AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents
TypeScript ai-agentai-agent-toolsai-agentsai-agents-frameworkmcpmcp-servermcp-toolsmcpsn8n-alternativeno-code-automationworkflowworkflow-automationworkflows
// Extracts `{{ ... }}` tokens from a string using brace counting rather than a
// regex. A naive /\{\{(.*?)\}\}/ stops at the first `}}` it sees, so expressions
// with nested braces (object literals, function calls) or string literals that
// contain `}}` get truncated. Every consumer that parses mentions (the engine's
// props-resolver and the flow-version expression rewriter) MUST share this so
// they tokenize identically — divergence silently breaks variable resolution.
//
// `inner` is the raw content between the braces, NOT trimmed: the rewriter
// reconstructs `{{<inner>}}` and trimming would drop a space, collapsing
// `{{ {a:1} }}` into `{{{a:1}}}` which re-tokenizes incorrectly. Callers that use
// `inner` as a variable name (the resolver) should trim it themselves.
export function extractMustacheTokens(input: string): MustacheToken[] {
const results: MustacheToken[] = []
let i = 0
while (i < input.length - 1) {
if (input[i] === '{' && input[i + 1] === '{') {
const start = i
let depth = 1
i += 2
while (i < input.length - 1 && depth > 0) {
if (input[i] === '{' && input[i + 1] === '{') {
depth++; i += 2
}
else if (input[i] === '}' && input[i + 1] === '}') {
depth--; i += 2
}
else {
i++
}
}
if (depth === 0) {
const token = input.slice(start, i)
const inner = token.slice(2, -2)
results.push({ token, inner, index: start })
}
}
else {
i++
}
}
return results
}
export type MustacheToken = {
token: string
inner: string
index: number
}