From 59a0dc67ba0cf649eb6f047a6da82a1237a9ff03 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Christian=20R=C3=A9millard?= <christian.remillard@gmail.com>
Date: Wed, 20 Dec 2023 14:20:19 -0500
Subject: [PATCH] refactor(parser): La fonction linkParser est maintenant dans
 /config

---
 config/linkParser.js | 21 +++++++++++++++++++++
 harvest.js           | 25 +++----------------------
 2 files changed, 24 insertions(+), 22 deletions(-)
 create mode 100644 config/linkParser.js

diff --git a/config/linkParser.js b/config/linkParser.js
new file mode 100644
index 0000000..0788622
--- /dev/null
+++ b/config/linkParser.js
@@ -0,0 +1,21 @@
+export function linkParser() {
+  return Array
+      .from(/** @type {NodeListOf<HTMLAnchorElement>} */(document.querySelectorAll('a[href]')))
+      // Exclude those inside a rss module
+      .filter(link => !(link.parentNode && link.parentNode instanceof HTMLAnchorElement && link.closest('.s-lg-rss')))
+      .map(link => ({
+          url: link.href,
+          text: link.tagName === 'IMG' ? link.getAttribute('alt') : link.innerText,
+          urlData: link.getAttribute('href'),
+          isNavigationRequest: true
+      }))
+      .concat(Array
+          .from(/** @type {NodeListOf<HTMLImageElement>} */(document.querySelectorAll('img[src]')))
+          .map(img => ({
+              url: img.src,
+              text: img.alt || img.title || null,
+              urlData: img.getAttribute('src'),
+              isNavigationRequest: false
+          }))
+      )
+}
\ No newline at end of file
diff --git a/harvest.js b/harvest.js
index 4a24f60..6e74856 100644
--- a/harvest.js
+++ b/harvest.js
@@ -1,14 +1,15 @@
 import readline from 'readline'
 import yargs from 'yargs'
 import { hideBin } from 'yargs/helpers'
+import { console, inspect } from 'corvee-core'
 import { Harvester } from 'corvee-harvester'
 import { fetchGuides } from './lib/fetch-guides.js'
 import { saveBrowsingContexts, saveReportCodes, saveRecords, saveSystemInfo } from './utils/index.js'
 import pageSnippetPlugin from './plugins/page-snippet.js'
-import { console, inspect } from 'corvee-core'
 
 import { harvesterConfig } from './config/index.js'
 import { savePageSnippets } from './utils/save-page-snippets.js'
+import { linkParser } from 'config/linkParser.js'
 
 const today = new Date();
 const year = today.getFullYear();
@@ -75,27 +76,7 @@ async function harvest() {
 
     harvester.addPlugins(pageSnippetPlugin)
 
-    harvester.setLinkParser(function linkParser() {
-        return Array
-            .from(/** @type {NodeListOf<HTMLAnchorElement>} */(document.querySelectorAll('a[href]')))
-            // Exclude those inside a rss module
-            .filter(link => !(link.parentNode && link.parentNode instanceof HTMLAnchorElement && link.closest('.s-lg-rss')))
-            .map(link => ({
-                url: link.href,
-                text: link.tagName === 'IMG' ? link.getAttribute('alt') : link.innerText,
-                urlData: link.getAttribute('href'),
-                isNavigationRequest: true
-            }))
-            .concat(Array
-                .from(/** @type {NodeListOf<HTMLImageElement>} */(document.querySelectorAll('img[src]')))
-                .map(img => ({
-                    url: img.src,
-                    text: img.alt || img.title || null,
-                    urlData: img.getAttribute('src'),
-                    isNavigationRequest: false
-                }))
-            )
-    })
+    harvester.setLinkParser(linkParser)
 
     await harvester.addUrl(links);