From 13341496429b3c7fc74674a78192e99ac1f3ca3f Mon Sep 17 00:00:00 2001
From: isaac hershenson <ihershenson@hmc.edu>
Date: Wed, 28 Aug 2024 22:14:16 -0700
Subject: [PATCH] wip

---
 scripts/check-link.js | 94 +++++++++++++++----------------------------
 1 file changed, 32 insertions(+), 62 deletions(-)

diff --git a/scripts/check-link.js b/scripts/check-link.js
index f3312da1..b525704e 100644
--- a/scripts/check-link.js
+++ b/scripts/check-link.js
@@ -1,75 +1,45 @@
-const { exec } = require('child_process');
 const fs = require('fs');
 const path = require('path');
-const markdownLinkCheck = require('markdown-link-check');
-
-const ignoredUrls = [
-  'https://(api|web)\\.smith\\.langchain\\.com/.*',
-  'https://x\\.com/.*'
-];
-
-function convertNotebookToMarkdown(filePath) {
-  return new Promise((resolve, reject) => {
-    const outputPath = filePath.replace('.ipynb', '.md');
-    exec(`jupyter nbconvert --to markdown "${filePath}" --output "${outputPath}"`, (error) => {
-      if (error) {
-        reject(error);
-      } else {
-        resolve(outputPath);
-      }
-    });
-  });
+const { execSync } = require('child_process');
+
+// Function to find all .ipynb files in the given directory
+function findIpynbFiles(dir) {
+  let results = [];
+  const files = fs.readdirSync(dir);
+  for (const file of files) {
+    const filePath = path.join(dir, file);
+    const stat = fs.statSync(filePath);
+    if (stat.isDirectory()) {
+      results = results.concat(findIpynbFiles(filePath));
+    } else if (path.extname(file) === '.ipynb') {
+      results.push(filePath);
+    }
+  }
+  return results;
 }
 
-function checkLinks(filePath) {
-  return new Promise((resolve, reject) => {
-    const markdown = fs.readFileSync(filePath, 'utf8');
-    markdownLinkCheck(markdown, {
-      ignorePatterns: ignoredUrls,
-      baseUrl: 'https://github.com', // Adjust this if needed
-      httpHeaders: [
-        {
-          urls: ['https://github.com'],
-          headers: {
-            'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/93.0.4577.63 Safari/537.36'
-          }
-        }
-      ]
-    }, (err, results) => {
-      if (err) {
-        reject(err);
-      } else {
-        const brokenLinks = results.filter(result => result.status === 'dead');
-        resolve(brokenLinks);
-      }
-    });
-  });
-}
+// Main function to check links
+function checkLinks() {
+  const ignorePatterns = [
+    'https://(api|web)\\.smith\\.langchain\\.com/.*',
+    'https://x\\.com/.*'
+  ];
 
-async function main() {
-  const changedFiles = process.env.CHANGED_FILES.split(' ');
-  const notebookFiles = changedFiles.filter(file => path.extname(file) === '.ipynb');
+  const ipynbFiles = findIpynbFiles('.');
+  console.log('Found .ipynb files:', ipynbFiles);
 
-  for (const file of notebookFiles) {
+  for (const file of ipynbFiles) {
+    console.log(`Checking links in ${file}`);
     try {
-      const markdownFile = await convertNotebookToMarkdown(file);
-      const brokenLinks = await checkLinks(markdownFile);
-
-      if (brokenLinks.length > 0) {
-        console.error(`Broken links found in ${file}:`);
-        brokenLinks.forEach(link => console.error(`- ${link.link}: ${link.statusCode}`));
-        process.exit(1);
+      execSync(`yarn run linkinator ${file} ${ignorePatterns.map(pattern => `--skip "${pattern}"`).join(' ')}`, { stdio: 'inherit' });
+    } catch (error) {
+      if (error.status === 5) {
+        console.log('Broken links found, but continuing...');
       } else {
-        console.log(`No broken links found in ${file}`);
+        throw error;
       }
-
-      // Clean up the temporary markdown file
-      fs.unlinkSync(markdownFile);
-    } catch (error) {
-      console.error(`Error processing ${file}:`, error);
-      process.exit(1);
     }
   }
 }
 
-main();
\ No newline at end of file
+checkLinks();
\ No newline at end of file