about summary refs log tree commit diff stats
path: root/077hash.cc
diff options
context:
space:
mode:
authorKartik K. Agaram <vc@akkartik.com>2016-02-16 17:46:37 -0800
committerKartik K. Agaram <vc@akkartik.com>2016-02-16 17:51:34 -0800
commit37421f8a8bbccc871a3d452d81f03ad9a58f23bf (patch)
treedc1374ae1d9e868f579f0f6904f8dac578c8f9df /077hash.cc
parentd8a530296a7bc1a3932e61bcae98463a559641e5 (diff)
downloadmu-37421f8a8bbccc871a3d452d81f03ad9a58f23bf.tar.gz
2664 - 'hash' any type
This is smart enough to only look at data and ignores internal
book-keeping like addresses, refcounts and floating-point details.
Though the last is not ideal yet.
Diffstat (limited to '077hash.cc')
-rw-r--r--077hash.cc342
1 files changed, 339 insertions, 3 deletions
diff --git a/077hash.cc b/077hash.cc
index c60c07e6..ea185cfa 100644
--- a/077hash.cc
+++ b/077hash.cc
@@ -1,4 +1,13 @@
-// From http://burtleburtle.net/bob/hash/hashfaq.html
+// A universal hash function that can handle objects of any type.
+//
+// The way it's currently implemented, two objects will have the same hash if
+// all their non-address fields (all the way down) expand to the same sequence
+// of scalar values. In particular, a container with all zero addresses hashes
+// to 0. Hopefully this won't be an issue because we are usually hashing
+// objects of a single type in any given hash table.
+//
+// Based on http://burtleburtle.net/bob/hash/hashfaq.html
+
 :(before "End Primitive Recipe Declarations")
 HASH,
 :(before "End Primitive Recipe Numbers")
@@ -9,14 +18,341 @@ case HASH: {
     raise_error << maybe(get(Recipe, r).name) << "'hash' takes exactly one ingredient rather than '" << inst.to_string() << "'\n" << end();
     break;
   }
+  break;
+}
+:(before "End Primitive Recipe Implementations")
+case HASH: {
+  reagent input = current_instruction().ingredients.at(0);  // copy
+  products.resize(1);
+  products.at(0).push_back(hash(0, input));
+  break;
+}
+
+//: in all the code below, the intermediate results of hashing are threaded through 'h'
+
+:(code)
+size_t hash(size_t h, reagent& r) {
+//?   cerr << debug_string(r) << '\n';
+  canonize(r);
+  if (is_mu_string(r))  // optimization
+    return hash_mu_string(h, r);
+  else if (is_mu_address(r))
+    return hash_mu_address(h, r);
+  else if (is_mu_scalar(r))
+    return hash_mu_scalar(h, r);
+  else if (is_mu_array(r))
+    return hash_mu_array(h, r);
+  else if (is_mu_container(r))
+    return hash_mu_container(h, r);
+  else if (is_mu_exclusive_container(r))
+    return hash_mu_exclusive_container(h, r);
+  assert(false);
+}
+
+size_t hash_mu_scalar(size_t h, const reagent& r) {
+  double input = is_literal(r) ? r.value : get_or_insert(Memory, r.value);
+  return hash_iter(h, static_cast<size_t>(input));
+}
+
+size_t hash_mu_address(size_t h, reagent& r) {
+  if (r.value == 0) return 0;
+  r.value = get_or_insert(Memory, r.value);
+  drop_from_type(r, "address");
+  if (r.properties.at(0).second->value == "shared") {
+    ++r.value;
+    drop_from_type(r, "shared");
+  }
+  return hash(h, r);
+}
+
+size_t hash_mu_string(size_t h, const reagent& r) {
+  string input = read_mu_string(get_or_insert(Memory, r.value));
+  for (long long int i = 0; i < SIZE(input); ++i) {
+    h = hash_iter(h, static_cast<size_t>(input.at(i)));
+//?     cerr << i << ": " << h << '\n';
+  }
+  return h;
+}
+
+size_t hash_mu_array(size_t h, const reagent& r) {
+  long long int size = get_or_insert(Memory, r.value);
+  reagent elem = r;
+  drop_from_type(elem, "array");
+  elem.type = new type_tree(*array_element(r.type));
+  for (long long int i=0, address = r.value+1; i < size; ++i, address += size_of(elem)) {
+    reagent tmp = elem;
+    tmp.value = address;
+    h = hash(h, tmp);
+//?     cerr << i << " (" << address << "): " << h << '\n';
+  }
+  return h;
+}
+
+bool is_mu_container(const reagent& r) {
+  if (r.type->value == 0) return false;
+  type_info& info = get(Type, r.type->value);
+  return info.kind == CONTAINER;
+}
+
+size_t hash_mu_container(size_t h, const reagent& r) {
+  assert(r.type->value);
+  type_info& info = get(Type, r.type->value);
+  long long int address = r.value;
+  long long int offset = 0;
+  for (long long int i = 0; i < SIZE(info.elements); ++i) {
+    reagent element = element_type(r, i);
+    element.set_value(address+offset);
+    h = hash(h, element);
+//?     cerr << i << ": " << h << '\n';
+    offset += size_of(info.elements.at(i));
+  }
+  return h;
+}
+
+bool is_mu_exclusive_container(const reagent& r) {
+  if (r.type->value == 0) return false;
+  type_info& info = get(Type, r.type->value);
+  return info.kind == EXCLUSIVE_CONTAINER;
+}
+
+size_t hash_mu_exclusive_container(size_t h, const reagent& r) {
+  assert(r.type->value);
+  long long int tag = get(Memory, r.value);
+  reagent variant = variant_type(r, tag);
+  variant.set_value(r.value + /*skip tag*/1);
+  h = hash(h, variant);
+  return h;
+}
+
+size_t hash_iter(size_t h, size_t input) {
+  h += input;
+  h += (h<<10);
+  h ^= (h>>6);
+
+  h += (h<<3);
+  h ^= (h>>11);
+  h += (h<<15);
+  return h;
+}
+
+:(scenario hash_container_checks_all_elements)
+container foo [
+  x:number
+  y:character
+]
+recipe main [
+  1:foo <- merge 34, 97/a
+  3:number <- hash 1:foo
+  reply-unless 3:number
+  4:foo <- merge 34, 98/a
+  6:number <- hash 4:foo
+  reply-unless 6:number
+  7:boolean <- equal 3:number, 6:number
+]
+# hash on containers includes all elements
++mem: storing 0 in location 7
+
+:(scenario hash_exclusive_container_checks_all_elements)
+exclusive-container foo [
+  x:bar
+  y:number
+]
+container bar [
+  a:number
+  b:number
+]
+recipe main [
+  1:foo <- merge 0/x, 34, 35
+  4:number <- hash 1:foo
+  reply-unless 4:number
+  5:foo <- merge 0/x, 34, 36
+  8:number <- hash 5:foo
+  reply-unless 8:number
+  9:boolean <- equal 4:number, 8:number
+]
+# hash on containers includes all elements
++mem: storing 0 in location 9
+
+//: These properties aren't necessary for hash, they just test that the
+//: current implementation works like we think it does.
+
+:(scenario hash_of_zero_address)
+recipe main [
+  1:address:number <- copy 0
+  2:number <- hash 1:address:number
+]
++mem: storing 0 in location 2
+
+//: This is probably too aggressive, but we need some way to avoid depending
+//: on the precise bit pattern of a floating-point number.
+:(scenario hash_of_numbers_ignores_fractional_part)
+recipe main [
+  1:number <- hash 1.5
+  2:number <- hash 1
+  3:boolean <- equal 1:number, 2:number
+]
++mem: storing 1 in location 3
+
+:(scenario hash_of_array_same_as_string)
+recipe main [
+  10:number <- copy 3
+  11:number <- copy 97
+  12:number <- copy 98
+  13:number <- copy 99
+  2:number <- hash 10:array:number/unsafe
+  reply-unless 2:number
+  3:address:shared:array:character <- new [abc]
+  4:number <- hash 3:address:shared:array:character
+  reply-unless 4:number
+  5:boolean <- equal 2:number, 4:number
+]
++mem: storing 1 in location 5
+
+:(scenario hash_ignores_address_value)
+recipe main [
+  1:address:shared:number <- new number:type
+  *1:address:shared:number <- copy 34
+  2:number <- hash 1:address:shared:number
+  3:address:shared:number <- new number:type
+  *3:address:shared:number <- copy 34
+  4:number <- hash 3:address:shared:number
+  5:boolean <- equal 2:number, 4:number
+]
+# different addresses hash to the same result as long as the values the point to do so
++mem: storing 1 in location 5
+
+:(scenario hash_ignores_address_refcount)
+recipe main [
+  1:address:shared:number <- new number:type
+  *1:address:shared:number <- copy 34
+  2:number <- hash 1:address:shared:number
+  reply-unless 2:number
+  # increment refcount
+  3:address:shared:number <- copy 1:address:shared:number
+  4:number <- hash 3:address:shared:number
+  reply-unless 4:number
+  5:boolean <- equal 2:number, 4:number
+]
+# hash doesn't change when refcount changes
++mem: storing 1 in location 5
+
+:(scenario hash_container_depends_only_on_elements)
+container foo [
+  x:number
+  y:character
+]
+container bar [
+  x:number
+  y:character
+]
+recipe main [
+  1:foo <- merge 34, 97/a
+  3:number <- hash 1:foo
+  reply-unless 3:number
+  4:bar <- merge 34, 97/a
+  6:number <- hash 4:bar
+  reply-unless 6:number
+  7:boolean <- equal 3:number, 6:number
+]
+# containers with identical elements return identical hashes
++mem: storing 1 in location 7
+
+:(scenario hash_container_depends_only_on_elements_2)
+container foo [
+  x:number
+  y:character
+  z:address:shared:number
+]
+recipe main [
+  1:address:shared:number <- new number:type
+  *1:address:shared:number <- copy 34
+  2:foo <- merge 34, 97/a, 1:address:shared:number
+  5:number <- hash 2:foo
+  reply-unless 5:number
+  6:address:shared:number <- new number:type
+  *6:address:shared:number <- copy 34
+  7:foo <- merge 34, 97/a, 6:address:shared:number
+  10:number <- hash 7:foo
+  reply-unless 10:number
+  11:boolean <- equal 5:number, 10:number
+]
+# containers with identical 'leaf' elements return identical hashes
++mem: storing 1 in location 11
+
+:(scenario hash_container_depends_only_on_elements_3)
+container foo [
+  x:number
+  y:character
+  z:bar
+]
+container bar [
+  x:number
+  y:number
+]
+recipe main [
+  1:foo <- merge 34, 97/a, 47, 48
+  6:number <- hash 1:foo
+  reply-unless 6:number
+  7:foo <- merge 34, 97/a, 47, 48
+  12:number <- hash 7:foo
+  reply-unless 12:number
+  13:boolean <- equal 6:number, 12:number
+]
+# containers with identical 'leaf' elements return identical hashes
++mem: storing 1 in location 13
+
+:(scenario hash_exclusive_container_ignores_tag)
+exclusive-container foo [
+  x:bar
+  y:number
+]
+container bar [
+  a:number
+  b:number
+]
+recipe main [
+  1:foo <- merge 0/x, 34, 35
+  4:number <- hash 1:foo
+  reply-unless 4:number
+  5:bar <- merge 34, 35
+  7:number <- hash 5:bar
+  reply-unless 7:number
+  8:boolean <- equal 4:number, 7:number
+]
+# hash on containers includes all elements
++mem: storing 1 in location 8
+
+//: An older version that supported only strings.
+//: Hash functions are subtle and easy to get wrong, so we keep the old
+//: version around and check that the new one is consistent with it.
+
+:(scenario hash_matches_old_version)
+recipe main [
+  1:address:shared:array:character <- new [abc]
+  2:number <- hash 1:address:shared:array:character
+  3:number <- hash_old 1:address:shared:array:character
+  4:number <- equal 2:number, 3:number
+]
++mem: storing 1 in location 4
+
+:(before "End Primitive Recipe Declarations")
+HASH_OLD,
+:(before "End Primitive Recipe Numbers")
+put(Recipe_ordinal, "hash_old", HASH_OLD);
+:(before "End Primitive Recipe Checks")
+case HASH_OLD: {
+  if (SIZE(inst.ingredients) != 1) {
+    raise_error << maybe(get(Recipe, r).name) << "'hash_old' takes exactly one ingredient rather than '" << inst.to_string() << "'\n" << end();
+    break;
+  }
   if (!is_mu_string(inst.ingredients.at(0))) {
-    raise_error << maybe(get(Recipe, r).name) << "'hash' currently only supports strings (address:shared:array:character), but got " << inst.ingredients.at(0).original_string << '\n' << end();
+    raise_error << maybe(get(Recipe, r).name) << "'hash_old' currently only supports strings (address:shared:array:character), but got " << inst.ingredients.at(0).original_string << '\n' << end();
     break;
   }
   break;
 }
 :(before "End Primitive Recipe Implementations")
-case HASH: {
+case HASH_OLD: {
   string input = read_mu_string(ingredients.at(0).at(0));
   size_t h = 0 ;