Steedos Analytics

Steedos Analytics — Questions & Dashboards File Format Guide

Overview | 概述

Analytics in Steedos consists of two file types:

• Questions (.question.yml) — individual report/chart definitions with MBQL queries
• Dashboards (.dashboard.yml) — visual containers arranging multiple question cards in a grid

Workflow: Create questions first, then create dashboards that reference them.

基于 @steedos-labs/analytics 包。工作流程：先创建 question，再创建 dashboard 引用它们。

Part 1: Questions | 问题

File Location & Naming | 文件路径与命名

<package>/main/default/questions/<问题名称>.question.yml

Examples:

• questions/单位总数.question.yml
• questions/媒体类型统计.question.yml

File Structure | 文件结构

name: <问题名称>
created_at: <ISO8601 datetime>
creator_id: <MongoDB ObjectId>
database_id: 1
dataset_query:
  database: 1
  type: query          # "query" for MBQL, "native" for SQL
  query:
    source-table: <object_name>
    # optional: filter, aggregation, breakout, limit, order-by
display: <display_type>
enable_embedding: false
entity_id: <MongoDB ObjectId>   # same as id
id: <MongoDB ObjectId>
parameter_mappings: []
parameters: []
query_type: query
result_metadata:
  - ...               # column metadata list
table_id: <object_name>
updated_at: <ISO8601 datetime>
visualization_settings: '{}'

ID Generation | ID 生成

All id and entity_id values are 24-character MongoDB-style hex ObjectIds. entity_id and id are always the same value within one file. Generate unique IDs — never reuse across different questions or dashboards.

Example: 68a5805bb74676d1c8cb3614

dataset_query — MBQL Query | MBQL 查询结构

Count (scalar) | 计数

dataset_query:
  database: 1
  type: query
  query:
    source-table: media_org_info
    aggregation:
      - - count

Count with filter | 带过滤的计数

dataset_query:
  database: 1
  type: query
  query:
    source-table: media_journalist
    filter:
      - '='
      - - field
        - media_journalist.open
        - null
      - true
    aggregation:
      - - count

Group by field (breakout) | 按字段分组

dataset_query:
  database: 1
  type: query
  query:
    source-table: media_infor
    aggregation:
      - - count
    breakout:
      - - field
        - media_infor.type
        - null

Table list with limit | 列表查询

dataset_query:
  database: 1
  type: query
  query:
    source-table: media_org_info
    limit: 10000

Native SQL query | 原生 SQL 查询

dataset_query:
  database: 1
  type: native
  native:
    query: "SELECT status, COUNT(*) as cnt FROM contracts GROUP BY status"

Display Types | 显示类型

visualization_settings | 可视化设置

For simple questions, use '{}' (empty JSON string).

Common examples:

# Scalar — show specific column
visualization_settings: '{"table.cell_column":"count"}'

# Bar chart — specify axes
visualization_settings: >-
  {"graph.dimensions":["type"],"graph.metrics":["count"],"graph.show_values":true,"graph.x_axis.labels_enabled":false,"graph.y_axis.labels_enabled":false}

# Table — specify pivot and cell columns
visualization_settings: >-
  {"table.pivot_column":"status","table.cell_column":"count"}

result_metadata | 结果列元数据

For aggregation (count) | 聚合查询结果

result_metadata:
  - base_type: type/BigInteger
    display_name: 计数
    effective_type: type/BigInteger
    field_ref:
      - aggregation
      - 0
    name: count
    semantic_type: type/Quantity
    source: aggregation

For breakout + count | 分组+计数结果

result_metadata:
  - id: type
    name: type
    display_name: 媒体类型
    base_type: type/Text
    effective_type: type/Text
    semantic_type: null
    source: breakout
    visibility_type: normal
    table_id: media_infor
    field_ref:
      - field
      - media_infor.type
      - null
  - base_type: type/BigInteger
    display_name: 计数
    effective_type: type/BigInteger
    field_ref:
      - aggregation
      - 0
    name: count
    semantic_type: type/Quantity
    source: aggregation

For table list query | 列表查询结果

Each output column needs an entry. id uses <table>.<field> format:

result_metadata:
  - id: media_org_info.name
    name: name
    display_name: 单位名称
    base_type: type/Text
    effective_type: type/Text
    semantic_type: null
    field_ref:
      - field
      - media_org_info.name
      - null
    source: fields
    visibility_type: normal
    table_id: media_org_info

base_type mapping | 字段类型映射

Question Examples | 问题示例

Scalar — count of records

name: 单位总数
created_at: 2025-08-20T07:59:23.569Z
creator_id: 689c34199c7714afa7502547
database_id: 1
dataset_query:
  query:
    source-table: media_org_info
    aggregation:
      - - count
  type: query
  database: 1
display: scalar
entity_id: 68a5805bb74676d1c8cb3614
id: 68a5805bb74676d1c8cb3614
parameter_mappings: []
parameters: []
query_type: query
result_metadata:
  - base_type: type/BigInteger
    display_name: 计数
    effective_type: type/BigInteger
    field_ref:
      - aggregation
      - 0
    name: count
    semantic_type: type/Quantity
    source: aggregation
table_id: media_org_info
updated_at: 2025-08-20T07:59:23.569Z
visualization_settings: '{}'

Bar chart — grouped count

name: 媒体类型统计
created_at: 2025-08-20T08:06:22.163Z
creator_id: 689c34199c7714afa7502547
database_id: 1
dataset_query:
  database: 1
  type: query
  query:
    source-table: media_infor
    aggregation:
      - - count
    breakout:
      - - field
        - media_infor.type
        - null
display: bar
enable_embedding: false
entity_id: 68a581feb74676d1c8cb361e
id: 68a581feb74676d1c8cb361e
parameter_mappings: []
parameters: []
query_type: query
result_metadata:
  - id: type
    name: type
    display_name: 媒体类型
    base_type: type/Text
    effective_type: type/Text
    semantic_type: null
    source: breakout
    visibility_type: normal
    table_id: media_infor
    field_ref:
      - field
      - media_infor.type
      - null
  - base_type: type/BigInteger
    display_name: 计数
    effective_type: type/BigInteger
    field_ref:
      - aggregation
      - 0
    name: count
    semantic_type: type/Quantity
    source: aggregation
table_id: media_infor
updated_at: 2025-08-20T08:06:22.163Z
visualization_settings: >-
  {"graph.x_axis.labels_enabled":false,"graph.y_axis.labels_enabled":false,"graph.dimensions":["type"],"graph.metrics":["count"],"graph.show_values":true}

Question Key Rules | 问题关键规则

1. File name = question name: 单位总数.question.yml
2. id == entity_id: always the same 24-char hex value
3. table_id: the Steedos object name (e.g. media_org_info)
4. database_id and dataset_query.database: always 1
5. query_type: always "query" (even for native SQL)
6. parameter_mappings and parameters: use [] unless participating in dashboard filtering
7. visualization_settings: must be a JSON string (quoted), not an object
8. Generate fresh unique ObjectIds — never copy IDs from other questions

Part 2: Dashboards | 仪表盘

File Location & Naming | 文件路径与命名

<package>/main/default/dashboards/<仪表盘名称>.dashboard.yml

Example: dashboards/日常看板.dashboard.yml

File Structure | 文件结构

name: <仪表盘名称>           # ⚠️ Required, MUST NOT be omitted | 必填，不能省略
archived: 'false'
auto_apply_filters: 'true'
enable_embedding: 'false'
entity_id: <MongoDB ObjectId>   # same as id
id: <MongoDB ObjectId>
show_in_getting_started: 'false'
parameters:                     # dashboard-level filter parameters (can be [])
  - ...
cards:                          # list of card placements
  - ...

Note: archived, auto_apply_filters, enable_embedding, show_in_getting_started are string values 'true'/'false', not YAML booleans.

Grid Layout | 网格布局

The dashboard uses an 18-column grid. Cards are positioned with:

Common layout patterns:

cards — Card Placement | 卡片布局定义

Each entry in cards places one question onto the dashboard:

cards:
  - card_id: <question id>          # references a question's id
    col: 0
    row: 0
    size_x: 6
    size_y: 3
    entity_id: <unique ObjectId>    # unique per card placement
    id: <same as entity_id>
    series: []
    visualization_settings: '{}'
    parameter_mappings: []          # or list if this card uses dashboard filters

parameter_mappings for a card | 卡片的参数映射

When a dashboard parameter filters a card, add parameter_mappings:

parameter_mappings:
  - parameter_id: <dashboard parameter id>   # matches parameters[].id
    card_id: <question id>                   # same as card_id above
    target:
      - dimension
      - - field
        - <object_name>.<field_name>
        - null

parameters — Dashboard Filter Parameters | 仪表盘过滤参数

parameters:
  - name: 单位名称
    slug: '%E5%8D%95%E4%BD%8D%E5%90%8D%E7%A7%B0'   # URL-encoded name
    id: 9e2d576d                                      # 8-char hex id
    type: string/=
    sectionId: string
    values_source_type: card        # or "static-list" or "connected-field"
    values_source_config:
      card_id: <question id>        # question that provides the value list
      value_field:
        - field
        - <object_name>.<field_name>
        - null
    filteringParameters:            # optional: other param ids that filter this one
      - f82fb543

Parameter types | 参数类型

values_source_type options | 数据来源类型

slug | 参数 slug

slug is the URL-encoded version of name. In Python: urllib.parse.quote('单位名称') → '%E5%8D%95%E4%BD%8D%E5%90%8D%E7%A7%B0'

For ASCII names, slug equals the name itself (e.g. name: status → slug: status).

Dashboard Complete Example | 仪表盘完整示例

name: 日常看板
archived: 'false'
auto_apply_filters: 'true'
enable_embedding: 'false'
entity_id: 68a58065b74676d1c8cb3615
id: 68a58065b74676d1c8cb3615
show_in_getting_started: 'false'
parameters:
  - name: 单位名称
    slug: '%E5%8D%95%E4%BD%8D%E5%90%8D%E7%A7%B0'
    id: 9e2d576d
    type: string/=
    sectionId: string
    values_source_type: card
    values_source_config:
      card_id: 68e75b39ae8aaf0623f11fb1
      value_field:
        - field
        - media_org_info.name
        - null
    filteringParameters:
      - f82fb543
  - name: 单位层级
    slug: '%E5%8D%95%E4%BD%8D%E5%B1%82%E7%BA%A7'
    id: f82fb543
    type: string/=
    sectionId: string
    values_source_type: static-list
    values_source_config:
      values:
        - 一级单位
        - 二级单位
        - 三级单位
cards:
  - card_id: 68a5805bb74676d1c8cb3614   # 单位总数 (scalar)
    col: 0
    row: 0
    size_x: 6
    size_y: 3
    entity_id: 68a580ebb74676d1c8cb3618
    id: 68a580ebb74676d1c8cb3618
    series: []
    visualization_settings: '{}'
    parameter_mappings:
      - parameter_id: 9e2d576d
        card_id: 68a5805bb74676d1c8cb3614
        target:
          - dimension
          - - field
            - media_org_info.media_matrix
            - null
      - parameter_id: f82fb543
        card_id: 68a5805bb74676d1c8cb3614
        target:
          - dimension
          - - field
            - media_org_info.organizational_level
            - null
  - card_id: 68a581feb74676d1c8cb361e   # 媒体类型统计 (bar chart, full width)
    col: 0
    row: 3
    size_x: 18
    size_y: 5
    entity_id: 68a58206b74676d1c8cb361f
    id: 68a58206b74676d1c8cb361f
    series: []
    visualization_settings: '{}'
    parameter_mappings:
      - parameter_id: 9e2d576d
        card_id: 68a581feb74676d1c8cb361e
        target:
          - dimension
          - - field
            - media_infor.media_org_info
            - null
      - parameter_id: f82fb543
        card_id: 68a581feb74676d1c8cb361e
        target:
          - dimension
          - - field
            - media_infor.organizational_level
            - null

Dashboard Key Rules | 仪表盘关键规则

1. File name = dashboard name: 日常看板.dashboard.yml
2. id == entity_id for both the dashboard and each card placement
3. Each card placement gets its own unique ObjectId (different from the question's id)
4. card_id references the question's id field (defined in .question.yml)
5. Grid is 18 columns wide — col + size_x must not exceed 18
6. parameter_mappings on each card must reference parameter_id values defined in the top-level parameters list
7. Boolean-like fields (archived, auto_apply_filters, enable_embedding, show_in_getting_started) are strings: 'true'/'false'
8. visualization_settings on each card is a JSON string '{}'
9. series is always [] unless using combo/multi-series charts

Workflow | 创建步骤

1. Create questions first: write all .question.yml files, note their id values
2. Plan the grid layout: decide rows, cols, sizes for each card
3. Define parameters (if filtering is needed): assign 8-char hex ids
4. Write the dashboard file: reference question ids in card_id, add parameter_mappings for each card that participates in filtering