conversation-migration

examples/oci_openai/migration/README.md

@@ -0,0 +1,75 @@
+
+---
+
+## OCI GenAI Conversation Migration – Quick Instructions
+
+### 1 Prepare Conversation IDs
+
+You can provide conversation IDs in **two ways**:
+
+1. **Comma-separated** via CLI:
+
+```text
+conv_123,conv_456,conv_789
+```
+
+2. **File input** – one conversation ID per line, e.g., `convs.txt`:
+
+```text
+conv_123
+conv_456
+conv_789
+```
+
+---
+
+### 2 Run the Migration
+
+**Basic Example:**
+
+```bash
+python migrate_conversations.py \
+  --region us-chicago-1 \
+  --profile BoatOc1 \
+  --compartment-id ocid1.compartment.oc1..xxx \
+  --project-id ocid1.generativeaiproject.oc1..xxx \
+  --conversation-store-id ocid1.conversationstore.oc1..xx \
+  --conversations conv_123,conv_456
+```
+
+**Or using a file:**
+
+```bash
+python migrate_conversations.py \
+  --region us-chicago-1 \
+  --profile BoatOc1 \
+  --compartment-id ocid1.compartment.oc1..xxx \
+  --project-id ocid1.generativeaiproject.oc1..xxx \
+  --conversation-store-id ocid1.conversationstore.oc1..xx \
+  --conv-file convs.txt
+```
+
+---
+
+### 3 Dry Run (Optional)
+
+To check inputs without actually migrating:
+
+```bash
+python migrate_conversations.py ... --dry-run
+```
+
+---
+
+### 4 Output
+
+After running, you will see:
+
+```
+conv_123 → new_conv_id_abc
+conv_456 → new_conv_id_def
+```
+
+✅ Successfully migrated conversations or `None` if migration failed.
+
+---

examples/oci_openai/migration/migrate_conversations_cli.py

@@ -0,0 +1,169 @@
+#!/usr/bin/env python3
+
+import time
+import argparse
+from oci_openai import OciOpenAI, OciSessionAuth
+
+# --------------------------------------------------
+# Fetch ALL messages from one conversation (with pagination)
+# --------------------------------------------------
+def get_all_items(client, conv_id):
+    all_items = []
+    start = None
+
+    while True:
+        params = {"limit": 100}
+        if start:
+            params["start"] = start
+
+        resp = client.conversations.items.list(conv_id, **params)
+
+        items = getattr(resp, "data", [])
+        has_more = getattr(resp, "has_more", False)
+        last_id = getattr(resp, "last_id", None)
+
+        all_items.extend(items)
+
+        if not has_more:
+            break
+        if not last_id:
+            print("⚠️ Warning: has_more=True but no last_id. Stop paging.")
+            break
+
+        start = last_id
+        time.sleep(0.1)
+
+    return all_items
+
+
+# --------------------------------------------------
+# Main Migration Function
+# --------------------------------------------------
+def migrate_conversations(conv_ids, project_id, store_id, region, profile, compartment_id, chunk_size=20, dry_run=False):
+    # --------------------------------------------------
+    # Source Client (Conversation Store)
+    # --------------------------------------------------
+    client_src = OciOpenAI(
+        service_endpoint=f"https://inference.generativeai.{region}.oci.oraclecloud.com",
+        auth=OciSessionAuth(profile_name=profile),
+        compartment_id=compartment_id,
+        default_headers={
+            "opc-conversation-store-id": store_id,
+            "opc-compartment-id": compartment_id
+        },
+    )
+
+    # --------------------------------------------------
+    # Destination Client (Project)
+    # --------------------------------------------------
+    client_dst = OciOpenAI(
+        service_endpoint=f"https://inference.generativeai.{region}.oci.oraclecloud.com",
+        auth=OciSessionAuth(profile_name=profile),
+        compartment_id=compartment_id,
+        default_headers={
+            "OpenAI-Project": project_id
+        },
+    )
+
+    results = {}
+
+    for conv_id in conv_ids:
+        print(f"🔄 Migrating: {conv_id}")
+
+        try:
+            # 1. Fetch source messages
+            source_items = get_all_items(client_src, conv_id)
+            print(f"   → Fetched {len(source_items)} items")
+
+            if not source_items:
+                print("   ⚠️ No messages, skipping\n")
+                results[conv_id] = None
+                continue
+
+            if dry_run:
+                print("   → Dry run mode: skipping creation/replay")
+                results[conv_id] = "dry_run"
+                continue
+
+            # 2. Create new conversation
+            new_conv = client_dst.conversations.create(
+                metadata={"original_conversation_id": conv_id},
+                items=[]
+            )
+            new_id = new_conv.id
+            print(f"   → Created new conversation: {new_id}")
+
+            # 3. Append messages in batches
+            for i in range(0, len(source_items), chunk_size):
+                chunk = source_items[i:i + chunk_size]
+                client_dst.conversations.items.create(new_id, items=chunk)
+                time.sleep(0.05)
+
+            print(f"   ✅ Success: {conv_id} → {new_id}\n")
+            results[conv_id] = new_id
+
+        except Exception as e:
+            print(f"   ❌ Failed: {e}\n")
+            results[conv_id] = None
+
+    return results
+
+
+# --------------------------------------------------
+# CLI
+# --------------------------------------------------
+def parse_args():
+    parser = argparse.ArgumentParser(
+        description="OCI GenAI Conversation Store → Project Migration CLI"
+    )
+    parser.add_argument("--region", required=True, help="OCI region, e.g., us-chicago-1")
+    parser.add_argument("--profile", default="BoatOc1", help="OCI config profile")
+    parser.add_argument("--compartment-id", required=True, help="Compartment or tenancy OCID")
+    parser.add_argument("--project-id", required=True, help="Target GenAI Project OCID")
+    parser.add_argument("--conversation-store-id", required=True, help="Source Conversation Store OCID")
+    parser.add_argument("--conversations", help="Comma-separated conversation IDs")
+    parser.add_argument("--conv-file", help="File containing conversation IDs (one per line)")
+    parser.add_argument("--dry-run", action="store_true", help="Do not actually migrate, just validate")
+    return parser.parse_args()
+
+
+def load_conversation_ids(args):
+    conv_ids = []
+
+    if args.conversations:
+        conv_ids.extend([c.strip() for c in args.conversations.split(",") if c.strip()])
+
+    if args.conv_file:
+        with open(args.conv_file, "r") as f:
+            for line in f:
+                line = line.strip()
+                if line:
+                    conv_ids.append(line)
+
+    if not conv_ids:
+        raise ValueError("No conversation IDs provided via --conversations or --conv-file")
+
+    return list(set(conv_ids))  # remove duplicates
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    conv_ids = load_conversation_ids(args)
+
+    print("🚀 Starting conversation migration...\n")
+    results = migrate_conversations(
+        conv_ids=conv_ids,
+        project_id=args.project_id,
+        store_id=args.conversation_store_id,
+        region=args.region,
+        profile=args.profile,
+        compartment_id=args.compartment_id,
+        chunk_size=20,
+        dry_run=args.dry_run
+    )
+
+    print("\n===================================")
+    print("🎉 Migration Finished")
+    print("===================================")
+    for old_id, new_id in results.items():
+        print(f"{old_id} → {new_id}")