From 46494d261cbedd3c798f584459e42ab7ee6ea1f4 Mon Sep 17 00:00:00 2001
From: "Hans Elias B. Josephsen" <me@hansihe.com>
Date: Wed, 27 Feb 2019 13:14:57 +0100
Subject: [PATCH] implement hashing for term

---
 rustler/src/term.rs                           | 35 +++++++++++++++++++
 rustler/src/types/local_pid.rs                |  2 +-
 rustler_tests/lib/rustler_test.ex             |  2 ++
 rustler_tests/native/rustler_test/src/lib.rs  |  2 ++
 .../native/rustler_test/src/test_term.rs      | 10 ++++++
 rustler_tests/test/term_test.exs              | 23 ++++++++++++
 6 files changed, 73 insertions(+), 1 deletion(-)

diff --git a/rustler/src/term.rs b/rustler/src/term.rs
index 3158978c..d9a3e149 100644
--- a/rustler/src/term.rs
+++ b/rustler/src/term.rs
@@ -4,6 +4,7 @@ use crate::wrapper::NIF_TERM;
 use crate::{Binary, Decoder, Env, NifResult};
 use std::cmp::Ordering;
 use std::fmt::{self, Debug};
+use std::hash::{Hash, Hasher};
 
 /// Term is used to represent all erlang terms. Terms are always lifetime limited by a Env.
 ///
@@ -95,6 +96,31 @@ impl<'a> Term<'a> {
         let raw_binary = unsafe { term_to_binary(self.env.as_c_arg(), self.as_c_arg()) }.unwrap();
         unsafe { OwnedBinary::from_raw(raw_binary) }
     }
+
+    /// Non-portable hash function that only guarantees the same hash for the same term within
+    /// one Erlang VM instance.
+    ///
+    /// It takes 32-bit salt values and generates hashes within 0..2^32-1.
+    pub fn hash_internal(&self, salt: u32) -> u32 {
+        unsafe {
+            rustler_sys::enif_hash(
+                rustler_sys::ErlNifHash::ERL_NIF_INTERNAL_HASH,
+                self.as_c_arg(),
+                salt as u64,
+            ) as u32
+        }
+    }
+
+    /// Portable hash function that gives the same hash for the same Erlang term regardless of
+    /// machine architecture and ERTS version.
+    ///
+    /// It generates hashes within 0..2^27-1.
+    pub fn hash_phash2(&self) -> u32 {
+        unsafe {
+            rustler_sys::enif_hash(rustler_sys::ErlNifHash::ERL_NIF_PHASH2, self.as_c_arg(), 0)
+                as u32
+        }
+    }
 }
 
 impl<'a> PartialEq for Term<'a> {
@@ -124,5 +150,14 @@ impl<'a> PartialOrd for Term<'a> {
     }
 }
 
+impl<'a> Hash for Term<'a> {
+    fn hash<H: Hasher>(&self, state: &mut H) {
+        // As far as I can see, there is really no way
+        // to get a seed from the hasher. This is definitely
+        // not optimal, but it's the best we can do for now.
+        state.write_u32(self.hash_internal(0));
+    }
+}
+
 unsafe impl<'a> Sync for Term<'a> {}
 unsafe impl<'a> Send for Term<'a> {}
diff --git a/rustler/src/types/local_pid.rs b/rustler/src/types/local_pid.rs
index c1b50779..d5ee1980 100644
--- a/rustler/src/types/local_pid.rs
+++ b/rustler/src/types/local_pid.rs
@@ -2,7 +2,7 @@ use crate::wrapper::{pid, ErlNifPid};
 use crate::{Decoder, Encoder, Env, Error, NifResult, Term};
 use std::mem::MaybeUninit;
 
-#[derive(Clone)]
+#[derive(Copy, Clone)]
 pub struct LocalPid {
     c: ErlNifPid,
 }
diff --git a/rustler_tests/lib/rustler_test.ex b/rustler_tests/lib/rustler_test.ex
index d3b16a1b..ec351e21 100644
--- a/rustler_tests/lib/rustler_test.ex
+++ b/rustler_tests/lib/rustler_test.ex
@@ -23,6 +23,8 @@ defmodule RustlerTest do
   def term_debug(_), do: err()
   def term_eq(_, _), do: err()
   def term_cmp(_, _), do: err()
+  def term_internal_hash(_, _), do: err()
+  def term_phash2_hash(_), do: err()
 
   def sum_map_values(_), do: err()
   def map_entries_sorted(_), do: err()
diff --git a/rustler_tests/native/rustler_test/src/lib.rs b/rustler_tests/native/rustler_test/src/lib.rs
index 8dadead8..402637f5 100644
--- a/rustler_tests/native/rustler_test/src/lib.rs
+++ b/rustler_tests/native/rustler_test/src/lib.rs
@@ -26,6 +26,8 @@ rustler::init!(
         test_term::term_debug,
         test_term::term_eq,
         test_term::term_cmp,
+        test_term::term_internal_hash,
+        test_term::term_phash2_hash,
         test_map::sum_map_values,
         test_map::map_entries_sorted,
         test_map::map_from_arrays,
diff --git a/rustler_tests/native/rustler_test/src/test_term.rs b/rustler_tests/native/rustler_test/src/test_term.rs
index e23eb4db..a499d80f 100644
--- a/rustler_tests/native/rustler_test/src/test_term.rs
+++ b/rustler_tests/native/rustler_test/src/test_term.rs
@@ -30,3 +30,13 @@ pub fn term_cmp<'a>(a: Term<'a>, b: Term<'a>) -> Atom {
         Ordering::Greater => atoms::greater(),
     }
 }
+
+#[rustler::nif]
+pub fn term_internal_hash(term: Term, salt: u32) -> u32 {
+    term.hash_internal(salt)
+}
+
+#[rustler::nif]
+pub fn term_phash2_hash(term: Term) -> u32 {
+    term.hash_phash2()
+}
diff --git a/rustler_tests/test/term_test.exs b/rustler_tests/test/term_test.exs
index 5c54c1d8..4a68f75e 100644
--- a/rustler_tests/test/term_test.exs
+++ b/rustler_tests/test/term_test.exs
@@ -48,4 +48,27 @@ defmodule RustlerTest.TermTest do
     # Other term types
     assert RustlerTest.term_cmp(5, :test) == :less
   end
+
+  test "term hash" do
+    assert RustlerTest.term_phash2_hash(:foobar) == :erlang.phash2(:foobar)
+    assert RustlerTest.term_phash2_hash("testing") == :erlang.phash2("testing")
+    assert RustlerTest.term_phash2_hash(42) == :erlang.phash2(42)
+
+    # Assume a certain distribution
+    unique =
+      0..100
+      |> Enum.map(&RustlerTest.term_phash2_hash(&1))
+      |> Enum.group_by(fn n -> n end, fn n -> n end)
+      |> map_size
+
+    assert unique > 50
+
+    unique =
+      0..100
+      |> Enum.map(&RustlerTest.term_internal_hash(&1, 0))
+      |> Enum.group_by(fn n -> n end, fn n -> n end)
+      |> map_size
+
+    assert unique > 50
+  end
 end