Change Hasher and HMAC APIs closer to std::hash model

- Implement Clone and std::io::Writer.
- Reduce the API to write() and finish(). Contrary to std::hash, finish() resets the hasher immediately.
- Add hmac::hmac() convenience fn.
- Replace hash::evpmd() with HashType methods.
- Add assertions as a crude check for failed calls into openssl.
- Add examples and some tests.

[breaking-change]

1 files changed, 235 insertions, 136 deletions

diff --git a/src/crypto/hash.rs b/src/crypto/hash.rs
index 6d751ade..1c2dc149 100644
--- a/src/crypto/hash.rs
+++ b/src/crypto/hash.rs
@@ -1,10 +1,10 @@
 use libc::c_uint;
-use std::ptr;
-use std::old_io;
 use std::iter::repeat;
+use std::old_io::{IoError, Writer};
 
 use ffi;
 
+/// Message digest (hash) type.
 #[derive(Copy)]
 pub enum HashType {
     MD5,
@@ -16,219 +16,318 @@ pub enum HashType {
     RIPEMD160
 }
 
-pub fn evpmd(t: HashType) -> (*const ffi::EVP_MD, u32) {
-    unsafe {
-        match t {
-            HashType::MD5 => (ffi::EVP_md5(), 16),
-            HashType::SHA1 => (ffi::EVP_sha1(), 20),
-            HashType::SHA224 => (ffi::EVP_sha224(), 28),
-            HashType::SHA256 => (ffi::EVP_sha256(), 32),
-            HashType::SHA384 => (ffi::EVP_sha384(), 48),
-            HashType::SHA512 => (ffi::EVP_sha512(), 64),
-            HashType::RIPEMD160 => (ffi::EVP_ripemd160(), 20),
+impl HashType {
+    /// Returns the length of the message digest.
+    #[inline]
+    pub fn md_len(&self) -> usize {
+        use self::HashType::*;
+        match *self {
+            MD5 => 16,
+            SHA1 => 20,
+            SHA224 => 28,
+            SHA256 => 32,
+            SHA384 => 48,
+            SHA512 => 64,
+            RIPEMD160 => 20,
         }
     }
-}
-
-pub struct HasherContext {
-    ptr: *mut ffi::EVP_MD_CTX
-}
-
-impl HasherContext {
-    pub fn new() -> HasherContext {
-        ffi::init();
 
+    /// Internal interface subject to removal.
+    #[inline]
+    pub fn evp_md(&self) -> *const ffi::EVP_MD {
         unsafe {
-            HasherContext { ptr: ffi::EVP_MD_CTX_create() }
+            use self::HashType::*;
+            match *self {
+                MD5 => ffi::EVP_md5(),
+                SHA1 => ffi::EVP_sha1(),
+                SHA224 => ffi::EVP_sha224(),
+                SHA256 => ffi::EVP_sha256(),
+                SHA384 => ffi::EVP_sha384(),
+                SHA512 => ffi::EVP_sha512(),
+                RIPEMD160 => ffi::EVP_ripemd160(),
+            }
         }
     }
 }
 
-impl Drop for HasherContext {
-    fn drop(&mut self) {
-        unsafe {
-            ffi::EVP_MD_CTX_destroy(self.ptr);
-        }
-    }
+#[derive(PartialEq, Copy)]
+enum State {
+    Reset,
+    Updated,
+    Finalized,
 }
 
-#[allow(dead_code)]
+use self::State::*;
+
+/// Provides message digest (hash) computation.
+///
+/// # Examples
+///
+/// Calculate a hash in one go.
+///
+/// ```
+/// use openssl::crypto::hash::{hash, HashType};
+/// let data = b"\x42\xF4\x97\xE0";
+/// let spec = b"\x7c\x43\x0f\x17\x8a\xef\xdf\x14\x87\xfe\xe7\x14\x4e\x96\x41\xe2";
+/// let res = hash(HashType::MD5, data);
+/// assert_eq!(res, spec);
+/// ```
+///
+/// Use the `Writer` trait to supply the input in chunks.
+///
+/// ```
+/// use std::old_io::Writer;
+/// use openssl::crypto::hash::{Hasher, HashType};
+/// let data = [b"\x42\xF4", b"\x97\xE0"];
+/// let spec = b"\x7c\x43\x0f\x17\x8a\xef\xdf\x14\x87\xfe\xe7\x14\x4e\x96\x41\xe2";
+/// let mut h = Hasher::new(HashType::MD5);
+/// h.write_all(data[0]);
+/// h.write_all(data[1]);
+/// let res = h.finish();
+/// assert_eq!(res, spec);
+/// ```
+///
+/// # Warning
+///
+/// Don't actually use MD5 and SHA-1 hashes, they're not secure anymore.
+///
+/// Don't ever hash passwords, use `crypto::pkcs5` or bcrypt/scrypt instead.
 pub struct Hasher {
-    evp: *const ffi::EVP_MD,
-    ctx: HasherContext,
-    len: u32,
-}
-
-impl old_io::Writer for Hasher {
-    fn write_all(&mut self, buf: &[u8]) -> old_io::IoResult<()> {
-        self.update(buf);
-        Ok(())
-    }
+    ctx: *mut ffi::EVP_MD_CTX,
+    md: *const ffi::EVP_MD,
+    type_: HashType,
+    state: State,
 }
 
 impl Hasher {
-    pub fn new(ht: HashType) -> Hasher {
-        let ctx = HasherContext::new();
-        Hasher::with_context(ctx, ht)
+    /// Creates a new `Hasher` with the specified hash type.
+    pub fn new(ty: HashType) -> Hasher {
+        ffi::init();
+
+        let ctx = unsafe {
+            let r = ffi::EVP_MD_CTX_create();
+            assert!(!r.is_null());
+            r
+        };
+        let md = ty.evp_md();
+
+        let mut h = Hasher { ctx: ctx, md: md, type_: ty, state: Finalized };
+        h.init();
+        h
     }
 
-    pub fn with_context(ctx: HasherContext, ht: HashType) -> Hasher {
-        let (evp, mdlen) = evpmd(ht);
+    #[inline]
+    fn init(&mut self) {
+        match self.state {
+            Reset => return,
+            Updated => { self.finalize(); },
+            Finalized => (),
+        }
         unsafe {
-            ffi::EVP_DigestInit_ex(ctx.ptr, evp, 0 as *const _);
+            let r = ffi::EVP_DigestInit_ex(self.ctx, self.md, 0 as *const _);
+            assert_eq!(r, 1);
         }
-
-        Hasher { evp: evp, ctx: ctx, len: mdlen }
+        self.state = Reset;
     }
 
-    /// Update this hasher with more input bytes
-    pub fn update(&mut self, data: &[u8]) {
+    #[inline]
+    fn update(&mut self, data: &[u8]) {
+        if self.state == Finalized {
+            self.init();
+        }
         unsafe {
-            ffi::EVP_DigestUpdate(self.ctx.ptr, data.as_ptr(), data.len() as c_uint);
+            let r = ffi::EVP_DigestUpdate(self.ctx, data.as_ptr(),
+                                          data.len() as c_uint);
+            assert_eq!(r, 1);
         }
+        self.state = Updated;
     }
 
-    /**
-     * Return the digest of all bytes added to this hasher since its last
-     * initialization
-     */
-    pub fn finalize(self) -> Vec<u8> {
-        let (res, _) = self.finalize_reuse();
+    #[inline]
+    fn finalize(&mut self) -> Vec<u8> {
+        if self.state == Finalized {
+            self.init();
+        }
+        let md_len = self.type_.md_len();
+        let mut res: Vec<u8> = repeat(0).take(md_len).collect();
+        unsafe {
+            let mut len = 0;
+            let r = ffi::EVP_DigestFinal_ex(self.ctx, res.as_mut_ptr(), &mut len);
+            assert_eq!(len as usize, md_len);
+            assert_eq!(r, 1);
+        }
+        self.state = Finalized;
         res
     }
 
-    /**
-     * Return the digest of all bytes added to this hasher since its last
-     * initialization and its context for reuse
-     */
-    pub fn finalize_reuse(self) -> (Vec<u8>, HasherContext) {
-        let mut res = repeat(0u8).take(self.len as usize).collect::<Vec<_>>();
-        unsafe {
-            ffi::EVP_DigestFinal_ex(self.ctx.ptr, res.as_mut_ptr(), ptr::null_mut())
+    /// Returns the hash of the data written since creation or
+    /// the last `finish` and resets the hasher.
+    #[inline]
+    pub fn finish(&mut self) -> Vec<u8> {
+        self.finalize()
+    }
+}
+
+impl Writer for Hasher {
+    #[inline]
+    fn write_all(&mut self, buf: &[u8]) -> Result<(), IoError> {
+        self.update(buf);
+        Ok(())
+    }
+}
+
+impl Clone for Hasher {
+    fn clone(&self) -> Hasher {
+        let ctx = unsafe {
+            let ctx = ffi::EVP_MD_CTX_create();
+            assert!(!ctx.is_null());
+            let r = ffi::EVP_MD_CTX_copy_ex(ctx, self.ctx);
+            assert_eq!(r, 1);
+            ctx
         };
-        (res, self.ctx)
+        Hasher { ctx: ctx, md: self.md, type_: self.type_, state: self.state }
     }
 }
 
-/**
- * Hashes the supplied input data using hash t, returning the resulting hash
- * value
- */
+impl Drop for Hasher {
+    fn drop(&mut self) {
+        unsafe {
+            if self.state != Finalized {
+                let mut buf: Vec<u8> = repeat(0).take(self.type_.md_len()).collect();
+                let mut len = 0;
+                ffi::EVP_DigestFinal_ex(self.ctx, buf.as_mut_ptr(), &mut len);
+            }
+            ffi::EVP_MD_CTX_destroy(self.ctx);
+        }
+    }
+}
+
+/// Computes the hash of the `data` with the hash `t`.
 pub fn hash(t: HashType, data: &[u8]) -> Vec<u8> {
     let mut h = Hasher::new(t);
-    h.update(data);
-    h.finalize()
+    let _ = h.write_all(data);
+    h.finish()
 }
 
 #[cfg(test)]
 mod tests {
     use serialize::hex::{FromHex, ToHex};
+    use super::{hash, Hasher, HashType};
+    use std::old_io::Writer;
 
-    struct HashTest {
-        input: Vec<u8>,
-        expected_output: String
+    fn hash_test(hashtype: HashType, hashtest: &(&str, &str)) {
+        let res = hash(hashtype, &*hashtest.0.from_hex().unwrap());
+        assert_eq!(res.to_hex(), hashtest.1);
     }
 
-    #[allow(non_snake_case)]
-    fn HashTest(input: &str, output: &str) -> HashTest {
-        HashTest { input: input.from_hex().unwrap(),
-                   expected_output: output.to_string() }
+    fn hash_recycle_test(h: &mut Hasher, hashtest: &(&str, &str)) {
+        let _ = h.write_all(&*hashtest.0.from_hex().unwrap());
+        let res = h.finish();
+        assert_eq!(res.to_hex(), hashtest.1);
     }
 
-    fn compare(calced_raw: Vec<u8>, hashtest: &HashTest) {
-        let calced = calced_raw.as_slice().to_hex().to_string();
+    // Test vectors from http://www.nsrl.nist.gov/testdata/
+    #[allow(non_upper_case_globals)]
+    const md5_tests: [(&'static str, &'static str); 13] = [
+        ("", "d41d8cd98f00b204e9800998ecf8427e"),
+        ("7F", "83acb6e67e50e31db6ed341dd2de1595"),
+        ("EC9C", "0b07f0d4ca797d8ac58874f887cb0b68"),
+        ("FEE57A", "e0d583171eb06d56198fc0ef22173907"),
+        ("42F497E0", "7c430f178aefdf1487fee7144e9641e2"),
+        ("C53B777F1C", "75ef141d64cb37ec423da2d9d440c925"),
+        ("89D5B576327B", "ebbaf15eb0ed784c6faa9dc32831bf33"),
+        ("5D4CCE781EB190", "ce175c4b08172019f05e6b5279889f2c"),
+        ("81901FE94932D7B9", "cd4d2f62b8cdb3a0cf968a735a239281"),
+        ("C9FFDEE7788EFB4EC9", "e0841a231ab698db30c6c0f3f246c014"),
+        ("66AC4B7EBA95E53DC10B", "a3b3cea71910d9af56742aa0bb2fe329"),
+        ("A510CD18F7A56852EB0319", "577e216843dd11573574d3fb209b97d8"),
+        ("AAED18DBE8938C19ED734A8D", "6f80fb775f27e0a4ce5c2f42fc72c5f1")
+    ];
 
-        if calced != hashtest.expected_output {
-            println!("Test failed - {} != {}", calced, hashtest.expected_output);
+    #[test]
+    fn test_md5() {
+        for test in md5_tests.iter() {
+            hash_test(HashType::MD5, test);
         }
-
-        assert!(calced == hashtest.expected_output);
     }
 
-    fn hash_test(hashtype: super::HashType, hashtest: &HashTest) {
-        let calced_raw = super::hash(hashtype, hashtest.input.as_slice());
-        compare(calced_raw, hashtest);
-    }
-
-    fn hash_reuse_test(ctx: super::HasherContext, hashtype: super::HashType,
-                       hashtest: &HashTest) -> super::HasherContext {
-        let mut h = super::Hasher::with_context(ctx, hashtype);
-        h.update(hashtest.input.as_slice());
-        let (calced_raw, ctx) = h.finalize_reuse();
-
-        compare(calced_raw, hashtest);
-
-        ctx
+    #[test]
+    fn test_md5_recycle() {
+        let mut h = Hasher::new(HashType::MD5);
+        for test in md5_tests.iter() {
+            hash_recycle_test(&mut h, test);
+        }
     }
 
-    pub fn hash_writer(t: super::HashType, data: &[u8]) -> Vec<u8> {
-        let mut h = super::Hasher::new(t);
-        h.write_all(data).unwrap();
-        h.finalize()
+    #[test]
+    fn test_finish_twice() {
+        let mut h = Hasher::new(HashType::MD5);
+        let _ = h.write_all(&*md5_tests[6].0.from_hex().unwrap());
+        let _ = h.finish();
+        let res = h.finish();
+        let null = hash(HashType::MD5, &[]);
+        assert_eq!(res, null);
     }
 
-    // Test vectors from http://www.nsrl.nist.gov/testdata/
     #[test]
-    fn test_md5() {
-        let tests = [
-            HashTest("", "d41d8cd98f00b204e9800998ecf8427e"),
-            HashTest("7F", "83acb6e67e50e31db6ed341dd2de1595"),
-            HashTest("EC9C", "0b07f0d4ca797d8ac58874f887cb0b68"),
-            HashTest("FEE57A", "e0d583171eb06d56198fc0ef22173907"),
-            HashTest("42F497E0", "7c430f178aefdf1487fee7144e9641e2"),
-            HashTest("C53B777F1C", "75ef141d64cb37ec423da2d9d440c925"),
-            HashTest("89D5B576327B", "ebbaf15eb0ed784c6faa9dc32831bf33"),
-            HashTest("5D4CCE781EB190", "ce175c4b08172019f05e6b5279889f2c"),
-            HashTest("81901FE94932D7B9", "cd4d2f62b8cdb3a0cf968a735a239281"),
-            HashTest("C9FFDEE7788EFB4EC9", "e0841a231ab698db30c6c0f3f246c014"),
-            HashTest("66AC4B7EBA95E53DC10B", "a3b3cea71910d9af56742aa0bb2fe329"),
-            HashTest("A510CD18F7A56852EB0319", "577e216843dd11573574d3fb209b97d8"),
-            HashTest("AAED18DBE8938C19ED734A8D", "6f80fb775f27e0a4ce5c2f42fc72c5f1")];
-
-        let mut ctx = super::HasherContext::new();
-
-        for test in tests.iter() {
-            ctx = hash_reuse_test(ctx, super::HashType::MD5, test);
+    fn test_clone() {
+        let i = 7;
+        let inp = md5_tests[i].0.from_hex().unwrap();
+        assert!(inp.len() > 2);
+        let p = inp.len() / 2;
+        let h0 = Hasher::new(HashType::MD5);
+
+        println!("Clone a new hasher");
+        let mut h1 = h0.clone();
+        let _ = h1.write_all(&inp[..p]);
+        {
+            println!("Clone an updated hasher");
+            let mut h2 = h1.clone();
+            let _ = h2.write_all(&inp[p..]);
+            let res = h2.finish();
+            assert_eq!(res.to_hex(), md5_tests[i].1);
         }
+        let _ = h1.write_all(&inp[p..]);
+        let res = h1.finish();
+        assert_eq!(res.to_hex(), md5_tests[i].1);
+
+        println!("Clone a finished hasher");
+        let mut h3 = h1.clone();
+        let _ = h3.write_all(&*md5_tests[i + 1].0.from_hex().unwrap());
+        let res = h3.finish();
+        assert_eq!(res.to_hex(), md5_tests[i + 1].1);
     }
 
     #[test]
     fn test_sha1() {
         let tests = [
-            HashTest("616263", "a9993e364706816aba3e25717850c26c9cd0d89d"),
+            ("616263", "a9993e364706816aba3e25717850c26c9cd0d89d"),
             ];
 
         for test in tests.iter() {
-            hash_test(super::HashType::SHA1, test);
+            hash_test(HashType::SHA1, test);
         }
     }
 
     #[test]
     fn test_sha256() {
         let tests = [
-            HashTest("616263", "ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad")
+            ("616263", "ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad")
             ];
 
         for test in tests.iter() {
-            hash_test(super::HashType::SHA256, test);
+            hash_test(HashType::SHA256, test);
         }
     }
 
     #[test]
     fn test_ripemd160() {
         let tests = [
-            HashTest("616263", "8eb208f7e05d987a9b044a8e98c6b087f15a0bfc")
+            ("616263", "8eb208f7e05d987a9b044a8e98c6b087f15a0bfc")
             ];
 
         for test in tests.iter() {
-            hash_test(super::HashType::RIPEMD160, test);
+            hash_test(HashType::RIPEMD160, test);
         }
     }
-
-    #[test]
-    fn test_writer() {
-        let tv = "rust-openssl".as_bytes();
-        let ht = super::HashType::RIPEMD160;
-        assert!(hash_writer(ht, tv) == super::hash(ht, tv));
-    }
 }