From d4e487e181e090991782384eee544d1dbd37c5fa Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Tue, 23 Jan 2024 21:50:10 +0800
Subject: [PATCH 1/2] set the state to failed if content is not fetched

---
 packages/api/src/jobs/save_page.ts | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/packages/api/src/jobs/save_page.ts b/packages/api/src/jobs/save_page.ts
index 07cbde2f3..191597873 100644
--- a/packages/api/src/jobs/save_page.ts
+++ b/packages/api/src/jobs/save_page.ts
@@ -279,7 +279,6 @@ export const savePageJob = async (data: Data, attemptsMade: number) => {
   const {
     userId,
     articleSavingRequestId,
-    state,
     labels,
     source,
     folder,
@@ -289,14 +288,17 @@ export const savePageJob = async (data: Data, attemptsMade: number) => {
     taskId,
     url,
   } = data
-  let isImported, isSaved
+  let isImported,
+    isSaved,
+    state = data.state
 
   try {
     console.log(`savePageJob: ${userId} ${url}`)
 
     // get the fetch result from cache
-    const { title, content, contentType, readabilityResult } =
-      await getCachedFetchResult(url)
+    const fetchedResult = await getCachedFetchResult(url)
+    const { title, contentType, readabilityResult } = fetchedResult
+    let content = fetchedResult.content
 
     // for pdf content, we need to upload the pdf
     if (contentType === 'application/pdf') {
@@ -329,9 +331,10 @@ export const savePageJob = async (data: Data, attemptsMade: number) => {
     }
 
     if (!content) {
-      throw new Error(
-        'Invalid SavePage job, fetch result missing required data'
-      )
+      console.log('content is not fetched', url)
+      // set the state to failed if we don't have content
+      content = 'Failed to fetch content'
+      state = ArticleSavingRequestStatus.Failed
     }
 
     const user = await userRepository.findById(userId)

From c1d5ec722f68b0acfcd236934859841625c55e7b Mon Sep 17 00:00:00 2001
From: Hongbo Wu <hongbo@omnivore.app>
Date: Tue, 23 Jan 2024 22:06:27 +0800
Subject: [PATCH 2/2] do not encode url when getting file data

---
 packages/api/src/jobs/save_page.ts | 8 ++------
 1 file changed, 2 insertions(+), 6 deletions(-)

diff --git a/packages/api/src/jobs/save_page.ts b/packages/api/src/jobs/save_page.ts
index 191597873..14500d959 100644
--- a/packages/api/src/jobs/save_page.ts
+++ b/packages/api/src/jobs/save_page.ts
@@ -114,7 +114,7 @@ const getUploadIdAndSignedUrl = async (
     }`,
     variables: {
       input: {
-        url,
+        url: encodeURI(url),
         contentType: 'application/pdf',
         clientRequestId: articleSavingRequestId,
       },
@@ -304,11 +304,7 @@ export const savePageJob = async (data: Data, attemptsMade: number) => {
     if (contentType === 'application/pdf') {
       const encodedUrl = encodeURI(url)
 
-      const uploadFileId = await uploadPdf(
-        encodedUrl,
-        userId,
-        articleSavingRequestId
-      )
+      const uploadFileId = await uploadPdf(url, userId, articleSavingRequestId)
       const uploadedPdf = await sendCreateArticleMutation(userId, {
         url: encodedUrl,
         articleSavingRequestId,