From 4d12eab0c3d7d76284accc73a8f6bcb5ab1bf32b Mon Sep 17 00:00:00 2001
From: Emmanuel Raviart <emmanuel@raviart.com>
Date: Fri, 19 May 2023 14:12:43 +0200
Subject: [PATCH] Export function parseDossierLegislatif

---
 src/lib/index.ts           |  1 +
 src/lib/parsers/dole.ts    | 69 ++++++++++++++++++++++++++++++
 src/lib/parsers/index.ts   |  1 +
 src/scripts/import_dole.ts | 88 ++++++++------------------------------
 4 files changed, 90 insertions(+), 69 deletions(-)
 create mode 100644 src/lib/parsers/dole.ts
 create mode 100644 src/lib/parsers/index.ts

diff --git a/src/lib/index.ts b/src/lib/index.ts
index 714004d..376357d 100644
--- a/src/lib/index.ts
+++ b/src/lib/index.ts
@@ -184,3 +184,4 @@ export {
   type TmLienTxt,
   type Versions,
 } from "./legal"
+export { parseDossierLegislatif } from "./parsers"
diff --git a/src/lib/parsers/dole.ts b/src/lib/parsers/dole.ts
new file mode 100644
index 0000000..719aed2
--- /dev/null
+++ b/src/lib/parsers/dole.ts
@@ -0,0 +1,69 @@
+import { auditChain, auditRequire, strictAudit } from "@auditors/core"
+import assert from "assert"
+import { XMLParser } from "fast-xml-parser"
+import he from "he"
+
+import { auditDossierLegislatif } from "$lib/auditors/dole"
+import type { DossierLegislatif, XmlHeader } from "$lib/legal"
+
+const xmlParser = new XMLParser({
+  attributeNamePrefix: "@",
+  ignoreAttributes: false,
+  stopNodes: [
+    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_1",
+    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_2",
+    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_3",
+    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_4",
+    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_5",
+    "DOSSIER_LEGISLATIF.CONTENU.EXPOSE_MOTIF",
+  ],
+  tagValueProcessor: (_tagName, tagValue) => he.decode(tagValue),
+})
+
+export function parseDossierLegislatif(
+  filePath: string,
+  xmlString: string,
+): DossierLegislatif | undefined {
+  try {
+    const xmlData = xmlParser.parse(xmlString)
+    for (const [key, element] of Object.entries(xmlData) as [
+      string,
+      DossierLegislatif | XmlHeader,
+    ][]) {
+      switch (key) {
+        case "?xml": {
+          const xmlHeader = element as XmlHeader
+          assert.strictEqual(xmlHeader["@encoding"], "UTF-8", filePath)
+          assert.strictEqual(xmlHeader["@version"], "1.0", filePath)
+          break
+        }
+        case "DOSSIER_LEGISLATIF": {
+          const [dossierLegislatif, error] = auditChain(
+            auditDossierLegislatif,
+            auditRequire,
+          )(strictAudit, element) as [DossierLegislatif, unknown]
+          assert.strictEqual(
+            error,
+            null,
+            `Unexpected format for DOSSIER_LEGISLATIF:\n${JSON.stringify(
+              dossierLegislatif,
+              null,
+              2,
+            )}\nError:\n${JSON.stringify(error, null, 2)}`,
+          )
+          return dossierLegislatif
+        }
+        default: {
+          console.warn(
+            `Unexpected root element "${key}" in XML file: ${filePath}`,
+          )
+          return undefined
+        }
+      }
+    }
+    return undefined
+  } catch (e) {
+    console.error("An error occurred while parsing XML file", filePath)
+    throw e
+  }
+}
diff --git a/src/lib/parsers/index.ts b/src/lib/parsers/index.ts
new file mode 100644
index 0000000..aea0d1f
--- /dev/null
+++ b/src/lib/parsers/index.ts
@@ -0,0 +1 @@
+export { parseDossierLegislatif } from "./dole"
diff --git a/src/scripts/import_dole.ts b/src/scripts/import_dole.ts
index afc2b0c..1ea07a5 100644
--- a/src/scripts/import_dole.ts
+++ b/src/scripts/import_dole.ts
@@ -1,31 +1,13 @@
-import { auditChain, auditRequire, strictAudit } from "@auditors/core"
 import assert from "assert"
-import { XMLParser } from "fast-xml-parser"
 import fs from "fs-extra"
-import he from "he"
 import path from "path"
 import type { JSONValue } from "postgres"
 import sade from "sade"
 
-import { auditDossierLegislatif } from "$lib/auditors/dole"
-import type { DossierLegislatif, XmlHeader } from "$lib/legal"
+import { parseDossierLegislatif } from "$lib/parsers"
 import { db } from "$lib/server/databases"
 import { walkDir } from "$lib/server/file_systems"
 
-const xmlParser = new XMLParser({
-  attributeNamePrefix: "@",
-  ignoreAttributes: false,
-  stopNodes: [
-    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_1",
-    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_2",
-    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_3",
-    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_4",
-    "DOSSIER_LEGISLATIF.CONTENU.CONTENU_DOSSIER_5",
-    "DOSSIER_LEGISLATIF.CONTENU.EXPOSE_MOTIF",
-  ],
-  tagValueProcessor: (_tagName, tagValue) => he.decode(tagValue),
-})
-
 async function importDole(
   dilaDir: string,
   { resume }: { resume?: string } = {},
@@ -65,57 +47,25 @@ async function importDole(
       const xmlString: string = await fs.readFile(filePath, {
         encoding: "utf8",
       })
-      const xmlData = xmlParser.parse(xmlString)
-      for (const [key, element] of Object.entries(xmlData) as [
-        string,
-        DossierLegislatif | XmlHeader,
-      ][]) {
-        switch (key) {
-          case "?xml": {
-            const xmlHeader = element as XmlHeader
-            assert.strictEqual(xmlHeader["@encoding"], "UTF-8", filePath)
-            assert.strictEqual(xmlHeader["@version"], "1.0", filePath)
-            break
-          }
-          case "DOSSIER_LEGISLATIF": {
-            const [dossierLegislatif, error] = auditChain(
-              auditDossierLegislatif,
-              auditRequire,
-            )(strictAudit, element) as [DossierLegislatif, unknown]
-            assert.strictEqual(
-              error,
-              null,
-              `Unexpected format for DOSSIER_LEGISLATIF:\n${JSON.stringify(
-                dossierLegislatif,
-                null,
-                2,
-              )}\nError:\n${JSON.stringify(error, null, 2)}`,
-            )
-            await db`
-              INSERT INTO dossier_legislatif (
-                id,
-                data
-              ) VALUES (
-                ${dossierLegislatif.META.META_COMMUN.ID},
-                ${db.json(dossierLegislatif as unknown as JSONValue)}
-              )
-              ON CONFLICT (id)
-              DO UPDATE SET
-                data = ${db.json(dossierLegislatif as unknown as JSONValue)}
-            `
-            dossierLegislatifRemainingIds.delete(
-              dossierLegislatif.META.META_COMMUN.ID,
-            )
-            break
-          }
-          default: {
-            console.warn(
-              `Unexpected root element "${key}" in XML file: ${filePath}`,
-            )
-            break iterXmlFiles
-          }
-        }
+      const dossierLegislatif = parseDossierLegislatif(filePath, xmlString)
+      if (dossierLegislatif === undefined) {
+        break iterXmlFiles
       }
+      await db`
+        INSERT INTO dossier_legislatif (
+          id,
+          data
+        ) VALUES (
+          ${dossierLegislatif.META.META_COMMUN.ID},
+          ${db.json(dossierLegislatif as unknown as JSONValue)}
+        )
+        ON CONFLICT (id)
+        DO UPDATE SET
+          data = ${db.json(dossierLegislatif as unknown as JSONValue)}
+      `
+      dossierLegislatifRemainingIds.delete(
+        dossierLegislatif.META.META_COMMUN.ID,
+      )
     } catch (e) {
       console.error("An error occurred while parsing XML file", filePath)
       throw e