| name | makrell-convert |
| description | Convert between Makrell formats and their conventional equivalents — JSON to/from MRON, HTML to/from MRML, CSV to/from MRTD. Use when migrating data or generating Makrell-format files from existing sources. |
| allowed-tools | Read Grep Glob Edit Write Bash |
Convert To and From Makrell Formats
You are helping convert data between Makrell formats and their conventional equivalents.
Format Pairs
| Makrell Format | Conventional Equivalent | Direction |
|---|
| MRON | JSON, TOML, YAML | Both ways |
| MRML | HTML, XML | Both ways |
| MRTD | CSV, TSV | Both ways |
JSON ↔ MRON
JSON to MRON rules
| JSON | MRON |
|---|
{ "key": "value" } | key "value" |
{ "key": 42 } | key 42 |
{ "key": true } | key true |
{ "key": null } | key null |
{ "key": [1, 2, 3] } | key [1 2 3] |
{ "key": { "a": 1 } } | key { a 1 } |
[{ "a": 1 }, { "a": 2 }] | [{ a 1 } { a 2 }] |
Key differences:
- Remove colons between keys and values
- Remove commas between entries and list elements
- Remove quotes around keys (unless they contain spaces or special characters)
- Keep quotes around string values
- Top-level MRON is implicitly an object — no outer
{}
Conversion procedure (JSON → MRON)
- If the top-level is an object, emit each key/value as a line (no outer braces)
- Replace
: between key and value with a space
- Remove all commas
- Keep string values quoted; leave numbers, booleans, null bare
- For nested objects, use
{ ... } with the same rules inside
- For arrays, use
[ ... ] with space-separated elements
Example
JSON:
{
"name": "Rena Holm",
"age": 29,
"active": true,
"tags": ["dev", "docs"],
"address": {
"city": "Bergen",
"country": "Norway"
}
}
MRON:
name "Rena Holm"
age 29
active true
tags ["dev" "docs"]
address {
city "Bergen"
country "Norway"
}
MRON → JSON procedure
- Wrap all content in
{ }
- Add
: after each key
- Add commas between entries and list elements
- Quote all keys
HTML ↔ MRML
HTML to MRML rules
| HTML | MRML |
|---|
<tag>content</tag> | {tag content} |
<tag attr="val">content</tag> | {tag [attr="val"] content} |
<tag /> | {tag} |
<tag><child /></tag> | {tag {child}} |
some <b>bold</b> text | some {b bold} text |
Key differences:
- Replace
<tag>...</tag> with {tag ...}
- Move attributes into
[...] after the tag name
- No closing tags — structure is expressed through nesting and
}
- Self-closing tags become
{tag} or {tag [attrs]}
- Inline elements work naturally:
text {b bold} more text
Conversion procedure (HTML → MRML)
- For each element, write
{tagname
- If it has attributes, write
[attr1="val1" attr2="val2"]
- Write child content (text and nested elements) after the tag/attributes
- Close with
}
- Void elements (
<br>, <hr>, <img>, <meta>, <link>, <input>) become {br}, {img [src="..."]}, etc.
Example
HTML:
<html>
<head>
<meta charset="utf-8">
<title>My Page</title>
</head>
<body>
<h1>Welcome</h1>
<p class="lead">This is <b>Makrell</b> markup.</p>
<ul>
<li>Item 1</li>
<li>Item 2</li>
</ul>
</body>
</html>
MRML:
{html
{head
{meta [charset="utf-8"]}
{title My Page}}
{body
{h1 Welcome}
{p [class="lead"] This is {b Makrell} markup.}
{ul
{li Item 1}
{li Item 2}}}}
MRML → HTML procedure
- For each
{tag ...}, emit <tag>...</tag>
- Move
[...] attributes into the opening tag
- Add closing tags for all non-void elements
- Self-close void elements
CSV ↔ MRTD
CSV to MRTD rules
| CSV | MRTD |
|---|
name,age,active | name age active (untyped) |
name,age,active | name:string age:int active:bool (typed) |
Ada,32,true | Ada 32 true |
"Rena Holm",29,false | "Rena Holm" 29 false |
Key differences:
- Replace commas with whitespace
- Optionally add type annotations to header:
:string, :int, :float, :bool
- Keep quotes around values that contain spaces
- Multiline rows use
( ) wrapping instead of quoting with embedded newlines
Conversion procedure (CSV → MRTD)
- Split the header row on commas; join with spaces
- Optionally infer types from data and add annotations (
:int, :float, :bool, :string)
- For each data row, split on commas and join with spaces
- Quote values that contain spaces or are not valid identifiers
- If a row is very long, consider
( ... ) multiline wrapping
Example
CSV:
name,age,city,active
Ada,32,Oslo,true
"Rena Holm",29,"Bergen sentrum",false
MRTD (typed):
name:string age:int city:string active:bool
Ada 32 Oslo true
"Rena Holm" 29 "Bergen sentrum" false
MRTD (untyped):
name age city active
Ada 32 Oslo true
"Rena Holm" 29 "Bergen sentrum" false
MRTD → CSV procedure
- Strip type annotations from header cells (e.g.,
name:string → name)
- Join header cells with commas
- Join each data row's cells with commas
- Quote any value containing commas (standard CSV quoting rules)
Guidelines
When converting:
- Ask about typing for MRTD — the user may want untyped or typed headers
- Preserve data fidelity — don't lose values, types, or structure during conversion
- Format readably — use consistent indentation in MRON and MRML output
- Handle edge cases — empty strings (
""), null values, nested arrays
- For large conversions, consider writing a script rather than converting manually
When the user says "convert this JSON to MRON" or similar:
- Read the source file
- Apply the conversion rules
- Write the result to the target file (or show it)
- Optionally validate with
makrellsharp check-mron/check-mrml/check-mrtd
