hakorune/tools/hako_check/analysis_consumer.hako

// tools/hako_check/analysis_consumer.hako — HakoAnalysisBuilderBox (MVP)
// Build a minimal Analysis IR from raw .hako source (no Rust parser needed).
// IR (MapBox): {
//   path: String,
//   uses: Array<String>,
//   boxes: Array<Map{name,is_static,methods:Array<Map{name,arity,span}}}>,
//   methods: Array<String> (qualified: Box.method/arity),
//   calls: Array<Map{from,to}},
//   entrypoints: Array<String>
// }

using selfhost.shared.common.string_helpers as Str
using tools.hako_parser.parser_core as HakoParserCoreBox

static box HakoAnalysisBuilderBox {
  build_from_source(text, path) { return me.build_from_source_flags(text, path, 0) }
  build_from_source_flags(text, path, no_ast) {
    local ir = new MapBox()
    ir.set("path", path)
    ir.set("uses", new ArrayBox())
    ir.set("boxes", new ArrayBox())
    ir.set("methods", new ArrayBox())
    ir.set("calls", new ArrayBox())
    ir.set("method_spans", new ArrayBox())
    ir.set("source", text)
    local eps = new ArrayBox(); eps.push("Main.main"); eps.push("main"); ir.set("entrypoints", eps)
    // debug disabled in strict environments
    local debug = 0

    // Prefer AST (Hako Parser) if possible
    local ast = null
    if no_ast == 0 { ast = HakoParserCoreBox.parse(text) }
    if ast != null {
      // uses (with fallback for backward compat)
      local uses = ast.get("uses"); if uses == null { uses = ast.get("uses_arr") }
      if uses != null { local ui=0; while ui<uses.size() { me._ensure_array(ir, "uses").push(uses.get(ui)); ui=ui+1 } }
      // methods (qualified: Box.method/arity) and boxes metadata
      local boxes_ast = ast.get("boxes"); if boxes_ast == null { boxes_ast = ast.get("boxes_arr") }
      if boxes_ast != null {
        local bi=0
        while bi<boxes_ast.size() {
          local b = boxes_ast.get(bi)
          local name = b.get("name")
          local is_static = b.get("is_static")
          local span_line = b.get("span_line")

          // Add box to IR boxes array for HC012 dead static box detection
          local box_info = new MapBox()
          box_info.set("name", name)
          box_info.set("is_static", is_static)
          box_info.set("methods", new ArrayBox())
          if span_line != null { box_info.set("span_line", span_line) }

          local ms = b.get("methods")
          if ms != null {
            local mi=0
            while mi<ms.size() {
              local m = ms.get(mi)
              local qual = name + "." + m.get("name") + "/" + me._itoa(m.get("arity"))
              me._ensure_array(ir, "methods").push(qual)
              // capture start line spans for later source mapping
              local spans = me._ensure_array(ir, "method_spans")
              local rec = new MapBox(); rec.set("name", qual); rec.set("line", m.get("span"))
              spans.push(rec)

              // Also add method metadata to box_info
              local method_info = new MapBox()
              method_info.set("name", m.get("name"))
              method_info.set("arity", m.get("arity"))
              method_info.set("span", m.get("span"))
              local box_info_methods = box_info.get("methods")
              box_info_methods.push(method_info)

              mi = mi + 1
            }
          }

          // Add box_info to IR boxes array
          me._ensure_array(ir, "boxes").push(box_info)
          bi = bi + 1
        }
      }
    }

    // 1) collect using lines（AST が空のときのみテキスト走査）
    local lines = me._split_lines(text)
    if ir.get("uses") == null || ir.get("uses").size() == 0 {
      local _i = 0
      while _i < lines.size() {
        local ln = me._ltrim(lines.get(_i))
        if ln.indexOf('using "') == 0 {
          // using "pkg.name" as Alias
          local q1 = ln.indexOf('"')
          local q2 = -1
          if q1 >= 0 { q2 = ln.indexOf('"', q1+1) }
          if q1 >= 0 && q2 > q1 { me._ensure_array(ir, "uses").push(ln.substring(q1+1, q2)) }
        }
        _i = _i + 1
      }
    }

    // 2) scan static/box and methods when AST did not populate any methods or boxes
    local need_method_scan = 1
    if ir.get("methods") != null { if ir.get("methods").size() > 0 { need_method_scan = 0 } }
    // Also scan if boxes array is empty (for HC012 dead static box detection)
    local need_box_scan = 0
    if ir.get("boxes") != null { if ir.get("boxes").size() == 0 { need_box_scan = 1 } }
    if need_method_scan == 1 || need_box_scan == 1 {
      // debug noop
      local boxes_arr = me._ensure_array(ir, "boxes")
      local methods_arr = me._ensure_array(ir, "methods")
      local cur_name = null
      local cur_is_static = 0
      local i2 = 0
      while i2 < lines.size() {
        local ln = me._ltrim(lines.get(i2))
        // static box Name {
        if ln.indexOf("static box ") == 0 {
          local rest = ln.substring("static box ".length())
          local sp = me._upto(rest, " {")
          cur_name = sp
          cur_is_static = 1
          local b = new MapBox(); b.set("name", cur_name); b.set("is_static", true); b.set("methods", new ArrayBox())
          boxes_arr.push(b)
          i2 = i2 + 1
          continue
        }
        // method foo(args) {
        if ln.indexOf("method ") == 0 {
          if cur_name == null { cur_name = "Main" }
          local rest = ln.substring("method ".length())
          local p = rest.indexOf("(")
          local mname = (p>0) ? rest.substring(0,p) : rest
          mname = me._rstrip(mname)
          local arity = me._count_commas_in_parens(rest)
          local method = new MapBox(); method.set("name", mname); method.set("arity", arity); method.set("span", (i2+1))
          // Safe push pattern: get box, then get its methods array, then push
          local last_box = boxes_arr.get(boxes_arr.size()-1)
          local box_methods = last_box.get("methods")
          box_methods.push(method)
          methods_arr.push(cur_name + "." + mname + "/" + me._itoa(arity))
          i2 = i2 + 1
          continue
        }
        // box boundary heuristic
        if ln == "}" { cur_name = null; cur_is_static = 0; }
        i2 = i2 + 1
      }
    }
    // Final fallback: super simple scan over raw text if still no methods
    if ir.get("methods").size() == 0 { me._scan_methods_fallback(text, ir) }

    // 3) calls: naive pattern Box.method( or Alias.method(  — arity推定付き
    // For MVP, we scan whole text and link within same file boxes only.
    // debug noop
    local calls_arr = me._ensure_array(ir, "calls")
    local i3 = 0
    while i3 < lines.size() {
      local ln = lines.get(i3)
      // source context: try to infer last seen method
      // We fallback to "Main.main" when unknown
      local src = me._last_method_for_line(ir, i3+1)
      local pos = 0
      local L = ln.length()
      local k = 0
      while k <= L {
        local dot = ln.indexOf(".", pos)
        if dot < 0 { break }
        // find ident before '.' and after '.'
        local lhs = me._scan_ident_rev(ln, dot-1)
        local rhs = me._scan_ident_fwd(ln, dot+1)
        if lhs != null && rhs != null {
          // infer arity by scanning parentheses immediately after method
          local ar = me._infer_call_arity(ln, dot + 1 + rhs.length())
          local tgt = lhs + "." + rhs + "/" + me._itoa(ar)
          // record
          local c = new MapBox(); c.set("from", src); c.set("to", tgt)
          calls_arr.push(c)
        }
        pos = dot + 1
        k = k + 1
      }
      i3 = i3 + 1
    }
    return ir
  }

  // utilities
  _ltrim(s) { return me._ltrim_chars(s, " \t") }
  _itoa(n) { local v=0+n; if v==0 { return "0" } local out=""; local digits="0123456789"; local tmp=""; while v>0 { local d=v%10; tmp=digits.substring(d,d+1)+tmp; v=v/10 } out=tmp; return out }
  _split_lines(s) {
    local arr = new ArrayBox(); if s == null { return arr }
    local n = s.length(); local last = 0; local i = 0
    loop (i < n) { local ch = s.substring(i,i+1); if ch == "\n" { arr.push(s.substring(last,i)); last = i+1 } i = i + 1 }
    if last <= n { arr.push(s.substring(last)) }
    return arr
  }
  _rstrip(s) {
    local n = s.length()
    local last = n
    // scan from end using reverse index
    local r = 0
    while r < n {
      local i4 = n-1-r
      local c = s.substring(i4, i4+1)
      if c != " " && c != "\t" { last = i4+1; break }
      if r == n-1 { last = 0 }
      r = r + 1
    }
    return s.substring(0, last)
  }
  _ltrim_chars(s, cs) {
    local n = s.length()
    local head = 0
    local idx = 0
    while idx < n {
      local ch = s.substring(idx, idx+1)
      if ch != " " && ch != "\t" { head = idx; break }
      if idx == n-1 { head = n }
      idx = idx + 1
    }
    return s.substring(head)
  }
  _upto(s, needle) {
    local p = s.indexOf(needle)
    if p < 0 { return me._rstrip(s) }
    return s.substring(0,p)
  }
  _count_commas_in_parens(rest) {
    // method foo(a,b,c) → 3 ; if empty → 0
    local p1 = rest.indexOf("("); local p2 = rest.indexOf(")", p1+1)
    if p1 < 0 || p2 < 0 || p2 <= p1+1 { return 0 }
    local inside = rest.substring(p1+1, p2)
    local cnt = 1; local n=inside.length(); local any=0
    local i5 = 0
    while i5 < n {
      local c = inside.substring(i5,i5+1)
      if c == "," { cnt = cnt + 1 }
      if c != " " && c != "\t" { any = 1 }
      i5 = i5 + 1
    }
    if any==0 { return 0 }
    return cnt
  }
  _scan_methods_fallback(text, ir) {
    if text == null { return 0 }
    local methods = ir.get("methods")
    local box_name = "Main"
    // find "static box Name" to prefer given name
    local pbox = text.indexOf("static box ")
    if pbox >= 0 {
      local after = pbox + "static box ".length()
      local name = ""
      local i = after
      loop (i < text.length()) {
        local ch = text.substring(i,i+1)
        if (ch >= "A" && ch <= "Z") || (ch >= "a" && ch <= "z") || ch == "_" || (ch >= "0" && ch <= "9") {
          name = name + ch
          i = i + 1
          continue
        }
        break
      }
      if name != "" { box_name = name }
    }
    // scan for "method " occurrences
    local pos = 0
    local n = text.length()
    loop (pos < n) {
      local k = text.indexOf("method ", pos)
      if k < 0 { break }
      local i = k + "method ".length()
      // read ident
      local mname = ""
      loop (i < n) {
        local ch2 = text.substring(i,i+1)
        if (ch2 >= "A" && ch2 <= "Z") || (ch2 >= "a" && ch2 <= "z") || ch2 == "_" || (ch2 >= "0" && ch2 <= "9") {
          mname = mname + ch2
          i = i + 1
          continue
        }
        break
      }
      // look ahead for params (...) and count commas
      local ar = 0
      local lp = text.indexOf("(", i)
      if lp >= 0 {
        local rp = text.indexOf(")", lp+1)
        if rp > lp+1 {
          local inside = text.substring(lp+1, rp)
          local any = 0; local c = 0; local j=0
          loop (j < inside.length()) {
            local ch3=inside.substring(j,j+1)
            if ch3 == "," { c = c + 1 }
            if ch3 != " " && ch3 != "\t" { any = 1 }
            j = j + 1
          }
          if any == 1 { ar = c + 1 }
        }
      }
      if mname != "" { methods.push(box_name + "." + mname + "/" + me._itoa(ar)) }
      pos = i
    }
    return methods.size()
  }
  _scan_ident_rev(s, i) {
    if i<0 { return null }
    local n = i
    local start = 0
    local rr = 0
    while rr <= n {
      local j = i - rr
      local c = s.substring(j, j+1)
      if me._is_ident_char(c) == 0 { start = j+1; break }
      if j == 0 { start = 0; break }
      rr = rr + 1
    }
    if start>i { return null }
    return s.substring(start, i+1)
  }
  _scan_ident_fwd(s, i) {
    local n=s.length(); if i>=n { return null }
    local endp = i
    local off = 0
    while off < n {
      local j = i + off
      if j >= n { break }
      local c = s.substring(j, j+1)
      if me._is_ident_char(c) == 0 { endp = j; break }
      if j == n-1 { endp = n; break }
      off = off + 1
    }
    if endp == i { return null }
    return s.substring(i, endp)
  }
  _is_ident_char(c) {
    if c == "_" { return 1 }
    if c >= "A" && c <= "Z" { return 1 }
    if c >= "a" && c <= "z" { return 1 }
    if c >= "0" && c <= "9" { return 1 }
    return 0
  }
  _last_method_for_line(ir, line_num) {
    // Conservative: return default entry when spans are not guaranteed to be maps
    // This avoids runtime errors when method_spans is absent or malformed in MVP builds.
    return "Main.main"
  }
  _infer_call_arity(line, after_name_pos) {
    // Count commas between the first '(' after method name and its matching ')'
    if line == null { return 0 }
    local n = line.length()
    local i = after_name_pos
    // seek to '('
    while i < n {
      local ch = line.substring(i,i+1)
      if ch == "(" { break }
      // stop early if we hit non-space before '('
      if ch != " " && ch != "\t" { return 0 }
      i = i + 1
    }
    if i >= n || line.substring(i,i+1) != "(" { return 0 }
    // scan until matching ')', no nesting assumed for MVP
    local any = 0; local commas = 0
    i = i + 1
    while i < n {
      local ch = line.substring(i,i+1)
      if ch == ")" { break }
      if ch == "," { commas = commas + 1 }
      if ch != " " && ch != "\t" { any = 1 }
      i = i + 1
    }
    if any == 0 { return 0 }
    return commas + 1
  }
  _ensure_array(m, key) {
    if m == null { return new ArrayBox() }
    // If key absent, create and return a new ArrayBox
    if m.has(key) == 0 { local arr = new ArrayBox(); m.set(key, arr); return arr }
    // Otherwise return the existing value
    return m.get(key)
  }
}

static box HakoAnalysisBuilderMain { method main(args) { return 0 } }