geminispace.info

commit c172c20e952abe9f34c2c78446b7bcdb81dbbd71
parent d012217757ba8dfb3f2e6dff0dbdc79aa867ca4e
Author: Natalie Pendragon <natpen@natpen.net>
Date:   Sun, 19 Jul 2020 09:23:46 -0400

[crawl] Update db model, and delete links before recreating

Diffstat:
M
gus/crawl.py
 | 
1
+
M
gus/lib/db_model.py
 | 
6
+++---

2 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/gus/crawl.py b/gus/crawl.py
@@ -232,6 +232,7 @@ def index_content(resource, response):
 
 def index_links(from_resource, contained_resources):
     from_page, created = Page.get_or_create(url=from_resource.indexable_url)
+    Link.delete().where(Link.from_page == from_page).execute()
     data = []
     for cr in contained_resources:
         should_skip = False
diff --git a/gus/lib/db_model.py b/gus/lib/db_model.py
@@ -44,15 +44,15 @@ class Link(Model):
     Hyperlinks between pages in Geminispace
     """
 
-    from_page = ForeignKeyField(Page, backref="outbound_links")
-    to_page = ForeignKeyField(Page, backref="backlinks")
+    from_page = ForeignKeyField(Page, backref="outbound_links", on_delete='CASCADE')
+    to_page = ForeignKeyField(Page, backref="backlinks", on_delete='CASCADE')
 
 class Crawl(Model):
     """
     Attempts to crawl a page.
     """
 
-    page = ForeignKeyField(Page, backref="crawls")
+    page = ForeignKeyField(Page, backref="crawls", on_delete='CASCADE')
     status = IntegerField()
     is_different = BooleanField()
     timestamp = DateTimeField()