volcago

README

volcago

Automatically generate code used by Cloud Firestore.

日本語ドキュメント

Installation

Recommend that you drop the binary from the release and use it.
Also, possible with go install ↓

$ go install github.com/go-generalize/volcago/cmd/volcago@latest

Usage

package task

import (
	"time"
)

//go:generate volcago Task

type Task struct {
	ID      string          `firestore:"-"           firestore_key:""`
	Desc    string          `firestore:"description" indexer:"suffix,like" unique:""`
	Done    bool            `firestore:"done"        indexer:"equal"`
	Count   int             `firestore:"count"`
	Created time.Time       `firestore:"created"`
	Indexes map[string]bool `firestore:"indexes"`
}

By writing a line starting with go:generate, the model for firestore will be automatically generated.

When used in SubCollection, add the argument -sub-collection.

If you want to use Meta information (such as CreatedAt and Version used in optimistic exclusive lock) together, you can invalidate it by removing the argument -disable-meta.

The format of the Meta structure is as follows.

type Meta struct {
	CreatedAt time.Time
	CreatedBy string
	UpdatedAt time.Time
	UpdatedBy string
	DeletedAt *time.Time
	DeletedBy string
	Version   int
}

Also, one element in the struct must have an element with firestore_key:"".
The type of this element must be string.
ID is automatically generated by setting firestore_key:"auto".

If you execute go generate in this state, the model will be generated in a file with the suffix _gen.go.

$ go generate

Unique constraint

If there is a tag called unique, a document for unique constraints will be generated in another collection called Unique.
Use when you do not want to allow duplicates such as phone numbers and email addresses.
The type of this element must be string.

Search diversity

The existence of a field called Indexes (map[string]bool type) enables n-gram search using xim.
Corresponding search is prefix/suffix/partial/exact match. (tag: prefix/suffix/like/equal)
Since xim uses only Unigram Bigram, it is prone to noise (eg, when 東京都 searches in 京都, it hits).

Search query

Task.Desc = "Hello, World!".

• Partial match search

param := &model.TaskSearchParam{
	Desc: model.NewQueryChainer().Filters("o, Wor", model.FilterTypeAddBiunigrams),
}

tasks, err := taskRepo.Search(ctx, param, nil)
if err != nil {
	// error handling
}

• Prefix match search

param := &model.TaskSearchParam{
	Desc: model.NewQueryChainer().Filters("Hell", model.FilterTypeAddPrefix),
}

tasks, err := taskRepo.Search(ctx, param, nil)
if err != nil {
	// error handling
}

• Suffix match search

param := &model.TaskSearchParam{
	Desc: model.NewQueryChainer().Filters("orld!", model.FilterTypeAddSuffix),
}

tasks, err := taskRepo.Search(ctx, param, nil)
if err != nil {
	// error handling
}

• Exact match search

chainer := model.NewQueryChainer
param := &model.TaskSearchParam{
	Desc: chainer().Filters("Hello, World!", model.FilterTypeAdd),
	Done: chainer().Filters(true, model.FilterTypeAddSomething), // Use Add Something when it is not a string.
}

tasks, err := taskRepo.Search(ctx, param, nil)
if err != nil {
	// error handling
}

Query builder

The code for the query builder called query_builder_gen.go is generated.

qb := model.NewQueryBuilder(taskRepo.GetCollection())
qb.GreaterThan("count", 3)
qb.LessThan("count", 8)

tasks, err := taskRepo.Search(ctx, nil, qb.Query())
if err != nil {
	// error handling
}

Strict update

Use a function called StrictUpdate.
By using this, firestore.Increment etc. can also be used.
Uniquely constrained fields are not available.

param := &model.TaskUpdateParam{
	Done:    false,
	Created: firestore.ServerTimestamp,
	Count:   firestore.Increment(1),
}
if err = taskRepo.StrictUpdate(ctx, id, param); err != nil {
	// error handling
}

Examples

Generated code example.

License

• Under the MIT License
• Copyright (C) 2021 go-generalize