From 09c584774b242374625d481ae043e8162d94ab52 Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@robocracy.org>
Date: Tue, 23 Apr 2019 18:38:32 -0700
Subject: add limits to match importers

---
 python/fatcat_tools/importers/arabesque.py | 12 +++++++++++-
 python/fatcat_tools/importers/common.py    |  3 +++
 python/fatcat_tools/importers/matched.py   | 14 +++++++++++++-
 3 files changed, 27 insertions(+), 2 deletions(-)

diff --git a/python/fatcat_tools/importers/arabesque.py b/python/fatcat_tools/importers/arabesque.py
index 4353795a..7838a7ff 100644
--- a/python/fatcat_tools/importers/arabesque.py
+++ b/python/fatcat_tools/importers/arabesque.py
@@ -5,7 +5,7 @@ import base64
 import sqlite3
 import itertools
 import fatcat_client
-from .common import EntityImporter, clean, make_rel_url
+from .common import EntityImporter, clean, make_rel_url, SANE_MAX_RELEASES, SANE_MAX_URLS
 
 
 def b32_hex(s):
@@ -130,6 +130,10 @@ class ArabesqueMatchImporter(EntityImporter):
 
         urls = [fatcat_client.FileEntityUrls(rel=rel, url=url) for (rel, url) in urls]
 
+        if len(urls) > SANE_MAX_URLS:
+            self.counts['skip-too-many-url'] += 1
+            return None
+
         fe = fatcat_client.FileEntity(
             sha1=b32_hex(row['final_sha1']),
             mimetype=row['final_mimetype'],
@@ -174,7 +178,13 @@ class ArabesqueMatchImporter(EntityImporter):
         # merge the existing into this one and update
         existing.urls = list(set([(u.rel, u.url) for u in fe.urls + existing.urls]))
         existing.urls = [fatcat_client.FileEntityUrls(rel=rel, url=url) for (rel, url) in existing.urls]
+        if len(existing.urls) > SANE_MAX_URLS:
+            self.counts['skip-update-too-many-url'] += 1
+            return None
         existing.release_ids = list(set(fe.release_ids + existing.release_ids))
+        if len(existing.release_ids) > SANE_MAX_RELEASES:
+            self.counts['skip-update-too-many-url'] += 1
+            return None
         existing.mimetype = existing.mimetype or fe.mimetype
         edit = self.api.update_file(existing.ident, existing, editgroup_id=self.get_editgroup_id())
         self._edits_inflight.append(edit)
diff --git a/python/fatcat_tools/importers/common.py b/python/fatcat_tools/importers/common.py
index beec99df..d291aeec 100644
--- a/python/fatcat_tools/importers/common.py
+++ b/python/fatcat_tools/importers/common.py
@@ -14,6 +14,9 @@ import fatcat_client
 from fatcat_client.rest import ApiException
 
 
+SANE_MAX_RELEASES = 200
+SANE_MAX_URLS = 100
+
 def clean(thing, force_xml=False):
     """
     This function is appropriate to be called on any random, non-markup string,
diff --git a/python/fatcat_tools/importers/matched.py b/python/fatcat_tools/importers/matched.py
index 5096cc97..0afd00e3 100644
--- a/python/fatcat_tools/importers/matched.py
+++ b/python/fatcat_tools/importers/matched.py
@@ -4,7 +4,7 @@ import json
 import sqlite3
 import itertools
 import fatcat_client
-from .common import EntityImporter, clean, make_rel_url
+from .common import EntityImporter, clean, make_rel_url, SANE_MAX_RELEASES, SANE_MAX_URLS
 
 
 class MatchedImporter(EntityImporter):
@@ -68,6 +68,9 @@ class MatchedImporter(EntityImporter):
         if len(release_ids) == 0:
             self.counts['skip-no-doi'] += 1
             return None
+        if len(release_ids) > SANE_MAX_RELEASES:
+            self.counts['skip-too-many-dois'] += 1
+            return None
 
         # parse URLs and CDX
         urls = set()
@@ -89,6 +92,9 @@ class MatchedImporter(EntityImporter):
         if len(urls) == 0:
             self.counts['skip-no-urls'] += 1
             return None
+        if len(urls) > SANE_MAX_URLS:
+            self.counts['skip-too-many-urls'] += 1
+            return None
 
         size = obj.get('size')
         if size:
@@ -126,7 +132,13 @@ class MatchedImporter(EntityImporter):
         # merge the existing into this one and update
         existing.urls = list(set([(u.rel, u.url) for u in fe.urls + existing.urls]))
         existing.urls = [fatcat_client.FileEntityUrls(rel=rel, url=url) for (rel, url) in existing.urls]
+        if len(existing.urls) > SANE_MAX_URLS:
+            self.counts['skip-update-too-many-url'] += 1
+            return None
         existing.release_ids = list(set(fe.release_ids + existing.release_ids))
+        if len(existing.release_ids) > SANE_MAX_RELEASES:
+            self.counts['skip-update-too-many-url'] += 1
+            return None
         existing.mimetype = existing.mimetype or fe.mimetype
         existing.size = existing.size or fe.size
         existing.md5 = existing.md5 or fe.md5
-- 
cgit v1.2.3