Show HN: 我们构建了一个针对 RedisGraph/FalkorDB 的类型安全 Python ORM
2 分•作者: hello-tmst•14 天前
我们厌倦了编写原始的 Cypher 查询——转义引号、零代码补全、重构噩梦——所以我们构建了 GraphORM：一个使用纯 Python 对象的、针对 RedisGraph/FalkorDB 的类型安全的 Python ORM。
它能做什么
不再需要脆弱的 Cypher 查询：
```cypher
query = """
MATCH (a:User {user_id: 1})-[r1:FRIEND]->(b:User)-[r2:FRIEND]->(c:User)
WHERE c.user_id <> 1 AND b.active = true
WITH b, count(r2) as friend_count
WHERE friend_count > 5
RETURN c, friend_count
ORDER BY friend_count DESC
LIMIT 10
"""
```
你可以编写类型安全的 Python 代码：
```python
stmt = select().match(
(UserA, FRIEND.alias("r1"), UserB),
(UserB, FRIEND.alias("r2"), UserC)
).where(
(UserA.user_id == 1) & (UserC.user_id != 1) & (UserB.active == True)
).with_(
UserB, count(FRIEND.alias("r2")).label("friend_count")
).where(
count(FRIEND.alias("r2")) > 5
).returns(
UserC, count(FRIEND.alias("r2")).label("friend_count")
).orderby(
count(FRIEND.alias("r2")).desc()
).limit(10)
```
主要特性：
• 使用 Python 类型提示的类型安全模式
• 流畅的查询构建器 (select().match().where().returns())
• 自动批量处理 (flush(batch_size=1000))
• 原子事务 (with graph.transaction(): ...)
• 零字符串转义——O'Connor 和 "The Builder" 都能正常工作
目标受众
• AI/LLM 代理开发者：将长期记忆存储为图 (User → Message → ToolCall)
• 网络爬虫工程师：用 12 行代码插入 1 万个页面 + 链接，而 Cypher 需要 80 行
• 社交网络构建者：使用 indegree()/outdegree() 查询“朋友的朋友”
• 数据工程师：跟踪血缘关系 (Dataset → Transform → Output)
• 刚接触图的 Python 开发者：避免 Cypher 学习曲线
数据插入：真正的变革
原始 Cypher 噩梦：
```
queries = [
"""CREATE (:User {email: "alice@example.com", name: "Alice O\'Connor"})""",
"""CREATE (:User {email: "bob@example.com", name: "Bob \"The Builder\""})"""
]
for q in queries:
graph.query(q) # 没有事务安全！
```
GraphORM 的幸福：
```python
alice = User(email="alice@example.com", name="Alice O'Connor")
bob = User(email="bob@example.com", name='Bob "The Builder"')
graph.add_node(alice)
graph.add_edge(Follows(alice, bob, since=1704067200))
graph.flush() # 一次网络调用，原子事务
```
30 秒内试用
```bash
pip install graphorm
```
```python
from graphorm import Node, Edge, Graph
class User(Node):
__primary_key__ = ["email"]
email: str
name: str
class Follows(Edge):
since: int
graph = Graph("social", host="localhost", port=6379)
graph.create()
alice = User(email="alice@example.com", name="Alice")
bob = User(email="bob@example.com", name="Bob")
graph.add_node(alice)
graph.add_edge(Follows(alice, bob, since=1704067200))
graph.flush()
```
GitHub: [https://github.com/hello-tmst/graphorm](https://github.com/hello-tmst/graphorm)
我们很乐意收到诚实的反馈：
• 这是否解决了您真正的痛点？
• 生产环境使用还缺少什么？
• 有任何 API 设计建议吗？
查看原文
We were tired of writing raw Cypher — escaping quotes, zero autocomplete, refactoring nightmares — so we built GraphORM: a type-safe Python ORM for RedisGraph/FalkorDB using pure Python objects.<p>What it does
Instead of fragile Cypher:<p><pre><code> query = """
MATCH (a:User {user_id: 1})-[r1:FRIEND]->(b:User)-[r2:FRIEND]->(c:User)
WHERE c.user_id <> 1 AND b.active = true
WITH b, count(r2) as friend_count
WHERE friend_count > 5
RETURN c, friend_count
ORDER BY friend_count DESC
LIMIT 10
"""
</code></pre>
You write type-safe Python:<p><pre><code> stmt = select().match(
(UserA, FRIEND.alias("r1"), UserB),
(UserB, FRIEND.alias("r2"), UserC)
).where(
(UserA.user_id == 1) & (UserC.user_id != 1) & (UserB.active == True)
).with_(
UserB, count(FRIEND.alias("r2")).label("friend_count")
).where(
count(FRIEND.alias("r2")) > 5
).returns(
UserC, count(FRIEND.alias("r2")).label("friend_count")
).orderby(
count(FRIEND.alias("r2")).desc()
).limit(10)
</code></pre>
Key features:
• Type-safe schema with Python type hints
• Fluent query builder (select().match().where().returns())
• Automatic batching (flush(batch_size=1000))
• Atomic transactions (with graph.transaction(): ...)
• Zero string escaping — O'Connor and "The Builder" just work<p>Target audience
• AI/LLM agent devs: store long-term memory as graphs (User → Message → ToolCall)
• Web crawler engineers: insert 10k pages + links in 12 lines vs 80 lines of Cypher
• Social network builders: query "friends of friends" with indegree()/outdegree()
• Data engineers: track lineage (Dataset → Transform → Output)
• Python devs new to graphs: avoid Cypher learning curve<p>Data insertion: the real game-changer<p>Raw Cypher nightmare:
queries = [
"""CREATE (:User {email: "alice@example.com", name: "Alice O\\'Connor"})""",
"""CREATE (:User {email: "bob@example.com", name: "Bob \\"The Builder\\""})"""
]
for q in queries:
graph.query(q) # No transaction safety!<p>GraphORM bliss:
alice = User(email="alice@example.com", name="Alice O'Connor")
bob = User(email="bob@example.com", name='Bob "The Builder"')
graph.add_node(alice)
graph.add_edge(Follows(alice, bob, since=1704067200))
graph.flush() # One network call, atomic transaction<p>Try it in 30 seconds
pip install graphorm<p><pre><code> from graphorm import Node, Edge, Graph
class User(Node):
__primary_key__ = ["email"]
email: str
name: str
class Follows(Edge):
since: int
graph = Graph("social", host="localhost", port=6379)
graph.create()
alice = User(email="alice@example.com", name="Alice")
bob = User(email="bob@example.com", name="Bob")
graph.add_node(alice)
graph.add_edge(Follows(alice, bob, since=1704067200))
graph.flush()
</code></pre>
GitHub: <a href="https://github.com/hello-tmst/graphorm" rel="nofollow">https://github.com/hello-tmst/graphorm</a><p>We'd love honest feedback:
• Does this solve a real pain point for you?
• What's missing for production use?
• Any API design suggestions?