h2co3/avocado

Strongly-typed MongoDB driver for Rust

Avocado, the strongly-typed MongoDB driver

Usage

• See the online documentation above, or open it locally:

• cargo doc --open

• Check out the examples/ folder

• More high-level information can be found on the project page.

• The schema_validation feature can be enabled (it's enabled by default), in which case the DatabaseExt::empty_collection() method becomes available. If a collection is created using this method, it will add a JSON schema validation pass and specify the schema as generated by magnet.

• The raw_uuid feature (also enabled by default) adds some useful extension methods to make it more convenient to work with UUIDs as the type of the _id field.

  This can potentially be slow if you are performing many insertions into a collection of a complex type. However, it dynamically ensures that other users/drivers can't put malformed data in the collection. Therefore it's probably more useful if you or somebody else are accessing a database from outside the Avocado driver too. It's also great for debugging Avocado itself.

Changelog

v0.6.0

• Fix #6 by adding a context dictionary to Error.

v0.5.0

• Fix #5 by adding an #[options(...)] attribute to the Doc derive proc-macro.

v0.4.0

• Fix #2 by adding a &self parameter to the options() methods on ops:: traits.

v0.3.3

• Fix a deprecation warning related to Uuid::from_random_bytes()
• Catch some errors that incorrectly pass through the MongoDB client's Cursor API

v0.3.2

• Added a remove_inner_doc() method to DocumentExt. This allows for the easy chaining of removal from hierarchically contained Documents in transform().
• Fixed a bug where inserting 0 entities into a collection failed.
• Fixed the incorrect ordering of the key and the value type in the error message generated by DocumentExt::remove_*() methods.

v0.3.1

• Added a DocumentExt trait to the prelude for convenient and idiomatic implementation of transform(raw: Document) -> Result<Bson> methods
• Hopefully fixed the code so that docs.rs can handle it now

v0.3.0

• Added Doc::id() and Doc::set_id() methods for the sake of better efficiency in some Collection methods
  • This means that single-element wrappers such as Box<Doc> and RefCell<Doc> can no longer implement Doc themselves
• Added Collection::find_one_and_delete(), Collection::find_one_and_replace(), and Collection::find_one_and_update() methods
• Added more documentation and clarified/improved existing docs
• Added more tests, including compile-time tests for cases when #[derive(Doc)] should fail, as well as testing that an optional _id is correctly allowed

Compile-time testing the derive macro

Due to a bug in compiletest_rs, running the tests that check the error messages of the #[derive(Doc)] proc-macro requires running cargo clean first, otherwise compilation will fail with E0464.

Therefore, the recommended way of running the tests is:

cargo clean && cargo test

TODO:

• Add weights property to text indices
• Add migrations
• Default Doc::Id to ObjectId and Query::Output and FindAndUpdate::Output to T, once #29661 is stabilized

avocado's Issues

How to upsert

Hello, the crate is amazing but the docs are missing somewhere.
For example, I have a vector of struct which implements Doc. Now I want to upsert_many: insert if does not exist and update content if exists.
Should I really need to create another struct-operation, implement Upsert trait for it and use it for this operation? Maybe there is a simple way for this?

Support for the new MongoDB driver

With the (alpha) release of the actually-official MongoDB Rust driver, will we see Avocado updated to support it?

No info about inserted_ids on insert_many when there were some exceptions

Hey ho! :)
It works, but now we don't have info about inserted_ids when ordered = false.

avocado/avocado/src/coll.rs

Lines 149 to 157 in e430353

	self.inner
	.insert_many(docs, options.into())
	.chain(&message)
	.and_then(|result| {
	if let Some(error) = result.bulk_write_exception {
	Err(Error::with_cause(message(), error))
	} else if let Some(ids) = result.inserted_ids {
	let ids = ids
	.into_iter()

So you check result and if there were bulk error, drop the request as failed. But InsertManyResult also contains info about inserted ids:

pub struct InsertManyResult {
    pub acknowledged: bool,
    pub inserted_ids: Option<BTreeMap<i64, Bson>>,
    pub bulk_write_exception: Option<BulkWriteException>,
}

And I can't understand in code that transaction is okay and how many rows were anyway inserted because of ordered = false.

I'm not sure what's the correct way to handle such situation. Maybe when ordered=false in insert many, don't return Err if result was BulkWriteException and mark it always as success?

Query trait and find options

Hi! I found that in Query trait method options (https://docs.rs/avocado/0.3.3/avocado/ops/trait.Query.html#method.options) is static method (does not have access to data).
It's useful to access some inner data to configure, for example, count and limit. I need this for pagination query where that params are dynamic for options but the same for filtering.

No ability to specify ordered for insert many

Hello, there is no way to specify for insert_many ordered option: https://docs.rs/mongodb/0.3.12/mongodb/coll/options/struct.InsertManyOptions.html#structfield.ordered

It's useful, when we insert vector of data and some fields could fail on insert (for example, because of unique index). When we set ordered to false, mongodb will ignore errors and try to insert as much as possible.

Allow overriding Doc methods with ability to use Doc procedural macro

Here is related question: #4
Now we can't redefine some Doc's methods if we use proc macro on structure.

Possible to use github releases?

Thanks for amazing library @H2CO3 Would it be possible to do changelog in github releases?

Statically typed filters?

Is it possible to write statically typed filters? (e.g. similar to what diesel has or something else), and if not, is there any plans for them?

Something like:

#[derive(Debug, Serialize, Deserialize, Doc, Clone)]
pub struct Example {
	#[serde(rename = "_id")]
	pub id: Uid<Example>,
	pub description: String,
}

...

let example_col: Collection<Example>  = shared.db.existing_collection();
let res: Example = example_col.find_one(Example::description.eq("abc"));

SSL?

There doesn't seem to be a feature or any documentation for SSL connections. Is that by design or does SSL just work if you need to use it?

Counters in the README are slightly off?

The "fuck counter" and the "goto counter" (does that one even make sense in Rust?) both show 1, but the only result is their entry in the README...
And the "unsafe counter" shows three, but the results are (again) the counter in the README, and then the #![deny(...)] entries which should make it impossible to have unsafe code...

I don't know if this is intentional, but it sounds like the counters don't really make sense. Just wanted to let you know :)

Doc for generic struct

Hi! Thanks again for amazing wrapper :)
I found, that it's not possible to derive Doc for generic structure:

> message: `Doc` can't be derived for a type that is generic over type parameters

#[derive(Doc, Serialize, Deserialize)]
pub struct Test<T: Serialize + Deserialize> {
   pub data<T>
}

We specify Output type for queries, maybe we also can specify somehow exact types for deserialization?

Does aggregate operation supported in avocado?