MeridianMERIDIAN

container

Structured data — JSON, CSV, query strings, key-value pairs

3 categories, 12 types: 10 a sibling in its category also accepts, 12 accepted by another category's rule, 1 that decompose into parts.

3 categories · 12 types

The 3 container categories

Each category's lookup value, against every other category's rule

CategoryTypesCasts toFor exampleWithin the groupAlso accepted by
object
7VARCHAR, JSON<p>Hello world</p>6 a sibling accepts, 1 misreadarray
array
4VARCHARapple,banana,cherry4 a sibling also acceptsobject
key_value
1VARCHARname=John&age=30&city=NewYor…—array, object

Where container categories overlap

Each type's lookup value, against every other category's rule

ValueWritten byAlso accepted by
name=John&age=30&city=NewYor…key_value.query_stringarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated; object.csv
John,30,trueobject.csvarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated
<p>Hello world</p>object.htmlarray.comma_separated → <p>Hello world</p>; array.pipe_separated → <p>Hello world</p>; array.semicolon_separated → <p>Hello world</p>; array.whitespace_separated → <p>Hello world</p>
{"user": "john", "age": 30}object.jsonarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated
[1, 2, 3]object.json_arrayarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated
(ROOT (S (NP (NN cat)) (VP (…object.s_expressionarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated
<root><name>John</name><age>…object.xmlarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated
name: John age: 30 active: t…object.yamlarray.comma_separated; array.pipe_separated; array.semicolon_separated; array.whitespace_separated
apple,banana,cherryarray.comma_separatedobject.csv
apple|banana|cherryarray.pipe_separatedobject.csv
apple;banana;cherryarray.semicolon_separatedobject.csv
apple banana cherryarray.whitespace_separatedobject.csv

Which container type is it?

7 of 12 container types in 2 categories accept it, and read it 2 ways.

TypeCategoryReads it as
HTML Content htmlobjectHello world
Comma-Separated Values (Simple) comma_separatedarray<p>Hello world</p>
Pipe-Separated Values pipe_separatedarray<p>Hello world</p>
Semicolon-Separated Values semicolon_separatedarray<p>Hello world</p>
Whitespace-Separated Values whitespace_separatedarray<p>Hello world</p>
CSV Record csvobject<p>Hello world</p>
XML Document xmlobject<p>Hello world</p>

What the engine extracts

  • object1 of 7tag_count · text_content

Use it

One value from each category, each inferred on its own

CLI · DuckDB
$ for v in '<p>Hello world</p>' 'apple,banana,cherry' 'name=John&age=30&city=NewYork'; do finetype infer -i "$v"; done

SELECT v, ft_infer(v) FROM (VALUES ('<p>Hello world</p>'), ('apple,banana,cherry'), ('name=John&age=30&city=NewYork')) t(v);

Type Registry