shape definiton being overwritten if column name === field name

[bradzacher]

I have the following query:

query {
  User(id:39) {
    fieldA {
      created_by {
        id
      }
    }
  }
}

The problem that I have is that in the definition: fieldA's underlying column name is created_by, which matches the field name of the relationship field for created_by.

I've analyzed the generated query, and it queries perfectly.
The problem is with the shape generation..

The problem occurs in this switch here:
https://github.com/stems/join-monster/blob/master/src/define-object-shape.js#L17-L47

The switch assumes everything has a completely unique name, hence everything is writing into fieldDefinition[child.fieldName].

the generate SQL AST that runs through this function looks like this:

[
    {
        "type": "column",
        "name": "id",
        "fieldName": "id",
        "as": "id"
    },
    {
        "args": {},
        "type": "table",
        "name": "user",
        "as": "created_by",
        "children": [
            {
                "type": "column",
                "name": "user_id",
                "fieldName": "user_id",
                "as": "user_id"
            },
            {
                "args": {},
                "type": "column",
                "name": "user_id",
                "fieldName": "id",
                "as": "id"
            }
        ],
        "fieldName": "created_by", /* created_by */
        "grabMany": false
    },
    {
        "type": "column",
        "name": "created_by",
        "fieldName": "created_by", /* created_by */
        "as": "created_by"
    }
]

so when the switch iterates over that AST, the shape definition for created_by with type === "table" is overwritten by the the definition for created_by with type === "column".

debug output:

  join-monster
  join-monster SQL_AST
  join-monster  { args: { id: 39 },
  type: 'table',
  name: 'user',
  as: 'Users',
  children:
   [ { type: 'column',
       name: 'user_id',
       fieldName: 'user_id',
       as: 'user_id' },
     SQLASTNode {
       args: {},
       type: 'table',
       name: 'some_table',
       as: 'submitted_',
       children:
        [ { type: 'column', name: 'id', fieldName: 'id', as: 'id' },
          SQLASTNode {
            args: {},
            type: 'table',
            name: 'user',
            as: 'created_by',
            children:
             [ { type: 'column',
                 name: 'user_id',
                 fieldName: 'user_id',
                 as: 'user_id' },
               SQLASTNode {
                 args: {},
                 type: 'column',
                 name: 'user_id',
                 fieldName: 'id',
                 as: 'id' } ],
            fieldName: 'created_by',
            grabMany: false,
            sqlJoin: [Function] } ],
       fieldName: 'submitted_some_table',
       grabMany: true,
       sqlBatch:
        { thisKey:
           { type: 'column',
             name: 'created_by',
             fieldName: 'created_by',
             as: 'created_by' },
          parentKey:
           { type: 'column',
             name: 'user_id',
             fieldName: 'user_id',
             as: 'user_id' } } } ],
  fieldName: 'Users',
  grabMany: true,
  where: [Function] } +0ms
  join-monster
  join-monster SQL
  join-monster  SELECT
  `Users`.`user_id` AS `user_id`
FROM user `Users`
WHERE `Users`.`user_id` = 39
   AND `Users`.`active` = 1 +6ms
  join-monster
  join-monster SHAPE_DEFINITION
  join-monster  [ { user_id: 'user_id' } ] +1ms
  join-monster
  join-monster RAW DATA
  join-monster  [ TextRow { user_id: 39 } ] +22ms
  join-monster 1 rows... +0ms
  join-monster
  join-monster SHAPED_DATA
  join-monster  [ { user_id: 39 } ] +2ms
  join-monster
  join-monster SQL_AST
  join-monster  SQLASTNode {
  args: {},
  type: 'table',
  name: 'some_table',
  as: 'submitted_',
  children:
   [ { type: 'column', name: 'id', fieldName: 'id', as: 'id' },
     SQLASTNode {
       args: {},
       type: 'table',
       name: 'user',
       as: 'created_by',
       children:
        [ { type: 'column',
            name: 'user_id',
            fieldName: 'user_id',
            as: 'user_id' },
          SQLASTNode {
            args: {},
            type: 'column',
            name: 'user_id',
            fieldName: 'id',
            as: 'id' } ],
       fieldName: 'created_by',
       grabMany: false,
       sqlJoin: [Function] },
     { type: 'column',
       name: 'created_by',
       fieldName: 'created_by',
       as: 'created_by' } ],
  fieldName: 'some_table',
  grabMany: true,
  sqlBatch:
   { thisKey:
      { type: 'column',
        name: 'created_by',
        fieldName: 'created_by',
        as: 'created_by' },
     parentKey:
      { type: 'column',
        name: 'user_id',
        fieldName: 'user_id',
        as: 'user_id' } } } +4ms
  join-monster
  join-monster SQL
  join-monster  SELECT
  `submitted_`.`id` AS `id`,
  `created_by`.`user_id` AS `created_by__user_id`,
  `created_by`.`user_id` AS `created_by__id`,
  `submitted_`.`created_by` AS `created_by`
FROM some_table `submitted_`
LEFT JOIN user `created_by` ON `submitted_`.created_by = `created_by`.user_id
WHERE `submitted_`.`created_by` IN (39) +1ms
  join-monster
  join-monster SHAPE_DEFINITION
  join-monster  [ { id: 'id', created_by: 'created_by' } ] +0ms
  join-monster
  join-monster RAW DATA
  join-monster  [ TextRow {
    id: 1459,
    created_by__user_id: 39,
    created_by__id: 39,
    created_by: 39 },
  TextRow {
    id: 1464,
    created_by__user_id: 39,
    created_by__id: 39,
    created_by: 39 } ] +4ms
  join-monster 2 rows... +0ms
  join-monster
  join-monster SHAPED_DATA
  join-monster  [ { id: 1459, created_by: 39 }, { id: 1464, created_by: 39 } ] +2ms

[nicoabie]

Can you provide a https://github.com/join-monster/join-monster-sscce ? I believe this could still be an edge case.