Pattern Matching

ThinkLang's match expression lets you branch on the shape and values of data, which is especially useful for handling AI results with varying confidence levels or categories.

Syntax

match subject {
  pattern => result
  pattern => result
  _ => default_result
}

Object Patterns

Match on object fields with comparison operators:

let sentiment = think<Confident<Sentiment>>("Analyze sentiment") with context: review

let response = match sentiment {
  { confidence: >= 0.9 } => "High confidence result"
  { confidence: >= 0.5 } => "Moderate confidence result"
  _ => "Low confidence — manual review needed"
}

print response

Supported comparison operators: >=, <=, ==, !=.

Literal Patterns

Match against exact values:

let category = infer<string>(message, "Classify as info, warning, or error")

let icon = match category {
  "error" => "[ERROR]"
  "warning" => "[WARN]"
  "info" => "[INFO]"
  _ => "[?]"
}

print icon

Multiple Field Patterns

Match on several fields at once:

type Analysis {
  risk: string
  score: float
}

let action = match analysis {
  { risk: "high", score: >= 0.8 } => "Immediate action required"
  { risk: "medium" } => "Monitor closely"
  { risk: "low" } => "No action needed"
  _ => "Unclassified"
}

All field conditions in a pattern must be satisfied for the arm to match.

Wildcard

The _ pattern matches anything. Place it last as a catch-all:

let result = match value {
  { status: "ok" } => "Success"
  _ => "Something else"
}

Match is an Expression

match returns a value, so you can assign its result to a variable or use it inline:

let label = match score {
  { value: >= 90 } => "A"
  { value: >= 80 } => "B"
  { value: >= 70 } => "C"
  _ => "F"
}

print label

Note for JS/TS Users

match is a language-level construct with no direct library equivalent. In JavaScript/TypeScript, use standard if/else or switch statements for the same branching logic:

if (sentiment.confidence >= 0.9) {
  console.log("High confidence result");
} else if (sentiment.confidence >= 0.5) {
  console.log("Moderate confidence result");
} else {
  console.log("Low confidence — manual review needed");
}