How to use variables to DEFINE SCHEMAFULL tables with shell and curl

[koakh]

sometimes is to boring to DRY and always repeat tablename in our schemas, sometime we want to copy one table shema and tweak a little and move on, don't worry about forget to replace tablename in one FIELD by mistake, and waste time to figure it out, or just like to refactor tablename in one place only, why not, we have great tools like multi cursor, search and replace etc, but nothing like a VARIABLE

surrealdb, until that day do everything I try, even when we hack and push it a lot,
even doing stupid things, like forget that this is database, it's not, it's a lot more

$variables are powerful beats, but can't be used with DEFINE,
to solve this, one can use shell variables and move on

old way repeating a lot of table names like 15 times in this minimal example

SQL="
  # tag:define
  DEFINE TABLE tag SCHEMAFULL;
  DEFINE FIELD name ON tag TYPE string;
  DEFINE FIELD meta_data.* ON tag TYPE object;
  DEFINE FIELD meta_data.field ON tag TYPE string;
  DEFINE FIELD created_at ON tag TYPE datetime VALUE time::now();
  DEFINE INDEX idx_name ON tag COLUMNS name UNIQUE;
  
  # tag:create
  CREATE tag:red CONTENT { name: 'Red', meta_data: { field: 'value' } };
  CREATE tag:green CONTENT { name: 'Green', meta_data: { field: 'value' } };
  CREATE tag:blue CONTENT { name: 'Blue', meta_data: { field: 'value' } };
  CREATE tag:white CONTENT { name: 'White', meta_data: { field: 'value' } };
  CREATE tag:black CONTENT { name: 'Black', meta_data: { field: 'value' } };
  
  # tag:info
  INFO FOR TABLE tag;
"

curl -s POST \
  --header 'Content-Type: application/json' \
  --header 'NS: test' \
  --header 'DB: test' \
  --user "root:root" \
  --data "${SQL}" \
  http://localhost:8000/sql | jq

old school way

TABLE_NAME="tag"
SQL="
  # ${TABLE_NAME}:define
  DEFINE TABLE ${TABLE_NAME} SCHEMAFULL;
  DEFINE FIELD name ON ${TABLE_NAME} TYPE string;
  DEFINE FIELD meta_data.* ON ${TABLE_NAME} TYPE object;
  DEFINE FIELD meta_data.field ON ${TABLE_NAME} TYPE string;
  DEFINE FIELD created_at ON ${TABLE_NAME} TYPE datetime VALUE time::now();
  DEFINE INDEX idx_name ON ${TABLE_NAME} COLUMNS name UNIQUE;
  
  # ${TABLE_NAME}:create
  CREATE ${TABLE_NAME}:red CONTENT { name: 'Red', meta_data: { field: 'value' } };
  CREATE ${TABLE_NAME}:green CONTENT { name: 'Green', meta_data: { field: 'value' } };
  CREATE ${TABLE_NAME}:blue CONTENT { name: 'Blue', meta_data: { field: 'value' } };
  CREATE ${TABLE_NAME}:white CONTENT { name: 'White', meta_data: { field: 'value' } };
  CREATE ${TABLE_NAME}:black CONTENT { name: 'Black', meta_data: { field: 'value' } };
  
  # ${TABLE_NAME}:info
  INFO FOR TABLE ${TABLE_NAME};
  
curl -s POST \
  --header 'Content-Type: application/json' \
  --header 'NS: test' \
  --header 'DB: test' \
  --user "root:root" \
  --data "${SQL}" \
  http://localhost:8000/sql | jq
  "

I think you get the point

if we wanna clone it, just copy pasta and change to TABLE_NAME="person"

and add and remove some columns

we can use same approach for other ends, just use and abuse it