From f0b113cf9739519bacd42aaa5fa3cc8aee0cd030 Mon Sep 17 00:00:00 2001
From: Cameron Pfiffer <cameron@pfiffer.org>
Date: Mon, 9 Mar 2026 22:29:20 -0700
Subject: [PATCH] feat: add fetch-history command to pull historical corpora
 for stylometric analysis

---
 src/void_tools/cli.py | 65 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 65 insertions(+)

diff --git a/src/void_tools/cli.py b/src/void_tools/cli.py
index cc83674..a4c9d40 100644
--- a/src/void_tools/cli.py
+++ b/src/void_tools/cli.py
@@ -769,6 +769,62 @@ def cmd_ingest_thread(args):
     except Exception as e:
         print(f"Error ingesting thread: {e}", file=sys.stderr)
 
+def cmd_fetch_history(args):
+    client = get_client()
+    try:
+        profile = client.app.bsky.actor.get_profile({'actor': args.handle})
+        did = profile.did
+        
+        posts = []
+        cursor = None
+        
+        import datetime
+        from dateutil import parser as date_parser
+        
+        target_date = date_parser.isoparse(args.before)
+        if target_date.tzinfo is None:
+            target_date = target_date.replace(tzinfo=datetime.UTC)
+
+        print(f"Fetching posts for {args.handle} before {target_date}...")
+        
+        while len(posts) < args.limit:
+            kwargs = {'actor': did, 'limit': 100}
+            if cursor:
+                kwargs['cursor'] = cursor
+                
+            response = client.app.bsky.feed.get_author_feed(kwargs)
+            
+            if not response.feed:
+                break
+                
+            for item in response.feed:
+                # We only want posts by the author, not reposts
+                if hasattr(item.post.record, 'text') and item.post.author.did == did:
+                    post_time = date_parser.isoparse(item.post.indexed_at)
+                    if post_time < target_date:
+                        posts.append({
+                            'uri': item.post.uri,
+                            'text': item.post.record.text,
+                            'indexed_at': item.post.indexed_at
+                        })
+                        
+                if len(posts) >= args.limit:
+                    break
+                    
+            if not response.cursor:
+                break
+            cursor = response.cursor
+            
+        import json
+        out_path = os.path.expanduser(os.path.join("~/void", args.out))
+        with open(out_path, 'w') as f:
+            json.dump(posts, f, indent=2)
+            
+        print(f"Successfully fetched {len(posts)} posts and saved to {out_path}")
+        
+    except Exception as e:
+        print(f"Error fetching history: {e}", file=sys.stderr)
+
 def main():
     parser = argparse.ArgumentParser(description="Void ATProto CLI Tools")
     subparsers = parser.add_subparsers(dest="command", required=True)
@@ -795,6 +851,13 @@ def main():
     ingest_parser = subparsers.add_parser("ingest-thread", help="Download an entire post thread and save it to MemFS")
     ingest_parser.add_argument("uri", help="The AT-URI of the root or target post")
     ingest_parser.add_argument("--depth", type=int, default=10, help="Maximum depth of replies to fetch")
+
+    # Fetch History
+    history_parser = subparsers.add_parser("fetch-history", help="Fetch a user's historical posts before a specific date")
+    history_parser.add_argument("handle", help="The user handle to fetch (e.g., void.comind.network)")
+    history_parser.add_argument("--before", help="ISO format date (e.g., 2026-02-20T00:00:00Z)", required=True)
+    history_parser.add_argument("--limit", type=int, default=1000, help="Maximum number of posts to fetch")
+    history_parser.add_argument("--out", help="Output JSON file path", default="historical_corpus.json")
     
     # Profile
     profile_parser = subparsers.add_parser("profile", help="Manage profile")
@@ -831,6 +894,8 @@ def main():
         cmd_search(args)
     elif args.command == "ingest-thread":
         cmd_ingest_thread(args)
+    elif args.command == "fetch-history":
+        cmd_fetch_history(args)
     elif args.command == "profile":
         cmd_profile(args)
     elif args.command == "annotate":
-- 
2.51.2