harn-stdlib 0.10.42

/**
 * Protocol-fragment recognition for text-tool responses.
 *
 * Chat templates can damage only part of a response tag. Keep that
 * classification policy here so the parser treats those fragments as
 * malformed protocol instead of visible assistant narration.
 */
import { TAGGED_DIALECT } from "std/llm/dialects"

fn __tool_parse_protocol_tag_spellings(canonical: string) -> list<string> {
  for dialect in TAGGED_DIALECT {
    if dialect.canonical == canonical {
      return dialect.spellings
    }
  }
  return []
}

fn __tool_parse_protocol_contains_tag_fragment(text: string, spelling: string) -> bool {
  for prefix in ["<", "</"] {
    const needle = prefix + spelling
    let rest = text
    while true {
      const found = rest.index_of(needle)
      if found < 0 {
        break
      }
      const next_at = found + len(needle)
      if next_at >= len(rest) {
        return true
      }
      const next = rest.slice(next_at, next_at + 1)
      if next == ">" || trim(next) == "" {
        return true
      }
      rest = rest.slice(next_at, len(rest))
    }
  }
  return false
}

pub fn has_response_protocol_fragment(text: string) -> bool {
  const lower = text.lower()
  if contains(lower, "_call>") {
    return true
  }
  for canonical in ["tool_call", "assistant_prose", "user_response", "done"] {
    for spelling in __tool_parse_protocol_tag_spellings(canonical) {
      if __tool_parse_protocol_contains_tag_fragment(lower, spelling) {
        return true
      }
    }
  }
  return false
}