From fc64ff6ca5ab5a58df0d8e199757a27e72a248d1 Mon Sep 17 00:00:00 2001
From: seeleng <seelengxd@gmail.com>
Date: Mon, 23 Sep 2024 03:25:30 +0800
Subject: [PATCH] feat: add image to process scripts

---
 backend/src/scrapers/cna/process.py      | 2 ++
 backend/src/scrapers/guardian/process.py | 2 ++
 2 files changed, 4 insertions(+)

diff --git a/backend/src/scrapers/cna/process.py b/backend/src/scrapers/cna/process.py
index 029d6c4f..67e4547f 100644
--- a/backend/src/scrapers/cna/process.py
+++ b/backend/src/scrapers/cna/process.py
@@ -43,6 +43,7 @@ class CNAArticle(BaseModel):
     field_summary: str | None = None
     description: str | None = None
     release_date: str
+    image_url: str
 
 
 async def process(category: str):
@@ -78,6 +79,7 @@ async def process(category: str):
                     source=ArticleSource.CNA,
                     body=body.strip(),
                     date=article.release_date,
+                    image_url=article.image_url,
                 )
                 with Session(engine) as session:
                     session.add(article_orm)
diff --git a/backend/src/scrapers/guardian/process.py b/backend/src/scrapers/guardian/process.py
index 9ef73856..e6be8a71 100644
--- a/backend/src/scrapers/guardian/process.py
+++ b/backend/src/scrapers/guardian/process.py
@@ -16,6 +16,7 @@ class GuardianArticleFields(BaseModel):
     model_config = ConfigDict(extra="allow")
     bodyText: str
     trailText: str | None = None
+    thumbnail: str | None = None
 
 
 class GuardianArticle(BaseModel):
@@ -37,6 +38,7 @@ class GuardianArticle(BaseModel):
             source=ArticleSource.GUARDIAN,
             body=article.fields.bodyText,
             date=article.webPublicationDate,
+            image_url=article.fields.thumbnail or "",
         )
         with Session(engine) as session:
             session.add(article_orm)