This is useful for build scripts to generate a UML diagram but a local go lang environment is not desirable.

Usage is like

docker run --rm planter postgres://user:password@database_host:dbname\?sslmode=disable

Txh for the code!

Hi, Thank you so very much for this very useful program!

Everything works well with one exception: all relations are duplicated and I cannot figure why. Here is a simple example from a much larger PostgreSQL database:

• I have a table TAG, with a PK tag_key
• I have a table CTG (= categories), with a PK ctg_key
• There is a relationship of many to 0/1 between TAG and CTG, defined using a foreign key tag_fk: each tag belongs or not to a category
• both table belong to a schema kuq

Here is the TAG DDL for TAG:

CREATE TABLE kuq.tag (
	tag_key int4 NOT NULL,
	ctg_key int4 NULL,
	tag varchar(60) NULL,
	CONSTRAINT tag_pk PRIMARY KEY (tag_key),
	CONSTRAINT tag_un UNIQUE (tag_key)
);
CREATE INDEX tag_tag_idx ON kuq.tag USING btree (tag);

ALTER TABLE kuq.tag ADD CONSTRAINT tag_fk FOREIGN KEY (ctg_key) REFERENCES kuq.ctg(ctg_key);

When I execute:

planter postgres://<connection>?sslmode=disable \
-s kuq    \
-t tag    \
-t ctg    \
-o test.uml

I get this PlantUML schema where the many to one link between TAG and CTG is duplicated:

And, in effect, the link is duplicated in the UML definition file:

@startuml

entity "ctg" {
  Tag Categories
  ..
  + ctg_key [PK]
  --
  ctg_key
  nam
  cmt
}

entity "tag" {
  Tags
  ..
  + tag_key [PK]
  --
  tag_key
  ctg_key
  tag
}

tag }-- ctg
tag }-- ctg

@enduml

What am I doing wrong? Best, Stephen

This pull request add a -x option that selects all tables not on the list. It uses the same logic passing a flag to FilterTables to determine the type of match.

• Add --title,-T option to add a diagram title
• Use Creole formatting to:
  • Make entity names bold
  • Make column names monospace
  • Make column descriptions italic
• Update test.sh to be more portable
  • Locate bash with /usr/bin/env
  • Specify server location with PGHOST fallback to localhost
  • Updated example with new output format

Hi, thanks for the great tool. I made a docker file which packages planter and plantuml in one docker image: https://github.com/kimmobrunfeldt/planter-docker. This makes using the tool a one-liner assuming docker is installed.

It exposes planter and plantuml directly, but also implements er command which allows selecting tables with a matching pattern. Useful for databases with tens or hundreds of tables.

exception planter postgres://user:123@localhost/adatabase?sslmode=disable 2019/11/30 16:16:48 failed to get columns of chat: failed to load table def: pq: column ad.adsrc does not exist

environment I am running PostgreSQL inside docker from the official docker image version 12.1-alpine I have discovered it local and tried it inside the container so docker is probably not the issue.

Database structure It also happens in the example from the repo.

It seems that CI has been failed for long time. This PR will fix it.

This PR includes following changes.

• Remove gom install to fix CI
  • planter no longer uses gom.
• Run test with Go version 1.14, 1.15 and 1.16 in CI.
  • According to Go Release policy, each major Go release is supported until there are two newer major releases. Now that 1.16 has been released, 1.16 and 1.15 are supported. However, I'm sure there are a lot of 1.14 users. So I added 1.14, 1.15 and 1.16 (and tip).

https://github.com/mattn/gom is a super awesome tool. However, today Go has Go's module system itself go mod. This pull request will remove Gomfile with respect.

Can we do revers feature? ;) PlantUml->PostgreSQL (like this one for MySQL https://github.com/grafov/plantuml2mysql) Would be awesome complete tool

I'd like to add test.yml to enable GitHub Actions for CI.

Look at https://github.com/hiroakis/planter/actions/runs/620985013. It shows the result of GitHub Actions CI on forked repository.

planter generates a UML like following.

@startuml
hide circle
skinparam linetype ortho

entity "bar" {
  + foo_id:bigint [PK][FK]
  --
  *bar_id:bigint
}

entity "foo" {
  + id:bigint [PK]
  --
  *name:text
  *created_at:timestamp with time zone
}

 bar ||-|| foo
@enduml

This pull request will show serial type if the column is associated with the sequence. Look at foo.id column.

@startuml
hide circle
skinparam linetype ortho

entity "bar" {
  + foo_id:bigint [PK][FK]
  --
  *bar_id:bigint
}

entity "foo" {
  + id:bigserial [PK]
  --
  *name:text
  *created_at:timestamp with time zone
}

 bar ||-|| foo
@enduml

Table definition

create table foo (id bigserial primary key, name text not null, created_at timestamp with time zone not null);
create table bar (foo_id bigint primary key, bar_id bigint not null, foreign key(foo_id) references foo (id));

   Column   |           Type           | Collation | Nullable |             Default
------------+--------------------------+-----------+----------+---------------------------------
 id         | bigint                   |           | not null | nextval('foo_id_seq'::regclass)
 name       | text                     |           | not null |
 created_at | timestamp with time zone |           | not null |
Indexes:
    "foo_pkey" PRIMARY KEY, btree (id)
Referenced by:
    TABLE "bar" CONSTRAINT "bar_foo_id_fkey" FOREIGN KEY (foo_id) REFERENCES foo(id)

 Column |  Type  | Collation | Nullable | Default
--------+--------+-----------+----------+---------
 foo_id | bigint |           | not null |
 bar_id | bigint |           | not null |
Indexes:
    "bar_pkey" PRIMARY KEY, btree (foo_id)
Foreign-key constraints:
    "bar_foo_id_fkey" FOREIGN KEY (foo_id) REFERENCES foo(id)

Hey!

I try to install the solution by command, without success.

go get -u github.com/achiku/planter
# github.com/alecthomas/kingpin
go-1/src/github.com/alecthomas/kingpin/values.go:142:25: undefined: str2duration.Str2Duration

Include partitioned tables, but ignore their partitions. Without this change, it would fail on database schemas involving partitioning.

Note the use of the row_to_json() function is to permit backward compatibility with PostgreSQL 9 and 10. The relispartition column does not exist until PostgreSQL 10 and the conparentid column does not exist until PostgreSQL 11. Directly referencing these columns, while simpler, would cause the query to fail in those earlier versions. While another approach would be to generate different SQL for different PostgreSQL versions, that seems like it would be more work. Note however that this use of JSON means it will no longer work on PostgreSQL 9.2 and earlier, since the necessary JSON support was introduced in 9.3. That's hopefully not a big issue, given 9.2 reached its end-of-support date almost 5 years ago.

For large schemas, it would be useful to generate the ERD without the list of columns to just show the relationships. Perhaps with an additional option to just show PKs and FKs Would this be considered?

Hi!

Great project, I'm really liking it! I just used to to run it on our live system and it's absolutely fantastic, specially for onboarding new folks :)

I just just wanted to flag with this ticket that I wasn't able to extract diagrams for tables that reference tables in other schemas. Not sure if the is a way or not? Nonetheless, still enjoying it a lot!

Best, Javier

Would like to see more info about columns

• Type
• If field[s] is indexed
• Comment

My personal rules:

• column definition: {name} {SQL type} [-- {comment}]
• Primary keys should be in the header of each table
• Foreign keys should be marked with + (will be displayed as green dot) with reference in comment
• Indexed fields should be marked as bold
• Comments starts with --
• Comments for table should be put in header of each table
• Relationships should be placed