Add chat-memory module: MessageWindowChatMemory, JDBC and Redis repositories, per-user conversation IDs
Real PostgreSQL 16 and Redis Stack 7.4; 24 tests write output/01-24. Covers the 20-message default window with no property, the 36-character conversation_id, tool messages dropped by JdbcChatMemoryRepository, concurrent add() on one conversation, a 1.x table under the 2.0 repository, the Redis repository silently backing off for a custom ChatMemory, and the removal of PromptChatMemoryAdvisor. Co-Authored-By: Claude Sonnet 5 <[email protected]> Claude-Session: https://claude.ai/code/session_01Ja4jkzrbQ4LQZBNrb5mkZE
This commit is contained in:
Claude
@@ -12,5 +12,6 @@ Runnable companion code for the Spring AI articles on [ankurm.com](https://ankur
|
||||
| [`tool-calling/`](tool-calling) | `@Tool` methods, `ToolCallingAdvisor` (the advisor-layer replacement for Spring AI 1.x's per-model tool loop), `returnDirect`, `ToolContext`, and `ToolSearchToolCallingAdvisor` for progressive disclosure across a 230-tool synthetic library -- every test driven by a hand-written `ScriptedChatModel`, no live model anywhere. Spring Boot 4.1.1, Spring AI 2.0.1, Java 25. | [Tool Calling in Spring AI 2.0](https://ankurm.com/spring-ai-2-0-tool-calling/) |
|
||||
| [`structured-output/`](structured-output) | `ChatClient.entity()` mapping LLM responses to Java records, lists and maps; `StructuredOutputValidationAdvisor` retrying non-conforming JSON with a real enum-constrained schema, including a captured run that exhausts every retry without throwing. Spring Boot 4.1.1, Spring AI 2.0.1, Java 25. | [Structured Output in Spring AI 2.0](https://ankurm.com/spring-ai-2-0-structured-output/) |
|
||||
| [`ollama-local/`](ollama-local) | Chat and embeddings against a real local `qwen2.5:0.5b`/`all-minilm`, no API key, driven by a Testcontainers-managed Ollama container started from a baked image; a confirmed model unload via `keep_alive: 0` and `/api/ps`, not a scripted model anywhere. Spring Boot 4.1.1, Spring AI 2.0.1, Testcontainers 2.0.5, Java 25. | [Run LLMs Locally with Spring AI and Ollama](https://ankurm.com/spring-ai-2-0-ollama-local/) |
|
||||
| [`chat-memory/`](chat-memory) | `MessageChatMemoryAdvisor`, `MessageWindowChatMemory`, the JDBC and Redis `ChatMemoryRepository`, per-user conversation IDs and a token-budget memory of our own, with the traps reproduced against a real PostgreSQL 16 and Redis Stack: a 36-character `conversation_id`, tool messages dropped on save, concurrent writers, a 1.x table under the 2.0 repository, and a Redis repository that silently steps aside for a custom `ChatMemory`. Spring Boot 4.1.1, Spring AI 2.0.1, Java 25. | [Chat Memory in Spring AI 2.0: JDBC, Redis and Windowed Conversations](https://ankurm.com/spring-ai-2-0-chat-memory-jdbc-redis-windowed-conversations/) |
|
||||
|
||||
Upgrading from Spring AI 1.x: [migration guide](https://ankurm.com/spring-ai-1-to-2-migration-guide/).
|
||||
|
||||
@@ -0,0 +1 @@
|
||||
target/
|
||||
@@ -0,0 +1,67 @@
|
||||
# chat-memory
|
||||
|
||||
Companion code for [Chat Memory in Spring AI 2.0: JDBC, Redis and Windowed Conversations](https://ankurm.com/spring-ai-2-0-chat-memory-jdbc-redis-windowed-conversations/), part of the [Spring AI series](../README.md) on ankurm.com.
|
||||
|
||||
`MessageChatMemoryAdvisor` + `MessageWindowChatMemory` + a `ChatMemoryRepository` (in-memory, PostgreSQL through JDBC, or Redis Stack), with per-user conversation IDs, a token-budget memory of our own, and every trap the article describes reproduced by a test that writes its own transcript.
|
||||
|
||||
There is no live model anywhere: [`ScriptedChatModel`](src/test/java/com/ankurm/chatmemory/support/ScriptedChatModel.java) records the prompts it is sent, which is what a memory test needs to prove -- *which messages reached the model*. The databases are real.
|
||||
|
||||
## Versions
|
||||
|
||||
| Component | Version |
|
||||
|---|---|
|
||||
| Spring Boot | 4.1.1 |
|
||||
| Spring AI | 2.0.1 (GA 2026-06-12 for 2.0.0; 2.0.1 on Maven Central 2026-08-20) |
|
||||
| PostgreSQL | 16.13 |
|
||||
| Redis Stack | 7.4.7 (RediSearch 2.10.20, RedisJSON) |
|
||||
| Jedis (pulled in by the Redis starter) | 7.4.1 |
|
||||
| Java | 25 (LTS) |
|
||||
|
||||
## Quickstart
|
||||
|
||||
```bash
|
||||
scripts/services-up.sh # PostgreSQL 16 (apt), Redis Stack (docker), plain Redis (apt) -- optional
|
||||
scripts/run-all.sh # runs all 24 tests and regenerates output/01 .. 24
|
||||
```
|
||||
|
||||
Tests that need a service which is not listening are **skipped with a message**, not failed, so the in-memory tests still run with nothing installed. Run the app itself with `mvn spring-boot:run` (in-memory), `-Dspring-boot.run.profiles=jdbc` or `=redis`; it needs `OPENAI_API_KEY` for real answers.
|
||||
|
||||
## What's here
|
||||
|
||||
| File | What it shows |
|
||||
|---|---|
|
||||
| [`config/MemoryConfig.java`](src/main/java/com/ankurm/chatmemory/config/MemoryConfig.java) | The `ChatMemory` bean (window size is not configurable by property) and the `ChatClient` with `MessageChatMemoryAdvisor` |
|
||||
| [`config/RedisMemoryConfig.java`](src/main/java/com/ankurm/chatmemory/config/RedisMemoryConfig.java) | Why the Redis repository is built by hand: the autoconfigured one silently steps aside for a custom `ChatMemory` |
|
||||
| [`memory/ConversationRegistry.java`](src/main/java/com/ankurm/chatmemory/memory/ConversationRegistry.java) | Who owns which conversation; `app.ownership.enforce=false` reproduces the leak |
|
||||
| [`memory/ConversationService.java`](src/main/java/com/ankurm/chatmemory/memory/ConversationService.java) | Passing `ChatMemory.CONVERSATION_ID` per request |
|
||||
| [`memory/TokenBudgetChatMemory.java`](src/main/java/com/ankurm/chatmemory/memory/TokenBudgetChatMemory.java) | A `ChatMemory` that prunes by tokens instead of by message count |
|
||||
| [`web/ChatController.java`](src/main/java/com/ankurm/chatmemory/web/ChatController.java) | Four endpoints; the user comes from an `X-User` header only so the tests need no login |
|
||||
| [`application*.yml`](src/main/resources) | Profiles `jdbc` and `redis`; the default excludes both repository autoconfigurations |
|
||||
| [`src/broken/PromptAdvisorFrom1x.java`](src/broken/PromptAdvisorFrom1x.java) | Not compiled by the build; `scripts/capture-1x-compile.sh` compiles it against 2.0.1 and 1.1.8 |
|
||||
|
||||
## Output files
|
||||
|
||||
Every file is written by the test named in the right column (or by the script), and every console block in the article is quoted from one of them.
|
||||
|
||||
| File | Written by |
|
||||
|---|---|
|
||||
| `01-window-semantics.txt` | `WindowSemanticsTest` |
|
||||
| `02-advisor-conversations.txt`, `03-missing-conversation-id.txt` | `AdvisorConversationTest` |
|
||||
| `04-autoconfigured-defaults.txt`, `05-no-window-property.txt` | `DefaultWindowTest` |
|
||||
| `06-token-growth.txt` | `TokenGrowthTest` |
|
||||
| `07-token-budget-memory.txt` | `TokenBudgetChatMemoryTest` |
|
||||
| `08` .. `14` (JDBC round trip, 36-character limit, dropped tool messages, lost update, overlapping saves, 1.x table upgrade, locked fix) | `JdbcPostgresTest` |
|
||||
| `15-redis-round-trip.txt`, `16-redis-lost-update.txt` | `RedisStackTest` |
|
||||
| `17-redis-silent-fallback.txt` | `RedisSilentFallbackTest` |
|
||||
| `18-redis-ttl-and-cap.txt` | `RedisCapsAndTtlTest` |
|
||||
| `19-redis-plain-fails.txt` | `RedisPlainFailureTest` |
|
||||
| `20-ownership-enforced.txt`, `21-ownership-unenforced.txt` | `OwnershipTest`, `OwnershipLeakTest` |
|
||||
| `22-config-keys.txt` | `ConfigKeysTest` |
|
||||
| `23-prompt-advisor-removed.txt` | `scripts/capture-1x-compile.sh` |
|
||||
| `24-memory-with-tool-calling.txt` | `ToolCallingMemoryTest` |
|
||||
|
||||
Two consecutive `mvn test` runs produce byte-identical files. `12-jdbc-overlapping-saves.txt` deliberately prints no exact count: two `saveAll` calls that overlap produce 3 or 6 messages depending on timing, and the file says so instead of pinning one run.
|
||||
|
||||
## Requirements
|
||||
|
||||
JDK 25, Maven. PostgreSQL 16 and Redis Stack are optional (see above); Docker is only used to run Redis Stack.
|
||||
@@ -0,0 +1,16 @@
|
||||
# MessageWindowChatMemory, maxMessages = 4 and 5
|
||||
|
||||
maxMessages = 4. One system message, then one user+assistant turn at a time:
|
||||
after turn 1: S:SYS-1 | U:u1 | A:a1
|
||||
after turn 2: S:SYS-1 | U:u2 | A:a2
|
||||
after turn 3: S:SYS-1 | U:u3 | A:a3
|
||||
after turn 4: S:SYS-1 | U:u4 | A:a4
|
||||
|
||||
A second, different system message arrives:
|
||||
after SYS-2: U:u4 | A:a4 | S:SYS-2
|
||||
|
||||
maxMessages = 5 (odd), no system message:
|
||||
after turn 1: U:u1 | A:a1
|
||||
after turn 2: U:u1 | A:a1 | U:u2 | A:a2
|
||||
after turn 3: U:u2 | A:a2 | U:u3 | A:a3
|
||||
after turn 4: U:u3 | A:a3 | U:u4 | A:a4
|
||||
@@ -0,0 +1,19 @@
|
||||
# MessageChatMemoryAdvisor: what the model is sent
|
||||
|
||||
conv-A, call 1: "My name is Priya"
|
||||
reply: Nice to meet you, Priya.
|
||||
model was sent: S:You are a terse assistant. | U:My name is Priya
|
||||
|
||||
conv-A, call 2: "What is my name?"
|
||||
reply: Your name is Priya.
|
||||
model was sent: S:You are a terse assistant. | U:My name is Priya | A:Nice to meet you, Priya. | U:What is my name?
|
||||
|
||||
conv-B, call 1: "What is my name?" (different conversation ID)
|
||||
reply: I do not know your name yet.
|
||||
model was sent: S:You are a terse assistant. | U:What is my name?
|
||||
|
||||
Stored for conv-A afterwards:
|
||||
USER My name is Priya
|
||||
ASSISTANT Nice to meet you, Priya.
|
||||
USER What is my name?
|
||||
ASSISTANT Your name is Priya.
|
||||
@@ -0,0 +1,5 @@
|
||||
# No conversation ID on the request
|
||||
|
||||
chat.prompt().user("hello").call().content() // no advisors(...) param
|
||||
threw java.lang.IllegalArgumentException
|
||||
message: conversationId cannot be null
|
||||
@@ -0,0 +1,5 @@
|
||||
# ChatMemoryAutoConfiguration, nothing configured
|
||||
|
||||
ChatMemory bean: MessageWindowChatMemory
|
||||
ChatMemoryRepository bean: InMemoryChatMemoryRepository
|
||||
15 turns (30 messages) added -> 20 stored, oldest kept: u6
|
||||
@@ -0,0 +1,7 @@
|
||||
# Guessing a window-size property
|
||||
|
||||
spring.ai.chat.memory.max-messages = 6 -> 30 messages added, 20 stored
|
||||
spring.ai.chat.memory.window-size = 6 -> 30 messages added, 20 stored
|
||||
spring.ai.chat.memory.repository.max-messages = 6 -> 30 messages added, 20 stored
|
||||
|
||||
Spring Boot ignores unknown keys silently, so none of these did anything.
|
||||
@@ -0,0 +1,18 @@
|
||||
# Input tokens per call, 30 turns, four pruning strategies
|
||||
|
||||
Encoding: o200k_base (JTokkit estimate). Each turn: ~25-token question, ~40-token answer, ~6-token system prompt.
|
||||
|
||||
turn unbounded window=10 window=20 budget=400tok
|
||||
1 26 26 26 26
|
||||
2 79 79 79 79
|
||||
5 238 238 238 238
|
||||
10 503 291 503 397
|
||||
11 556 291 556 397
|
||||
15 768 291 556 397
|
||||
20 1033 291 556 397
|
||||
21 1086 291 556 397
|
||||
25 1298 291 556 397
|
||||
30 1563 291 556 397
|
||||
|
||||
sum 23835 7935 13765 10426
|
||||
sum as % of unbounded: window=10 33%, window=20 58%, budget=400tok 44%
|
||||
@@ -0,0 +1,4 @@
|
||||
# TokenBudgetChatMemory, budget = 120 tokens
|
||||
|
||||
three short turns: S:Be terse. | U:short question 1 | A:short answer 1 | U:short question 2 | A:short answer 2 | U:short question 3 | A:short answer 3
|
||||
then one pasted trace: 7 messages kept: S:Be terse. | U:short question 2 | A:short answer 2 | U:short question 3 | A:short answer 3 | U:<pasted stack trace, 89 tokens> | A:Check the datasource URL.
|
||||
@@ -0,0 +1,16 @@
|
||||
# JdbcChatMemoryRepository on PostgreSQL 16
|
||||
|
||||
Table created by initialize-schema=always:
|
||||
conversation_id character varying (36)
|
||||
content text
|
||||
type character varying (10)
|
||||
timestamp timestamp without time zone
|
||||
sequence_id bigint
|
||||
|
||||
After one call, rows for the conversation:
|
||||
seq=0 USER My name is Priya
|
||||
seq=1 ASSISTANT Nice to meet you, Priya.
|
||||
|
||||
New repository + new memory over the same database, then "What is my name?":
|
||||
reply: Your name is Priya.
|
||||
findConversationIds() returns every conversation in the table: [6f0a1c9e-2b7d-4c55-9a53-0d1f3e7a2b10]
|
||||
@@ -0,0 +1,7 @@
|
||||
# conversation_id is VARCHAR(36)
|
||||
|
||||
conversation ID: "user:[email protected]:thread:2026-09-24" (43 characters)
|
||||
model calls made for this request: 0 (the user message is saved before the model is called)
|
||||
thrown: org.springframework.dao.DataIntegrityViolationException
|
||||
root cause: ERROR: value too long for type character varying(36)
|
||||
rows stored for it afterwards: 0
|
||||
@@ -0,0 +1,13 @@
|
||||
# What JdbcChatMemoryRepository.saveAll keeps
|
||||
|
||||
saveAll() given 4 messages:
|
||||
USER Weather in Mumbai?
|
||||
ASSISTANT (tool call weather)
|
||||
TOOL (tool call weather)
|
||||
ASSISTANT 31C and humid.
|
||||
|
||||
findByConversationId() returns 2:
|
||||
USER Weather in Mumbai?
|
||||
ASSISTANT 31C and humid.
|
||||
|
||||
Logged by the repository: WARN JdbcChatMemoryRepository does not support tool call messages. Some messages were filtered out for conversation: tool-conv
|
||||
@@ -0,0 +1,8 @@
|
||||
# Two concurrent add() calls on one conversation
|
||||
|
||||
seeded with: u0, a0
|
||||
two writers each add one message; both read the conversation before either writes it back
|
||||
first writer to save: adds a 3-message list, saved
|
||||
second writer to save: adds a different 3-message list, replaces the first
|
||||
stored afterwards: [u0, a0, <second writer's message>] (3 messages, 4 were sent)
|
||||
the first writer's message is GONE, the second writer's is kept
|
||||
@@ -0,0 +1,7 @@
|
||||
# Two saveAll() calls released at the same instant
|
||||
|
||||
seeded with u0, a0; two writers add from-A and from-B; both saveAll() calls start together
|
||||
The outcome differs from run to run, so the exact count is not printed here. Across repeated runs on
|
||||
PostgreSQL 16 (READ COMMITTED) it is one of two things, and never the 4 messages that were sent:
|
||||
3 messages -> one writer's message is gone
|
||||
6 messages -> both transactions inserted, history is duplicated, sequence_id values collide
|
||||
@@ -0,0 +1,16 @@
|
||||
# A Spring AI 1.x table under the 2.0 repository
|
||||
|
||||
A 1.1.x-shaped table (no sequence_id column) holds 2 rows for "old-conv".
|
||||
|
||||
2.0.1 repository, findByConversationId("old-conv"):
|
||||
ERROR: column "sequence_id" does not exist
|
||||
|
||||
Migration: add the column, backfill it from the timestamp order, then make it NOT NULL:
|
||||
ALTER TABLE SPRING_AI_CHAT_MEMORY ADD COLUMN sequence_id BIGINT
|
||||
UPDATE SPRING_AI_CHAT_MEMORY m SET sequence_id = s.rn FROM (SELECT ctid AS c, row_number() OVER (PARTITION ...
|
||||
ALTER TABLE SPRING_AI_CHAT_MEMORY ALTER COLUMN sequence_id SET NOT NULL
|
||||
CREATE INDEX IF NOT EXISTS SPRING_AI_CHAT_MEMORY_CONVERSATION_ID_SEQUENCE_ID_IDX ON SPRING_AI_CHAT_MEMORY(c...
|
||||
|
||||
findByConversationId("old-conv") after the migration:
|
||||
USER My name is Priya
|
||||
ASSISTANT Nice to meet you, Priya.
|
||||
@@ -0,0 +1,4 @@
|
||||
# Same two writers, per-conversation lock
|
||||
|
||||
stored afterwards: 4 messages
|
||||
contains from-A: true, from-B: true
|
||||
@@ -0,0 +1,16 @@
|
||||
# RedisChatMemoryRepository on Redis Stack
|
||||
|
||||
ChatMemoryRepository bean: RedisChatMemoryRepository
|
||||
call 1 reply: Nice to meet you, Priya.
|
||||
|
||||
Keys under the default prefix "chat-memory:" (3: one JSON document per message, plus a counter):
|
||||
chat-memory:6f0a1c9e-2b7d-4c55-9a53-0d1f3e7a2b10:<n>
|
||||
chat-memory:6f0a1c9e-2b7d-4c55-9a53-0d1f3e7a2b10:<n>
|
||||
chat-memory:counter:6f0a1c9e-2b7d-4c55-9a53-0d1f3e7a2b10
|
||||
|
||||
Fields of one document: [content, conversation_id, metadata, timestamp, type]
|
||||
TTL on each key (spring.ai.chat.memory.repository.redis.time-to-live=24h): between 86390 and 86400 s -> true
|
||||
|
||||
Same conversation, second call "What is my name?":
|
||||
reply: Your name is Priya.
|
||||
stored: U:My name is Priya | A:Nice to meet you, Priya. | U:What is my name? | A:Your name is Priya.
|
||||
@@ -0,0 +1,4 @@
|
||||
# Two concurrent add() calls, Redis repository
|
||||
|
||||
seeded with u0, a0; two writers add one message each, both read before either writes back
|
||||
stored afterwards: [u0, a0, <second writer's message>] (3 messages, 4 were sent)
|
||||
@@ -0,0 +1,11 @@
|
||||
# Redis starter + your own ChatMemory bean
|
||||
|
||||
application has: spring-ai-starter-model-chat-memory-repository-redis, profile "redis", Redis Stack reachable,
|
||||
and its own @Bean ChatMemory (needed to set a window other than 20)
|
||||
|
||||
ChatMemoryRepository bean in the context: InMemoryChatMemoryRepository
|
||||
Redis keys under "chat-memory:" before / after one chat call: 0 / 0
|
||||
|
||||
Condition report for the autoconfigured Redis repository bean:
|
||||
@ConditionalOnMissingBean (types: org.springframework.ai.chat.memory.repository.redis.RedisChatMemoryRepository,org.springframework.ai.chat.memory.ChatMemory,org.springframework.ai.chat.memory.ChatMemoryRepository; SearchStrategy: all) found beans of type 'org.springframework.ai.chat.memory.ChatMemory' chatMemory
|
||||
WARN/ERROR log lines mentioning Redis or ChatMemory, captured over the whole test class: 0
|
||||
@@ -0,0 +1,8 @@
|
||||
# Redis time-to-live = 2s, max-messages-per-conversation = 4
|
||||
|
||||
saved 2 messages; readable now: U:u1 | A:a1
|
||||
3 seconds later: (gone: Redis expired the keys)
|
||||
|
||||
saved 6 messages with max-messages-per-conversation=4, read back 4:
|
||||
U:u1 | A:a1 | U:u2 | A:a2
|
||||
keys in Redis for it: 7
|
||||
@@ -0,0 +1,6 @@
|
||||
# RedisChatMemoryRepository against plain Redis 7.0 (no modules)
|
||||
|
||||
server: redis_version:7.0.15
|
||||
RedisChatMemoryRepository.builder()...initializeSchema(true).build() threw:
|
||||
java.lang.IllegalStateException
|
||||
root cause: JedisDataException: ERR unknown command 'FT._LIST', with args beginning with:
|
||||
@@ -0,0 +1,7 @@
|
||||
# app.ownership.enforce=true (default)
|
||||
|
||||
alice starts a conversation and says "My name is Priya"
|
||||
|
||||
bob: GET /conversations/<alice's id>/messages -> 403
|
||||
bob: POST /conversations/<alice's id>/messages -> 403
|
||||
alice: GET /conversations/<her id>/messages -> 200, 2 messages
|
||||
@@ -0,0 +1,7 @@
|
||||
# app.ownership.enforce=false
|
||||
|
||||
alice starts a conversation and says "My name is Priya"
|
||||
|
||||
bob: GET /conversations/<alice's id>/messages -> 200
|
||||
body: [{"role":"USER","text":"My name is Priya"},{"role":"ASSISTANT","text":"Nice to meet you, Priya."}]
|
||||
bob: POST "What is my name?" -> 200 {"reply":"Your name is Priya."}
|
||||
@@ -0,0 +1,11 @@
|
||||
# Do the chat-memory property keys bind?
|
||||
|
||||
spring.ai.chat.memory.repository.jdbc.initialize-schema=always -> initializeSchema = ALWAYS
|
||||
|
||||
redis.host / port / time-to-live / max-messages-per-conversation = cache.internal / 6390 / PT24H / 200
|
||||
|
||||
Three guesses, each of which looks right and binds nothing:
|
||||
spring.ai.chat.memory.redis.port=6390 (no "repository" segment)
|
||||
spring.ai.chat.memory.repository.redis.ttl=24h (the key is time-to-live)
|
||||
spring.data.redis.port=6390 (Spring Data Redis's key; this repository builds its own Jedis client)
|
||||
-> port = 6379, time-to-live = null
|
||||
@@ -0,0 +1,21 @@
|
||||
# PromptChatMemoryAdvisor: Spring AI 1.1.8 vs 2.0.1
|
||||
|
||||
$ javac PromptAdvisorFrom1x.java # classpath: spring-ai-client-chat 2.0.1
|
||||
PromptAdvisorFrom1x.java:3: error: cannot find symbol
|
||||
import org.springframework.ai.chat.client.advisor.PromptChatMemoryAdvisor;
|
||||
^
|
||||
symbol: class PromptChatMemoryAdvisor
|
||||
location: package org.springframework.ai.chat.client.advisor
|
||||
PromptAdvisorFrom1x.java:10: error: cannot find symbol
|
||||
return PromptChatMemoryAdvisor.builder(chatMemory).build();
|
||||
^
|
||||
symbol: variable PromptChatMemoryAdvisor
|
||||
location: class PromptAdvisorFrom1x
|
||||
2 errors
|
||||
|
||||
$ javac PromptAdvisorFrom1x.java # classpath: spring-ai-client-chat 1.1.8 first
|
||||
PromptAdvisorFrom1x.java:10: warning: [removal] PromptChatMemoryAdvisor in org.springframework.ai.chat.client.advisor has been deprecated and marked for removal
|
||||
return PromptChatMemoryAdvisor.builder(chatMemory).build();
|
||||
^
|
||||
1 warning
|
||||
compiled: PromptAdvisorFrom1x.class
|
||||
@@ -0,0 +1,15 @@
|
||||
# MessageChatMemoryAdvisor + ToolCallingAdvisor
|
||||
|
||||
Advisor.DEFAULT_CHAT_MEMORY_PRECEDENCE_ORDER = HIGHEST_PRECEDENCE + 200
|
||||
ToolCallingAdvisor.DEFAULT_ORDER = HIGHEST_PRECEDENCE + 300
|
||||
|
||||
answer: It is 31C in Mumbai.
|
||||
model calls: 2
|
||||
|
||||
What the model was sent on each call:
|
||||
call 1: U:Weather in Mumbai?
|
||||
call 2: U:Weather in Mumbai? | A:[tool call currentWeather] | T:[tool result "31C in Mumbai"]
|
||||
|
||||
Stored in memory afterwards:
|
||||
USER Weather in Mumbai?
|
||||
ASSISTANT It is 31C in Mumbai.
|
||||
@@ -0,0 +1,82 @@
|
||||
<?xml version="1.0" encoding="UTF-8"?>
|
||||
<project xmlns="http://maven.apache.org/POM/4.0.0"
|
||||
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
|
||||
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 https://maven.apache.org/xsd/maven-4.0.0.xsd">
|
||||
<modelVersion>4.0.0</modelVersion>
|
||||
|
||||
<parent>
|
||||
<groupId>org.springframework.boot</groupId>
|
||||
<artifactId>spring-boot-starter-parent</artifactId>
|
||||
<version>4.1.1</version>
|
||||
<relativePath/>
|
||||
</parent>
|
||||
|
||||
<groupId>com.ankurm</groupId>
|
||||
<artifactId>chat-memory</artifactId>
|
||||
<version>1.0.0</version>
|
||||
<name>chat-memory</name>
|
||||
<description>Chat memory in Spring AI 2.0: MessageWindowChatMemory, JDBC and Redis ChatMemoryRepository, per-user conversation IDs, pruning.</description>
|
||||
|
||||
<properties>
|
||||
<java.version>25</java.version>
|
||||
<spring-ai.version>2.0.1</spring-ai.version>
|
||||
</properties>
|
||||
|
||||
<dependencyManagement>
|
||||
<dependencies>
|
||||
<dependency>
|
||||
<groupId>org.springframework.ai</groupId>
|
||||
<artifactId>spring-ai-bom</artifactId>
|
||||
<version>${spring-ai.version}</version>
|
||||
<type>pom</type>
|
||||
<scope>import</scope>
|
||||
</dependency>
|
||||
</dependencies>
|
||||
</dependencyManagement>
|
||||
|
||||
<dependencies>
|
||||
<dependency>
|
||||
<groupId>org.springframework.boot</groupId>
|
||||
<artifactId>spring-boot-starter-webmvc</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.springframework.ai</groupId>
|
||||
<artifactId>spring-ai-starter-model-openai</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.springframework.ai</groupId>
|
||||
<artifactId>spring-ai-starter-model-chat-memory-repository-jdbc</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.springframework.ai</groupId>
|
||||
<artifactId>spring-ai-starter-model-chat-memory-repository-redis</artifactId>
|
||||
</dependency>
|
||||
<dependency>
|
||||
<groupId>org.postgresql</groupId>
|
||||
<artifactId>postgresql</artifactId>
|
||||
<scope>runtime</scope>
|
||||
</dependency>
|
||||
|
||||
<dependency>
|
||||
<groupId>org.springframework.boot</groupId>
|
||||
<artifactId>spring-boot-starter-test</artifactId>
|
||||
<scope>test</scope>
|
||||
</dependency>
|
||||
</dependencies>
|
||||
|
||||
<build>
|
||||
<plugins>
|
||||
<plugin>
|
||||
<groupId>org.springframework.boot</groupId>
|
||||
<artifactId>spring-boot-maven-plugin</artifactId>
|
||||
</plugin>
|
||||
<plugin>
|
||||
<groupId>org.apache.maven.plugins</groupId>
|
||||
<artifactId>maven-surefire-plugin</artifactId>
|
||||
<configuration>
|
||||
<argLine>-Duser.timezone=UTC -Dstdout.encoding=UTF-8 -Dfile.encoding=UTF-8</argLine>
|
||||
</configuration>
|
||||
</plugin>
|
||||
</plugins>
|
||||
</build>
|
||||
</project>
|
||||
Executable
+31
@@ -0,0 +1,31 @@
|
||||
#!/usr/bin/env bash
|
||||
# Compiles src/broken/PromptAdvisorFrom1x.java against the 2.0.1 classpath, then against the
|
||||
# 1.1.8 spring-ai-client-chat + spring-ai-model jars, and writes both compiler results to
|
||||
# output/23-prompt-advisor-removed.txt. Needs the module built once online (mvn -q test-compile).
|
||||
set -euo pipefail
|
||||
cd "$(dirname "$0")/.."
|
||||
OUT=output/23-prompt-advisor-removed.txt
|
||||
mvn -q -B dependency:build-classpath -Dmdep.outputFile=target/classpath.txt >/dev/null 2>&1
|
||||
CP2="$(cat target/classpath.txt)"
|
||||
WORK="$(mktemp -d)"
|
||||
for a in spring-ai-client-chat spring-ai-model; do
|
||||
# Maven Central occasionally answers 429/5xx to a burst of requests: retry with a pause.
|
||||
for attempt in 1 2 3 4 5; do
|
||||
curl -sfL -o "$WORK/$a-1.1.8.jar" "https://repo1.maven.org/maven2/org/springframework/ai/$a/1.1.8/$a-1.1.8.jar" && break
|
||||
sleep $((attempt * 3))
|
||||
done
|
||||
done
|
||||
CP1="$WORK/spring-ai-client-chat-1.1.8.jar:$WORK/spring-ai-model-1.1.8.jar:$CP2"
|
||||
# put the 1.1.8 jars FIRST so their classes win over the 2.0.1 ones on the classpath
|
||||
{
|
||||
echo "# PromptChatMemoryAdvisor: Spring AI 1.1.8 vs 2.0.1"
|
||||
echo
|
||||
echo "\$ javac PromptAdvisorFrom1x.java # classpath: spring-ai-client-chat 2.0.1"
|
||||
javac -proc:none -d "$WORK/out2" -cp "$CP2" src/broken/PromptAdvisorFrom1x.java 2>&1 | grep -v JAVA_TOOL | sed "s#src/broken/##" || true
|
||||
echo
|
||||
echo "\$ javac PromptAdvisorFrom1x.java # classpath: spring-ai-client-chat 1.1.8 first"
|
||||
mkdir -p "$WORK/out1"
|
||||
if javac -proc:none -d "$WORK/out1" -cp "$CP1" src/broken/PromptAdvisorFrom1x.java 2>&1 | grep -v JAVA_TOOL | sed "s#src/broken/##"; then :; fi
|
||||
ls "$WORK/out1" | sed 's/^/compiled: /'
|
||||
} > "$OUT"
|
||||
cat "$OUT"
|
||||
Executable
+13
@@ -0,0 +1,13 @@
|
||||
#!/usr/bin/env bash
|
||||
# Regenerates every file under output/. The test suite writes 01-22 and 24 itself through the
|
||||
# Transcript helper; 23 comes from scripts/capture-1x-compile.sh.
|
||||
#
|
||||
# Backing services: scripts/services-up.sh starts PostgreSQL 16, Redis Stack (docker) and a plain
|
||||
# Redis. Tests that need one that is not listening are SKIPPED with a message, not failed, so the
|
||||
# in-memory tests (01-07, 20-22, 24) still run without any of them.
|
||||
set -euo pipefail
|
||||
cd "$(dirname "$0")/.."
|
||||
rm -rf target
|
||||
mvn -q -B test
|
||||
scripts/capture-1x-compile.sh >/dev/null
|
||||
ls output
|
||||
Executable
+23
@@ -0,0 +1,23 @@
|
||||
#!/usr/bin/env bash
|
||||
# Starts the three backing services the tests need, with no docker-compose:
|
||||
# * PostgreSQL 16 on 127.0.0.1:5440 (apt: postgresql-16; runs as the postgres user)
|
||||
# * Redis Stack on 127.0.0.1:6390 (docker: redis/redis-stack-server -- Redis + RediSearch + RedisJSON)
|
||||
# * plain Redis on 127.0.0.1:6391 (apt: redis-server, NO modules; only used to capture a failure)
|
||||
# If you already have these, skip this script and set CM_PG_URL / CM_REDIS_STACK_PORT / CM_REDIS_PLAIN_PORT.
|
||||
set -euo pipefail
|
||||
|
||||
PGDATA="${PGDATA:-/tmp/pgchat/data}"; PGPORT="${PGPORT:-5440}"; SOCKDIR="$(dirname "$PGDATA")"
|
||||
BIN="$(ls -d /usr/lib/postgresql/*/bin 2>/dev/null | sort -V | tail -1)"
|
||||
as_pg() { if [ "$(id -u)" = 0 ]; then su postgres -c "$*"; else bash -c "$*"; fi; }
|
||||
mkdir -p "$SOCKDIR"; [ "$(id -u)" = 0 ] && chown postgres "$SOCKDIR"
|
||||
[ -d "$PGDATA/base" ] || as_pg "'$BIN/initdb' -D '$PGDATA' -A trust >'$SOCKDIR/initdb.log' 2>&1"
|
||||
as_pg "'$BIN/pg_ctl' -D '$PGDATA' status" >/dev/null 2>&1 || \
|
||||
as_pg "'$BIN/pg_ctl' -D '$PGDATA' -o \"-p $PGPORT -c listen_addresses=127.0.0.1 -c unix_socket_directories=$SOCKDIR\" -l '$SOCKDIR/pg.log' -w start"
|
||||
PSQL="$BIN/psql -h $SOCKDIR -p $PGPORT -U postgres -v ON_ERROR_STOP=1 -q"
|
||||
as_pg "$PSQL -d postgres -tc \"select 1 from pg_roles where rolname='chat'\"" | grep -q 1 || as_pg "$PSQL -d postgres -c \"create role chat login superuser password 'chat'\""
|
||||
as_pg "$PSQL -d postgres -tc \"select 1 from pg_database where datname='chatdb'\"" | grep -q 1 || as_pg "$PSQL -d postgres -c 'create database chatdb owner chat'"
|
||||
|
||||
docker rm -f cm-redis-stack >/dev/null 2>&1 || true
|
||||
docker run -d --name cm-redis-stack -p 6390:6379 redis/redis-stack-server:latest >/dev/null
|
||||
redis-server --port 6391 --save "" --appendonly no --daemonize yes >/dev/null
|
||||
echo "postgres 127.0.0.1:$PGPORT (chatdb / chat / chat), redis-stack :6390, plain redis :6391"
|
||||
@@ -0,0 +1,12 @@
|
||||
// Deliberately NOT part of the build (src/broken is not a source root). Compiled by
|
||||
// scripts/capture-1x-compile.sh to capture the compiler's own message.
|
||||
import org.springframework.ai.chat.client.advisor.PromptChatMemoryAdvisor;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
|
||||
class PromptAdvisorFrom1x {
|
||||
|
||||
Object build(ChatMemory chatMemory) {
|
||||
// Spring AI 1.x: memory rendered into the system prompt as text
|
||||
return PromptChatMemoryAdvisor.builder(chatMemory).build();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,12 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import org.springframework.boot.SpringApplication;
|
||||
import org.springframework.boot.autoconfigure.SpringBootApplication;
|
||||
|
||||
@SpringBootApplication
|
||||
public class ChatMemoryApplication {
|
||||
|
||||
public static void main(String[] args) {
|
||||
SpringApplication.run(ChatMemoryApplication.class, args);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,44 @@
|
||||
package com.ankurm.chatmemory.config;
|
||||
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.client.advisor.MessageChatMemoryAdvisor;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
|
||||
import org.springframework.ai.chat.model.ChatModel;
|
||||
import org.springframework.beans.factory.annotation.Value;
|
||||
import org.springframework.context.annotation.Bean;
|
||||
import org.springframework.context.annotation.Configuration;
|
||||
|
||||
/**
|
||||
* Two beans, and both are choices the autoconfiguration would otherwise make for you:
|
||||
*
|
||||
* <ul>
|
||||
* <li>{@link ChatMemory}: Spring AI 2.0.1 autoconfigures a {@code MessageWindowChatMemory}
|
||||
* with a hard-coded window of 20 messages and no property to change it, so a window of
|
||||
* any other size means declaring this bean yourself. It sits on whichever
|
||||
* {@link ChatMemoryRepository} bean exists (in-memory, JDBC or Redis).</li>
|
||||
* <li>{@link ChatClient}: one advisor, {@link MessageChatMemoryAdvisor}, that loads the
|
||||
* conversation before every call and appends the new turn after it.</li>
|
||||
* </ul>
|
||||
*/
|
||||
@Configuration
|
||||
public class MemoryConfig {
|
||||
|
||||
@Bean
|
||||
ChatMemory chatMemory(ChatMemoryRepository repository,
|
||||
@Value("${app.memory.max-messages:20}") int maxMessages) {
|
||||
return MessageWindowChatMemory.builder()
|
||||
.chatMemoryRepository(repository)
|
||||
.maxMessages(maxMessages)
|
||||
.build();
|
||||
}
|
||||
|
||||
@Bean
|
||||
ChatClient chatClient(ChatModel chatModel, ChatMemory chatMemory) {
|
||||
return ChatClient.builder(chatModel)
|
||||
.defaultSystem("You are a terse assistant.")
|
||||
.defaultAdvisors(MessageChatMemoryAdvisor.builder(chatMemory).build())
|
||||
.build();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,55 @@
|
||||
package com.ankurm.chatmemory.config;
|
||||
|
||||
import java.time.Duration;
|
||||
|
||||
import org.springframework.ai.chat.memory.repository.redis.RedisChatMemoryRepository;
|
||||
import org.springframework.ai.model.chat.memory.repository.redis.autoconfigure.RedisChatMemoryRepositoryProperties;
|
||||
import org.springframework.boot.autoconfigure.condition.ConditionalOnProperty;
|
||||
import org.springframework.context.annotation.Bean;
|
||||
import org.springframework.context.annotation.Configuration;
|
||||
import org.springframework.context.annotation.Profile;
|
||||
import redis.clients.jedis.RedisClient;
|
||||
|
||||
/**
|
||||
* Builds the {@link RedisChatMemoryRepository} by hand from the autoconfigured properties and
|
||||
* client. Why this class exists at all is the second-most surprising thing in the article: in
|
||||
* Spring AI 2.0.1 the autoconfigured Redis repository bean is
|
||||
* {@code @ConditionalOnMissingBean(RedisChatMemoryRepository, ChatMemory, ChatMemoryRepository)}.
|
||||
* The moment the application declares its own {@code ChatMemory} bean (which it must, to change
|
||||
* the window from 20), the autoconfigured Redis repository silently steps aside and the
|
||||
* {@code InMemoryChatMemoryRepository} default takes over. Nothing is logged and everything
|
||||
* appears to work until the first restart.
|
||||
*
|
||||
* <p>Set {@code app.memory.redis.explicit-repository=false} to remove this class from the
|
||||
* context and reproduce that.
|
||||
*/
|
||||
@Configuration
|
||||
@Profile("redis")
|
||||
@ConditionalOnProperty(name = "app.memory.redis.explicit-repository", havingValue = "true", matchIfMissing = true)
|
||||
public class RedisMemoryConfig {
|
||||
|
||||
@Bean
|
||||
RedisChatMemoryRepository redisChatMemoryRepository(RedisClient jedisClient, RedisChatMemoryRepositoryProperties props) {
|
||||
RedisChatMemoryRepository.Builder builder = RedisChatMemoryRepository.builder()
|
||||
.jedisClient(jedisClient)
|
||||
.indexName(props.getIndexName())
|
||||
.keyPrefix(props.getKeyPrefix());
|
||||
Duration ttl = props.getTimeToLive();
|
||||
if (ttl != null) {
|
||||
builder.timeToLive(ttl);
|
||||
}
|
||||
if (props.getInitializeSchema() != null) {
|
||||
builder.initializeSchema(props.getInitializeSchema());
|
||||
}
|
||||
if (props.getMaxMessagesPerConversation() != null) {
|
||||
builder.maxMessagesPerConversation(props.getMaxMessagesPerConversation());
|
||||
}
|
||||
if (props.getMaxConversationIds() != null) {
|
||||
builder.maxConversationIds(props.getMaxConversationIds());
|
||||
}
|
||||
if (props.getMetadataFields() != null) {
|
||||
builder.metadataFields(props.getMetadataFields());
|
||||
}
|
||||
return builder.build();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,57 @@
|
||||
package com.ankurm.chatmemory.memory;
|
||||
|
||||
import java.util.List;
|
||||
import java.util.Map;
|
||||
import java.util.UUID;
|
||||
import java.util.concurrent.ConcurrentHashMap;
|
||||
|
||||
import org.springframework.beans.factory.annotation.Value;
|
||||
import org.springframework.stereotype.Component;
|
||||
|
||||
/**
|
||||
* Who owns which conversation. {@code ChatMemory} has no idea: it stores whatever it is handed
|
||||
* under whatever ID it is handed, so "user A must not read user B's history" is entirely the
|
||||
* application's job. This map is the smallest honest version of that job. In production it is a
|
||||
* table ({@code conversation_id, user_id}) next to the memory table.
|
||||
*
|
||||
* <p>Set {@code app.ownership.enforce=false} to switch the check off and reproduce the leak the
|
||||
* article shows.
|
||||
*/
|
||||
@Component
|
||||
public class ConversationRegistry {
|
||||
|
||||
private final Map<String, String> ownerByConversation = new ConcurrentHashMap<>();
|
||||
private final boolean enforce;
|
||||
|
||||
public ConversationRegistry(@Value("${app.ownership.enforce:true}") boolean enforce) {
|
||||
this.enforce = enforce;
|
||||
}
|
||||
|
||||
/** A UUID is exactly 36 characters, the width of the JDBC repository's conversation_id column. */
|
||||
public String create(String user) {
|
||||
String id = UUID.randomUUID().toString();
|
||||
ownerByConversation.put(id, user);
|
||||
return id;
|
||||
}
|
||||
|
||||
public void requireOwner(String user, String conversationId) {
|
||||
if (!enforce) {
|
||||
return;
|
||||
}
|
||||
if (!user.equals(ownerByConversation.get(conversationId))) {
|
||||
throw new NotYourConversationException(conversationId);
|
||||
}
|
||||
}
|
||||
|
||||
public List<String> conversationsOf(String user) {
|
||||
return ownerByConversation.entrySet().stream()
|
||||
.filter(e -> e.getValue().equals(user))
|
||||
.map(Map.Entry::getKey)
|
||||
.sorted()
|
||||
.toList();
|
||||
}
|
||||
|
||||
public void forget(String conversationId) {
|
||||
ownerByConversation.remove(conversationId);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,51 @@
|
||||
package com.ankurm.chatmemory.memory;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.stereotype.Service;
|
||||
|
||||
/**
|
||||
* The only class that talks to both the {@link ChatClient} and the {@link ChatMemory}. The
|
||||
* conversation ID reaches the advisor as a request parameter ({@link ChatMemory#CONVERSATION_ID});
|
||||
* the advisor reads it back out of the request context.
|
||||
*/
|
||||
@Service
|
||||
public class ConversationService {
|
||||
|
||||
private final ChatClient chatClient;
|
||||
private final ChatMemory chatMemory;
|
||||
private final ConversationRegistry registry;
|
||||
|
||||
public ConversationService(ChatClient chatClient, ChatMemory chatMemory, ConversationRegistry registry) {
|
||||
this.chatClient = chatClient;
|
||||
this.chatMemory = chatMemory;
|
||||
this.registry = registry;
|
||||
}
|
||||
|
||||
public String start(String user) {
|
||||
return registry.create(user);
|
||||
}
|
||||
|
||||
public String send(String user, String conversationId, String text) {
|
||||
registry.requireOwner(user, conversationId);
|
||||
return chatClient.prompt()
|
||||
.user(text)
|
||||
.advisors(a -> a.param(ChatMemory.CONVERSATION_ID, conversationId))
|
||||
.call()
|
||||
.content();
|
||||
}
|
||||
|
||||
public List<Message> history(String user, String conversationId) {
|
||||
registry.requireOwner(user, conversationId);
|
||||
return chatMemory.get(conversationId);
|
||||
}
|
||||
|
||||
public void delete(String user, String conversationId) {
|
||||
registry.requireOwner(user, conversationId);
|
||||
chatMemory.clear(conversationId);
|
||||
registry.forget(conversationId);
|
||||
}
|
||||
}
|
||||
+12
@@ -0,0 +1,12 @@
|
||||
package com.ankurm.chatmemory.memory;
|
||||
|
||||
import org.springframework.http.HttpStatus;
|
||||
import org.springframework.web.bind.annotation.ResponseStatus;
|
||||
|
||||
@ResponseStatus(HttpStatus.FORBIDDEN)
|
||||
public class NotYourConversationException extends RuntimeException {
|
||||
|
||||
public NotYourConversationException(String conversationId) {
|
||||
super("conversation " + conversationId + " does not belong to this user");
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,88 @@
|
||||
package com.ankurm.chatmemory.memory;
|
||||
|
||||
import java.util.ArrayList;
|
||||
import java.util.List;
|
||||
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.MessageType;
|
||||
import org.springframework.ai.tokenizer.TokenCountEstimator;
|
||||
|
||||
/**
|
||||
* A {@link ChatMemory} that prunes by tokens instead of by message count. {@code
|
||||
* MessageWindowChatMemory} keeps "the last N messages", which is a poor proxy for cost: ten
|
||||
* short messages and ten pasted stack traces are both "10". This one drops the oldest
|
||||
* non-system messages until the whole conversation fits a token budget, and never leaves a
|
||||
* reply at the front with its question gone.
|
||||
*
|
||||
* <p>It is a {@code ChatMemory}, so it drops into {@code MessageChatMemoryAdvisor.builder(...)}
|
||||
* exactly where the window did. The token count is an estimate (the estimator you pass in), not
|
||||
* the provider's own billing count.
|
||||
*/
|
||||
public final class TokenBudgetChatMemory implements ChatMemory {
|
||||
|
||||
private final ChatMemoryRepository repository;
|
||||
private final TokenCountEstimator estimator;
|
||||
private final int maxTokens;
|
||||
|
||||
public TokenBudgetChatMemory(ChatMemoryRepository repository, TokenCountEstimator estimator, int maxTokens) {
|
||||
this.repository = repository;
|
||||
this.estimator = estimator;
|
||||
this.maxTokens = maxTokens;
|
||||
}
|
||||
|
||||
@Override
|
||||
public void add(String conversationId, List<Message> messages) {
|
||||
List<Message> all = new ArrayList<>(repository.findByConversationId(conversationId));
|
||||
all.addAll(messages);
|
||||
repository.saveAll(conversationId, prune(all));
|
||||
}
|
||||
|
||||
@Override
|
||||
public List<Message> get(String conversationId) {
|
||||
return repository.findByConversationId(conversationId);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void clear(String conversationId) {
|
||||
repository.deleteByConversationId(conversationId);
|
||||
}
|
||||
|
||||
List<Message> prune(List<Message> all) {
|
||||
List<Message> kept = new ArrayList<>(all);
|
||||
while (tokens(kept) > maxTokens) {
|
||||
int oldest = firstIndexOfNonSystem(kept);
|
||||
if (oldest < 0) {
|
||||
break; // only system messages left: nothing more can be dropped
|
||||
}
|
||||
kept.remove(oldest);
|
||||
// A reply whose question was just dropped is noise; drop leading non-user messages too.
|
||||
while (!kept.isEmpty()) {
|
||||
int first = firstIndexOfNonSystem(kept);
|
||||
if (first < 0 || kept.get(first).getMessageType() == MessageType.USER) {
|
||||
break;
|
||||
}
|
||||
kept.remove(first);
|
||||
}
|
||||
}
|
||||
return kept;
|
||||
}
|
||||
|
||||
private int tokens(List<Message> messages) {
|
||||
int total = 0;
|
||||
for (Message m : messages) {
|
||||
total += estimator.estimate(m.getText());
|
||||
}
|
||||
return total;
|
||||
}
|
||||
|
||||
private static int firstIndexOfNonSystem(List<Message> messages) {
|
||||
for (int i = 0; i < messages.size(); i++) {
|
||||
if (messages.get(i).getMessageType() != MessageType.SYSTEM) {
|
||||
return i;
|
||||
}
|
||||
}
|
||||
return -1;
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,57 @@
|
||||
package com.ankurm.chatmemory.web;
|
||||
|
||||
import java.util.List;
|
||||
import java.util.Map;
|
||||
|
||||
import com.ankurm.chatmemory.memory.ConversationService;
|
||||
import org.springframework.web.bind.annotation.DeleteMapping;
|
||||
import org.springframework.web.bind.annotation.GetMapping;
|
||||
import org.springframework.web.bind.annotation.PathVariable;
|
||||
import org.springframework.web.bind.annotation.PostMapping;
|
||||
import org.springframework.web.bind.annotation.RequestBody;
|
||||
import org.springframework.web.bind.annotation.RequestHeader;
|
||||
import org.springframework.web.bind.annotation.RestController;
|
||||
|
||||
/**
|
||||
* The user comes from an {@code X-User} header purely so the tests need no login. In a real
|
||||
* service it is {@code Principal#getName()} from Spring Security -- never a header the client
|
||||
* chooses.
|
||||
*/
|
||||
@RestController
|
||||
public class ChatController {
|
||||
|
||||
public record SendRequest(String text) {
|
||||
}
|
||||
|
||||
public record MessageView(String role, String text) {
|
||||
}
|
||||
|
||||
private final ConversationService conversations;
|
||||
|
||||
public ChatController(ConversationService conversations) {
|
||||
this.conversations = conversations;
|
||||
}
|
||||
|
||||
@PostMapping("/conversations")
|
||||
Map<String, String> start(@RequestHeader("X-User") String user) {
|
||||
return Map.of("conversationId", conversations.start(user));
|
||||
}
|
||||
|
||||
@PostMapping("/conversations/{id}/messages")
|
||||
Map<String, String> send(@RequestHeader("X-User") String user, @PathVariable String id,
|
||||
@RequestBody SendRequest request) {
|
||||
return Map.of("reply", conversations.send(user, id, request.text()));
|
||||
}
|
||||
|
||||
@GetMapping("/conversations/{id}/messages")
|
||||
List<MessageView> history(@RequestHeader("X-User") String user, @PathVariable String id) {
|
||||
return conversations.history(user, id).stream()
|
||||
.map(m -> new MessageView(m.getMessageType().name(), m.getText()))
|
||||
.toList();
|
||||
}
|
||||
|
||||
@DeleteMapping("/conversations/{id}")
|
||||
void delete(@RequestHeader("X-User") String user, @PathVariable String id) {
|
||||
conversations.delete(user, id);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,14 @@
|
||||
spring:
|
||||
datasource:
|
||||
url: ${CM_PG_URL:jdbc:postgresql://127.0.0.1:5440/chatdb}
|
||||
username: chat
|
||||
password: chat
|
||||
ai:
|
||||
chat:
|
||||
memory:
|
||||
repository:
|
||||
jdbc:
|
||||
initialize-schema: always
|
||||
autoconfigure:
|
||||
exclude:
|
||||
- org.springframework.ai.model.chat.memory.repository.redis.autoconfigure.RedisChatMemoryRepositoryAutoConfiguration
|
||||
@@ -0,0 +1,14 @@
|
||||
spring:
|
||||
ai:
|
||||
chat:
|
||||
memory:
|
||||
repository:
|
||||
redis:
|
||||
host: 127.0.0.1
|
||||
port: ${CM_REDIS_STACK_PORT:6390}
|
||||
time-to-live: 24h
|
||||
max-messages-per-conversation: 200
|
||||
autoconfigure:
|
||||
exclude:
|
||||
- org.springframework.boot.jdbc.autoconfigure.DataSourceAutoConfiguration
|
||||
- org.springframework.ai.model.chat.memory.repository.jdbc.autoconfigure.JdbcChatMemoryRepositoryAutoConfiguration
|
||||
@@ -0,0 +1,25 @@
|
||||
spring:
|
||||
application:
|
||||
name: chat-memory
|
||||
ai:
|
||||
model:
|
||||
chat: openai
|
||||
openai:
|
||||
api-key: ${OPENAI_API_KEY:sk-not-set}
|
||||
chat:
|
||||
model: gpt-5-mini
|
||||
# Default profile = the in-memory repository. Spring AI's JDBC and Redis repository
|
||||
# autoconfigurations are on the classpath (both starters are dependencies) and would otherwise
|
||||
# demand a DataSource / a Redis server at startup, so they are excluded here and re-enabled per
|
||||
# profile in application-jdbc.yml and application-redis.yml.
|
||||
autoconfigure:
|
||||
exclude:
|
||||
- org.springframework.boot.jdbc.autoconfigure.DataSourceAutoConfiguration
|
||||
- org.springframework.ai.model.chat.memory.repository.jdbc.autoconfigure.JdbcChatMemoryRepositoryAutoConfiguration
|
||||
- org.springframework.ai.model.chat.memory.repository.redis.autoconfigure.RedisChatMemoryRepositoryAutoConfiguration
|
||||
|
||||
app:
|
||||
memory:
|
||||
max-messages: 20
|
||||
ownership:
|
||||
enforce: true
|
||||
@@ -0,0 +1,100 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
import static org.assertj.core.api.Assertions.assertThatThrownBy;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import com.ankurm.chatmemory.support.ScriptedChatModel;
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.prompt.Prompt;
|
||||
import org.springframework.beans.factory.annotation.Autowired;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.context.annotation.Import;
|
||||
|
||||
/**
|
||||
* The advisor end to end on the in-memory repository: which messages reach the model on each
|
||||
* call, what ends up in storage, and what happens when the conversation ID is left out.
|
||||
*/
|
||||
@SpringBootTest(properties = "spring.ai.model.chat=none")
|
||||
@Import(TestModelConfig.class)
|
||||
class AdvisorConversationTest {
|
||||
|
||||
@Autowired
|
||||
ChatClient chat;
|
||||
@Autowired
|
||||
ChatMemory memory;
|
||||
@Autowired
|
||||
ScriptedChatModel model;
|
||||
|
||||
private String ask(String conversationId, String text) {
|
||||
return chat.prompt().user(text)
|
||||
.advisors(a -> a.param(ChatMemory.CONVERSATION_ID, conversationId))
|
||||
.call().content();
|
||||
}
|
||||
|
||||
private static String sent(Prompt prompt) {
|
||||
return Show.inline(prompt.getInstructions());
|
||||
}
|
||||
|
||||
@Test
|
||||
void memoryIsPerConversationAndCarriesTheHistoryIntoEveryPrompt() {
|
||||
try (Transcript t = new Transcript("02-advisor-conversations.txt", "MessageChatMemoryAdvisor: what the model is sent")) {
|
||||
String a = "conv-A";
|
||||
String b = "conv-B";
|
||||
|
||||
t.line("conv-A, call 1: \"My name is Priya\"");
|
||||
t.line(" reply: %s", ask(a, "My name is Priya"));
|
||||
t.line(" model was sent: %s", sent(model.lastPrompt()));
|
||||
|
||||
t.blank().line("conv-A, call 2: \"What is my name?\"");
|
||||
String replyA = ask(a, "What is my name?");
|
||||
t.line(" reply: %s", replyA);
|
||||
t.line(" model was sent: %s", sent(model.lastPrompt()));
|
||||
|
||||
t.blank().line("conv-B, call 1: \"What is my name?\" (different conversation ID)");
|
||||
String replyB = ask(b, "What is my name?");
|
||||
t.line(" reply: %s", replyB);
|
||||
t.line(" model was sent: %s", sent(model.lastPrompt()));
|
||||
|
||||
t.blank().line("Stored for conv-A afterwards:");
|
||||
memory.get(a).forEach(m -> t.line(" %s", Show.one(m)));
|
||||
|
||||
assertThat(replyA).isEqualTo("Your name is Priya.");
|
||||
assertThat(replyB).isEqualTo("I do not know your name yet.");
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void leavingOutTheConversationIdIsAnErrorNotASharedDefault() {
|
||||
try (Transcript t = new Transcript("03-missing-conversation-id.txt", "No conversation ID on the request")) {
|
||||
t.line("chat.prompt().user(\"hello\").call().content() // no advisors(...) param");
|
||||
Throwable thrown = null;
|
||||
try {
|
||||
chat.prompt().user("hello").call().content();
|
||||
}
|
||||
catch (RuntimeException e) {
|
||||
thrown = e;
|
||||
}
|
||||
assertThat(thrown).isNotNull();
|
||||
t.line(" threw %s", thrown.getClass().getName());
|
||||
t.line(" message: %s", thrown.getMessage());
|
||||
assertThatThrownBy(() -> chat.prompt().user("hello").call().content())
|
||||
.isInstanceOf(RuntimeException.class);
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void oneCallOnAFreshConversationSendsOnlyTheSystemPromptAndTheUserTurn() {
|
||||
List<Prompt> before = List.copyOf(model.prompts());
|
||||
ask("fresh", "hi");
|
||||
Prompt p = model.lastPrompt();
|
||||
assertThat(model.prompts()).hasSize(before.size() + 1);
|
||||
assertThat(p.getInstructions()).hasSize(2);
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,59 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.Map;
|
||||
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.model.chat.memory.repository.jdbc.autoconfigure.JdbcChatMemoryRepositoryProperties;
|
||||
import org.springframework.ai.model.chat.memory.repository.redis.autoconfigure.RedisChatMemoryRepositoryProperties;
|
||||
import org.springframework.boot.context.properties.bind.Bindable;
|
||||
import org.springframework.boot.context.properties.bind.Binder;
|
||||
import org.springframework.boot.context.properties.source.MapConfigurationPropertySource;
|
||||
|
||||
/**
|
||||
* Spring Boot silently ignores property keys nothing binds to. This test binds the keys the
|
||||
* application's YAML uses to the real properties classes, then binds two plausible-looking wrong
|
||||
* spellings to show that they change nothing.
|
||||
*/
|
||||
class ConfigKeysTest {
|
||||
|
||||
private static <T> T bind(String prefix, T target, Map<String, String> props) {
|
||||
new Binder(new MapConfigurationPropertySource(props)).bind(prefix, Bindable.ofInstance(target));
|
||||
return target;
|
||||
}
|
||||
|
||||
@Test
|
||||
void theKeysInTheYamlBindAndTheGuessesDoNot() {
|
||||
try (Transcript t = new Transcript("22-config-keys.txt", "Do the chat-memory property keys bind?")) {
|
||||
JdbcChatMemoryRepositoryProperties jdbc = bind("spring.ai.chat.memory.repository.jdbc",
|
||||
new JdbcChatMemoryRepositoryProperties(), Map.of("spring.ai.chat.memory.repository.jdbc.initialize-schema", "always"));
|
||||
t.line("spring.ai.chat.memory.repository.jdbc.initialize-schema=always -> initializeSchema = %s", jdbc.getInitializeSchema());
|
||||
assertThat(jdbc.getInitializeSchema().name()).isEqualTo("ALWAYS");
|
||||
|
||||
RedisChatMemoryRepositoryProperties redis = bind("spring.ai.chat.memory.repository.redis",
|
||||
new RedisChatMemoryRepositoryProperties(), Map.of(
|
||||
"spring.ai.chat.memory.repository.redis.host", "cache.internal",
|
||||
"spring.ai.chat.memory.repository.redis.port", "6390",
|
||||
"spring.ai.chat.memory.repository.redis.time-to-live", "24h",
|
||||
"spring.ai.chat.memory.repository.redis.max-messages-per-conversation", "200"));
|
||||
t.blank().line("redis.host / port / time-to-live / max-messages-per-conversation = %s / %d / %s / %d",
|
||||
redis.getHost(), redis.getPort(), redis.getTimeToLive(), redis.getMaxMessagesPerConversation());
|
||||
assertThat(redis.getPort()).isEqualTo(6390);
|
||||
|
||||
RedisChatMemoryRepositoryProperties guess = bind("spring.ai.chat.memory.repository.redis",
|
||||
new RedisChatMemoryRepositoryProperties(), Map.of(
|
||||
"spring.ai.chat.memory.redis.port", "6390",
|
||||
"spring.ai.chat.memory.repository.redis.ttl", "24h",
|
||||
"spring.data.redis.port", "6390"));
|
||||
t.blank().line("Three guesses, each of which looks right and binds nothing:");
|
||||
t.line(" spring.ai.chat.memory.redis.port=6390 (no \"repository\" segment)");
|
||||
t.line(" spring.ai.chat.memory.repository.redis.ttl=24h (the key is time-to-live)");
|
||||
t.line(" spring.data.redis.port=6390 (Spring Data Redis's key; this repository builds its own Jedis client)");
|
||||
t.line(" -> port = %d, time-to-live = %s", guess.getPort(), guess.getTimeToLive());
|
||||
assertThat(guess.getPort()).isEqualTo(6379);
|
||||
assertThat(guess.getTimeToLive()).isNull();
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,62 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.UserMessage;
|
||||
import org.springframework.ai.model.chat.memory.autoconfigure.ChatMemoryAutoConfiguration;
|
||||
import org.springframework.boot.autoconfigure.AutoConfigurations;
|
||||
import org.springframework.boot.test.context.runner.ApplicationContextRunner;
|
||||
|
||||
/** What you get with no configuration at all -- and what happens to the properties you might expect to exist. */
|
||||
class DefaultWindowTest {
|
||||
|
||||
private final ApplicationContextRunner runner = new ApplicationContextRunner()
|
||||
.withConfiguration(AutoConfigurations.of(ChatMemoryAutoConfiguration.class));
|
||||
|
||||
private static int storedAfter30Messages(ChatMemory memory) {
|
||||
for (int n = 1; n <= 15; n++) {
|
||||
memory.add("c", List.of(new UserMessage("u" + n), new AssistantMessage("a" + n)));
|
||||
}
|
||||
List<Message> kept = memory.get("c");
|
||||
return kept.size();
|
||||
}
|
||||
|
||||
@Test
|
||||
void autoconfigurationGivesAnInMemoryRepositoryAndAWindowOfTwenty() {
|
||||
try (Transcript t = new Transcript("04-autoconfigured-defaults.txt", "ChatMemoryAutoConfiguration, nothing configured")) {
|
||||
runner.run(ctx -> {
|
||||
ChatMemory memory = ctx.getBean(ChatMemory.class);
|
||||
ChatMemoryRepository repo = ctx.getBean(ChatMemoryRepository.class);
|
||||
t.line("ChatMemory bean: %s", memory.getClass().getSimpleName());
|
||||
t.line("ChatMemoryRepository bean: %s", repo.getClass().getSimpleName());
|
||||
int kept = storedAfter30Messages(memory);
|
||||
t.line("15 turns (30 messages) added -> %d stored, oldest kept: %s", kept, memory.get("c").getFirst().getText());
|
||||
assertThat(memory).isInstanceOf(org.springframework.ai.chat.memory.MessageWindowChatMemory.class);
|
||||
assertThat(kept).isEqualTo(20);
|
||||
});
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void thereIsNoSpringAiPropertyForTheWindowSize() {
|
||||
try (Transcript t = new Transcript("05-no-window-property.txt", "Guessing a window-size property")) {
|
||||
for (String key : List.of("spring.ai.chat.memory.max-messages", "spring.ai.chat.memory.window-size",
|
||||
"spring.ai.chat.memory.repository.max-messages")) {
|
||||
runner.withPropertyValues(key + "=6").run(ctx -> {
|
||||
int kept = storedAfter30Messages(ctx.getBean(ChatMemory.class));
|
||||
t.line("%-52s = 6 -> 30 messages added, %d stored", key, kept);
|
||||
assertThat(kept).isEqualTo(20);
|
||||
});
|
||||
}
|
||||
t.blank().line("Spring Boot ignores unknown keys silently, so none of these did anything.");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,295 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
import static org.assertj.core.api.Assertions.assertThatThrownBy;
|
||||
|
||||
import java.util.List;
|
||||
import java.util.Map;
|
||||
import java.util.concurrent.BrokenBarrierException;
|
||||
import java.util.concurrent.CyclicBarrier;
|
||||
import java.util.concurrent.TimeUnit;
|
||||
import java.util.concurrent.TimeoutException;
|
||||
import java.util.stream.Collectors;
|
||||
|
||||
import com.ankurm.chatmemory.support.InterleavingRepository;
|
||||
import com.ankurm.chatmemory.support.ScriptedChatModel;
|
||||
import com.ankurm.chatmemory.support.Services;
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.BeforeAll;
|
||||
import org.junit.jupiter.api.BeforeEach;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.junit.jupiter.api.extension.ExtendWith;
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.client.advisor.MessageChatMemoryAdvisor;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
|
||||
import org.springframework.ai.chat.memory.repository.jdbc.JdbcChatMemoryRepository;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.ToolResponseMessage;
|
||||
import org.springframework.ai.chat.messages.UserMessage;
|
||||
import org.springframework.ai.chat.prompt.Prompt;
|
||||
import org.springframework.beans.factory.annotation.Autowired;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.boot.test.system.CapturedOutput;
|
||||
import org.springframework.boot.test.system.OutputCaptureExtension;
|
||||
import org.springframework.context.annotation.Import;
|
||||
import org.springframework.jdbc.core.JdbcTemplate;
|
||||
|
||||
/** The autoconfigured JdbcChatMemoryRepository against a real PostgreSQL 16 (scripts/services-up.sh). */
|
||||
@SpringBootTest(properties = { "spring.ai.model.chat=none", "spring.profiles.active=jdbc" })
|
||||
@Import(TestModelConfig.class)
|
||||
@ExtendWith(OutputCaptureExtension.class)
|
||||
class JdbcPostgresTest {
|
||||
|
||||
@BeforeAll
|
||||
static void needPostgres() {
|
||||
Services.require("PostgreSQL", Services.PG_PORT);
|
||||
}
|
||||
|
||||
@Autowired
|
||||
ChatClient chat;
|
||||
@Autowired
|
||||
ChatMemory memory;
|
||||
@Autowired
|
||||
ChatMemoryRepository repository;
|
||||
@Autowired
|
||||
JdbcTemplate jdbc;
|
||||
@Autowired
|
||||
ScriptedChatModel model;
|
||||
|
||||
@BeforeEach
|
||||
void clean() {
|
||||
jdbc.update("DELETE FROM SPRING_AI_CHAT_MEMORY");
|
||||
}
|
||||
|
||||
private String ask(String id, String text) {
|
||||
return chat.prompt().user(text).advisors(a -> a.param(ChatMemory.CONVERSATION_ID, id)).call().content();
|
||||
}
|
||||
|
||||
@Test
|
||||
void historySurvivesARestartBecauseItLivesInThePostgresTable() {
|
||||
try (Transcript t = new Transcript("08-jdbc-round-trip.txt", "JdbcChatMemoryRepository on PostgreSQL 16")) {
|
||||
List<Map<String, Object>> columns = jdbc.queryForList("""
|
||||
SELECT column_name, data_type, character_maximum_length
|
||||
FROM information_schema.columns WHERE table_name = 'spring_ai_chat_memory' ORDER BY ordinal_position""");
|
||||
t.line("Table created by initialize-schema=always:");
|
||||
columns.forEach(c -> t.line(" %-16s %-28s %s", c.get("column_name"), c.get("data_type"),
|
||||
c.get("character_maximum_length") == null ? "" : "(" + c.get("character_maximum_length") + ")"));
|
||||
|
||||
String id = "6f0a1c9e-2b7d-4c55-9a53-0d1f3e7a2b10";
|
||||
ask(id, "My name is Priya");
|
||||
t.blank().line("After one call, rows for the conversation:");
|
||||
jdbc.queryForList("SELECT sequence_id, type, content FROM SPRING_AI_CHAT_MEMORY WHERE conversation_id = ? ORDER BY sequence_id", id)
|
||||
.forEach(r -> t.line(" seq=%s %-9s %s", r.get("sequence_id"), r.get("type"), r.get("content")));
|
||||
|
||||
// "Restart": a brand-new repository and memory over the same database. Nothing is shared in the JVM.
|
||||
JdbcChatMemoryRepository fresh = JdbcChatMemoryRepository.builder().jdbcTemplate(jdbc)
|
||||
.dialect(new org.springframework.ai.chat.memory.repository.jdbc.PostgresChatMemoryRepositoryDialect()).build();
|
||||
ChatMemory freshMemory = MessageWindowChatMemory.builder().chatMemoryRepository(fresh).build();
|
||||
ChatClient freshClient = ChatClient.builder(model)
|
||||
.defaultAdvisors(MessageChatMemoryAdvisor.builder(freshMemory).build()).build();
|
||||
String reply = freshClient.prompt().user("What is my name?")
|
||||
.advisors(a -> a.param(ChatMemory.CONVERSATION_ID, id)).call().content();
|
||||
t.blank().line("New repository + new memory over the same database, then \"What is my name?\":");
|
||||
t.line(" reply: %s", reply);
|
||||
assertThat(reply).isEqualTo("Your name is Priya.");
|
||||
t.line(" findConversationIds() returns every conversation in the table: %s", fresh.findConversationIds());
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void aConversationIdLongerThan36CharactersFailsBeforeTheModelIsCalled() {
|
||||
try (Transcript t = new Transcript("09-jdbc-conversation-id-limit.txt", "conversation_id is VARCHAR(36)")) {
|
||||
String tooLong = "user:[email protected]:thread:2026-09-24"; // 43 characters
|
||||
t.line("conversation ID: \"%s\" (%d characters)", tooLong, tooLong.length());
|
||||
int callsBefore = model.prompts().size();
|
||||
Throwable thrown = null;
|
||||
try {
|
||||
ask(tooLong, "hello");
|
||||
}
|
||||
catch (RuntimeException e) {
|
||||
thrown = e;
|
||||
}
|
||||
assertThat(thrown).isNotNull();
|
||||
Throwable root = thrown;
|
||||
while (root.getCause() != null) {
|
||||
root = root.getCause();
|
||||
}
|
||||
t.line("model calls made for this request: %d (the user message is saved before the model is called)", model.prompts().size() - callsBefore);
|
||||
t.line("thrown: %s", thrown.getClass().getName());
|
||||
t.line("root cause: %s", root.getMessage().lines().findFirst().orElse(""));
|
||||
t.line("rows stored for it afterwards: %d",
|
||||
jdbc.queryForObject("SELECT count(*) FROM SPRING_AI_CHAT_MEMORY", Integer.class));
|
||||
assertThat(model.prompts().size() - callsBefore).isZero();
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void toolCallMessagesAreDroppedOnSave(CapturedOutput output) {
|
||||
try (Transcript t = new Transcript("10-jdbc-drops-tool-messages.txt", "What JdbcChatMemoryRepository.saveAll keeps")) {
|
||||
String id = "tool-conv";
|
||||
AssistantMessage withToolCall = AssistantMessage.builder().content("")
|
||||
.toolCalls(List.of(new AssistantMessage.ToolCall("call-1", "function", "weather", "{\"city\":\"Mumbai\"}"))).build();
|
||||
ToolResponseMessage toolResult = ToolResponseMessage.builder()
|
||||
.responses(List.of(new ToolResponseMessage.ToolResponse("call-1", "weather", "31C"))).build();
|
||||
List<Message> turn = List.of(new UserMessage("Weather in Mumbai?"), withToolCall, toolResult,
|
||||
new AssistantMessage("31C and humid."));
|
||||
t.line("saveAll() given 4 messages:");
|
||||
turn.forEach(m -> t.line(" %-9s %s", m.getMessageType(), m.getText().isEmpty() ? "(tool call weather)" : m.getText()));
|
||||
repository.saveAll(id, turn);
|
||||
List<Message> back = repository.findByConversationId(id);
|
||||
t.blank().line("findByConversationId() returns %d:", back.size());
|
||||
back.forEach(m -> t.line(" %s", Show.one(m)));
|
||||
String warning = output.getAll().lines().filter(l -> l.contains("JdbcChatMemoryRepository")).findFirst().orElse("(none)");
|
||||
t.blank().line("Logged by the repository: WARN %s", warning.replaceAll("^.*JdbcChatMemoryRepository\\s+:\\s+", ""));
|
||||
assertThat(back).hasSize(2);
|
||||
}
|
||||
}
|
||||
|
||||
private List<String> twoWritersOnOneConversation(ChatMemory memoryUnderTest) throws Exception {
|
||||
repository.saveAll("shared", List.of(new UserMessage("u0"), new AssistantMessage("a0"))); // seed without the barrier
|
||||
Thread a = Thread.ofPlatform().start(() -> memoryUnderTest.add("shared", new UserMessage("from-A")));
|
||||
Thread b = Thread.ofPlatform().start(() -> memoryUnderTest.add("shared", new UserMessage("from-B")));
|
||||
a.join();
|
||||
b.join();
|
||||
return memoryUnderTest.get("shared").stream().map(Message::getText).collect(Collectors.toList());
|
||||
}
|
||||
|
||||
@Test
|
||||
void twoWritersOnOneConversationLoseAMessageBecauseSaveAllReplacesTheWholeConversation() throws Exception {
|
||||
try (Transcript t = new Transcript("11-jdbc-lost-update.txt", "Two concurrent add() calls on one conversation")) {
|
||||
InterleavingRepository racy = new InterleavingRepository(repository, false);
|
||||
List<String> stored = twoWritersOnOneConversation(
|
||||
MessageWindowChatMemory.builder().chatMemoryRepository(racy).build());
|
||||
String first = racy.saveOrder.getFirst();
|
||||
String last = racy.saveOrder.getLast();
|
||||
t.line("seeded with: u0, a0");
|
||||
t.line("two writers each add one message; both read the conversation before either writes it back");
|
||||
t.line("first writer to save: adds a 3-message list, saved");
|
||||
t.line("second writer to save: adds a different 3-message list, replaces the first");
|
||||
t.line("stored afterwards: %s (%d messages, 4 were sent)",
|
||||
stored.toString().replace(last, "<second writer's message>"), stored.size());
|
||||
t.line("the first writer's message is %s, the second writer's is %s",
|
||||
stored.contains(first) ? "kept" : "GONE", stored.contains(last) ? "kept" : "gone");
|
||||
assertThat(stored).containsExactly("u0", "a0", last);
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void saveAllsThatOverlapInTimeAreWorseThanALostUpdate() throws Exception {
|
||||
try (Transcript t = new Transcript("12-jdbc-overlapping-saves.txt", "Two saveAll() calls released at the same instant")) {
|
||||
InterleavingRepository racy = new InterleavingRepository(repository, true);
|
||||
List<String> stored = twoWritersOnOneConversation(
|
||||
MessageWindowChatMemory.builder().chatMemoryRepository(racy).build());
|
||||
t.line("seeded with u0, a0; two writers add from-A and from-B; both saveAll() calls start together");
|
||||
t.line("The outcome differs from run to run, so the exact count is not printed here. Across repeated runs on");
|
||||
t.line("PostgreSQL 16 (READ COMMITTED) it is one of two things, and never the 4 messages that were sent:");
|
||||
t.line(" 3 messages -> one writer's message is gone");
|
||||
t.line(" 6 messages -> both transactions inserted, history is duplicated, sequence_id values collide");
|
||||
assertThat(stored.size()).isIn(3, 6);
|
||||
}
|
||||
}
|
||||
|
||||
/** The smallest possible fix that works inside one JVM: one lock per conversation. */
|
||||
private static final class LockingChatMemory implements ChatMemory {
|
||||
private final ChatMemory delegate;
|
||||
private final java.util.concurrent.ConcurrentHashMap<String, Object> locks = new java.util.concurrent.ConcurrentHashMap<>();
|
||||
|
||||
LockingChatMemory(ChatMemory delegate) {
|
||||
this.delegate = delegate;
|
||||
}
|
||||
|
||||
@Override
|
||||
public void add(String id, List<Message> messages) {
|
||||
synchronized (locks.computeIfAbsent(id, k -> new Object())) {
|
||||
delegate.add(id, messages);
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public List<Message> get(String id) {
|
||||
return delegate.get(id);
|
||||
}
|
||||
|
||||
@Override
|
||||
public void clear(String id) {
|
||||
delegate.clear(id);
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void aPerConversationLockRemovesTheLossOnASingleInstance() throws Exception {
|
||||
try (Transcript t = new Transcript("14-jdbc-lost-update-locked.txt", "Same two writers, per-conversation lock")) {
|
||||
MessageWindowChatMemory racy = MessageWindowChatMemory.builder()
|
||||
.chatMemoryRepository(new InterleavingRepository(repository, false)).build();
|
||||
List<String> stored = twoWritersOnOneConversation(new LockingChatMemory(racy));
|
||||
t.line("stored afterwards: %d messages", stored.size());
|
||||
t.line("contains from-A: %s, from-B: %s", stored.contains("from-A"), stored.contains("from-B"));
|
||||
assertThat(stored).hasSize(4).contains("from-A", "from-B");
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void aOneXTableStopsWorkingAfterTheUpgrade() {
|
||||
Services.require("PostgreSQL", Services.PG_PORT);
|
||||
try (Transcript t = new Transcript("13-jdbc-1x-table-upgrade.txt", "A Spring AI 1.x table under the 2.0 repository")) {
|
||||
jdbc.execute("DROP SCHEMA IF EXISTS legacy1x CASCADE");
|
||||
jdbc.execute("CREATE SCHEMA legacy1x");
|
||||
jdbc.execute("""
|
||||
CREATE TABLE legacy1x.SPRING_AI_CHAT_MEMORY (
|
||||
conversation_id VARCHAR(36) NOT NULL,
|
||||
content TEXT NOT NULL,
|
||||
type VARCHAR(10) NOT NULL CHECK (type IN ('USER', 'ASSISTANT', 'SYSTEM', 'TOOL')),
|
||||
"timestamp" TIMESTAMP NOT NULL)""");
|
||||
jdbc.update("INSERT INTO legacy1x.SPRING_AI_CHAT_MEMORY VALUES ('old-conv', 'My name is Priya', 'USER', now() - interval '2 minutes')");
|
||||
jdbc.update("INSERT INTO legacy1x.SPRING_AI_CHAT_MEMORY VALUES ('old-conv', 'Nice to meet you, Priya.', 'ASSISTANT', now() - interval '1 minute')");
|
||||
t.line("A 1.1.x-shaped table (no sequence_id column) holds 2 rows for \"old-conv\".");
|
||||
|
||||
var ds = new org.springframework.jdbc.datasource.DriverManagerDataSource(
|
||||
"jdbc:postgresql://127.0.0.1:" + Services.PG_PORT + "/chatdb?currentSchema=legacy1x", "chat", "chat");
|
||||
JdbcTemplate legacy = new JdbcTemplate(ds);
|
||||
JdbcChatMemoryRepository repo = JdbcChatMemoryRepository.builder().jdbcTemplate(legacy)
|
||||
.dialect(new org.springframework.ai.chat.memory.repository.jdbc.PostgresChatMemoryRepositoryDialect()).build();
|
||||
|
||||
t.blank().line("2.0.1 repository, findByConversationId(\"old-conv\"):");
|
||||
Throwable read = null;
|
||||
try {
|
||||
repo.findByConversationId("old-conv");
|
||||
}
|
||||
catch (RuntimeException e) {
|
||||
read = e;
|
||||
}
|
||||
assertThat(read).isNotNull();
|
||||
t.line(" %s", firstLine(read));
|
||||
|
||||
t.blank().line("Migration: add the column, backfill it from the timestamp order, then make it NOT NULL:");
|
||||
String[] steps = {
|
||||
"ALTER TABLE SPRING_AI_CHAT_MEMORY ADD COLUMN sequence_id BIGINT",
|
||||
"UPDATE SPRING_AI_CHAT_MEMORY m SET sequence_id = s.rn FROM (SELECT ctid AS c, row_number() OVER (PARTITION BY conversation_id ORDER BY \"timestamp\") - 1 AS rn FROM SPRING_AI_CHAT_MEMORY) s WHERE m.ctid = s.c",
|
||||
"ALTER TABLE SPRING_AI_CHAT_MEMORY ALTER COLUMN sequence_id SET NOT NULL",
|
||||
"CREATE INDEX IF NOT EXISTS SPRING_AI_CHAT_MEMORY_CONVERSATION_ID_SEQUENCE_ID_IDX ON SPRING_AI_CHAT_MEMORY(conversation_id, sequence_id)"
|
||||
};
|
||||
for (String step : steps) {
|
||||
legacy.execute(step);
|
||||
t.line(" %s", step.length() > 110 ? step.substring(0, 107) + "..." : step);
|
||||
}
|
||||
List<Message> migrated = repo.findByConversationId("old-conv");
|
||||
t.blank().line("findByConversationId(\"old-conv\") after the migration:");
|
||||
migrated.forEach(m -> t.line(" %s", Show.one(m)));
|
||||
assertThat(migrated).extracting(Message::getText).containsExactly("My name is Priya", "Nice to meet you, Priya.");
|
||||
jdbc.execute("DROP SCHEMA legacy1x CASCADE");
|
||||
}
|
||||
}
|
||||
|
||||
private static String firstLine(Throwable t) {
|
||||
Throwable root = t;
|
||||
while (root.getCause() != null) {
|
||||
root = root.getCause();
|
||||
}
|
||||
return root.getMessage().lines().findFirst().orElse("");
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,41 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import com.ankurm.chatmemory.support.Http;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.beans.factory.annotation.Value;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.context.annotation.Import;
|
||||
|
||||
/** The same endpoints with the ownership check off: a conversation ID is a bearer token for someone's history. */
|
||||
@SpringBootTest(webEnvironment = SpringBootTest.WebEnvironment.RANDOM_PORT,
|
||||
properties = { "spring.ai.model.chat=none", "app.ownership.enforce=false" })
|
||||
@Import(TestModelConfig.class)
|
||||
class OwnershipLeakTest {
|
||||
|
||||
@Value("${local.server.port}")
|
||||
int port;
|
||||
|
||||
@Test
|
||||
void anyoneWhoHoldsTheConversationIdCanReadAndContinueIt() {
|
||||
try (Transcript t = new Transcript("21-ownership-unenforced.txt", "app.ownership.enforce=false")) {
|
||||
Http http = new Http(port);
|
||||
String id = Http.field(http.post("alice", "/conversations", "").body(), "conversationId");
|
||||
http.post("alice", "/conversations/" + id + "/messages", "{\"text\":\"My name is Priya\"}");
|
||||
t.line("alice starts a conversation and says \"My name is Priya\"");
|
||||
|
||||
Http.Reply read = http.get("bob", "/conversations/" + id + "/messages");
|
||||
t.blank().line("bob: GET /conversations/<alice's id>/messages -> %d", read.status());
|
||||
t.line(" body: %s", read.body());
|
||||
Http.Reply write = http.post("bob", "/conversations/" + id + "/messages", "{\"text\":\"What is my name?\"}");
|
||||
t.line("bob: POST \"What is my name?\" -> %d %s", write.status(), write.body());
|
||||
|
||||
assertThat(read.status()).isEqualTo(200);
|
||||
assertThat(read.body()).contains("My name is Priya");
|
||||
assertThat(write.body()).contains("Your name is Priya.");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,42 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import com.ankurm.chatmemory.support.Http;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.beans.factory.annotation.Value;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.context.annotation.Import;
|
||||
|
||||
/** Per-user conversations over HTTP, with the ownership check on (the default). */
|
||||
@SpringBootTest(webEnvironment = SpringBootTest.WebEnvironment.RANDOM_PORT, properties = "spring.ai.model.chat=none")
|
||||
@Import(TestModelConfig.class)
|
||||
class OwnershipTest {
|
||||
|
||||
@Value("${local.server.port}")
|
||||
int port;
|
||||
|
||||
@Test
|
||||
void anotherUserGetsA403ForSomeoneElsesConversationId() {
|
||||
try (Transcript t = new Transcript("20-ownership-enforced.txt", "app.ownership.enforce=true (default)")) {
|
||||
Http http = new Http(port);
|
||||
String id = Http.field(http.post("alice", "/conversations", "").body(), "conversationId");
|
||||
http.post("alice", "/conversations/" + id + "/messages", "{\"text\":\"My name is Priya\"}");
|
||||
t.line("alice starts a conversation and says \"My name is Priya\"");
|
||||
|
||||
Http.Reply read = http.get("bob", "/conversations/" + id + "/messages");
|
||||
t.blank().line("bob: GET /conversations/<alice's id>/messages -> %d", read.status());
|
||||
Http.Reply write = http.post("bob", "/conversations/" + id + "/messages", "{\"text\":\"What is my name?\"}");
|
||||
t.line("bob: POST /conversations/<alice's id>/messages -> %d", write.status());
|
||||
Http.Reply own = http.get("alice", "/conversations/" + id + "/messages");
|
||||
t.line("alice: GET /conversations/<her id>/messages -> %d, %d messages", own.status(),
|
||||
own.body().split("\"role\"").length - 1);
|
||||
|
||||
assertThat(read.status()).isEqualTo(403);
|
||||
assertThat(write.status()).isEqualTo(403);
|
||||
assertThat(own.status()).isEqualTo(200);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,68 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.List;
|
||||
import java.util.Set;
|
||||
|
||||
import com.ankurm.chatmemory.support.Services;
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.BeforeAll;
|
||||
import org.junit.jupiter.api.BeforeEach;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.UserMessage;
|
||||
import org.springframework.beans.factory.annotation.Autowired;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.context.annotation.Import;
|
||||
import redis.clients.jedis.RedisClient;
|
||||
|
||||
/** time-to-live and max-messages-per-conversation, with a 2-second TTL so the expiry is observable. */
|
||||
@SpringBootTest(properties = { "spring.ai.model.chat=none", "spring.profiles.active=redis",
|
||||
"spring.ai.chat.memory.repository.redis.time-to-live=2s",
|
||||
"spring.ai.chat.memory.repository.redis.max-messages-per-conversation=4" })
|
||||
@Import(TestModelConfig.class)
|
||||
class RedisCapsAndTtlTest {
|
||||
|
||||
@BeforeAll
|
||||
static void needRedisStack() {
|
||||
Services.require("Redis Stack", Services.REDIS_STACK_PORT);
|
||||
}
|
||||
|
||||
@Autowired
|
||||
ChatMemoryRepository repository;
|
||||
@Autowired
|
||||
RedisClient redis;
|
||||
|
||||
@BeforeEach
|
||||
void clean() {
|
||||
Set<String> keys = redis.keys("chat-memory:*");
|
||||
if (!keys.isEmpty()) {
|
||||
redis.del(keys.toArray(new String[0]));
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void conversationsExpireAndTheRepositoryHasItsOwnMessageCap() throws Exception {
|
||||
try (Transcript t = new Transcript("18-redis-ttl-and-cap.txt", "Redis time-to-live = 2s, max-messages-per-conversation = 4")) {
|
||||
repository.saveAll("ttl-conv", List.of(new UserMessage("u1"), new AssistantMessage("a1")));
|
||||
t.line("saved 2 messages; readable now: %s", Show.inline(repository.findByConversationId("ttl-conv")));
|
||||
Thread.sleep(3000);
|
||||
List<Message> after = repository.findByConversationId("ttl-conv");
|
||||
t.line("3 seconds later: %s", after.isEmpty() ? "(gone: Redis expired the keys)" : Show.inline(after));
|
||||
assertThat(after).isEmpty();
|
||||
|
||||
repository.saveAll("cap-conv", List.of(new UserMessage("u1"), new AssistantMessage("a1"), new UserMessage("u2"),
|
||||
new AssistantMessage("a2"), new UserMessage("u3"), new AssistantMessage("a3")));
|
||||
List<Message> capped = repository.findByConversationId("cap-conv");
|
||||
t.blank().line("saved 6 messages with max-messages-per-conversation=4, read back %d:", capped.size());
|
||||
t.line(" %s", Show.inline(capped));
|
||||
Set<String> keys = redis.keys("chat-memory:*");
|
||||
t.line("keys in Redis for it: %d", keys.size());
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,42 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import com.ankurm.chatmemory.support.Services;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.BeforeAll;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.memory.repository.redis.RedisChatMemoryRepository;
|
||||
import redis.clients.jedis.RedisClient;
|
||||
|
||||
/** The Redis repository is not built on plain Redis: it needs RediSearch and RedisJSON. Point it at a plain server and read the error. */
|
||||
class RedisPlainFailureTest {
|
||||
|
||||
@BeforeAll
|
||||
static void needPlainRedis() {
|
||||
Services.require("plain Redis", Services.REDIS_PLAIN_PORT);
|
||||
}
|
||||
|
||||
@Test
|
||||
void plainRedisWithoutModulesFailsAtStartup() {
|
||||
try (Transcript t = new Transcript("19-redis-plain-fails.txt", "RedisChatMemoryRepository against plain Redis 7.0 (no modules)")) {
|
||||
RedisClient plain = RedisClient.create("redis://127.0.0.1:" + Services.REDIS_PLAIN_PORT);
|
||||
t.line("server: %s", plain.info("server").lines().filter(l -> l.startsWith("redis_version")).findFirst().orElse("?"));
|
||||
Throwable thrown = null;
|
||||
try {
|
||||
RedisChatMemoryRepository.builder().jedisClient(plain).initializeSchema(true).build();
|
||||
}
|
||||
catch (RuntimeException e) {
|
||||
thrown = e;
|
||||
}
|
||||
assertThat(thrown).isNotNull();
|
||||
Throwable root = thrown;
|
||||
while (root.getCause() != null) {
|
||||
root = root.getCause();
|
||||
}
|
||||
t.line("RedisChatMemoryRepository.builder()...initializeSchema(true).build() threw:");
|
||||
t.line(" %s", thrown.getClass().getName());
|
||||
t.line(" root cause: %s: %s", root.getClass().getSimpleName(), root.getMessage());
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,82 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.Set;
|
||||
|
||||
import com.ankurm.chatmemory.support.Services;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.BeforeAll;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.junit.jupiter.api.extension.ExtendWith;
|
||||
import org.springframework.boot.test.system.CapturedOutput;
|
||||
import org.springframework.boot.test.system.OutputCaptureExtension;
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.beans.factory.annotation.Autowired;
|
||||
import org.springframework.boot.autoconfigure.condition.ConditionEvaluationReport;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.context.ConfigurableApplicationContext;
|
||||
import org.springframework.context.annotation.Import;
|
||||
import redis.clients.jedis.RedisClient;
|
||||
|
||||
/**
|
||||
* The Redis starter, the Redis profile, a running Redis Stack -- and the history goes to the JVM heap.
|
||||
* Same application as RedisStackTest with one property flipped: app.memory.redis.explicit-repository=false.
|
||||
*/
|
||||
@SpringBootTest(properties = { "spring.ai.model.chat=none", "spring.profiles.active=redis",
|
||||
"app.memory.redis.explicit-repository=false" })
|
||||
@Import(TestModelConfig.class)
|
||||
@ExtendWith(OutputCaptureExtension.class)
|
||||
class RedisSilentFallbackTest {
|
||||
|
||||
@BeforeAll
|
||||
static void needRedisStack() {
|
||||
Services.require("Redis Stack", Services.REDIS_STACK_PORT);
|
||||
}
|
||||
|
||||
@Autowired
|
||||
ChatClient chat;
|
||||
@Autowired
|
||||
ChatMemoryRepository repository;
|
||||
@Autowired
|
||||
ConfigurableApplicationContext context;
|
||||
|
||||
@Test
|
||||
void aCustomChatMemoryBeanMakesTheRedisRepositoryStepAsideWithoutAWord(CapturedOutput output) {
|
||||
try (Transcript t = new Transcript("17-redis-silent-fallback.txt", "Redis starter + your own ChatMemory bean")) {
|
||||
RedisClient probe = RedisClient.create("redis://127.0.0.1:" + Services.REDIS_STACK_PORT);
|
||||
Set<String> leftovers = probe.keys("chat-memory:*");
|
||||
if (!leftovers.isEmpty()) {
|
||||
probe.del(leftovers.toArray(new String[0]));
|
||||
}
|
||||
Set<String> before = probe.keys("chat-memory:*");
|
||||
t.line("application has: spring-ai-starter-model-chat-memory-repository-redis, profile \"redis\", Redis Stack reachable,");
|
||||
t.line("and its own @Bean ChatMemory (needed to set a window other than 20)");
|
||||
t.blank().line("ChatMemoryRepository bean in the context: %s", repository.getClass().getSimpleName());
|
||||
|
||||
chat.prompt().user("hello").advisors(a -> a.param(ChatMemory.CONVERSATION_ID, "fallback-conv")).call().content();
|
||||
Set<String> after = probe.keys("chat-memory:*");
|
||||
t.line("Redis keys under \"chat-memory:\" before / after one chat call: %d / %d", before.size(), after.size());
|
||||
|
||||
ConditionEvaluationReport report = ConditionEvaluationReport.get(context.getBeanFactory());
|
||||
String outcome = report.getConditionAndOutcomesBySource().entrySet().stream()
|
||||
.filter(e -> e.getKey().endsWith("RedisChatMemoryRepositoryAutoConfiguration#redisChatMemoryRepository"))
|
||||
.flatMap(e -> e.getValue().stream())
|
||||
.map(o -> o.getOutcome().getMessage()).findFirst().orElse("(no condition report)");
|
||||
t.blank().line("Condition report for the autoconfigured Redis repository bean:");
|
||||
t.line(" %s", outcome.replace("SearchStrategy: all", "SearchStrategy: all"));
|
||||
long warnings = output.getAll().lines()
|
||||
.filter(l -> (l.contains(" WARN ") || l.contains(" ERROR ")) && (l.contains("Redis") || l.contains("ChatMemory")))
|
||||
.count();
|
||||
t.line("WARN/ERROR log lines mentioning Redis or ChatMemory, captured over the whole test class: %d", warnings);
|
||||
assertThat(warnings).isZero();
|
||||
|
||||
assertThat(repository.getClass().getSimpleName()).isEqualTo("InMemoryChatMemoryRepository");
|
||||
assertThat(after).hasSameSizeAs(before);
|
||||
assertThat(outcome).contains("org.springframework.ai.chat.memory.ChatMemory");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,108 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.ArrayList;
|
||||
import java.util.List;
|
||||
import java.util.Set;
|
||||
|
||||
import com.ankurm.chatmemory.support.InterleavingRepository;
|
||||
import com.ankurm.chatmemory.support.ScriptedChatModel;
|
||||
import com.ankurm.chatmemory.support.Services;
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.TestModelConfig;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.BeforeAll;
|
||||
import org.junit.jupiter.api.BeforeEach;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.UserMessage;
|
||||
import org.springframework.beans.factory.annotation.Autowired;
|
||||
import org.springframework.boot.test.context.SpringBootTest;
|
||||
import org.springframework.context.annotation.Import;
|
||||
import redis.clients.jedis.RedisClient;
|
||||
|
||||
/** RedisChatMemoryRepository against a real Redis Stack 7.4 (Redis + RediSearch + RedisJSON). */
|
||||
@SpringBootTest(properties = { "spring.ai.model.chat=none", "spring.profiles.active=redis" })
|
||||
@Import(TestModelConfig.class)
|
||||
class RedisStackTest {
|
||||
|
||||
@BeforeAll
|
||||
static void needRedisStack() {
|
||||
Services.require("Redis Stack", Services.REDIS_STACK_PORT);
|
||||
}
|
||||
|
||||
@Autowired
|
||||
ChatClient chat;
|
||||
@Autowired
|
||||
ChatMemory memory;
|
||||
@Autowired
|
||||
ChatMemoryRepository repository;
|
||||
@Autowired
|
||||
ScriptedChatModel model;
|
||||
@Autowired
|
||||
RedisClient redis;
|
||||
|
||||
@BeforeEach
|
||||
void clean() {
|
||||
Set<String> keys = redis.keys("chat-memory:*");
|
||||
if (!keys.isEmpty()) {
|
||||
redis.del(keys.toArray(new String[0]));
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
@SuppressWarnings("unchecked")
|
||||
void historyLivesInRedisJsonDocumentsWithAnExpiry() {
|
||||
try (Transcript t = new Transcript("15-redis-round-trip.txt", "RedisChatMemoryRepository on Redis Stack")) {
|
||||
t.line("ChatMemoryRepository bean: %s", repository.getClass().getSimpleName());
|
||||
String id = "6f0a1c9e-2b7d-4c55-9a53-0d1f3e7a2b10";
|
||||
String reply = chat.prompt().user("My name is Priya").advisors(a -> a.param(ChatMemory.CONVERSATION_ID, id)).call().content();
|
||||
t.line("call 1 reply: %s", reply);
|
||||
|
||||
List<String> keys = new ArrayList<>(redis.keys("chat-memory:*"));
|
||||
keys.sort(String::compareTo);
|
||||
t.blank().line("Keys under the default prefix \"chat-memory:\" (%d: one JSON document per message, plus a counter):", keys.size());
|
||||
for (String key : keys) {
|
||||
t.line(" %s", key.replaceAll("\\d{10,}", "<n>"));
|
||||
}
|
||||
Object doc = redis.jsonGet(keys.get(0));
|
||||
List<String> fields = new ArrayList<>(((java.util.Map<String, Object>) doc).keySet());
|
||||
fields.sort(String::compareTo);
|
||||
t.blank().line("Fields of one document: %s", fields);
|
||||
long ttl = redis.ttl(keys.getFirst());
|
||||
t.line("TTL on each key (spring.ai.chat.memory.repository.redis.time-to-live=24h): between 86390 and 86400 s -> %s",
|
||||
ttl > 86390 && ttl <= 86400);
|
||||
assertThat(ttl).isBetween(86390L, 86400L);
|
||||
|
||||
t.blank().line("Same conversation, second call \"What is my name?\":");
|
||||
t.line(" reply: %s", chat.prompt().user("What is my name?").advisors(a -> a.param(ChatMemory.CONVERSATION_ID, id)).call().content());
|
||||
t.line(" stored: %s", Show.inline(memory.get(id)));
|
||||
assertThat(memory.get(id)).hasSize(4);
|
||||
assertThat(repository.findConversationIds()).containsExactly(id);
|
||||
}
|
||||
}
|
||||
|
||||
@Test
|
||||
void twoWritersOnOneConversationLoseAMessageInRedisToo() throws Exception {
|
||||
try (Transcript t = new Transcript("16-redis-lost-update.txt", "Two concurrent add() calls, Redis repository")) {
|
||||
repository.saveAll("shared", List.of(new UserMessage("u0"), new AssistantMessage("a0")));
|
||||
InterleavingRepository racy = new InterleavingRepository(repository, false);
|
||||
MessageWindowChatMemory racyMemory = MessageWindowChatMemory.builder().chatMemoryRepository(racy).build();
|
||||
Thread a = Thread.ofPlatform().start(() -> racyMemory.add("shared", new UserMessage("from-A")));
|
||||
Thread b = Thread.ofPlatform().start(() -> racyMemory.add("shared", new UserMessage("from-B")));
|
||||
a.join();
|
||||
b.join();
|
||||
List<String> stored = racyMemory.get("shared").stream().map(Message::getText).toList();
|
||||
String last = racy.saveOrder.getLast();
|
||||
t.line("seeded with u0, a0; two writers add one message each, both read before either writes back");
|
||||
t.line("stored afterwards: %s (%d messages, 4 were sent)", stored.toString().replace(last, "<second writer's message>"), stored.size());
|
||||
assertThat(stored).containsExactly("u0", "a0", last);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,48 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import com.ankurm.chatmemory.memory.TokenBudgetChatMemory;
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import com.knuddels.jtokkit.api.EncodingType;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.memory.InMemoryChatMemoryRepository;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.MessageType;
|
||||
import org.springframework.ai.chat.messages.SystemMessage;
|
||||
import org.springframework.ai.chat.messages.UserMessage;
|
||||
import org.springframework.ai.tokenizer.JTokkitTokenCountEstimator;
|
||||
import org.springframework.ai.tokenizer.TokenCountEstimator;
|
||||
|
||||
/** The custom token-budget memory: one long message costs what it costs, however few messages there are. */
|
||||
class TokenBudgetChatMemoryTest {
|
||||
|
||||
private static final TokenCountEstimator ESTIMATOR = new JTokkitTokenCountEstimator(EncodingType.O200K_BASE);
|
||||
|
||||
@Test
|
||||
void aLongMessageEvictsMoreHistoryThanAWindowOfTheSameCountWould() {
|
||||
try (Transcript t = new Transcript("07-token-budget-memory.txt", "TokenBudgetChatMemory, budget = 120 tokens")) {
|
||||
TokenBudgetChatMemory memory = new TokenBudgetChatMemory(new InMemoryChatMemoryRepository(), ESTIMATOR, 120);
|
||||
memory.add("c", new SystemMessage("Be terse."));
|
||||
for (int n = 1; n <= 3; n++) {
|
||||
memory.add("c", List.of(new UserMessage("short question " + n), new AssistantMessage("short answer " + n)));
|
||||
}
|
||||
t.line("three short turns: %s", Show.inline(memory.get("c")));
|
||||
|
||||
String pasted = "java.lang.IllegalStateException: Failed to load ApplicationContext ".repeat(8);
|
||||
memory.add("c", List.of(new UserMessage(pasted), new AssistantMessage("Check the datasource URL.")));
|
||||
List<Message> after = memory.get("c");
|
||||
t.line("then one pasted trace: %d messages kept: %s", after.size(),
|
||||
Show.inline(after).replace(pasted, "<pasted stack trace, " + ESTIMATOR.estimate(pasted) + " tokens>"));
|
||||
|
||||
assertThat(after.getFirst().getMessageType()).isEqualTo(MessageType.SYSTEM);
|
||||
assertThat(after.get(1).getMessageType()).isEqualTo(MessageType.USER);
|
||||
assertThat(after.stream().mapToInt(m -> ESTIMATOR.estimate(m.getText())).sum()).isLessThanOrEqualTo(120);
|
||||
assertThat(after.size()).isLessThan(9);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,109 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.ArrayList;
|
||||
import java.util.List;
|
||||
import java.util.function.Supplier;
|
||||
|
||||
import com.ankurm.chatmemory.memory.TokenBudgetChatMemory;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import com.knuddels.jtokkit.api.EncodingType;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.client.advisor.MessageChatMemoryAdvisor;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.InMemoryChatMemoryRepository;
|
||||
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.model.ChatModel;
|
||||
import org.springframework.ai.chat.model.ChatResponse;
|
||||
import org.springframework.ai.chat.model.Generation;
|
||||
import org.springframework.ai.chat.prompt.Prompt;
|
||||
import org.springframework.ai.tokenizer.JTokkitTokenCountEstimator;
|
||||
import org.springframework.ai.tokenizer.TokenCountEstimator;
|
||||
import reactor.core.publisher.Flux;
|
||||
|
||||
/**
|
||||
* The cost side of memory: how many input tokens each call carries as a conversation grows,
|
||||
* under four pruning strategies. Token counts are JTokkit estimates with the o200k_base encoding
|
||||
* (the family OpenAI's current models use), not a provider's invoice.
|
||||
*/
|
||||
class TokenGrowthTest {
|
||||
|
||||
private static final int TURNS = 30;
|
||||
private static final TokenCountEstimator ESTIMATOR = new JTokkitTokenCountEstimator(EncodingType.O200K_BASE);
|
||||
|
||||
private static final String ANSWER = "The staging deploy failed because the readiness probe timed out before the new pods "
|
||||
+ "finished warming their caches, so the rollout was halted and the previous ReplicaSet kept serving traffic.";
|
||||
|
||||
/** Answers with a fixed paragraph and records how many tokens each prompt carried. */
|
||||
private static final class CountingModel implements ChatModel {
|
||||
final List<Integer> promptTokens = new ArrayList<>();
|
||||
|
||||
@Override
|
||||
public ChatResponse call(Prompt prompt) {
|
||||
int total = 0;
|
||||
for (Message m : prompt.getInstructions()) {
|
||||
total += ESTIMATOR.estimate(m.getText());
|
||||
}
|
||||
promptTokens.add(total);
|
||||
return new ChatResponse(List.of(new Generation(new AssistantMessage(ANSWER))));
|
||||
}
|
||||
|
||||
@Override
|
||||
public Flux<ChatResponse> stream(Prompt prompt) {
|
||||
return Flux.just(call(prompt));
|
||||
}
|
||||
}
|
||||
|
||||
private static List<Integer> run(Supplier<ChatMemory> memoryFactory) {
|
||||
CountingModel model = new CountingModel();
|
||||
ChatMemory memory = memoryFactory.get();
|
||||
ChatClient client = ChatClient.builder(model)
|
||||
.defaultSystem("You are a terse assistant.")
|
||||
.defaultAdvisors(MessageChatMemoryAdvisor.builder(memory).build())
|
||||
.build();
|
||||
for (int turn = 1; turn <= TURNS; turn++) {
|
||||
client.prompt()
|
||||
.user("Question " + turn + " about the pipeline: why did stage " + turn + " fail on the staging cluster last night?")
|
||||
.advisors(a -> a.param(ChatMemory.CONVERSATION_ID, "c"))
|
||||
.call().content();
|
||||
}
|
||||
return model.promptTokens;
|
||||
}
|
||||
|
||||
private static MessageWindowChatMemory window(int max) {
|
||||
return MessageWindowChatMemory.builder().chatMemoryRepository(new InMemoryChatMemoryRepository()).maxMessages(max).build();
|
||||
}
|
||||
|
||||
@Test
|
||||
void inputTokensPerCallGrowLinearlyUntilThePruningStrategyCapsThem() {
|
||||
List<Integer> unbounded = run(() -> window(10_000));
|
||||
List<Integer> w10 = run(() -> window(10));
|
||||
List<Integer> w20 = run(() -> window(20));
|
||||
List<Integer> budget = run(() -> new TokenBudgetChatMemory(new InMemoryChatMemoryRepository(), ESTIMATOR, 400));
|
||||
|
||||
try (Transcript t = new Transcript("06-token-growth.txt", "Input tokens per call, 30 turns, four pruning strategies")) {
|
||||
t.line("Encoding: o200k_base (JTokkit estimate). Each turn: ~25-token question, ~40-token answer, ~6-token system prompt.");
|
||||
t.blank().line("%-6s %10s %10s %10s %14s", "turn", "unbounded", "window=10", "window=20", "budget=400tok");
|
||||
for (int turn : new int[] {1, 2, 5, 10, 11, 15, 20, 21, 25, 30}) {
|
||||
t.line("%-6d %10d %10d %10d %14d", turn, unbounded.get(turn - 1), w10.get(turn - 1), w20.get(turn - 1), budget.get(turn - 1));
|
||||
}
|
||||
long su = unbounded.stream().mapToLong(Integer::longValue).sum();
|
||||
long s10 = w10.stream().mapToLong(Integer::longValue).sum();
|
||||
long s20 = w20.stream().mapToLong(Integer::longValue).sum();
|
||||
long sb = budget.stream().mapToLong(Integer::longValue).sum();
|
||||
t.blank().line("%-6s %10d %10d %10d %14d", "sum", su, s10, s20, sb);
|
||||
t.line("sum as %% of unbounded: window=10 %.0f%%, window=20 %.0f%%, budget=400tok %.0f%%",
|
||||
100.0 * s10 / su, 100.0 * s20 / su, 100.0 * sb / su);
|
||||
|
||||
assertThat(unbounded.get(29)).isGreaterThan(unbounded.get(0) * 20);
|
||||
assertThat(w10.get(29)).isEqualTo(w10.get(15));
|
||||
assertThat(w20.get(29)).isEqualTo(w20.get(25));
|
||||
assertThat(budget.get(29)).isLessThanOrEqualTo(400 + 90);
|
||||
assertThat(s10).isLessThan(su);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,74 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import com.ankurm.chatmemory.support.ScriptedToolModel;
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.client.ChatClient;
|
||||
import org.springframework.ai.chat.client.advisor.MessageChatMemoryAdvisor;
|
||||
import org.springframework.ai.chat.client.advisor.ToolCallingAdvisor;
|
||||
import org.springframework.ai.chat.memory.ChatMemory;
|
||||
import org.springframework.ai.chat.memory.InMemoryChatMemoryRepository;
|
||||
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.support.ToolCallbacks;
|
||||
import org.springframework.ai.tool.annotation.Tool;
|
||||
import org.springframework.ai.tool.annotation.ToolParam;
|
||||
|
||||
/**
|
||||
* Memory and tool calling together. The 2.0 advisor precedence puts the memory advisor
|
||||
* (HIGHEST_PRECEDENCE + 200) outside the ToolCallingAdvisor (+ 300), so the memory sees one
|
||||
* user message going in and one final answer coming out -- not the tool round trips between them.
|
||||
*/
|
||||
class ToolCallingMemoryTest {
|
||||
|
||||
static class Weather {
|
||||
@Tool(description = "Current temperature for a city")
|
||||
String currentWeather(@ToolParam(description = "city name") String city) {
|
||||
return "31C in " + city;
|
||||
}
|
||||
}
|
||||
|
||||
private static ScriptedToolModel model() {
|
||||
return ScriptedToolModel.builder()
|
||||
.thenCallTools(new AssistantMessage.ToolCall("call-1", "function", "currentWeather", "{\"city\":\"Mumbai\"}"))
|
||||
.thenRespond("It is 31C in Mumbai.")
|
||||
.build();
|
||||
}
|
||||
|
||||
@Test
|
||||
void memoryOutsideTheToolLoopStoresOnlyTheQuestionAndTheFinalAnswer() {
|
||||
try (Transcript t = new Transcript("24-memory-with-tool-calling.txt", "MessageChatMemoryAdvisor + ToolCallingAdvisor")) {
|
||||
ScriptedToolModel model = model();
|
||||
ChatMemory memory = MessageWindowChatMemory.builder().chatMemoryRepository(new InMemoryChatMemoryRepository()).build();
|
||||
ChatClient client = ChatClient.builder(model)
|
||||
.defaultToolCallbacks(ToolCallbacks.from(new Weather()))
|
||||
.defaultAdvisors(ToolCallingAdvisor.builder().build(), MessageChatMemoryAdvisor.builder(memory).build())
|
||||
.build();
|
||||
|
||||
t.line("Advisor.DEFAULT_CHAT_MEMORY_PRECEDENCE_ORDER = HIGHEST_PRECEDENCE + %d",
|
||||
org.springframework.ai.chat.client.advisor.api.Advisor.DEFAULT_CHAT_MEMORY_PRECEDENCE_ORDER - org.springframework.core.Ordered.HIGHEST_PRECEDENCE);
|
||||
t.line("ToolCallingAdvisor.DEFAULT_ORDER = HIGHEST_PRECEDENCE + %d",
|
||||
ToolCallingAdvisor.DEFAULT_ORDER - org.springframework.core.Ordered.HIGHEST_PRECEDENCE);
|
||||
t.blank();
|
||||
String answer = client.prompt().user("Weather in Mumbai?")
|
||||
.advisors(a -> a.param(ChatMemory.CONVERSATION_ID, "c")).call().content();
|
||||
t.line("answer: %s", answer);
|
||||
t.line("model calls: %d", model.callCount());
|
||||
t.blank().line("What the model was sent on each call:");
|
||||
for (int i = 0; i < model.capturedPrompts().size(); i++) {
|
||||
t.line(" call %d: %s", i + 1, Show.inline(model.capturedPrompts().get(i).getInstructions()));
|
||||
}
|
||||
List<Message> stored = memory.get("c");
|
||||
t.blank().line("Stored in memory afterwards:");
|
||||
stored.forEach(m -> t.line(" %s", Show.one(m)));
|
||||
assertThat(stored).hasSize(2);
|
||||
assertThat(stored.get(1).getText()).isEqualTo("It is 31C in Mumbai.");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,62 @@
|
||||
package com.ankurm.chatmemory;
|
||||
|
||||
import static org.assertj.core.api.Assertions.assertThat;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import com.ankurm.chatmemory.support.Show;
|
||||
import com.ankurm.chatmemory.support.Transcript;
|
||||
import org.junit.jupiter.api.Test;
|
||||
import org.springframework.ai.chat.memory.InMemoryChatMemoryRepository;
|
||||
import org.springframework.ai.chat.memory.MessageWindowChatMemory;
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.SystemMessage;
|
||||
import org.springframework.ai.chat.messages.UserMessage;
|
||||
|
||||
/**
|
||||
* What {@code MessageWindowChatMemory} actually keeps, measured. No ChatClient, no advisor, no
|
||||
* model: the memory object on its own, fed one turn at a time.
|
||||
*/
|
||||
class WindowSemanticsTest {
|
||||
|
||||
private static MessageWindowChatMemory window(int max) {
|
||||
return MessageWindowChatMemory.builder()
|
||||
.chatMemoryRepository(new InMemoryChatMemoryRepository())
|
||||
.maxMessages(max)
|
||||
.build();
|
||||
}
|
||||
|
||||
private static void turn(MessageWindowChatMemory memory, int n) {
|
||||
memory.add("c", List.of(new UserMessage("u" + n), new AssistantMessage("a" + n)));
|
||||
}
|
||||
|
||||
@Test
|
||||
void windowKeepsTheSystemMessageAndTrimsToAUserBoundary() {
|
||||
try (Transcript t = new Transcript("01-window-semantics.txt", "MessageWindowChatMemory, maxMessages = 4 and 5")) {
|
||||
MessageWindowChatMemory four = window(4);
|
||||
four.add("c", new SystemMessage("SYS-1"));
|
||||
t.line("maxMessages = 4. One system message, then one user+assistant turn at a time:");
|
||||
for (int n = 1; n <= 4; n++) {
|
||||
turn(four, n);
|
||||
t.line(" after turn %d: %s", n, Show.inline(four.get("c")));
|
||||
}
|
||||
assertThat(Show.inline(four.get("c"))).isEqualTo("S:SYS-1 | U:u4 | A:a4");
|
||||
|
||||
four.add("c", new SystemMessage("SYS-2"));
|
||||
t.blank().line("A second, different system message arrives:");
|
||||
t.line(" after SYS-2: %s", Show.inline(four.get("c")));
|
||||
assertThat(four.get("c")).noneMatch(m -> m.getText().equals("SYS-1"));
|
||||
|
||||
MessageWindowChatMemory five = window(5);
|
||||
t.blank().line("maxMessages = 5 (odd), no system message:");
|
||||
for (int n = 1; n <= 4; n++) {
|
||||
turn(five, n);
|
||||
t.line(" after turn %d: %s", n, Show.inline(five.get("c")));
|
||||
}
|
||||
List<Message> kept = five.get("c");
|
||||
assertThat(kept).hasSize(4);
|
||||
assertThat(kept.getFirst().getText()).isEqualTo("u3");
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,46 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import java.net.URI;
|
||||
import java.net.http.HttpClient;
|
||||
import java.net.http.HttpRequest;
|
||||
import java.net.http.HttpResponse;
|
||||
import java.util.regex.Matcher;
|
||||
import java.util.regex.Pattern;
|
||||
|
||||
/** A synchronous JDK HttpClient -- one send() per request, so a transcript line is one request. */
|
||||
public final class Http {
|
||||
|
||||
public record Reply(int status, String body) {
|
||||
}
|
||||
|
||||
private final HttpClient client = HttpClient.newHttpClient();
|
||||
private final String base;
|
||||
|
||||
public Http(int port) {
|
||||
this.base = "http://127.0.0.1:" + port;
|
||||
}
|
||||
|
||||
public Reply post(String user, String path, String json) {
|
||||
return send(HttpRequest.newBuilder(URI.create(base + path)).header("X-User", user)
|
||||
.header("Content-Type", "application/json").POST(HttpRequest.BodyPublishers.ofString(json)).build());
|
||||
}
|
||||
|
||||
public Reply get(String user, String path) {
|
||||
return send(HttpRequest.newBuilder(URI.create(base + path)).header("X-User", user).GET().build());
|
||||
}
|
||||
|
||||
private Reply send(HttpRequest request) {
|
||||
try {
|
||||
HttpResponse<String> r = client.send(request, HttpResponse.BodyHandlers.ofString());
|
||||
return new Reply(r.statusCode(), r.body());
|
||||
}
|
||||
catch (Exception e) {
|
||||
throw new IllegalStateException(e);
|
||||
}
|
||||
}
|
||||
|
||||
public static String field(String json, String name) {
|
||||
Matcher m = Pattern.compile("\"" + name + "\":\"([^\"]*)\"").matcher(json);
|
||||
return m.find() ? m.group(1) : null;
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,87 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import java.util.List;
|
||||
import java.util.concurrent.BrokenBarrierException;
|
||||
import java.util.concurrent.CopyOnWriteArrayList;
|
||||
import java.util.concurrent.CountDownLatch;
|
||||
import java.util.concurrent.CyclicBarrier;
|
||||
import java.util.concurrent.TimeUnit;
|
||||
import java.util.concurrent.TimeoutException;
|
||||
import java.util.concurrent.atomic.AtomicBoolean;
|
||||
|
||||
import org.springframework.ai.chat.memory.ChatMemoryRepository;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
|
||||
/**
|
||||
* Holds every reader at a barrier until both have read, reproducing the interleaving a busy
|
||||
* database produces by chance. {@code overlapWrites} additionally releases both writers into
|
||||
* {@code saveAll} at the same instant; otherwise the second writer waits for the first to finish.
|
||||
*/
|
||||
public final class InterleavingRepository implements ChatMemoryRepository {
|
||||
private final ChatMemoryRepository delegate;
|
||||
private final boolean overlapWrites;
|
||||
private final CyclicBarrier bothHaveRead = new CyclicBarrier(2);
|
||||
private final CyclicBarrier bothAboutToWrite = new CyclicBarrier(2);
|
||||
private final CountDownLatch firstWriterDone = new CountDownLatch(1);
|
||||
private final AtomicBoolean firstWriter = new AtomicBoolean(true);
|
||||
public final List<String> saveOrder = new CopyOnWriteArrayList<>();
|
||||
|
||||
public InterleavingRepository(ChatMemoryRepository delegate, boolean overlapWrites) {
|
||||
this.delegate = delegate;
|
||||
this.overlapWrites = overlapWrites;
|
||||
}
|
||||
|
||||
@Override
|
||||
public List<String> findConversationIds() {
|
||||
return delegate.findConversationIds();
|
||||
}
|
||||
|
||||
@Override
|
||||
public List<Message> findByConversationId(String id) {
|
||||
List<Message> read = delegate.findByConversationId(id);
|
||||
await(bothHaveRead);
|
||||
return read;
|
||||
}
|
||||
|
||||
@Override
|
||||
public void saveAll(String id, List<Message> messages) {
|
||||
String writer = messages.getLast().getText();
|
||||
if (overlapWrites) {
|
||||
await(bothAboutToWrite);
|
||||
delegate.saveAll(id, messages);
|
||||
return;
|
||||
}
|
||||
if (firstWriter.compareAndSet(true, false)) {
|
||||
saveOrder.add(writer);
|
||||
delegate.saveAll(id, messages);
|
||||
firstWriterDone.countDown();
|
||||
}
|
||||
else {
|
||||
try {
|
||||
firstWriterDone.await(1500, TimeUnit.MILLISECONDS);
|
||||
}
|
||||
catch (InterruptedException e) {
|
||||
throw new IllegalStateException(e);
|
||||
}
|
||||
saveOrder.add(writer);
|
||||
delegate.saveAll(id, messages);
|
||||
}
|
||||
}
|
||||
|
||||
@Override
|
||||
public void deleteByConversationId(String id) {
|
||||
delegate.deleteByConversationId(id);
|
||||
}
|
||||
|
||||
private static void await(CyclicBarrier barrier) {
|
||||
try {
|
||||
barrier.await(1500, TimeUnit.MILLISECONDS);
|
||||
}
|
||||
catch (TimeoutException | BrokenBarrierException e) {
|
||||
// the other writer is blocked behind a lock (or the barrier already gave up): carry on
|
||||
}
|
||||
catch (InterruptedException e) {
|
||||
throw new IllegalStateException(e);
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,75 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import java.util.Collections;
|
||||
import java.util.List;
|
||||
import java.util.concurrent.CopyOnWriteArrayList;
|
||||
import java.util.regex.Matcher;
|
||||
import java.util.regex.Pattern;
|
||||
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
import org.springframework.ai.chat.messages.MessageType;
|
||||
import org.springframework.ai.chat.model.ChatModel;
|
||||
import org.springframework.ai.chat.model.ChatResponse;
|
||||
import org.springframework.ai.chat.model.Generation;
|
||||
import org.springframework.ai.chat.prompt.Prompt;
|
||||
import reactor.core.publisher.Flux;
|
||||
|
||||
/**
|
||||
* A scripted {@link ChatModel}: no network, no API key, no randomness. It records every
|
||||
* {@link Prompt} it is sent, which is the whole point -- what a memory test needs to prove is
|
||||
* <em>which messages reached the model</em>, and that is exactly what a recorded prompt shows.
|
||||
*
|
||||
* <p>Its only "intelligence" is one rule that makes memory observable: if the last user message
|
||||
* asks for the user's name, it looks for "my name is X" in the <em>earlier</em> messages of the
|
||||
* same prompt and repeats X, or says it does not know. Nothing here says anything about what a
|
||||
* real model would reply; it proves what the model would have been given.
|
||||
*/
|
||||
public class ScriptedChatModel implements ChatModel {
|
||||
|
||||
private static final Pattern NAME = Pattern.compile("(?i)my name is (\\w+)");
|
||||
|
||||
private final List<Prompt> prompts = new CopyOnWriteArrayList<>();
|
||||
|
||||
public List<Prompt> prompts() {
|
||||
return Collections.unmodifiableList(prompts);
|
||||
}
|
||||
|
||||
public Prompt lastPrompt() {
|
||||
return prompts.getLast();
|
||||
}
|
||||
|
||||
@Override
|
||||
public ChatResponse call(Prompt prompt) {
|
||||
prompts.add(prompt);
|
||||
return new ChatResponse(List.of(new Generation(new AssistantMessage(reply(prompt)))));
|
||||
}
|
||||
|
||||
/** {@code ChatModel#stream} does not delegate to {@code call}; it throws unless overridden. */
|
||||
@Override
|
||||
public Flux<ChatResponse> stream(Prompt prompt) {
|
||||
return Flux.just(call(prompt));
|
||||
}
|
||||
|
||||
private static String reply(Prompt prompt) {
|
||||
List<Message> messages = prompt.getInstructions();
|
||||
Message last = messages.getLast();
|
||||
String text = last.getText();
|
||||
Matcher self = NAME.matcher(text);
|
||||
if (self.find()) {
|
||||
return "Nice to meet you, " + self.group(1) + ".";
|
||||
}
|
||||
if (text.toLowerCase().contains("what is my name")) {
|
||||
for (Message m : messages.subList(0, messages.size() - 1)) {
|
||||
if (m.getMessageType() == MessageType.USER) {
|
||||
Matcher earlier = NAME.matcher(m.getText());
|
||||
if (earlier.find()) {
|
||||
return "Your name is " + earlier.group(1) + ".";
|
||||
}
|
||||
}
|
||||
}
|
||||
return "I do not know your name yet.";
|
||||
}
|
||||
return "Noted.";
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,117 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import org.springframework.ai.chat.messages.AssistantMessage;
|
||||
import org.springframework.ai.chat.model.ChatModel;
|
||||
import org.springframework.ai.chat.model.ChatResponse;
|
||||
import org.springframework.ai.chat.model.Generation;
|
||||
import org.springframework.ai.chat.prompt.ChatOptions;
|
||||
import org.springframework.ai.chat.prompt.Prompt;
|
||||
import org.springframework.ai.model.tool.ToolCallingChatOptions;
|
||||
|
||||
import java.util.ArrayDeque;
|
||||
import java.util.ArrayList;
|
||||
import java.util.Deque;
|
||||
import java.util.List;
|
||||
import java.util.Queue;
|
||||
import java.util.concurrent.CopyOnWriteArrayList;
|
||||
|
||||
/**
|
||||
* A hand-written {@link ChatModel} that returns a pre-programmed queue of {@link ChatResponse}s,
|
||||
* one per {@link #call(Prompt)} invocation, instead of calling a real LLM API. {@code ChatModel}
|
||||
* has exactly one abstract method -- confirmed with {@code javap} against
|
||||
* {@code spring-ai-model-2.0.1.jar}, everything else on the interface has a default implementation
|
||||
* -- so this is the entire surface a deterministic test double needs to implement.
|
||||
*
|
||||
* <p>{@link org.springframework.ai.chat.client.advisor.ToolCallingAdvisor ToolCallingAdvisor} (and
|
||||
* its subclass, {@code ToolSearchToolCallingAdvisor}) drive the tool-calling loop by calling the
|
||||
* underlying {@link ChatModel} once per round: once to get the model's first response (which may
|
||||
* contain tool calls), then once more per round of tool results fed back in, until a response
|
||||
* comes back with no tool calls. Queuing responses here lets a test assert the exact shape of that
|
||||
* loop -- how many rounds it took, what tool calls appeared, what the final answer was -- without
|
||||
* an API key, network access, or the nondeterminism of an actual model.
|
||||
*
|
||||
* <p>Every {@link Prompt} the advisor sends is recorded in {@link #capturedPrompts()} so a test can
|
||||
* also assert on what the advisor sent back on the next round -- in particular, that a
|
||||
* {@code ToolResponseMessage} was appended after tool execution.
|
||||
*/
|
||||
public class ScriptedToolModel implements ChatModel {
|
||||
|
||||
private final Queue<ChatResponse> script;
|
||||
private final List<Prompt> capturedPrompts = new CopyOnWriteArrayList<>();
|
||||
|
||||
private ScriptedToolModel(Deque<ChatResponse> script) {
|
||||
this.script = script;
|
||||
}
|
||||
|
||||
public static Builder builder() {
|
||||
return new Builder();
|
||||
}
|
||||
|
||||
@Override
|
||||
public ChatResponse call(Prompt prompt) {
|
||||
this.capturedPrompts.add(prompt);
|
||||
ChatResponse next = this.script.poll();
|
||||
if (next == null) {
|
||||
throw new IllegalStateException(
|
||||
"ScriptedChatModel ran out of queued responses after " + this.capturedPrompts.size()
|
||||
+ " calls. Prompts so far: " + this.capturedPrompts);
|
||||
}
|
||||
return next;
|
||||
}
|
||||
|
||||
public List<Prompt> capturedPrompts() {
|
||||
return List.copyOf(this.capturedPrompts);
|
||||
}
|
||||
|
||||
/**
|
||||
* {@code DefaultChatClientUtils} builds every outgoing {@link Prompt}'s options from
|
||||
* {@code chatModel.getOptions().mutate()} -- not {@code getDefaultOptions()}, which is a
|
||||
* separate default method nobody in the request-building path actually calls. Confirmed by
|
||||
* disassembling both: {@link ChatModel#getOptions()}'s default body is a bare
|
||||
* {@code ChatOptions.builder().build()}, a plain {@link ChatOptions} that is not a
|
||||
* {@link ToolCallingChatOptions}. Since {@code ToolCallingAdvisor.adviseCall} starts with an
|
||||
* {@code instanceof ToolCallingChatOptions} check on that exact object and falls straight
|
||||
* through to the underlying model with no tool loop at all when it fails, leaving this method's
|
||||
* default in place silently turns every tool call in this repository into a no-op -- confirmed
|
||||
* the hard way, by a first version of this class that overrode {@code getDefaultOptions()}
|
||||
* instead and watched every test below get back an empty answer after exactly one model call.
|
||||
* Real providers (OpenAI, Anthropic) return their own {@code ToolCallingChatOptions}
|
||||
* implementation from {@code getOptions()} for the same reason.
|
||||
*/
|
||||
@Override
|
||||
public ChatOptions getOptions() {
|
||||
return ToolCallingChatOptions.builder().build();
|
||||
}
|
||||
|
||||
public int callCount() {
|
||||
return this.capturedPrompts.size();
|
||||
}
|
||||
|
||||
public static final class Builder {
|
||||
|
||||
private final Deque<ChatResponse> script = new ArrayDeque<>();
|
||||
|
||||
private Builder() {
|
||||
}
|
||||
|
||||
/** Queues a plain-text final answer with no tool calls -- ends the tool-calling loop. */
|
||||
public Builder thenRespond(String text) {
|
||||
this.script.add(new ChatResponse(List.of(new Generation(new AssistantMessage(text)))));
|
||||
return this;
|
||||
}
|
||||
|
||||
/** Queues an assistant turn that calls one or more tools, continuing the loop. */
|
||||
public Builder thenCallTools(AssistantMessage.ToolCall... toolCalls) {
|
||||
AssistantMessage message = AssistantMessage.builder()
|
||||
.content("")
|
||||
.toolCalls(List.of(toolCalls))
|
||||
.build();
|
||||
this.script.add(new ChatResponse(List.of(new Generation(message))));
|
||||
return this;
|
||||
}
|
||||
|
||||
public ScriptedToolModel build() {
|
||||
return new ScriptedToolModel(new ArrayDeque<>(this.script));
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,29 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import java.net.InetSocketAddress;
|
||||
import java.net.Socket;
|
||||
|
||||
import org.junit.jupiter.api.Assumptions;
|
||||
|
||||
/** Skips a test class (with a pointer to the script that fixes it) when a backing service is not running. */
|
||||
public final class Services {
|
||||
|
||||
public static final int PG_PORT = 5440;
|
||||
public static final int REDIS_STACK_PORT = 6390;
|
||||
public static final int REDIS_PLAIN_PORT = 6391;
|
||||
|
||||
private Services() {
|
||||
}
|
||||
|
||||
public static void require(String name, int port) {
|
||||
boolean up;
|
||||
try (Socket s = new Socket()) {
|
||||
s.connect(new InetSocketAddress("127.0.0.1", port), 500);
|
||||
up = true;
|
||||
}
|
||||
catch (Exception e) {
|
||||
up = false;
|
||||
}
|
||||
Assumptions.assumeTrue(up, name + " is not listening on 127.0.0.1:" + port + " -- run scripts/services-up.sh first");
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,31 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import java.util.List;
|
||||
|
||||
import org.springframework.ai.chat.messages.Message;
|
||||
|
||||
/** One-line renderings of messages so transcripts stay short and diff cleanly. */
|
||||
public final class Show {
|
||||
|
||||
private Show() {
|
||||
}
|
||||
|
||||
public static String one(Message m) {
|
||||
return String.format("%-9s %s", m.getMessageType(), m.getText());
|
||||
}
|
||||
|
||||
public static String inline(List<Message> messages) {
|
||||
return messages.stream().map(Show::brief).reduce((a, b) -> a + " | " + b).orElse("(empty)");
|
||||
}
|
||||
|
||||
private static String brief(Message m) {
|
||||
char kind = m.getMessageType().name().charAt(0);
|
||||
if (m instanceof org.springframework.ai.chat.messages.AssistantMessage a && a.hasToolCalls()) {
|
||||
return kind + ":[tool call " + a.getToolCalls().getFirst().name() + "]";
|
||||
}
|
||||
if (m instanceof org.springframework.ai.chat.messages.ToolResponseMessage r) {
|
||||
return kind + ":[tool result " + r.getResponses().getFirst().responseData() + "]";
|
||||
}
|
||||
return kind + ":" + m.getText();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,14 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import org.springframework.boot.test.context.TestConfiguration;
|
||||
import org.springframework.context.annotation.Bean;
|
||||
|
||||
/** Puts one shared {@link ScriptedChatModel} in the context so a test can inspect what it was sent. */
|
||||
@TestConfiguration(proxyBeanMethods = false)
|
||||
public class TestModelConfig {
|
||||
|
||||
@Bean
|
||||
ScriptedChatModel scriptedChatModel() {
|
||||
return new ScriptedChatModel();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,47 @@
|
||||
package com.ankurm.chatmemory.support;
|
||||
|
||||
import java.io.IOException;
|
||||
import java.io.PrintWriter;
|
||||
import java.io.StringWriter;
|
||||
import java.nio.file.Files;
|
||||
import java.nio.file.Path;
|
||||
|
||||
/**
|
||||
* Writes a numbered transcript under {@code output/} (repository root, not {@code docs/}) and
|
||||
* echoes it to the console. Every console block quoted in the article comes out of one of these
|
||||
* files verbatim.
|
||||
*/
|
||||
public final class Transcript implements AutoCloseable {
|
||||
|
||||
private final Path path;
|
||||
private final StringWriter buffer = new StringWriter();
|
||||
private final PrintWriter out = new PrintWriter(buffer);
|
||||
|
||||
public Transcript(String fileName, String title) {
|
||||
this.path = Path.of("output", fileName);
|
||||
out.println("# " + title);
|
||||
out.println();
|
||||
}
|
||||
|
||||
public Transcript line(String format, Object... args) {
|
||||
out.println(args.length == 0 ? format : String.format(format, args));
|
||||
return this;
|
||||
}
|
||||
|
||||
public Transcript blank() {
|
||||
out.println();
|
||||
return this;
|
||||
}
|
||||
|
||||
@Override
|
||||
public void close() {
|
||||
out.flush();
|
||||
try {
|
||||
Files.createDirectories(path.getParent());
|
||||
Files.writeString(path, buffer.toString());
|
||||
} catch (IOException e) {
|
||||
throw new IllegalStateException("could not write " + path, e);
|
||||
}
|
||||
System.out.print(buffer);
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user