@socialgouv/fiches-travail-data-types 4.532.0 → 4.533.0

package/build/fetch-data/parseDom.js

@@ -34,9 +34,7 @@ function getCleanSrc(src) {
 }
 const formatPicture = (node) => {
     let comment;
-    node.parentElement
-        .childNodes
-        .forEach(function (childNode) {
+    node.parentElement.childNodes.forEach(function (childNode) {
         if (childNode.nodeName === "#comment" || childNode.nodeType === 8) {
             if (childNode.data.match(SRC_REGEX)) {
                 comment = childNode;
@@ -52,9 +50,7 @@ const formatPicture = (node) => {
         }
     }
     let image;
-    node
-        .childNodes
-        .forEach(function (childNode) {
+    node.childNodes.forEach(function (childNode) {
         if (childNode.nodeName === "IMG") {
             image = childNode;
         }
@@ -150,20 +146,29 @@ function parseDom(dom, id, url) {
         }
     }
     const title = titleElement.textContent.trim();
-    const dateRaw = $(dom.window.document, "meta[property*=modified_time]") || $(dom.window.document, "meta[property$=published_time]");
+    const dateRaw = $(dom.window.document, "meta[property*=modified_time]") ||
+        $(dom.window.document, "meta[property$=published_time]");
     const [year, month, day] = dateRaw.getAttribute("content").split("-");
     let intro = $(article, ".main-article__chapo") || "";
-    intro = intro && intro.innerHTML.replace(/\n/g, "").replace(/\s+/g, " ").trim();
-    const description = $(dom.window.document, "meta[name=description]")?.getAttribute("content") ?? "";
+    intro =
+        intro && intro.innerHTML.replace(/\n/g, "").replace(/\s+/g, " ").trim();
+    // clean script tags and everything inside it
+    intro = intro.replace(/<script[^>]*>([\s\S]*?)<\/script>/g, "");
+    const description = $(dom.window.document, "meta[name=description]")?.getAttribute("content") ??
+        "";
     const sections = [];
     const sectionTag = getSectionTag(article);
     // First pass is only to get a potential untitled section at the top of the article
     // This section has neither anchor nor title
     let nextArticleElement = $(article, ".main-article__texte > *");
     const untitledSection = {
-        anchor: "", html: "", text: "", title: title,
+        anchor: "",
+        html: "",
+        text: "",
+        title: title,
     };
-    while (nextArticleElement && nextArticleElement.tagName.toLowerCase() !== sectionTag) {
+    while (nextArticleElement &&
+        nextArticleElement.tagName.toLowerCase() !== sectionTag) {
         if (nextArticleElement.textContent) {
             if (!untitledSection.description) {
                 untitledSection.description = "temp description";
@@ -172,7 +177,8 @@ function parseDom(dom, id, url) {
                 .replace(/\n+/g, "")
                 .replace(/>\s+</g, "><")
                 .replace(/\s+/g, " ");
-            untitledSection.text += " " + nextArticleElement.textContent.replace(/\s+/g, " ").trim();
+            untitledSection.text +=
+                " " + nextArticleElement.textContent.replace(/\s+/g, " ").trim();
         }
         nextArticleElement = nextArticleElement.nextElementSibling;
     }
@@ -198,7 +204,10 @@ function parseDom(dom, id, url) {
             sections.push({
                 anchor: el.getAttribute("id") || (0, cdtn_slugify_1.default)(el.textContent),
                 description: sectionText.slice(0, 200).trim(),
-                html: html.replace(/\n+/g, "").replace(/>\s+</g, "><").replace(/\s+/g, " "),
+                html: html
+                    .replace(/\n+/g, "")
+                    .replace(/>\s+</g, "><")
+                    .replace(/\s+/g, " "),
                 references: getReferences(sectionText),
                 text: sectionText,
                 title: el.textContent.trim(),
@@ -209,7 +218,13 @@ function parseDom(dom, id, url) {
         throw new got_1.ParseError(`No sections`);
     }
     return {
-        date: `${day}/${month}/${year}`, description, intro, pubId: id, sections, title, url,
+        date: `${day}/${month}/${year}`,
+        description,
+        intro,
+        pubId: id,
+        sections,
+        title,
+        url,
     };
 }
 exports.parseDom = parseDom;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@socialgouv/fiches-travail-data-types",
-  "version": "4.532.0",
+  "version": "4.533.0",
   "main": "build/index.js",
   "module": "build/index.js",
   "files": [