flower

flower takes a SQL dump file in input and produces a PlantUML ERD diagram of the schema.

Features:

• List all tables with fields
• Indicates field types (best effort, 'unstable')
• Indicates mandatory/nullable fields
• Indicates primary keys
• Displays link between tables (0,1,N ↔ 0,1,N)

Ideas:

• Optimise parsing (atm pretty naive, remove some un-needed for loops and models, etc)
• Verify SQL schema parsing works with several schema-dump versions
• Add more output formatters, like Mermaidjs and potentially others?
• Fix whitespaces around type fields

Quick start

bazel run //:flower -- parse --input $(pwd)/data/simple.sql --output $(pwd)/bin/out.plantuml --log-level warn

Tests

bazel test //...

Syntax

> bazel run //:flower -- parse --help
Run parse command

Usage:
  flower parse [flags]

Flags:
      --distance string       Distance: 'substring' or 'levenshtein' (optional)
  -h, --help                  help for parse
      --input string          Path to SQL file to read (required)
      --max-partition int     Maximum partition size (optional, default: 20) (default 20)
      --output string         Path to PlantUML file to write (including '.plantuml' extension) (required)
      --split-unconnected     Split the disconnected data graph into connected data graph, one per file. (optional, default: false)
      --weight-distance int   Weight coefficient to put over word distance weight, (optional, default: 1) (default 1)
      --weight-edge int       Weight coefficient to put between vertexes linked by edges (optional, default: 1) (default 1)

• --input is the input file path
• --output is the ouput file path, if several files are generated we will use the name to generate them
• --split-unconnected is a parameter to split an unconnected datagraph into connected parts, each file being a connected graph. The goal is to semantically split huge datagraph into several files independant from each others/
• --distance is an optional experimental parameter proposing to split a connected datagraph in several parts, using a distance for optimization computation
  • --max-partition is an optional parameter proposing to split the ERD in several parts of N elements, using a distance for optimization computation
  • --weight-edge is an optional parameter, to use alongside distance, representing the relative weight of the coefficient used for priorizing links
  • --weight-distance is an optional parameter, to use alongside distance, representing the relative weight of the coefficient used for priorizing the words distance

Warning

The partitioning using distances is very experimental and doesn't work very well. I suggest using a very high max-partition (20/30+) and experiment with several parition size and configuration weight values to see if the ouput is coherent with your data or not.

Example

Using file from ./data/simple.sql:

> flower parse --input data/simple.sql --output bin/out.plantuml

Docker build/compile

// build container
docker build . --tag flower:alpha
// run container
docker run -it <image_id>
// build
// make build-darwin-amd64 TODO?
// run
docker exec a7100df87d8b go run main.go -- parse --help

Licence

GNU GPL v3.0