];
+    get phoneNumbers(): TextCandidate<PhoneNumber>[];
+    get zipCodes(): TextCandidate<AddressZipCode>[];
+    get addresses(): TextCandidate<Address>[];
+    get swedishAddresses(): TextCandidate<SwedishAddress>[];
+    get personalIdentityNumbers(): TextCandidate<SwedishPersonalIdentityNumber>[];
+    get coordinationNumbers(): TextCandidate<SwedishCoordinationNumber>[];
+    get organizationNumbers(): TextCandidate<SwedishOrganizationNumber>[];
+    get euVatNumbers(): TextCandidate<EuVatNumber>[];
+    get leiCodes(): TextCandidate<LeiCode>[];
+    get dunsNumbers(): TextCandidate<DunsNumber>[];
+    get ibans(): TextCandidate<Iban>[];
+    get bics(): TextCandidate<Bic>[];
+    get bankgiroNumbers(): TextCandidate<SwedishBankgiroNumber>[];
+    get postgiroNumbers(): TextCandidate<SwedishPostgiroNumber>[];
+    get bankAccounts(): TextCandidate<SwedishBankAccount>[];
+    get ocrReferences(): TextCandidate<SwedishOcrReferenceNumber>[];
+    get swedishSwishNumbers(): TextCandidate<SwedishSwishNumber>[];
+    get isins(): TextCandidate<Isin>[];
+    get countries(): TextCandidate<Country>[];
+    get counties(): TextCandidate<SwedishCounty>[];
+    get municipalities(): TextCandidate<SwedishMunicipality>[];
+    get propertyDesignations(): TextCandidate<SwedishPropertyDesignation>[];
+    /** Returns all candidates in the given category. */
+    byCategory(category: TextCandidateCategory): TextCandidate[];
+    /**
+     * Replaces each resolved candidate's span in `text` with its `maskedForm`.
+     */
+    maskAll(text: string): string;
+    /**
+     * Replaces each resolved candidate's span in `text` with `replacement`.
+     */
+    redactAll(text: string, replacement?: string): string;
+    /**
+     * Replaces each resolved candidate's span in `text` using a custom mapper.
+     * Replacements are applied right-to-left on the original text so positions
+     * remain valid.
+     */
+    replaceAll(text: string, replacer: (c: TextCandidate) => string): string;
+}
+
+/**
+ * Configuration for the text scanner controlling which categories to scan and
+ * the minimum confidence threshold.
+ */
+interface TextScannerOptions {
+    /**
+     * When set, only categories in this set are scanned. Mutually exclusive
+     * with {@link excludeCategories}.
+     */
+    includeCategories?: ReadonlySet<TextCandidateCategory>;
+    /**
+     * When set, categories in this set are skipped. Ignored if
+     * {@link includeCategories} is set.
+     */
+    excludeCategories?: ReadonlySet<TextCandidateCategory>;
+    /**
+     * Candidates below this confidence level are discarded from results.
+     * Defaults to {@link TextMatchConfidence.Low} (keep everything).
+     */
+    minimumConfidence?: TextMatchConfidence;
+}
+
+/**
+ * A scan function returns all candidates found in `text` for one type.
+ */
+type TextScanFunction<T = unknown> = (text: string) => readonly TextCandidate<T>[];
+/**
+ * A registration that wires one type's `findCandidatesInText` function into a
+ * scanner.
+ */
+interface RegisteredScanner<T = unknown> {
+    readonly typeName: string;
+    readonly category: TextCandidateCategory;
+    readonly scan: TextScanFunction<T>;
+}
+/**
+ * Aggregate scanner that runs every registered type's `findCandidatesInText`
+ * and merges the results.
+ *
+ * In contrast to the .NET `TextScanner` (which hard-codes a fixed list of
+ * types in its constructor), the JS port uses a registry. Type modules call
+ * {@link TextScanner.register} (or push to {@link defaultTextScanner}) so the
+ * scanner is extensible without editing this file.
+ */
+declare class TextScanner {
+    private readonly registrations;
+    /**
+     * Registers a type's scan function. Called once per type, normally at
+     * module load time.
+     */
+    register<T>(reg: RegisteredScanner<T>): void;
+    /** Currently registered scanners. Mainly useful for tests. */
+    get registered(): readonly RegisteredScanner[];
+    /**
+     * Scans `text` for all registered types matching the given `options`.
+     * Returns an aggregate result with overlap resolution.
+     */
+    scan(text: string, options?: TextScannerOptions): TextScanResult;
+}
+/**
+ * The default scanner used by `@budiauktioner/buildi-primitives`. Type modules with
+ * `findCandidatesInText` register themselves with this instance at module
+ * load time. Applications that need an isolated scanner can construct their
+ * own {@link TextScanner}.
+ */
+declare const defaultTextScanner: TextScanner;
+
+export { type RegisteredScanner, TextCandidate, TextCandidateCategory, TextMatchConfidence, type TextScanFunction, TextScanResult, TextScanner, type TextScannerOptions, defaultTextScanner, registerDefaultTextScanners };

package/dist/text-scanning/index.d.ts

@@ -0,0 +1,153 @@
+import { T as TextCandidate, a as TextCandidateCategory, b as TextMatchConfidence } from '../text-candidate-qg2bD9PY.js';
+import { I as Iban, B as Bic, d as SwedishBankgiroNumber, f as SwedishPostgiroNumber, b as SwedishBankAccount, e as SwedishOcrReferenceNumber, g as SwedishSwishNumber } from '../swedish-swish-number-DDG3mNDg.js';
+import { c as AddressZipCode, A as Address, S as SwedishAddress } from '../swedish-address-GMzjqgxb.js';
+import { a as PhoneNumber } from '../phone-number-C24--ggy.js';
+import { I as Isin } from '../isin-EeVPpNWu.js';
+import { a as Country } from '../country-p0n0EVQ9.js';
+import { S as SwedishCounty, a as SwedishMunicipality } from '../swedish-municipality-nDRHYRAd.js';
+import { S as SwedishOrganizationNumber, E as EuVatNumber, L as LeiCode, D as DunsNumber } from '../swedish-organization-number-DM8dpXwR.js';
+import { b as SwedishPersonalIdentityNumber, S as SwedishCoordinationNumber } from '../swedish-personal-identity-number-BiIg4io3.js';
+import { S as SwedishPropertyDesignation } from '../swedish-property-designation-BM-gdd9x.js';
+import { EmailAddress, Url } from '../web/index.js';
+import '../validation/index.js';
+import '../swedish-organization-type-CZOdETwG.js';
+
+/** Registers every built-in `findCandidatesInText` implementation on the shared scanner. */
+declare function registerDefaultTextScanners(): void;
+
+/**
+ * The result of scanning unstructured text for structured value types.
+ * Contains all found candidates, a resolved non-overlapping subset, and
+ * methods for bulk replacement / masking.
+ *
+ * Text scanning is heuristic-based. Candidates may be false positives, and
+ * valid occurrences may be missed.
+ */
+declare class TextScanResult {
+    /** All candidates sorted by `startIndex`, including overlapping ones. */
+    readonly all: readonly TextCandidate[];
+    /**
+     * Non-overlapping subset after resolution. When candidates overlap or one
+     * contains another, the winner is chosen by: containment (longer enclosing
+     * match wins), then higher {@link TextMatchConfidence}, then longer span,
+     * then category priority.
+     */
+    readonly resolvedCandidates: readonly TextCandidate[];
+    constructor(candidates: readonly TextCandidate[]);
+    get totalCount(): number;
+    /** Returns the number of candidates in the given category. */
+    countByCategory(category: TextCandidateCategory): number;
+    /** Returns the number of candidates at or above the given confidence level. */
+    countByConfidence(minimum: TextMatchConfidence): number;
+    /** Returns all candidates whose `typeName` equals `typeName`. */
+    byTypeName<T = unknown>(typeName: string): TextCandidate<T>[];
+    get emails(): TextCandidate<EmailAddress>[];
+    get urls(): TextCandidate<Url>[];
+    get phoneNumbers(): TextCandidate<PhoneNumber>[];
+    get zipCodes(): TextCandidate<AddressZipCode>[];
+    get addresses(): TextCandidate<Address>[];
+    get swedishAddresses(): TextCandidate<SwedishAddress>[];
+    get personalIdentityNumbers(): TextCandidate<SwedishPersonalIdentityNumber>[];
+    get coordinationNumbers(): TextCandidate<SwedishCoordinationNumber>[];
+    get organizationNumbers(): TextCandidate<SwedishOrganizationNumber>[];
+    get euVatNumbers(): TextCandidate<EuVatNumber>[];
+    get leiCodes(): TextCandidate<LeiCode>[];
+    get dunsNumbers(): TextCandidate<DunsNumber>[];
+    get ibans(): TextCandidate<Iban>[];
+    get bics(): TextCandidate<Bic>[];
+    get bankgiroNumbers(): TextCandidate<SwedishBankgiroNumber>[];
+    get postgiroNumbers(): TextCandidate<SwedishPostgiroNumber>[];
+    get bankAccounts(): TextCandidate<SwedishBankAccount>[];
+    get ocrReferences(): TextCandidate<SwedishOcrReferenceNumber>[];
+    get swedishSwishNumbers(): TextCandidate<SwedishSwishNumber>[];
+    get isins(): TextCandidate<Isin>[];
+    get countries(): TextCandidate<Country>[];
+    get counties(): TextCandidate<SwedishCounty>[];
+    get municipalities(): TextCandidate<SwedishMunicipality>[];
+    get propertyDesignations(): TextCandidate<SwedishPropertyDesignation>[];
+    /** Returns all candidates in the given category. */
+    byCategory(category: TextCandidateCategory): TextCandidate[];
+    /**
+     * Replaces each resolved candidate's span in `text` with its `maskedForm`.
+     */
+    maskAll(text: string): string;
+    /**
+     * Replaces each resolved candidate's span in `text` with `replacement`.
+     */
+    redactAll(text: string, replacement?: string): string;
+    /**
+     * Replaces each resolved candidate's span in `text` using a custom mapper.
+     * Replacements are applied right-to-left on the original text so positions
+     * remain valid.
+     */
+    replaceAll(text: string, replacer: (c: TextCandidate) => string): string;
+}
+
+/**
+ * Configuration for the text scanner controlling which categories to scan and
+ * the minimum confidence threshold.
+ */
+interface TextScannerOptions {
+    /**
+     * When set, only categories in this set are scanned. Mutually exclusive
+     * with {@link excludeCategories}.
+     */
+    includeCategories?: ReadonlySet<TextCandidateCategory>;
+    /**
+     * When set, categories in this set are skipped. Ignored if
+     * {@link includeCategories} is set.
+     */
+    excludeCategories?: ReadonlySet<TextCandidateCategory>;
+    /**
+     * Candidates below this confidence level are discarded from results.
+     * Defaults to {@link TextMatchConfidence.Low} (keep everything).
+     */
+    minimumConfidence?: TextMatchConfidence;
+}
+
+/**
+ * A scan function returns all candidates found in `text` for one type.
+ */
+type TextScanFunction<T = unknown> = (text: string) => readonly TextCandidate<T>[];
+/**
+ * A registration that wires one type's `findCandidatesInText` function into a
+ * scanner.
+ */
+interface RegisteredScanner<T = unknown> {
+    readonly typeName: string;
+    readonly category: TextCandidateCategory;
+    readonly scan: TextScanFunction<T>;
+}
+/**
+ * Aggregate scanner that runs every registered type's `findCandidatesInText`
+ * and merges the results.
+ *
+ * In contrast to the .NET `TextScanner` (which hard-codes a fixed list of
+ * types in its constructor), the JS port uses a registry. Type modules call
+ * {@link TextScanner.register} (or push to {@link defaultTextScanner}) so the
+ * scanner is extensible without editing this file.
+ */
+declare class TextScanner {
+    private readonly registrations;
+    /**
+     * Registers a type's scan function. Called once per type, normally at
+     * module load time.
+     */
+    register<T>(reg: RegisteredScanner<T>): void;
+    /** Currently registered scanners. Mainly useful for tests. */
+    get registered(): readonly RegisteredScanner[];
+    /**
+     * Scans `text` for all registered types matching the given `options`.
+     * Returns an aggregate result with overlap resolution.
+     */
+    scan(text: string, options?: TextScannerOptions): TextScanResult;
+}
+/**
+ * The default scanner used by `@budiauktioner/buildi-primitives`. Type modules with
+ * `findCandidatesInText` register themselves with this instance at module
+ * load time. Applications that need an isolated scanner can construct their
+ * own {@link TextScanner}.
+ */
+declare const defaultTextScanner: TextScanner;
+
+export { type RegisteredScanner, TextCandidate, TextCandidateCategory, TextMatchConfidence, type TextScanFunction, TextScanResult, TextScanner, type TextScannerOptions, defaultTextScanner, registerDefaultTextScanners };