@tricoteuses/tisseuse 0.22.2 → 0.22.3

package/dist/lib/legi/europe_observation.d.ts

@@ -0,0 +1,23 @@
+import { CheerioAPI } from 'cheerio';
+/** Whether an article number can be verified at all, before reading anything. */
+export declare function isVerifiableEuropeanArticle(article: string): boolean;
+/** A loaded observation, so that many anchors are read in a single parse. */
+export type EuropeanObservationDocument = CheerioAPI;
+/**
+ * The canonical ELI fragment of an article inside one observation of a
+ * European act, or `undefined` when the observation does not prove it.
+ *
+ * A fragment is only admitted when the anchor exists **exactly once**, belongs
+ * to the enacting terms of the act, is the container Cellar gives an article,
+ * and is titled with the cited number. Everything else abstains: a fragment
+ * that cannot be verified in the observation is a misleading link.
+ *
+ * The check reads the stored observation, not the page Exploratrice renders
+ * from it. Both forms are measured by the tests, on the RGPD observation: the
+ * Cellar XHTML (`fixtures/rgpd-cellar.html`) and the captured public page
+ * (`fixtures/rgpd-exploratrice.html.gz`) carry the same anchors, under the
+ * same container and with the same heading.
+ */
+export declare function observationArticleFragment($: EuropeanObservationDocument, article: string): string | undefined;
+/** {@link observationArticleFragment} on the HTML of an observation. */
+export declare function europeanObservationArticleFragment(html: string, article: string): string | undefined;

package/dist/lib/legi/europe_observation.test.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/lib/legi/resolver.d.ts

@@ -1,4 +1,67 @@
 import { PublicationComponentRow } from './publication_components.js';
+/**
+ * Identity criteria of a cited European act, as written in the source. Only
+ * the criteria actually cited are provided: a missing date never becomes the
+ * resolution date of the document.
+ */
+export interface EuropeanActReference {
+    /** Number as cited, for example « 2016/679 » or « 95/46/CE ». */
+    numero: string;
+    /** Cited nature: « règlement », « directive » or « décision ». */
+    nature?: string;
+    /** Date of the act, explicitly cited in the source (AAAA-MM-JJ). */
+    date?: string;
+    /**
+     * Treaty regime written in parentheses, as written (« UE », « CE »,
+     * « UE, Euratom »). A resolver must confront it with the act it designates,
+     * or abstain: it is part of the designation the source claims.
+     */
+    regime?: string;
+    /**
+     * Institution the source names right after the act (« du Parlement européen
+     * et du Conseil »). Confronted with the act's own designation when the
+     * resolver can read it, otherwise abstained on.
+     */
+    institution?: string;
+}
+/**
+ * Verdict of the identity confrontation between a citation's written criteria
+ * and the designation the resolved act carries (`europe.document.title`).
+ */
+export type EuropeanActIdentity = {
+    /**
+     * CELEX of the single consultable act the citation designates, or `[]` when
+     * the citation is refused. Ambiguity is expressed the same way: nothing is
+     * established.
+     */
+    celexIds: string[];
+    /**
+     * Regimes written by the citation that the act's designation carries under
+     * another nomenclature of the same treaty (`CE`/`CEE`/`UE`): a link kept by
+     * equivalence, not by equality, and therefore announced rather than silent.
+     * Empty when every written criterion is matched literally, or none is
+     * written.
+     */
+    regimesDecales: string[];
+};
+/**
+ * A verified position inside one observation of a European act. The fragment is
+ * the canonical ELI subdivision (for example `art_5`) and carries no version:
+ * the observation identifies the version the fragment was checked in, and the
+ * empreinte attests the bytes it was checked in.
+ */
+export interface EuropeanObservationFragment {
+    /** Observation of the act the fragment was verified in. */
+    observation: string;
+    /** Canonical ELI fragment, without `#`. */
+    fragment: string;
+    /**
+     * SHA-256 of the HTML read for that observation. The page rendering the
+     * observation recomputes it and drops the fragment on a difference: a
+     * fragment is never honoured on an observation that is not the attested one.
+     */
+    empreinte: string;
+}
 export interface LegiPublicationComponentsProof {
     candidates: PublicationComponentRow[];
     bodyCid: string;
@@ -93,7 +156,39 @@ export declare abstract class LegiResolver {
         parentPath: string;
     }): Promise<LegiSectionInfo[]>;
     abstract getTextIdFromSegmentId(segmentId: string): Promise<string | undefined>;
+    /**
+     * Outcome of the identity confrontation for a cited European act: which act
+     * the citation designates, and which of its written criteria the act's own
+     * designation carries under another nomenclature of the same treaty.
+     *
+     * The verdict is `undefined` when the citation writes a criterion this
+     * grammar cannot read: the resolver abstains rather than ignore it.
+     */
+    getEuropeanActIdentity(reference: EuropeanActReference): Promise<EuropeanActIdentity | undefined>;
+    /**
+     * Compatibility projection over `getEuropeanActIdentity`: the consultable
+     * CELEX candidates of a cited act, `[]` when the citation is refused and
+     * `undefined` when the resolver abstains.
+     */
+    getEuropeanActCandidates(reference: EuropeanActReference): Promise<string[] | undefined>;
+    /**
+     * Compatibility wrapper over `getEuropeanActCandidates`: the consultation
+     * URL of the single consultable act cited by its number alone, or
+     * `undefined` when that number is unknown, ambiguous or has no document.
+     */
     getEuropeanTextUrl(titleId: string): Promise<string | undefined>;
+    /**
+     * Fragment of a cited article inside one observation of the act, or
+     * `undefined` when no observation proves it. A resolver must verify the
+     * anchor in the observation it names, or abstain: an unproven fragment is a
+     * misleading link.
+     */
+    getEuropeanArticleFragment(reference: {
+        /** CELEX of the established act. */
+        celexId: string;
+        /** Article number as cited, a plain integer. */
+        article: string;
+    }): Promise<EuropeanObservationFragment | undefined>;
     abstract getUniqueActiveArticlesByNum(options: {
         date: string;
         num: string | undefined;

package/dist/lib/linkers/bascule_render_decisions.test.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/lib/linkers/html_european.test.d.ts

@@ -0,0 +1 @@
+export {};

package/dist/lib/linkers/link_decision_metadata.d.ts

@@ -9,6 +9,11 @@ export declare const linkDecisionReferenceClasses: {
         readonly linked: "lien_texte_externe";
         readonly undecidable: "reference_texte_indecidable";
     };
+    /** European act: the act is linked, never one of its articles. */
+    readonly european_text: {
+        readonly linked: "lien_texte_europeen";
+        readonly undecidable: "reference_texte_europeen_indecidable";
+    };
 };
 export type LinkDecisionReferenceKind = keyof typeof linkDecisionReferenceClasses;
 export declare function isLinkDecisionWarning(code: string): boolean;

package/dist/lib/links.d.ts

@@ -14,4 +14,33 @@ export declare function citedArticleNumFromTextCitation(textCitation: string): s
  * `undefined` when this link kind has no equivalent for the given type.
  */
 export declare function urlFromArticleVersionsId(linkType: LinkType, baseUrl: string, articleId: string): string | undefined;
+/**
+ * Canonical ELI subdivision fragment: `art_5`, `cpt_III`, `anx_1`. The URL is
+ * composed from data, so the fragment is checked before being appended: a
+ * fragment that is not in this shape would designate nothing (or, once escaped
+ * by the client, something else).
+ */
+export declare const europeanFragmentRegExp: RegExp;
+/**
+ * Page of a European act in Exploratrice, keyed by its CELEX identifier. The
+ * route is the same for every link kind: no Git rendering of the act exists.
+ *
+ * A verified fragment carries three inseparable parts:
+ *
+ * - `?version=` pins the observation it was verified in;
+ * - `empreinte=` gives the SHA-256 of the observation's HTML, so the page can
+ *   recompute it and drop the fragment on any difference;
+ * - `ancre=` names the subdivision, so the page can neutralise the browser jump
+ *   when the fragment is not attested (the `#` itself never reaches the server).
+ *
+ * Without a verified empreinte the fragment is dropped, not escaped: the link
+ * degrades to the act, never to another position. A fragment on an observation
+ * whose bytes are not attested would be exactly the misleading link this chain
+ * exists to prevent.
+ */
+export declare function europeanDocumentUrl(celex: string, fragment?: {
+    empreinte: string;
+    observation: string;
+    fragment: string;
+}): string;
 export declare function urlFromLegalId(linkType: LinkType, baseUrl: string, id: string): string;

package/dist/lib/text_parsers/ast.d.ts

@@ -182,6 +182,12 @@ export type TextAstText = {
     nature: LawNature;
     legislation?: "international" | "UE";
     ofTheSaid?: boolean;
+    /**
+     * Treaty regime written in parentheses after the nature, as written
+     * (« UE », « CE », « UE, Euratom »). Kept unjudged: whether it proves which
+     * act is cited is decided where the act is resolved.
+     */
+    regimeEuropeen?: string;
     /**
      * For the texts found in the Légifrance datasets, the title
      * is the Légifrance one ⇒ it may differ from the title given in

package/dist/link-decision-metadata.js

@@ -7,6 +7,10 @@ var e = ["data-tisseuse-warning", "data-tisseuse-candidates"], t = {
 	text: {
 		linked: "lien_texte_externe",
 		undecidable: "reference_texte_indecidable"
+	},
+	european_text: {
+		linked: "lien_texte_europeen",
+		undecidable: "reference_texte_europeen_indecidable"
 	}
 }, n = new Set([
 	"article_futur_a_date_reference",
@@ -17,10 +21,12 @@ var e = ["data-tisseuse-warning", "data-tisseuse-candidates"], t = {
 	"reference_historique_explicite",
 	"plusieurs_candidats_indepartageables",
 	"cid_non_etabli",
-	"donnees_insuffisantes"
+	"donnees_insuffisantes",
+	"nomenclature_decalee"
 ]), r = {
 	article: /^(?:LEGIARTI|JORFARTI)\d{12}$/,
-	text: /^(?:LEGITEXT|JORFTEXT)\d{12}$/
+	text: /^(?:LEGITEXT|JORFTEXT)\d{12}$/,
+	european_text: /^[23]\d{4}[RLD]\d{4}$/
 };
 function i(e) {
 	return n.has(e);