feat: Support specifying BufferStrategy when open a database

[zhanglei1949]

User could specify the expected memory level to open the database:

• M_LAZY: SyncToFile
• M_FULL: InMemory
• M_HUGE: HugePagePreferred

Fix #174
Fix #180

[greptile-apps]

Your free trial has ended. If you'd like to continue receiving code reviews, you can add a payment method here.

[qodo-code-review]

Review Summary by Qodo

Add memory level configuration support to database initialization

✨ Enhancement

Walkthroughs

Description

• Add memory_level parameter to Database initialization
• Support three memory level options: InMemory, SyncToFile, HugePagePreferred
• Enable flexible string aliases (camelCase, lowercase, snake_case) for memory levels
• Add comprehensive test coverage for all memory level configurations

Diagram

flowchart LR
  User["User specifies memory_level"] -->|Python API| PyDB["PyDatabase constructor"]
  PyDB -->|Parse string| Parser["parse_memory_level function"]
  Parser -->|Convert to enum| Config["NeugDBConfig.memory_level"]
  Config -->|Pass to| Open["NeugDB::Open"]
  Open -->|Log| Logging["VLOG with memory level info"]

Loading

File Changes

1. tools/python_bind/neug/database.py ✨ Enhancement +7/-0

Add memory_level parameter to Python Database class

• Add memory_level parameter to Database.__init__() with default value "InMemory"
• Document three memory level options and their behaviors in docstring
• Pass memory_level to underlying PyDatabase C++ binding

tools/python_bind/neug/database.py

---

2. tools/python_bind/src/py_database.h ✨ Enhancement +4/-1

Add memory_level support to PyDatabase constructor

• Add memory_level parameter to PyDatabase constructor with default "InMemory"
• Parse memory level string and set config.memory_level before opening database
• Declare private parse_memory_level() method for string-to-enum conversion

tools/python_bind/src/py_database.h

---

3. tools/python_bind/src/py_database.cc ✨ Enhancement +21/-1

Implement memory level string parsing logic

• Include neug/config.h header for MemoryLevel enum
• Implement parse_memory_level() function supporting multiple string aliases
• Support canonical forms (InMemory, SyncToFile, HugePagePreferred) and variants (lowercase,
 snake_case)
• Throw exception for invalid memory level values

tools/python_bind/src/py_database.cc

---

View more (2)

4. src/main/neug_db.cc ✨ Enhancement +2/-1

Add memory level to database opening log

• Enhance logging in NeugDB::Open() to include memory level information
• Display configured memory level when opening database for debugging purposes

src/main/neug_db.cc

---

5. tools/python_bind/tests/test_db_init.py 🧪 Tests +72/-0

Add comprehensive memory level configuration tests

• Add test for default memory level behavior (InMemory)
• Add tests verifying all three memory level aliases (InMemory, SyncToFile, HugePagePreferred)
• Add test for invalid memory level error handling
• Total of 5 new test cases covering all memory level configurations

tools/python_bind/tests/test_db_init.py

---

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (1) 📘 Rule violations (0) 📎 Requirement gaps (0)

1. Enum std::to_string fails 🐞 Bug ≡ Correctness

Description

NeugDB::Open calls std::to_string(config_.memory_level) even though memory_level is a scoped
enum (enum class MemoryLevel), which does not implicitly convert to a numeric type. This will
cause a C++ compilation failure in src/main/neug_db.cc.

Code

src/main/neug_db.cc[R108-109]

+  VLOG(1) << "Opening NeuGDB at " << work_dir_
+          << ", memory level: " << std::to_string(config_.memory_level);

Evidence

NeugDBConfig::memory_level is declared as MemoryLevel, and MemoryLevel is an enum class, so
it cannot be passed to std::to_string without an explicit cast. Other parts of the codebase
already use static_cast<int>(memory_level) before std::to_string, showing the expected pattern.

src/main/neug_db.cc[103-111]
include/neug/config.h.in[48-53]
src/storages/graph/edge_table.cc[570-574]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`std::to_string(config_.memory_level)` does not compile because `config_.memory_level` is a scoped enum (`enum class MemoryLevel`).

## Issue Context
`MemoryLevel` is defined as `enum class MemoryLevel : uint8_t`, and scoped enums do not implicitly convert to integers.

## Fix Focus Areas
- src/main/neug_db.cc[107-110]

## Suggested change
Update the log line to cast the enum to an integer (or implement a dedicated stringifier):

```cpp
VLOG(1) << "Opening NeuGDB at " << work_dir_
       << ", memory level: "
       << std::to_string(static_cast<int>(config_.memory_level));
```

(Optionally, replace the numeric value with a human-readable string via a small helper mapping `MemoryLevel` -> name.)

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

ⓘ The new review experience is currently in Beta. Learn more

[longbinlai]

这个pr 更改了 python接口，那么至少应该改下这个文档：https://github.com/alibaba/neug/tree/main/doc/source/reference/python_api. 另外，我看到这个配置是 "InMemory"且是默认的配置，那么这个配置和 没有指定 path时候的 in-memory 模式到底是什么关系？这个有点confuse

[zhanglei1949]

这个pr 更改了 python接口，那么至少应该改下这个文档：https://github.com/alibaba/neug/tree/main/doc/source/reference/python_api. 另外，我看到这个配置是 "InMemory"且是默认的配置，那么这个配置和 没有指定 path时候的 in-memory 模式到底是什么关系？这个有点confuse

已经将参数名字更改为BufferStrategy. 默认改为M_HUGE

[longbinlai]

LGTM