hakorune/apps/macros/examples/loop_normalize_macro.nyash

// loop_normalize_macro.nyash
// MVP: identity expansion with (json, ctx) signature.
// Next steps: normalize `loop(cond){ body }` into carrier-based LoopForm.

static box MacroBoxSpec {
  name() { return "LoopNormalize" }

  expand(json, ctx) {
    // MVP normalizer: detect Loop nodes with canonical key order
    // "kind":"Loop","condition":<json>,"body":[ ... ] and rewrite them
    // into a normalized form using JsonBuilder (keys ordered as condition/body).

    local JB = include "apps/lib/json_builder.nyash"

    // helpers
    local s = json
    local out = ""
    local i = 0

    // parse a JSON value starting at i and return "<end_index>#<json>"
    function parse_value(s, i) {
      local n = s.length()
      if i >= n { return ("" + i) + "#" + "" }
      local ch = s.substring(i, i+1)
      // string
      if ch == "\"" {
        local j = i + 1
        loop(j < n) {
          local c = s.substring(j, j+1)
          if c == "\\" {
            j = j + 2
            continue
          }
          if c == "\"" {
            j = j + 1
            break
          }
          j = j + 1
        }
        return ("" + j) + "#" + s.substring(i, j)
      }
      // object
      if ch == "{" {
        local depth = 1
        local j = i + 1
        local in_str = false
        loop(j < n && depth > 0) {
          local c = s.substring(j, j+1)
          if c == "\"" {
            // toggle string (respect escape)
            local k = j - 1
            local esc = false
            if k >= 0 && s.substring(k, k+1) == "\\" { esc = true }
            if not esc { in_str = not in_str }
            j = j + 1
            continue
          }
          if not in_str {
            if c == "{" { depth = depth + 1 }
            if c == "}" { depth = depth - 1 }
          }
          j = j + 1
        }
        return ("" + j) + "#" + s.substring(i, j)
      }
      // array
      if ch == "[" {
        local depth = 1
        local j = i + 1
        local in_str = false
        loop(j < n && depth > 0) {
          local c = s.substring(j, j+1)
          if c == "\"" {
            local k = j - 1
            local esc = false
            if k >= 0 && s.substring(k, k+1) == "\\" { esc = true }
            if not esc { in_str = not in_str }
            j = j + 1
            continue
          }
          if not in_str {
            if c == "[" { depth = depth + 1 }
            if c == "]" { depth = depth - 1 }
          }
          j = j + 1
        }
        return ("" + j) + "#" + s.substring(i, j)
      }
      // number/true/false/null: read until delimiter
      local j = i
      loop(j < n) {
        local c = s.substring(j, j+1)
        if c == "," || c == "]" || c == "}" || c == "\n" || c == "\r" || c == "\t" || c == " " { break }
        j = j + 1
      }
      return ("" + j) + "#" + s.substring(i, j)
    }

    function pair_idx(pair) {
      // parse decimal at start until '#'
      local i = 0
      local n = pair.length()
      local val = 0
      loop(i < n) {
        local ch = pair.substring(i, i+1)
        if ch == "#" { break }
        local d = 0
        if ch == "0" { d = 0 }
        if ch == "1" { d = 1 }
        if ch == "2" { d = 2 }
        if ch == "3" { d = 3 }
        if ch == "4" { d = 4 }
        if ch == "5" { d = 5 }
        if ch == "6" { d = 6 }
        if ch == "7" { d = 7 }
        if ch == "8" { d = 8 }
        if ch == "9" { d = 9 }
        val = val * 10 + d
        i = i + 1
      }
      return val
    }

    function pair_json(pair) {
      // substring after first '#'
      local i = 0
      local n = pair.length()
      loop(i < n) {
        if pair.substring(i, i+1) == "#" { return pair.substring(i+1, n) }
        i = i + 1
      }
      return ""
    }

    // Extract assignment target variable name from an Assignment JSON object string.
    // Returns "" if not found.
    function extract_assign_target_var(e2) {
      // pattern: "target":{"kind":"Variable","name":"<NAME>"}
      local pat = "\"target\":{\"kind\":\"Variable\",\"name\":\""
      local pos = -1
      local j = 0
      loop(j + pat.length() <= e2.length()) {
        if e2.substring(j, j + pat.length()) == pat { pos = j + pat.length(); break }
        j = j + 1
      }
      if pos < 0 { return "" }
      // read until next unescaped quote
      local name = ""
      local k = pos
      loop(k < e2.length()) {
        local c = e2.substring(k, k+1)
        if c == "\\" {
          // skip escape and next char
          if k + 1 < e2.length() { name = name + e2.substring(k, k+2); k = k + 2; continue }
        }
        if c == "\"" { break }
        name = name + c
        k = k + 1
      }
      return name
    }

    // pattern tokens
    local t_kind_loop = "\"kind\":\"Loop\""
    local t_cond = "\"condition\":"
    local t_body = "\"body\":"

    loop(i < s.length()) {
      // try to detect a Loop object start
      if i + 6 < s.length() && s.substring(i, i+1) == "{" {
        // look ahead inside this object to see if it begins with kind:Loop
        local val = parse_value(s, i)
        local endi = pair_idx(val)
        local obj = pair_json(val)

        // quick check: contains kind:"Loop" (manual scan)
        {
          local found = 0
          local k = 0
          loop(k + t_kind_loop.length() <= obj.length()) {
            if obj.substring(k, k + t_kind_loop.length()) == t_kind_loop {
              found = 1
              break
            }
            k = k + 1
          }
          if found == 0 {
            out = out + obj
            i = endi
            continue
          }
        }

        // Now attempt to parse condition and body assuming canonical order
        // Find condition token within object string
        local oj = 0
        local pos_c = -1
        loop(oj + t_cond.length() <= obj.length()) {
          if obj.substring(oj, oj + t_cond.length()) == t_cond {
            pos_c = oj + t_cond.length()
            break
          }
          oj = oj + 1
        }
        local pos_b = -1
        local kk = 0
        loop(kk + t_body.length() <= obj.length()) {
          if obj.substring(kk, kk + t_body.length()) == t_body {
            pos_b = kk + t_body.length()
            break
          }
          kk = kk + 1
        }
        if pos_c >= 0 && pos_b >= 0 {
          // extract values
          local cond_pair = parse_value(obj, pos_c)
          local cond_json = pair_json(cond_pair)
          // move after condition to find body array
          // ensure we re-scan from pos_b to robustly pick body
          local body_pair = parse_value(obj, pos_b)
          local body_json = pair_json(body_pair)
          // if body_json is not array, keep identity
          if body_json.substring(0,1) == "[" {
            // Reorder body: move Assignment nodes to the tail (carrier-like normalization)
            local inner = body_json.substring(1, body_json.length()-1)
            local elems = []
            local p2 = 0
            local n2 = inner.length()
            local in_str2 = false
            local depth_obj2 = 0
            local depth_arr2 = 0
            local start2 = 0
            loop(p2 < n2) {
              local c2 = inner.substring(p2, p2+1)
              if c2 == "\"" {
                local k2 = p2 - 1
                local esc2 = false
                if k2 >= 0 && inner.substring(k2, k2+1) == "\\" { esc2 = true }
                if not esc2 { in_str2 = not in_str2 }
              } else if not in_str2 {
                if c2 == "{" { depth_obj2 = depth_obj2 + 1 }
                if c2 == "}" { depth_obj2 = depth_obj2 - 1 }
                if c2 == "[" { depth_arr2 = depth_arr2 + 1 }
                if c2 == "]" { depth_arr2 = depth_arr2 - 1 }
                if c2 == "," && depth_obj2 == 0 && depth_arr2 == 0 {
                  elems.push(inner.substring(start2, p2))
                  start2 = p2 + 1
                }
              }
              p2 = p2 + 1
            }
            if start2 < n2 { elems.push(inner.substring(start2, n2)) }

            // Classify with original indices
            local assigns = []    // list of [idx,json]
            local others = []     // list of [idx,json]
            local tagA = "\"kind\":\"Assignment\""
            local t = 0
            loop(t < elems.length()) {
              local e2 = elems.get(t)
              // trim
              local a = 0
              local b = e2.length()
              loop(a < b && (e2.substring(a,a+1)==" " || e2.substring(a,a+1)=="\n" || e2.substring(a,a+1)=="\t" || e2.substring(a,a+1)=="\r")) { a = a + 1 }
              loop(b > a && (e2.substring(b-1,b)==" " || e2.substring(b-1,b)=="\n" || e2.substring(b-1,b)=="\t" || e2.substring(b-1,b)=="\r")) { b = b - 1 }
              e2 = e2.substring(a,b)
              // contains tagA?
              local found = 0
              local q = 0
              loop(q + tagA.length() <= e2.length()) {
                if e2.substring(q, q + tagA.length()) == tagA {
                  found = 1
                  break
                }
                q = q + 1
              }
              if found == 1 { assigns.push([t, e2]) } else { others.push([t, e2]) }
              t = t + 1
            }

            // Only reorder when all others appear before all assigns in the original order
            local ok = 1
            if assigns.length() > 0 && others.length() > 0 {
              // max index of others, min index of assigns
              local max_o = others.get(0).get(0)
              local i2 = 1
              loop(i2 < others.length()) { if others.get(i2).get(0) > max_o { max_o = others.get(i2).get(0) } i2 = i2 + 1 }
              local min_a = assigns.get(0).get(0)
              i2 = 1
              loop(i2 < assigns.length()) { if assigns.get(i2).get(0) < min_a { min_a = assigns.get(i2).get(0) } i2 = i2 + 1 }
              if not (max_o <= min_a) { ok = 0 }
            }

            // MVP-2 gate: skip when Break/Continue exists (conservative)
            if ok == 1 {
              local has_ctrl = 0
              local tagBr = "\"kind\":\"Break\""
              local tagCt = "\"kind\":\"Continue\""
              t = 0
              loop(t < elems.length()) {
                local e3 = elems.get(t)
                // cheap contains
                local p = 0
                loop(p + tagBr.length() <= e3.length()) {
                  if e3.substring(p, p + tagBr.length()) == tagBr { has_ctrl = 1; break }
                  p = p + 1
                }
                if has_ctrl == 0 {
                  p = 0
                  loop(p + tagCt.length() <= e3.length()) {
                    if e3.substring(p, p + tagCt.length()) == tagCt { has_ctrl = 1; break }
                    p = p + 1
                  }
                }
                if has_ctrl == 1 { break }
                t = t + 1
              }
              if has_ctrl == 1 { ok = 0 }
            }

            // MVP-2 gate: allow up to 2 unique assignment targets; else keep original
            if ok == 1 {
              local uniq = []
              t = 0
              local too_many = 0
              loop(t < assigns.length()) {
                local aj = assigns.get(t).get(1)
                local nm = extract_assign_target_var(aj)
                if nm == "" {
                  // unknown structure → conservative: abort reorder
                  too_many = 1
                  break
                }
                // check if nm already recorded
                local seen = 0
                local u = 0
                loop(u < uniq.length()) {
                  if uniq.get(u) == nm { seen = 1; break }
                  u = u + 1
                }
                if seen == 0 { uniq.push(nm) }
                if uniq.length() > 2 { too_many = 1; break }
                t = t + 1
              }
              if too_many == 1 { ok = 0 }
            }

            // Rebuild body (others then assigns) only when ok; otherwise keep original
            local body_new = "["
            local first = 1
            t = 0
            if ok == 1 {
              loop(t < others.length()) {
                if first == 1 { first = 0 } else { body_new = body_new + "," }
                body_new = body_new + others.get(t).get(1)
                t = t + 1
              }
              t = 0
              loop(t < assigns.length()) {
                if first == 1 { first = 0 } else { body_new = body_new + "," }
                body_new = body_new + assigns.get(t).get(1)
                t = t + 1
              }
            } else {
              // keep original order
              local back = 0
              loop(back < elems.length()) {
                if first == 1 { first = 0 } else { body_new = body_new + "," }
                body_new = body_new + elems.get(back)
                back = back + 1
              }
            }
            body_new = body_new + "]"

            // rebuild Loop string directly (canonical key order: condition, body)
            local loop_norm = "{\"kind\":\"Loop\",\"condition\":" + cond_json + ",\"body\":" + body_new + "}"
            out = out + loop_norm
            i = endi
            continue
          }
        }
        // fallback: copy as-is if parsing failed
        out = out + obj
        i = endi
        continue
      }
      // default: copy through one char
      out = out + s.substring(i, i+1)
      i = i + 1
    }
    return out
  }
}