toolcall.go

package main

import (
	"regexp"
	"strings"
)

const zeroWidthSpace = "\u200B"

// reBrokenClose matches: ]</ + (zero-width-space) + whitespace + <tool_calls>
// e.g. "]\u200B\n<tool_calls>" — model tries to close but inserts zero-width space
var reBrokenClose = regexp.MustCompile(
	`\]</` + zeroWidthSpace + `\s*` + regexp.QuoteMeta("<tool_calls>"),
)

// reBrokenWhitespace matches: ]</ + whitespace + <tool_calls>
// e.g. "]</  \n<tool_calls>" — model uses whitespace instead of tag name
var reBrokenWhitespace = regexp.MustCompile(
	`\]</\s+` + regexp.QuoteMeta("<tool_calls>"),
)

// reUnclosedTag matches: ]</ + any tag name (letters) + whitespace + <tool_calls>
// e.g. "]</tagname\n<tool_calls>" — model writes a tag name but never closes it
var reUnclosedTag = regexp.MustCompile(
	`\]</[a-zA-Z]+\s*` + regexp.QuoteMeta("<tool_calls>"),
)

// FixToolCallXML repairs malformed XML tool call blocks generated by some
// models (e.g. DeepSeek) that omit proper closing tags.
//
// The typical failure mode:
//
//	<tool_calls>
//	<invoke name="edit">
//	<parameter name="edits" string="false">[{"newText":"...","oldText":"..."}]</​
//	<tool_calls>
//	<invoke name="edit">...
//
// Instead of </parameter></invoke></tool_calls>, the model writes a broken
// closing tag (zero-width space + / + whitespace) and starts a new block.
// This function replaces the broken pattern with proper closing tags
// followed by the new block opener.
func FixToolCallXML(text string) string {
	if !strings.Contains(text, "<tool_calls>") {
		return text
	}

	replacement := `]</parameter></invoke></tool_calls>` + "\n" + `<tool_calls>`

	text = reBrokenClose.ReplaceAllString(text, replacement)
	text = reBrokenWhitespace.ReplaceAllString(text, replacement)
	text = reUnclosedTag.ReplaceAllString(text, replacement)

	return text
}