Tbls

tbls is a CI-Friendly tool for document a database, written in Go.

Key features of tbls are:

• Document a database automatically in GFM format. Output database schema in many formats.
• Single binary = CI-Friendly.
• Support many databases.
• Work as linter for database

Table of Contents

• Quick Start
• Install
• Getting Started
  • Document a database
  • Diff database and ( document or database )
  • Lint a database
  • Measure document coverage
  • Continuous Integration
• Configuration
  • Name
  • Description
  • Labels
  • DSN
    • Support Datasource
  • Document path
  • Document format
  • ER diagram
  • Filter tables
  • Lint
  • Comments
  • Relations
  • Dictionary
  • Personalized Templates
  • Required Version
• Expand environment variables
• Output formats
• Command arguments
• Environment variables

Quick Start

Document a database with one command.

$ tbls doc postgres://dbuser:[email protected]:5432/dbname

Using docker image.

$ docker run --rm -v $PWD:/work -w /work ghcr.io/k1low/tbls doc postgres://dbuser:[email protected]:5432/dbname

Install

deb:

$ export TBLS_VERSION=X.X.X
$ curl -o tbls.deb -L https://github.com/k1LoW/tbls/releases/download/v$TBLS_VERSION/tbls_$TBLS_VERSION-1_amd64.deb
$ dpkg -i tbls.deb

RPM:

$ export TBLS_VERSION=X.X.X
$ yum install https://github.com/k1LoW/tbls/releases/download/v$TBLS_VERSION/tbls_$TBLS_VERSION-1_amd64.rpm

Homebrew:

$ brew install k1LoW/tap/tbls

MacPorts:

$ sudo port install tbls

Manually:

Download binary from releases page

go install:

$ go install github.com/k1LoW/[email protected]

Docker:

$ docker pull ghcr.io/k1low/tbls:latest

On GitHub Actions:

# .github/workflows/doc.yml
name: Document

on:
  push:
    branches:
      - main

jobs:
  doc:
    runs-on: ubuntu-latest
    steps:
      -
        name: Checkout .tbls.yml
        uses: actions/[email protected]
      -
        uses: k1low/[email protected]
      -
        name: Run tbls for generate database document
        run: tbls doc

GitHub Actions for tbls is here.

Temporary:

$ source <(curl https://raw.githubusercontent.com/k1LoW/tbls/main/use)

$ curl -sL https://raw.githubusercontent.com/k1LoW/tbls/main/use > /tmp/use-tbls.tmp && . /tmp/use-tbls.tmp

Getting Started

Document a database

Add .tbls.yml ( or tbls.yml ) file to your repository.

# .tbls.yml

# DSN (Database Source Name) to connect database
dsn: postgres://dbuser:[email protected]:5432/dbname

# Path to generate document
# Default is `dbdoc`
docPath: doc/schema

Notice: If you are using a symbol such as # < in database password, URL-encode the password

Run tbls doc to analyzes the database and generate document in GitHub Friendly Markdown format.

$ tbls doc

Commit .tbls.yml and the document.

$ git add .tbls.yml doc/schema
$ git commit -m 'Add database document'
$ git push origin main

View the document on GitHub.

Sample document

Diff database and ( document or database )

Update database schema.

$ psql -U dbuser -d dbname -h hostname -p 5432 -c 'ALTER TABLE users ADD COLUMN phone_number varchar(15);'
Password for user dbuser:
ALTER TABLE

tbls diff shows the difference between database schema and generated document.

$ tbls diff
diff postgres://dbuser:*****@hostname:5432/dbname doc/schema/README.md
--- postgres://dbuser:*****@hostname:5432/dbname
+++ doc/schema/README.md
@@ -4,7 +4,7 @@

 | Name | Columns | Comment | Type |
 | ---- | ------- | ------- | ---- |
-| [users](users.md) | 7 | Users table | BASE TABLE |
+| [users](users.md) | 6 | Users table | BASE TABLE |
 | [user_options](user_options.md) | 4 | User options table | BASE TABLE |
 | [posts](posts.md) | 8 | Posts table | BASE TABLE |
 | [comments](comments.md) | 6 | Comments<br>Multi-line<br>table<br>comment | BASE TABLE |
diff postgres://dbuser:*****@hostname:5432/dbname doc/schema/users.md
--- postgres://dbuser:*****@hostname:5432/dbname
+++ doc/schema/users.md
@@ -14,7 +14,6 @@
 | email | varchar(355) |  | false |  |  | ex. [email protected] |
 | created | timestamp without time zone |  | false |  |  |  |
 | updated | timestamp without time zone |  | true |  |  |  |
-| phone_number | varchar(15) |  | true |  |  |  |

 ## Constraints

And, tbls diff support for diff checking between database and other database

$ tbls diff postgres://dbuser:*****@local:5432/dbname postgres://dbuser:*****@production:5432/dbname

Notice: tbls diff shows the difference Markdown documents only.

Re-generating database documentation

Existing documentation can re-generated using either --force or --rm-dist flag.

--force forces overwrite of the existing documents. It does not, however, remove files of removed tables.

$ tbls doc --force

--rm-dist removes files in docPath before generating the documents.

$ tbls doc --rm-dist

Lint a database

Add linting rule to .tbls.yml following

# .tbls.yml
lint:
  requireColumnComment:
    enabled: true
    exclude:
      - id
      - created
      - updated
  columnCount:
    enabled: true
    max: 10

Run tbls lint to check the database according to lint: rules

$ tbls lint
users.username: column comment required.
users.password: column comment required.
users.phone_number: column comment required.
posts.user_id: column comment required.
posts.title: column comment required.
posts.labels: column comment required.
comments.post_id: column comment required.
comment_stars.user_id: column comment required.
post_comments.comment: column comment required.
posts: too many columns. [12/10]
comments: too many columns. [11/10]

11 detected

Measure document coverage

tbls coverage measure and show document coverage ( description, comments ).

$ tbls coverage
Table                       Coverage
All tables                  16.1%
 public.users               20%
 public.user_options        37.5%
 public.posts               35.3%
 public.comments            14.3%
 public.comment_stars       0%
 public.logs                12.5%
 public.post_comments       87.5%
 public.post_comment_stars  0%
 public.CamelizeTable       0%
 public.hyphen-table        0%
 administrator.blogs        0%
 backup.blogs               0%
 backup.blog_options        0%
 time.bar                   0%
 time.hyphenated-table      0%
 time.referencing           0%

Continuous Integration

Continuous integration using tbls.

1. Commit the document using tbls doc.
2. Update the database schema in the development cycle.
3. Check for document updates by running tbls diff or tbls lint in CI.
4. Return to 1.

Example: Travis CI

# .travis.yml
language: go

install:
  - source <(curl -sL https://raw.githubusercontent.com/k1LoW/tbls/main/use)
script:
  - tbls diff
  - tbls lint

Tips: If your CI based on Debian/Ubuntu (/bin/sh -> dash), you can use following install command curl -sL https://raw.githubusercontent.com/k1LoW/tbls/main/use > use-tbls.tmp && . ./use-tbls.tmp && rm ./use-tbls.tmp

Tips: If the order of the columns does not match, you can use the --sort option.

Configuration

Name

name: is used to specify the database name of the document.

# .tbls.yml
name: mydatabase

Description

desc: is used to specify the database description.

# .tbls.yml
desc: This is My Database

Labels

labels: is used to label the database or tables.

label database:

# .tbls.yml
labels:
  - cmdb
  - analytics

label tables:

# .tbls.yml
comments:
  -
    table: users
    labels:
      - user
      - privacy data

label columns:

# .tbls.yml
comments:
  -
    table: users
    columnLabels:
      email:
        - secure
        - encrypted

DSN

dsn: (Data Source Name) is used to connect to database.

# .tbls.yml
dsn: my://dbuser:[email protected]:3306/dbname

Support Datasource

tbls support following databases/datasources.

PostgreSQL:

# .tbls.yml
dsn: postgres://dbuser:[email protected]:5432/dbname

# .tbls.yml
dsn: pg://dbuser:[email protected]:5432/dbname

When you want to disable SSL mode, add "?sslmode=disable" For example:

dsn: pg://dbuser:[email protected]:5432/dbname?sslmode=disable

MySQL:

# .tbls.yml
dsn: mysql://dbuser:[email protected]:3306/dbname

# .tbls.yml
dsn: my://dbuser:[email protected]:3306/dbname

When you want to hide AUTO_INCREMENT clause on the table definitions, add "?hide_auto_increment". For example:

dsn: my://dbuser:[email protected]:3306/dbname?hide_auto_increment

MariaDB:

# .tbls.yml
dsn: mariadb://dbuser:[email protected]:3306/dbname

# .tbls.yml
dsn: maria://dbuser:[email protected]:3306/dbname

SQLite:

# .tbls.yml
dsn: sqlite:///path/to/dbname.db

# .tbls.yml
dsn: sq:///path/to/dbname.db

BigQuery:

# .tbls.yml
dsn: bigquery://project-id/dataset-id?creds=/path/to/google_application_credentials.json

# .tbls.yml
dsn: bq://project-id/dataset-id?creds=/path/to/google_application_credentials.json

To set GOOGLE_APPLICATION_CREDENTIALS environment variable, you can use

1. export GOOGLE_APPLICATION_CREDENTIALS or export GOOGLE_APPLICATION_CREDENTIALS_JSON
2. Add query to DSN
   • ?google_application_credentials=/path/to/client_secrets.json
   • ?credentials=/path/to/client_secrets.json
   • ?creds=/path/to/client_secrets.json

Required permissions: bigquery.datasets.get bigquery.tables.get bigquery.tables.list

Also, you can use impersonate service account using environment variables below.

• GOOGLE_IMPERSONATE_SERVICE_ACCOUNT: Email of service account
• GOOGLE_IMPERSONATE_SERVICE_ACCOUNT_LIFETIME: You can use impersonate service account within this lifetime. This value must be readable from k1LoW/duration .

Cloud Spanner:

# .tbls.yml
dsn: spanner://project-id/instance-id/dbname?creds=/path/to/google_application_credentials.json

To set GOOGLE_APPLICATION_CREDENTIALS environment variable, you can use

1. export GOOGLE_APPLICATION_CREDENTIALS or export GOOGLE_APPLICATION_CREDENTIALS_JSON
2. Add query to DSN
   • ?google_application_credentials=/path/to/client_secrets.json
   • ?credentials=/path/to/client_secrets.json
   • ?creds=/path/to/client_secrets.json

Also, you can use impersonate service account using environment variables below.

• GOOGLE_IMPERSONATE_SERVICE_ACCOUNT: Email of service account
• GOOGLE_IMPERSONATE_SERVICE_ACCOUNT_LIFETIME: You can use impersonate service account within this lifetime. This value must be readable from k1LoW/duration .

Amazon Redshift:

# .tbls.yml
dsn: redshift://dbuser:[email protected]:5432/dbname

# .tbls.yml
dsn: rs://dbuser:[email protected]:5432/dbname

Microsoft SQL Server:

# .tbls.yml
dsn: mssql://DbUser:[email protected]:1433/testdb

# .tbls.yml
dsn: sqlserver://DbUser:[email protected]:1433/testdb

# .tbls.yml
dsn: ms://DbUser:[email protected]:1433/testdb

Amazon DynamoDB:

# .tbls.yml
dsn: dynamodb://us-west-2

# .tbls.yml
dsn: dynamo://ap-northeast-1?aws_access_key_id=XXXXXxxxxxxxXXXXXXX&aws_secret_access_key=XXXXXxxxxxxxXXXXXXX

To set AWS credentials, you can use

1. Use default credential provider chain of AWS SDK for Go
2. Add query to DSN
   • ?aws_access_key_id=XXXXXxxxxxxxXXXXXXX&aws_secret_access_key=XXXXXxxxxxxxXXXXXXX

Snowflake (Experimental):

---
# .tbls.yml
dsn: snowflake://user:[email protected]/mydb/myschema

See also: https://pkg.go.dev/github.com/snowflakedb/gosnowflake

MongoDB:

# .tbls.yml
dsn: mongodb://mongoadmin:[email protected]:27017/test

# .tbls.yml
dsn: mongodb://mongoadmin:[email protected]:27017/test?sampleSize=20

JSON:

The JSON file output by the tbls out -t json command can be read as a datasource.

---
# .tbls.yml
dsn: json://path/to/testdb.json

HTTP:

---
# .tbls.yml
dsn: https://hostname/path/to/testdb.json

---
# .tbls.yml
dsn:
  url: https://hostname/path/to/testdb.json
  headers:
    Authorization: token GITHUB_OAUTH_TOKEN

GitHub:

---
# .tbls.yml
dsn: github://k1LoW/tbls/sample/mysql/schema.json

Document path

tbls doc generates document in the directory specified by docPath:.

# .tbls.yml
# Default is `dbdoc`
docPath: doc/schema

Document format

format: is used to change the document format.

# .tbls.yml
format:
  # Adjust the column width of Markdown format table
  # Default is false
  adjust: true
  # Sort the order of table list and columns
  # Default is false
  sort: false
  # Display sequential numbers in table rows
  # Default is false
  number: false
  # The comments for each table in the Tables section of the index page will display the text up to the first double newline (first paragraph).
  # Default is false
  showOnlyFirstParagraph: true
  # Hide table columns without values
  # Default is false
  hideColumnsWithoutValues: true
  # It can be boolean or array
  # hideColumnsWithoutValues: ["Parents", "Children"]

ER diagram

tbls doc generate ER diagram images at the same time.

# .tbls.yml
er:
  # Skip generation of ER diagram
  # Default is false
  skip: false
  # ER diagram image format (`png`, `jpg`, `svg`, `mermaid`)
  # Default is `svg`
  format: svg
  # Add table/column comment to ER diagram
  # Default is false
  comment: true
  # Hide relation definition from ER diagram
  # Default is false
  hideDef: true
  # Show column settings in ER diagram. If this section is not set, all columns will be displayed (default).
  showColumnTypes:
    # Show related columns
    related: true
    # Show primary key columns
    primary: true
  # Distance between tables that display relations in the ER
  # Default is 1
  distance: 2
  # ER diagram (png/jpg) font (font name, font file, font path or keyword)
  # Default is "" ( system default )
  font: M+

It is also possible to personalize the output by providing your own templates. See the Personalized Templates section below.

Lint

tbls lint work as linter for database.

# .tbls.yml
lint:
  # require table comment
  requireTableComment:
    enabled: true
    # all commented, or all uncommented.
    allOrNothing: false
  # require column comment
  requireColumnComment:
    enabled: true
    # all commented, or all uncommented.
    allOrNothing: true
    # exclude columns from warnings
    exclude:
      - id
      - created_at
      - updated_at
    # exclude tables from warnings
    excludeTables:
      - logs
      - comment_stars
  # require index comment
  requireIndexComment:
    enabled: true
    # all commented, or all uncommented.
    allOrNothing: false
    # exclude indexes from warnings
    exclude:
      - user_id_idx
    # exclude tables from warnings
    excludeTables:
      - logs
      - comment_stars
  # require constraint comment
  requireConstraintComment:
    enabled: true
    # all commented, or all uncommented.
    allOrNothing: false
    # exclude constrains from warnings
    exclude:
      - unique_user_name
    # exclude tables from warnings
    excludeTables:
      - logs
      - comment_stars
  # require trigger comment
  requireTriggerComment:
    enabled: true
    # all commented, or all uncommented.
    allOrNothing: false
    # exclude triggers from warnings
    exclude:
      - update_count
    # exclude tables from warnings
    excludeTables:
      - logs
      - comment_stars
  # require table labels
  requireTableLabels:
    enabled: true
    # all commented, or all uncommented.
    allOrNothing: false
    # exclude tables from warnings
    exclude:
      - logs
  # find a table that has no relation
  unrelatedTable:
    enabled: true
    # all related, or all unrelated.
    allOrNothing: true
    # exclude tables from warnings
    exclude:
      - logs
  # check max column count
  columnCount:
    enabled: true
    max: 10
    # exclude tables from warnings
    exclude:
      - user_options
  # require columns
  requireColumns:
    enabled: true
    columns:
      -
        name: created
      -
        name: updated
        exclude:
          - logs
          - CamelizeTable
  # check duplicate relations
  duplicateRelations:
    enabled: true
  # check if the foreign key columns have an index
  requireForeignKeyIndex:
    enabled: true
    exclude:
      - comments.user_id
  # checks if labels are in BigQuery style ( https://cloud.google.com/resource-manager/docs/creating-managing-labels#requirements )
  labelStyleBigQuery:
    enabled: true
    exclude:
      - schema_migrations

Filter tables

include: and exclude: are used to filter target tables from tbls *.

# .tbls.yml
include:
  - some_prefix_*
exclude:
  - some_prefix_logs
  - CamelizeTable

lintExclude: is used to exclude tables from tbls lint.

# .tbls.yml
lintExclude:
  - CamelizeTable

Filter logic

1. Add tables from include
2. Remove tables from exclude
   • Check for include/exclude overlaps
   • If include is more specific than exclude (i.e. schema.MyTable > schema.* or schema.MyT* > schema.* ), include the table(s). If include is equally or less specific than exclude, exclude wins.
3. Result

Comments

comments: is used to add table/column comment to database document without ALTER TABLE.

For example, you can add comment about VIEW TABLE or SQLite tables/columns.

# .tbls.yml
comments:
  -
    table: users
    # table comment
    tableComment: Users table
    # column comments
    columnComments:
      email: Email address as login id. ex. [email protected]
    # labels for tables
    labels:
      - privary data
      - backup:true
  -
    table: post_comments
    tableComment: post and comments View table
    columnComments:
      id: comments.id
      title: posts.title
      post_user: posts.users.username
      comment_user: comments.users.username
      created: comments.created
      updated: comments.updated
  -
    table: posts
    # index comments
    indexComments:
      posts_user_id_idx: user.id index
    # constraints comments
    constraintComments:
      posts_id_pk: PRIMARY KEY
    # triggers comments
    triggerComments:
      update_posts_updated: Update updated when posts update

Relations

relations: is used to add or override table relation to database document without FOREIGN KEY.

You can create ER diagrams with relations without having foreign key constraints.

relations:
  -
    table: logs
    columns:
      - user_id
    parentTable: users
    parentColumns:
      - id
    # Relation definition
    # Default is `Additional Relation`
    def: logs->users
  -
    table: logs
    columns:
      - post_id
    parentTable: posts
    parentColumns:
      - id
  -
    table: logs
    columns:
      - comment_id
    parentTable: comments
    parentColumns:
      - id
  -
    table: logs
    columns:
      - comment_star_id
    parentTable: comment_stars
    parentColumns:
      - id

Override relations

If you want to override an existing relation, set the override: to true.

relations:
  -
    table: posts
    columns:
      - user_id
    cardinality: zero or one
    parentTable: users
    parentColumns:
      - id
    parentCardinality: one or more
    override: true
    def: posts->users

Automatically detect relations

detectVirtualRelations: if enabled, automatically detect relations from table and column names.

detectVirtualRelations:
  enabled: true
  strategy: default

default strategy:

detectVirtualRelations:
  enabled: true
  strategy: default

• some_table.user_id -> users.id
• some_table.post_id -> posts.id

singularTableName strategy:

detectVirtualRelations:
  enabled: true
  strategy: singularTableName

• some_table.user_id -> user.id
• some_table.post_id -> post.id

Dictionary

dict: is used to replace title/table header of database document

# .tbls.yml
---
dict:
  Tables: 
  Description: 
  Columns: 
  Indexes: INDEX
  Constraints: 
  Triggers: 
  Relations: ER
  Name: 
  Comment: 
  Type: 
  Default: 
  Children: 
  Parents: 
  Definition: 
  Table Definition: 

Personalized Templates

It is possible to provide your own templates to personalize the documentation generated by tbls by adding a templates: section to your configuration. For example:

templates:
  dot:
    schema: 'templates/schema.dot.tmpl'
    table: 'templates/table.dot.tmpl'
  puml:
    schema: 'templates/schema.puml.tmpl'
    table: 'templates/table.puml.tmpl'
  md:
    index: 'templates/index.md.tmpl'
    table: 'templates/table.md.tmpl'

A good starting point to design your own template is to modify a copy the default ones for Dot, PlantUML and markdown.

Required Version

The requiredVersion setting defines a version constraint string. This defines which version of tbls can be used in the configuration.

requiredVersion: '>= 1.42, < 2'

Expand environment variables

All configuration values can be set by expanding the environment variables.

# .tbls.yml
dsn: my://${MYSQL_USER}:${MYSQL_PASSWORD}@hostname:3306/${MYSQL_DATABASE}

Output formats

tbls out output in various formats.

Markdown:

$ tbls out -t md -o schema.md

DOT:

$ tbls out -t dot -o schema.dot

PlantUML:

$ tbls out -t plantuml -o schema.puml

Mermaid:

$ tbls out -t mermaid -o schema.mmd

Image (svg, png, jpg):

$ tbls out -t svg --table users --distance 2 -o users.svg

JSON:

$ tbls out -t json -o schema.json

Tips: tbls doc can load schema.json as DSN.

$ tbls doc json:///path/to/schema.json

YAML:

$ tbls out -t yaml -o schema.yaml

Excel:

$ tbls out -t xlsx -o schema.xlsx

.tbls.yml:

$ tbls out -t config -o .tbls.new.yml

Command arguments

tbls subcommands ( doc,diff, etc) accepts arguments and options

$ tbls doc my://root:[email protected]:3306/testdb doc/schema

You can check available arguments and options using tbls help [COMMAND].

$ tbls help doc
'tbls doc' analyzes a database and generate document in GitHub Friendly Markdown format.

Usage:
  tbls doc [DSN] [DOC_PATH] [flags]

Flags:
  -j, --adjust-table       adjust column width of table
  -b, --base-url string    base url for links
  -c, --config string      config file path
  -t, --er-format string   ER diagrams output format (png, svg, jpg, mermaid). default: svg
  -f, --force              force
  -h, --help               help for doc
      --rm-dist            remove files in docPath before generating documents
      --sort               sort
      --when string        command execute condition
      --without-er         no generate ER diagrams

Output Schema data

tbls doc also output schema data ( schema.json ) to same directory as the generated schema document.

To disable output of schema data, set disableOutputSchema: to true in .tbls.yml file.

Environment variables

tbls accepts environment variables TBLS_DSN and TBLS_DOC_PATH

$ env TBLS_DSN=my://root:[email protected]:3306/testdb TBLS_DOC_PATH=doc/schema tbls doc