@tricoteuses/senat 2.8.6 → 2.9.1

package/lib/model/texte.ts

@@ -1,174 +0,0 @@
-import { JSDOM } from "jsdom"
-
-import { AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP, AKN_IDENTIFICATION_STRUCTURE_REGEXP } from "../scripts/datautil"
-import {
-  Alinea,
-  Article,
-  Division,
-  DivisionContent,
-  DivisionTag,
-  DivisionType,
-  FlatTexte, Step,
-  Version,
-} from "../types/texte"
-
-function buildWorklow (metaElement: HTMLMetaElement): Step[] {
-  const stepElements = metaElement.querySelectorAll("workflow step")
-  const steps: Step[] = []
-  for (const stepElement of stepElements) {
-    const identification = stepElement.getAttribute("href") ?? ""
-    const identificationParts = AKN_WORKFLOW_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups
-    steps.push({
-      eId: stepElement.getAttribute("eId")!,
-      date: stepElement.getAttribute("date") ? new Date(stepElement.getAttribute("date") ?? "") : null,
-      type: identificationParts?.type || null,
-      session: identificationParts?.session || null,
-      numero: identificationParts?.numTexte || null,
-      version: identificationParts?.version ? identificationParts.version as Version : null,
-      outcome: stepElement.getAttribute("outcome"),
-    })
-  }
-  return steps
-}
-
-function buildDivision (node: Node, index: number): Division {
-  const eId = (node as Element).getAttribute("eId")!
-  const tag = node.nodeName as DivisionTag
-  const level = DivisionType[tag]
-  const titleNode = (node as Element).querySelector("num")
-  const subtitleNode = (node as Element).querySelector("heading")
-  const headings: DivisionContent[] = [
-    ...(titleNode ? [{
-      text: titleNode.textContent?.trim() ?? null,
-      html: titleNode.innerHTML?.trim() ?? null,
-    }] : []),
-    ...(subtitleNode ? [{
-      text: subtitleNode.textContent?.trim() ?? null,
-      html: subtitleNode.innerHTML?.trim() ?? null,
-    }] : []),
-  ]
-  const division = {
-    index,
-    eId,
-    tag,
-    level,
-    headings,
-  }
-  if (tag === "article") {
-    (division as Article).alineas = []
-  }
-  return division
-}
-
-function buildAlinea (contentNode: Node, alineaNode: Node): Alinea {
-  const eId = (alineaNode as Element).getAttribute("eId")!
-  const heading = {
-    text: (alineaNode as Element).querySelector("num")?.textContent ?? null,
-  }
-  const pastille = (alineaNode as Element).getAttribute("data:pastille") ?? null
-  return {
-    eId,
-    heading,
-    text: (contentNode as Element).textContent?.trim() ?? null,
-    html: (contentNode as Element).innerHTML?.trim() ?? null,
-    pastille,
-  }
-}
-
-function buildEmptyArticle (index: number): Article {
-  return {
-    index: index,
-    eId: "",
-    tag: "article",
-    level: DivisionType["article"],
-    headings: [],
-    alineas: [],
-  }
-}
-
-function flattenTexte (texteContentRoot: Node): Division[] {
-  const divisions: Division[] = []
-  let divisionIndex = 0
-  const iter = (node: Node) => {
-    if (node.nodeName === "content") {
-      return
-    }
-    switch (node.nodeName) {
-      case "tome":
-      case "part":
-      case "book":
-      case "title":
-      case "subtitle":
-      case "chapter":
-      case "section":
-      case "subsection":
-      case "paragraph":
-      case "article":
-        divisions.push(buildDivision(node, divisionIndex++))
-        break
-    }
-    if (node.nodeName === "alinea") {
-      Array.from(node.childNodes)
-        // Find direct content children programmatically
-        // because `:scope` selector does not work
-        // https://github.com/jsdom/jsdom/issues/2998
-        .filter((alineaChildNode: Node) => alineaChildNode.nodeName === "content")
-        .forEach((alineaContentNode: Node) => {
-          // Hypothesis: alineas should always be enclosed in articles
-          let lastArticle = divisions.findLast(division => division.tag === "article") as Article
-          if (!lastArticle) {
-            lastArticle = buildEmptyArticle(divisionIndex++)
-            divisions.push(lastArticle)
-          }
-          lastArticle.alineas.push(buildAlinea(alineaContentNode, node))
-        })
-    }
-    if (node.hasChildNodes()) {
-      node.childNodes.forEach((childNode: Node) => iter(childNode))
-    }
-  }
-  iter(texteContentRoot)
-  return divisions
-}
-
-export function transformTexte (document: Document): FlatTexte | null {
-  let transformedTexte = null
-  const metaElement = document.querySelector("meta")
-  const identification = metaElement?.querySelector("FRBRExpression FRBRuri")?.getAttribute("value") ?? ""
-  const identificationParts = AKN_IDENTIFICATION_STRUCTURE_REGEXP.exec(identification)?.groups
-  const bodyElement = document.querySelector("body")
-  try {
-    transformedTexte = {
-      title: metaElement?.querySelector("FRBRalias[name='intitule-court']")
-        ?.getAttribute("value") || null,
-      signetDossierLegislatifSenat:
-        metaElement?.querySelector("FRBRalias[name='signet-dossier-legislatif-senat']")
-          ?.getAttribute("value") || null,
-      urlSenat: metaElement?.querySelector("FRBRalias[name='url-senat']")?.getAttribute("value") || null,
-      urlAssemblee: metaElement?.querySelector("FRBRalias[name='url-AN']")?.getAttribute("value") || null,
-      type: identificationParts?.type || null,
-      session: identificationParts?.session || null,
-      numero: identificationParts?.numTexte ? parseInt(identificationParts.numTexte) : null,
-      version: identificationParts?.version ? identificationParts.version as Version : null,
-      workflow: metaElement ? buildWorklow(metaElement) : [],
-      divisions: bodyElement ? flattenTexte(bodyElement) : [],
-    }
-  } catch (error: any) {
-    console.error(`Could not parse texte with error ${error}`)
-  }
-  return transformedTexte
-}
-
-export function parseTexte (texteXml: string): FlatTexte | null {
-  const { document } = (new JSDOM(texteXml, {
-    contentType: "text/xml",
-  })).window
-  return transformTexte(document)
-}
-
-// Prevent from memory leak
-// https://github.com/jsdom/jsdom/issues/2583#issuecomment-559520814
-export async function parseTexteFromFile (xmlFilePath: string): Promise<FlatTexte | null> {
-  const { document } = (await JSDOM.fromFile(xmlFilePath, { contentType: "text/xml" })).window
-  return transformTexte(document)
-}