🥮 mooncakes.io
    Sign in

    jeeflow-persist

    jeeflow dynamic table persist (DynamicTableWriter + PersistPostInterceptor + Meta)

    Download zip
    Author
    Version
    0.1.30
    License
    Apache-2.0
    Repository
    Last updated
    15 days ago
    Downloads
    341

    #jeeflow-persist

    Workflow → business-table persistence for jeeflow — ARCHIVE / SYNC with field permissions.

    mooncakes License

    PersistPostInterceptor copies form data out of running workflow instances into your own business tables, driven entirely by the process definition — the role jeeflow-persist plays for the MoonBit build of jeeflow-moon.

    #Quickstart

    let provider = @persist.InMemoryMetaProvider::new()
    let meta = @persist.TableMeta::make("expense", "报销")
    meta.add_field(@persist.FieldMeta::make("note", "VARCHAR(200)", "备注"))
    provider.register(meta)

    let interceptor = @persist.PersistPostInterceptor::make(provider, my_writer)
    ctx.register_interceptor(interceptor.as_interceptor())   // order = 100 (post)

    Process definition drives everything:

    {"persistMode": "ARCHIVE", "relTableName": "expense"}

    • ARCHIVE — at the end node, on FINISHED + agree (submitType=1), insert the full f_* form once; idempotent by the process_instance_id key.
    • SYNC — insert at start → update at each task (business fields filtered by the target node's field.PERMISSION_*: readonly/hidden never write through) → final-state update at the end; a {nodeId}_{state} status column is probed automatically.
    • Safety — dynamic table names are validated (alnum+underscore, wf_/sys_/… prefixes refused) and failures are loud, never silent.

    #License

    Apache-2.0 — part of jeeflow-moon.

    DynamicTableWriter

    source
    pub(open) trait DynamicTableWriter {
      async fn insert(Self, String, Map[String, Json]) -> Int64 raise 
    JeeflowError
      async fn update(Self, String, Int64, Map[String, Json]) -> Unit raise 
    JeeflowError
      async fn exists(Self, String, Int64) -> Bool raise 
    JeeflowError
      async fn exists_by_key(Self, String, String, Int64) -> Bool raise 
    JeeflowError
      async fn update_by_key(Self, String, Map[String, Json], String, Int64) -> Unit raise 
    JeeflowError
      async fn columns(Self, String, Array[String]) -> Array[String] raise 
    JeeflowError
    }

    DynamicTableWriter trait（MoonBit：大接口 trait 形态）

    issues/149：六法全部 async —— 本栈写真库只有 moonmysql 的 async 连接这一条路，同步形状 发不出查询（145-6 记的机理）。async 由 trait 声明承载：impl … with fn 处不重复写 async（实测写了判 parse error），⇒ InMemoryTableWriter 六支 impl 与 &DynamicTableWriter 的六处调用点都不用改；唯一要跟着改的是同步函数里调 writer 的 put_state_field。 同形参照＝core/spi/repository.mbt 的 ProcessRepository（async trait ＋ repository-mysql 真库实现）。 ⚠ 列探测（columns）归属裁决＝留在 writer：参考实现 java persist/DynamicTableWriter.java:20 的 filterColumns 就挂在 writer 上，而 IDynamicMetaProvider 管字段权限/显示元数据——两者在 java 本就并存，故不并入 DynamicMetaProviderFns（146 缺口二），也不留"同一件事两个端口"。

    DynamicMetaProviderFns

    source
    pub(all) struct DynamicMetaProviderFns {
      load_table_meta : async (String) -> TableMeta? raise 
    JeeflowError
    }

    IDynamicMetaProvider 的 MoonBit 同位（issues/146 缺口二）。 java 侧这个端口就定义在 jeeflow-persist/meta（不在 core），消费点是 MetaTableWriter/Reader ⇒ 本栈同位同样落在 persist：归档/SYNC 的写侧与回显读侧共用一份元数据（storageType 语义两侧一致）。 load_table_meta 是 async：集成方的元数据来源是库（mldong dev_schema / information_schema）， 同步形状就会重演缺口一——启动期快照、改表结构引擎看不见。 未注入 ⇒ 引擎行为与既往一致（InMemory 档照常，缺省不启用外部元数据）。

    FieldMeta

    source
    pub(all) struct FieldMeta {
      column_name : String
      column_type : String
      display_name : String
      nullable : Bool
      default_value : String?
      permission : Int
    }

    字段元数据（动态表列）

    FieldMeta::is_editable

    source
    fn FieldMeta::is_editable(self : FieldMeta) -> Bool

    FieldMeta::is_visible

    source
    fn FieldMeta::is_visible(self : FieldMeta) -> Bool

    FieldMeta::make

    source
    fn FieldMeta::make(column_name : String, column_type : String, display_name : String) -> FieldMeta

    InMemoryMetaProvider

    source
    pub(all) struct InMemoryMetaProvider {
      tables : Map[String, TableMeta]
    }

    IDynamicMetaProvider 内存实现

    InMemoryMetaProvider::as_provider

    source

    内存档 → 端口（纯内存读 ⇒ 经 @spi.async_of_sync 桥，注册点零警告）

    InMemoryMetaProvider::get_table_meta

    source
    fn InMemoryMetaProvider::get_table_meta(self : InMemoryMetaProvider, table_name : String) -> TableMeta?

    InMemoryMetaProvider::list_table_names

    source
    fn InMemoryMetaProvider::list_table_names(self : InMemoryMetaProvider) -> Array[String]

    InMemoryMetaProvider::new

    source

    InMemoryMetaProvider::register

    source
    fn InMemoryMetaProvider::register(self : InMemoryMetaProvider, meta : TableMeta) -> Unit

    InMemoryTableWriter

    source
    pub(all) struct InMemoryTableWriter {
      id_counter : Int64
      tables : Map[String, Map[Int64, Map[String, Json]]]
    }

    InMemoryTableWriter::columns

    source
    async fn InMemoryTableWriter::columns(self : InMemoryTableWriter, _table_name : String, candidates : Array[String]) -> Array[String] raise 
    JeeflowError

    InMemoryTableWriter::exists

    source
    async fn InMemoryTableWriter::exists(self : InMemoryTableWriter, table_name : String, id : Int64) -> Bool raise 
    JeeflowError

    InMemoryTableWriter::exists_by_key

    source
    async fn InMemoryTableWriter::exists_by_key(self : InMemoryTableWriter, table_name : String, key : String, value : Int64) -> Bool raise 
    JeeflowError

    InMemoryTableWriter::get_row

    source
    fn InMemoryTableWriter::get_row(self : InMemoryTableWriter, table_name : String, id : Int64) -> Map[String, Json]?

    InMemoryTableWriter::get_row_by_key

    source
    fn InMemoryTableWriter::get_row_by_key(self : InMemoryTableWriter, table_name : String, key : String, value : Int64) -> Map[String, Json]?

    InMemoryTableWriter::get_table_rows

    source
    fn InMemoryTableWriter::get_table_rows(self : InMemoryTableWriter, table_name : String) -> Array[Map[String, Json]]

    InMemoryTableWriter::insert

    source
    async fn InMemoryTableWriter::insert(self : InMemoryTableWriter, table_name : String, data : Map[String, Json]) -> Int64 raise 
    JeeflowError

    InMemoryTableWriter::new

    source

    InMemoryTableWriter::update

    source
    async fn InMemoryTableWriter::update(self : InMemoryTableWriter, table_name : String, id : Int64, data : Map[String, Json]) -> Unit raise 
    JeeflowError

    InMemoryTableWriter::update_by_key

    source
    async fn InMemoryTableWriter::update_by_key(self : InMemoryTableWriter, table_name : String, data : Map[String, Json], key : String, value : Int64) -> Unit raise 
    JeeflowError

    PersistMode

    source
    pub enum PersistMode {
      Archive
      Sync
    } derive(Eq, 
    Debug    )

    持久化模式

    PersistMode::equal

    source
    fn PersistMode::equal(PersistMode, PersistMode) -> Bool
    automatically derived

    PersistMode::from_str

    source
    fn PersistMode::from_str(s : String) -> PersistMode?

    PersistMode::not_equal

    source
    fn PersistMode::not_equal(x : PersistMode, y : PersistMode) -> Bool

    PersistMode::to_repr

    source
    automatically derived

    PersistPostInterceptor

    source
    pub(all) struct PersistPostInterceptor {
      meta : DynamicMetaProviderFns
      table_writer : &DynamicTableWriter
    }

    PersistPostInterceptor::as_interceptor

    source
    装配为引擎拦截器（order=100，后置）

    PersistPostInterceptor::intercept

    source

    PersistPostInterceptor::make

    source

    PersistPostInterceptor::make_from_provider

    source
    端口档装配（issues/146 缺口二）：集成方自备元数据来源（dev_schema/库内省）走这一支

    TableMeta

    source
    pub(all) struct TableMeta {
      table_name : String
      display_name : String
      fields : Array[FieldMeta]
    }

    TableMeta::add_field

    source
    fn TableMeta::add_field(self : TableMeta, field : FieldMeta) -> Unit

    TableMeta::get_field

    source
    fn TableMeta::get_field(self : TableMeta, column_name : String) -> FieldMeta?

    TableMeta::make

    source
    fn TableMeta::make(table_name : String, display_name : String) -> TableMeta

    fill_system_fields

    source
    fn fill_system_fields(data : Map[String, Json], operator : String, is_insert : Bool) -> Unit

    系统字段填充（时间走 Clock 注入口，D-M0-3）

    is_table_name_safe

    source
    fn is_table_name_safe(name : String) -> Bool

    动态表名安全检查：仅字母数字下划线 + 保留前缀拒绝

    Source Files

    Powered by MoonBit

    Site sourceReport issuePackagesBuild queueSkillsStatistics

    © 2026 mooncakes.io