From 252c9c3ce77e0997e6cf6be04cead45d849f46f4 Mon Sep 17 00:00:00 2001
From: K Sadov <ksadov@gmail.com>
Date: Sun, 17 Mar 2024 19:29:48 -0700
Subject: [PATCH] add wildcard search based on
 https://github.com/jsvine/waybackpack/pull/17

---
 waybackpack/asset.py | 10 +++++-----
 waybackpack/cli.py   |  2 +-
 waybackpack/pack.py  | 29 ++++++++++++++++-------------
 3 files changed, 22 insertions(+), 19 deletions(-)

diff --git a/waybackpack/asset.py b/waybackpack/asset.py
index b014c29..95a798b 100644
--- a/waybackpack/asset.py
+++ b/waybackpack/asset.py
@@ -32,12 +32,12 @@
 
 
 class Asset(object):
-    def __init__(self, original_url, timestamp):
+    def __init__(self, snapshot):
         # Ensure timestamp is only numeric
-        if re.match(r"^[0-9]+\Z", timestamp) is None:
-            raise RuntimeError("invalid timestamp {!r}".format(timestamp))
-        self.timestamp = timestamp
-        self.original_url = original_url
+        self.timestamp = snapshot['timestamp']
+        self.original_url = snapshot['original']
+        if re.match(r"^[0-9]+\Z", self.timestamp) is None:
+            raise RuntimeError("invalid timestamp {!r}".format(self.timestamp))
 
     def get_archive_url(self, raw=False):
         flag = "id_" if raw else ""
diff --git a/waybackpack/cli.py b/waybackpack/cli.py
index 5405ede..ce41f87 100644
--- a/waybackpack/cli.py
+++ b/waybackpack/cli.py
@@ -173,7 +173,7 @@ def main():
 
         timestamps = [snap["timestamp"] for snap in snapshots]
 
-        pack = Pack(search_url, timestamps=timestamps, session=session)
+        pack = Pack(search_url, snapshots=snapshots)
 
         time.sleep(args.delay)
 
diff --git a/waybackpack/pack.py b/waybackpack/pack.py
index b7ca6a4..0d3841e 100644
--- a/waybackpack/pack.py
+++ b/waybackpack/pack.py
@@ -2,6 +2,7 @@
 import os
 import platform
 import time
+import urllib
 
 from .asset import Asset
 from .cdx import search
@@ -42,7 +43,7 @@ def replace_invalid_chars(path, fallback_char="_"):
 
 
 class Pack(object):
-    def __init__(self, url, timestamps=None, uniques_only=False, session=None):
+    def __init__(self, url, snapshots=None, uniques_only=False, session=None):
 
         self.url = url
         prefix = "http://" if urlparse(url).scheme == "" else ""
@@ -51,15 +52,12 @@ def __init__(self, url, timestamps=None, uniques_only=False, session=None):
 
         self.session = session or Session()
 
-        if timestamps is None:
-            self.timestamps = [
-                snap["timestamp"]
-                for snap in search(url, uniques_only=uniques_only, session=self.session)
-            ]
-        else:
-            self.timestamps = timestamps
-
-        self.assets = [Asset(self.url, ts) for ts in self.timestamps]
+        self.snapshots = snapshots or search(
+            url,
+            uniques_only=uniques_only,
+            session=self.session
+        )
+        self.assets = [Asset(snapshot) for snapshot in self.snapshots]
 
     def download_to(
         self,
@@ -83,9 +81,14 @@ def download_to(
                 logger.info("Sleeping {0} seconds".format(delay))
                 time.sleep(delay)
 
-            path_head, path_tail = os.path.split(self.parsed_url.path)
-            if path_tail == "":
-                path_tail = "index.html"
+            path = urllib.parse.urlparse(asset.original_url).path[1:]
+
+            if path:
+                path_head, path_tail = path.rsplit('/', 1)
+                if not path_tail:
+                    path_tail = 'index.html'
+            else:
+                path_head, path_tail = '', 'index.html'
 
             filedir = os.path.join(
                 directory,