From a1afd0211f85467527c1342061b62c5d5c0d4952 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <github@sweeting.me>
Date: Fri, 17 May 2024 20:11:00 -0700
Subject: [PATCH] fix abid calculation

---
 archivebox/abid_utils/abid.py   |  9 ++++++---
 archivebox/abid_utils/models.py | 10 +++++-----
 archivebox/api/models.py        |  2 +-
 archivebox/config.py            |  2 +-
 archivebox/core/settings.py     |  1 +
 5 files changed, 14 insertions(+), 10 deletions(-)

diff --git a/archivebox/abid_utils/abid.py b/archivebox/abid_utils/abid.py
index 1d3d28f9..48597813 100644
--- a/archivebox/abid_utils/abid.py
+++ b/archivebox/abid_utils/abid.py
@@ -108,9 +108,12 @@ def uri_hash(uri: Union[str, bytes]) -> str:
 
     # only hash the domain part of URLs
     if '://' in uri_str:
-        domain = urlparse(uri_str).host
-        if domain:
-            url_str = domain
+        try:
+            domain = urlparse(uri_str).netloc
+            if domain:
+                uri_str = domain
+        except AttributeError:
+            pass
     
     uri_bytes = uri_str.encode('utf-8')
 
diff --git a/archivebox/abid_utils/models.py b/archivebox/abid_utils/models.py
index 06fb6c79..de8b3c87 100644
--- a/archivebox/abid_utils/models.py
+++ b/archivebox/abid_utils/models.py
@@ -82,11 +82,11 @@ class ABIDModel(models.Model):
 
     def save(self, *args: Any, **kwargs: Any) -> None:
         if hasattr(self, 'abid'):
-            # self.abid = ABID.parse(self.abid) if self.abid else self.calculate_abid()
-            self.abid = self.calculate_abid()
+            # self.abid = ABID.parse(self.abid) if self.abid else self.get_abid()
+            self.abid = self.get_abid()
         else:
             print(f'[!] WARNING: {self.__class__.__name__}.abid is not a DB field so ABID will not be persisted!')
-            self.abid = self.calculate_abid()
+            self.abid = self.get_abid()
         
         super().save(*args, **kwargs)
 
@@ -141,7 +141,7 @@ class ABIDModel(models.Model):
         """
         ULIDParts(timestamp='01HX9FPYTR', url='E4A5CCD9', subtype='00', randomness='ZYEBQE')
         """
-        return ABID.parse(self.abid) if getattr(self, 'abid', None) else self.calculate_abid()
+        return ABID.parse(self.abid) if getattr(self, 'abid', None) else self.get_abid()
 
     @property
     def ULID(self) -> ULID:
@@ -274,7 +274,7 @@ def find_obj_from_abid_rand(rand: Union[ABID, str], model=None) -> List[ABIDMode
                 )
 
             for obj in qs:
-                if obj.calculate_abid() == abid:
+                if obj.get_abid() == abid:
                     # found exact match, no need to keep iterating
                     return [obj]
                 partial_matches.append(obj)
diff --git a/archivebox/api/models.py b/archivebox/api/models.py
index 8d286a8b..177b275f 100644
--- a/archivebox/api/models.py
+++ b/archivebox/api/models.py
@@ -56,7 +56,7 @@ class APIToken(ABIDModel):
         return {
             "TYPE":             "APIToken",    
             "uuid":             str(self.id),
-            "abid":             str(self.calculate_abid()),
+            "abid":             str(self.get_abid()),
             "user_id":          str(self.user.id),
             "user_username":    self.user.username,
             "token":            self.token,
diff --git a/archivebox/config.py b/archivebox/config.py
index da2d1b04..1637023b 100644
--- a/archivebox/config.py
+++ b/archivebox/config.py
@@ -37,7 +37,7 @@ from sqlite3 import dbapi2 as sqlite3
 from hashlib import md5
 from pathlib import Path
 from datetime import datetime, timezone
-from typing import Optional, Type, Tuple, Dict, Union, List
+from typing import Optional, Type, Tuple, Dict, Union, List, Any
 from subprocess import run, PIPE, DEVNULL
 from configparser import ConfigParser
 from collections import defaultdict
diff --git a/archivebox/core/settings.py b/archivebox/core/settings.py
index d540d2be..ab197eea 100644
--- a/archivebox/core/settings.py
+++ b/archivebox/core/settings.py
@@ -10,6 +10,7 @@ from pathlib import Path
 from django.utils.crypto import get_random_string
 
 from ..config import (
+    CONFIG,
     DEBUG,
     SECRET_KEY,
     ALLOWED_HOSTS,