1 files changed, 260 insertions, 53 deletions

diff --git a/ext/digest/digest.c b/ext/digest/digest.c
index 0b910e8c47..28f6022754 100644
--- a/ext/digest/digest.c
+++ b/ext/digest/digest.c
@@ -29,6 +29,69 @@ RUBY_EXTERN void Init_digest_base(void);
  * Document-module: Digest
  *
  * This module provides a framework for message digest libraries.
+ *
+ * You may want to look at OpenSSL::Digest as it supports more algorithms.
+ *
+ * A cryptographic hash function is a procedure that takes data and returns a
+ * fixed bit string: the hash value, also known as _digest_. Hash functions
+ * are also called one-way functions, it is easy to compute a digest from
+ * a message, but it is infeasible to generate a message from a digest.
+ *
+ * == Examples
+ *
+ *   require 'digest'
+ *
+ *   # Compute a complete digest
+ *   Digest::SHA256.digest 'message'       #=> "\xABS\n\x13\xE4Y..."
+ *
+ *   sha256 = Digest::SHA256.new
+ *   sha256.digest 'message'               #=> "\xABS\n\x13\xE4Y..."
+ *
+ *   # Other encoding formats
+ *   Digest::SHA256.hexdigest 'message'    #=> "ab530a13e459..."
+ *   Digest::SHA256.base64digest 'message' #=> "q1MKE+RZFJgr..."
+ *
+ *   # Compute digest by chunks
+ *   md5 = Digest::MD5.new
+ *   md5.update 'message1'
+ *   md5 << 'message2'                     # << is an alias for update
+ *
+ *   md5.hexdigest                         #=> "94af09c09bb9..."
+ *
+ *   # Compute digest for a file
+ *   sha256 = Digest::SHA256.file 'testfile'
+ *   sha256.hexdigest
+ *
+ * Additionally digests can be encoded in "bubble babble" format as a sequence
+ * of consonants and vowels which is more recognizable and comparable than a
+ * hexadecimal digest.
+ *
+ *   require 'digest/bubblebabble'
+ *
+ *   Digest::SHA256.bubblebabble 'message' #=> "xopoh-fedac-fenyh-..."
+ *
+ * See the bubble babble specification at
+ * http://web.mit.edu/kenta/www/one/bubblebabble/spec/jrtrjwzi/draft-huima-01.txt.
+ *
+ * == Digest algorithms
+ *
+ * Different digest algorithms (or hash functions) are available:
+ *
+ * MD5::
+ *  See RFC 1321 The MD5 Message-Digest Algorithm
+ * RIPEMD-160::
+ *   As Digest::RMD160.
+ *   See http://homes.esat.kuleuven.be/~bosselae/ripemd160.html.
+ * SHA1::
+ *   See FIPS 180 Secure Hash Standard.
+ * SHA2 family::
+ *   See FIPS 180 Secure Hash Standard which defines the following algorithms:
+ *   * SHA512
+ *   * SHA384
+ *   * SHA256
+ *
+ * The latest versions of the FIPS publications can be found here:
+ * http://csrc.nist.gov/publications/PubsFIPS.html.
  */
 
 static VALUE
@@ -36,7 +99,7 @@ hexencode_str_new(VALUE str_digest)
 {
     char *digest;
     size_t digest_len;
-    int i;
+    size_t i;
     VALUE str;
     char *p;
     static const char hex[] = {
@@ -52,7 +115,7 @@ hexencode_str_new(VALUE str_digest)
         rb_raise(rb_eRuntimeError, "digest string too long");
     }
 
-    str = rb_str_new(0, digest_len * 2);
+    str = rb_usascii_str_new(0, digest_len * 2);
 
     for (i = 0, p = RSTRING_PTR(str); i < digest_len; i++) {
         unsigned char byte = digest[i];
@@ -61,6 +124,8 @@ hexencode_str_new(VALUE str_digest)
         p[i + i + 1] = hex[byte & 0x0f];
     }
 
+    RB_GC_GUARD(str_digest);
+
     return str;
 }
 
@@ -76,6 +141,8 @@ rb_digest_s_hexencode(VALUE klass, VALUE str)
     return hexencode_str_new(str);
 }
 
+NORETURN(static void rb_digest_instance_method_unimpl(VALUE self, const char *method));
+
 /*
  * Document-module: Digest::Instance
  *
@@ -83,6 +150,13 @@ rb_digest_s_hexencode(VALUE klass, VALUE str)
  * object to calculate message digest values.
  */
 
+static void
+rb_digest_instance_method_unimpl(VALUE self, const char *method)
+{
+    rb_raise(rb_eRuntimeError, "%s does not implement %s()",
+             rb_obj_classname(self), method);
+}
+
 /*
  * call-seq:
  *     digest_obj.update(string) -> digest_obj
@@ -97,7 +171,9 @@ rb_digest_s_hexencode(VALUE klass, VALUE str)
 static VALUE
 rb_digest_instance_update(VALUE self, VALUE str)
 {
-    rb_raise(rb_eRuntimeError, "%s does not implement update()", RSTRING_PTR(rb_inspect(self)));
+    rb_digest_instance_method_unimpl(self, "update");
+
+    UNREACHABLE;
 }
 
 /*
@@ -115,7 +191,9 @@ rb_digest_instance_update(VALUE self, VALUE str)
 static VALUE
 rb_digest_instance_finish(VALUE self)
 {
-    rb_raise(rb_eRuntimeError, "%s does not implement finish()", RSTRING_PTR(rb_inspect(self)));
+    rb_digest_instance_method_unimpl(self, "finish");
+
+    UNREACHABLE;
 }
 
 /*
@@ -129,7 +207,9 @@ rb_digest_instance_finish(VALUE self)
 static VALUE
 rb_digest_instance_reset(VALUE self)
 {
-    rb_raise(rb_eRuntimeError, "%s does not implement reset()", RSTRING_PTR(rb_inspect(self)));
+    rb_digest_instance_method_unimpl(self, "reset");
+
+    UNREACHABLE;
 }
 
 /*
@@ -170,10 +250,7 @@ rb_digest_instance_digest(int argc, VALUE *argv, VALUE self)
         value = rb_funcall(self, id_finish, 0);
         rb_funcall(self, id_reset, 0);
     } else {
-        VALUE clone = rb_obj_clone(self);
-
-        value = rb_funcall(clone, id_finish, 0);
-        rb_funcall(clone, id_reset, 0);
+        value = rb_funcall(rb_obj_clone(self), id_finish, 0);
     }
 
     return value;
@@ -218,10 +295,7 @@ rb_digest_instance_hexdigest(int argc, VALUE *argv, VALUE self)
         value = rb_funcall(self, id_finish, 0);
         rb_funcall(self, id_reset, 0);
     } else {
-        VALUE clone = rb_obj_clone(self);
-
-        value = rb_funcall(clone, id_finish, 0);
-        rb_funcall(clone, id_reset, 0);
+        value = rb_funcall(rb_obj_clone(self), id_finish, 0);
     }
 
     return hexencode_str_new(value);
@@ -231,8 +305,8 @@ rb_digest_instance_hexdigest(int argc, VALUE *argv, VALUE self)
  * call-seq:
  *     digest_obj.hexdigest! -> string
  *
- * Returns the resulting hash value and resets the digest to the
- * initial state.
+ * Returns the resulting hash value in a hex-encoded form and resets
+ * the digest to the initial state.
  */
 static VALUE
 rb_digest_instance_hexdigest_bang(VALUE self)
@@ -266,7 +340,7 @@ rb_digest_instance_inspect(VALUE self)
 {
     VALUE str;
     size_t digest_len = 32;	/* about this size at least */
-    char *cname;
+    const char *cname;
 
     cname = rb_obj_classname(self);
 
@@ -300,7 +374,8 @@ rb_digest_instance_equal(VALUE self, VALUE other)
         str2 = rb_digest_instance_digest(0, 0, other);
     } else {
         str1 = rb_digest_instance_to_s(self);
-        str2 = other;
+        str2 = rb_check_string_type(other);
+        if (NIL_P(str2)) return Qfalse;
     }
 
     /* never blindly assume that subclass methods return strings */
@@ -308,8 +383,8 @@ rb_digest_instance_equal(VALUE self, VALUE other)
     StringValue(str2);
 
     if (RSTRING_LEN(str1) == RSTRING_LEN(str2) &&
-	rb_str_cmp(str1, str2) == 0) {
-	return Qtrue;
+        rb_str_cmp(str1, str2) == 0) {
+        return Qtrue;
     }
     return Qfalse;
 }
@@ -331,7 +406,7 @@ rb_digest_instance_digest_length(VALUE self)
 
     /* never blindly assume that #digest() returns a string */
     StringValue(digest);
-    return INT2NUM(RSTRING_LEN(digest));
+    return LONG2NUM(RSTRING_LEN(digest));
 }
 
 /*
@@ -358,7 +433,9 @@ rb_digest_instance_length(VALUE self)
 static VALUE
 rb_digest_instance_block_length(VALUE self)
 {
-    rb_raise(rb_eRuntimeError, "%s does not implement block_length()", RSTRING_PTR(rb_inspect(self)));
+    rb_digest_instance_method_unimpl(self, "block_length");
+
+    UNREACHABLE;
 }
 
 /*
@@ -409,7 +486,14 @@ rb_digest_class_s_digest(int argc, VALUE *argv, VALUE klass)
 static VALUE
 rb_digest_class_s_hexdigest(int argc, VALUE *argv, VALUE klass)
 {
-    return hexencode_str_new(rb_funcall2(klass, id_digest, argc, argv));
+    return hexencode_str_new(rb_funcallv(klass, id_digest, argc, argv));
+}
+
+/* :nodoc: */
+static VALUE
+rb_digest_class_init(VALUE self)
+{
+    return self;
 }
 
 /*
@@ -417,29 +501,101 @@ rb_digest_class_s_hexdigest(int argc, VALUE *argv, VALUE klass)
  *
  * This abstract class provides a common interface to message digest
  * implementation classes written in C.
+ *
+ * ==Write a Digest subclass in C
+ * Digest::Base provides a common interface to message digest
+ * classes written in C. These classes must provide a struct
+ * of type rb_digest_metadata_t:
+ *  typedef int (*rb_digest_hash_init_func_t)(void *);
+ *  typedef void (*rb_digest_hash_update_func_t)(void *, unsigned char *, size_t);
+ *  typedef int (*rb_digest_hash_finish_func_t)(void *, unsigned char *);
+ *
+ *  typedef struct {
+ *    int api_version;
+ *    size_t digest_len;
+ *    size_t block_len;
+ *    size_t ctx_size;
+ *    rb_digest_hash_init_func_t init_func;
+ *    rb_digest_hash_update_func_t update_func;
+ *    rb_digest_hash_finish_func_t finish_func;
+ *  } rb_digest_metadata_t;
+ *
+ * This structure must be set as an instance variable named +metadata+
+ * (without the +@+ in front of the name). By example:
+ *   static const rb_digest_metadata_t sha1 = {
+ *      RUBY_DIGEST_API_VERSION,
+ *      SHA1_DIGEST_LENGTH,
+ *      SHA1_BLOCK_LENGTH,
+ *      sizeof(SHA1_CTX),
+ *      (rb_digest_hash_init_func_t)SHA1_Init,
+ *      (rb_digest_hash_update_func_t)SHA1_Update,
+ *      (rb_digest_hash_finish_func_t)SHA1_Finish,
+ *  };
+ *
+ *
+ *  rb_ivar_set(cDigest_SHA1, rb_intern("metadata"),
+ *		rb_digest_make_metadata(&sha1));
  */
 
+#ifdef DIGEST_USE_RB_EXT_RESOLVE_SYMBOL
+static const rb_data_type_t metadata_type = {
+    "digest/metadata",
+    {0},
+};
+
+RUBY_FUNC_EXPORTED VALUE
+rb_digest_wrap_metadata(const rb_digest_metadata_t *meta)
+{
+    return rb_obj_freeze(TypedData_Wrap_Struct(0, &metadata_type, (void *)meta));
+}
+#endif
+
+static rb_digest_metadata_t *
+get_metadata_ptr(VALUE obj)
+{
+    rb_digest_metadata_t *algo;
+
+#ifdef DIGEST_USE_RB_EXT_RESOLVE_SYMBOL
+    if (!rb_typeddata_is_kind_of(obj, &metadata_type)) return 0;
+    algo = RTYPEDDATA_DATA(obj);
+#else
+# undef RUBY_UNTYPED_DATA_WARNING
+# define RUBY_UNTYPED_DATA_WARNING 0
+    Data_Get_Struct(obj, rb_digest_metadata_t, algo);
+#endif
+
+    return algo;
+}
+
 static rb_digest_metadata_t *
 get_digest_base_metadata(VALUE klass)
 {
     VALUE p;
-    VALUE obj;
+    VALUE obj = Qnil;
     rb_digest_metadata_t *algo;
 
-    for (p = klass; p; p = RCLASS(p)->super) {
+    for (p = klass; !NIL_P(p); p = rb_class_superclass(p)) {
         if (rb_ivar_defined(p, id_metadata)) {
             obj = rb_ivar_get(p, id_metadata);
             break;
         }
     }
 
-    if (!p)
+    if (NIL_P(p))
         rb_raise(rb_eRuntimeError, "Digest::Base cannot be directly inherited in Ruby");
 
-    Data_Get_Struct(obj, rb_digest_metadata_t, algo);
+    algo = get_metadata_ptr(obj);
+    if (!algo) {
+        if (p == klass)
+            rb_raise(rb_eTypeError, "%"PRIsVALUE"::metadata is not initialized properly",
+                     klass);
+        else
+            rb_raise(rb_eTypeError, "%"PRIsVALUE"(%"PRIsVALUE")::metadata is not initialized properly",
+                     klass, p);
+    }
 
     switch (algo->api_version) {
-      case 2:
+      case 3:
         break;
 
       /*
@@ -453,6 +609,27 @@ get_digest_base_metadata(VALUE klass)
     return algo;
 }
 
+static rb_digest_metadata_t *
+get_digest_obj_metadata(VALUE obj)
+{
+    return get_digest_base_metadata(rb_obj_class(obj));
+}
+
+static const rb_data_type_t digest_type = {
+    "digest",
+    {0, RUBY_TYPED_DEFAULT_FREE, 0,},
+    0, 0,
+    (RUBY_TYPED_FREE_IMMEDIATELY|RUBY_TYPED_WB_PROTECTED),
+};
+
+static inline void
+algo_init(const rb_digest_metadata_t *algo, void *pctx)
+{
+    if (algo->init_func(pctx) != 1) {
+        rb_raise(rb_eRuntimeError, "Digest initialization failed.");
+    }
+}
+
 static VALUE
 rb_digest_base_alloc(VALUE klass)
 {
@@ -461,15 +638,14 @@ rb_digest_base_alloc(VALUE klass)
     void *pctx;
 
     if (klass == rb_cDigest_Base) {
-	rb_raise(rb_eNotImpError, "Digest::Base is an abstract class");
+        rb_raise(rb_eNotImpError, "Digest::Base is an abstract class");
     }
 
     algo = get_digest_base_metadata(klass);
 
-    pctx = xmalloc(algo->ctx_size);
-    algo->init_func(pctx);
-
-    obj = Data_Wrap_Struct(klass, 0, free, pctx);
+    obj = rb_data_typed_object_zalloc(klass, algo->ctx_size, &digest_type);
+    pctx = RTYPEDDATA_DATA(obj);
+    algo_init(algo, pctx);
 
     return obj;
 }
@@ -485,44 +661,57 @@ rb_digest_base_copy(VALUE copy, VALUE obj)
 
     rb_check_frozen(copy);
 
-    algo = get_digest_base_metadata(rb_obj_class(copy));
+    algo = get_digest_obj_metadata(copy);
+    if (algo != get_digest_obj_metadata(obj))
+        rb_raise(rb_eTypeError, "different algorithms");
 
-    Data_Get_Struct(obj, void, pctx1);
-    Data_Get_Struct(copy, void, pctx2);
+    TypedData_Get_Struct(obj, void, &digest_type, pctx1);
+    TypedData_Get_Struct(copy, void, &digest_type, pctx2);
     memcpy(pctx2, pctx1, algo->ctx_size);
 
     return copy;
 }
 
-/* :nodoc: */
+/*
+ * call-seq: digest_base.reset -> digest_base
+ *
+ * Reset the digest to its initial state and return +self+.
+ */
 static VALUE
 rb_digest_base_reset(VALUE self)
 {
     rb_digest_metadata_t *algo;
     void *pctx;
 
-    algo = get_digest_base_metadata(rb_obj_class(self));
+    algo = get_digest_obj_metadata(self);
 
-    Data_Get_Struct(self, void, pctx);
+    TypedData_Get_Struct(self, void, &digest_type, pctx);
 
-    algo->init_func(pctx);
+    algo_init(algo, pctx);
 
     return self;
 }
 
-/* :nodoc: */
+/*
+ * call-seq:
+ *   digest_base.update(string) -> digest_base
+ *   digest_base << string -> digest_base
+ *
+ * Update the digest using given _string_ and return +self+.
+ */
 static VALUE
 rb_digest_base_update(VALUE self, VALUE str)
 {
     rb_digest_metadata_t *algo;
     void *pctx;
 
-    algo = get_digest_base_metadata(rb_obj_class(self));
+    algo = get_digest_obj_metadata(self);
 
-    Data_Get_Struct(self, void, pctx);
+    TypedData_Get_Struct(self, void, &digest_type, pctx);
 
     StringValue(str);
     algo->update_func(pctx, (unsigned char *)RSTRING_PTR(str), RSTRING_LEN(str));
+    RB_GC_GUARD(str);
 
     return self;
 }
@@ -535,56 +724,75 @@ rb_digest_base_finish(VALUE self)
     void *pctx;
     VALUE str;
 
-    algo = get_digest_base_metadata(rb_obj_class(self));
+    algo = get_digest_obj_metadata(self);
 
-    Data_Get_Struct(self, void, pctx);
+    TypedData_Get_Struct(self, void, &digest_type, pctx);
 
     str = rb_str_new(0, algo->digest_len);
     algo->finish_func(pctx, (unsigned char *)RSTRING_PTR(str));
 
     /* avoid potential coredump caused by use of a finished context */
-    algo->init_func(pctx);
+    algo_init(algo, pctx);
 
     return str;
 }
 
-/* :nodoc: */
+/*
+ * call-seq: digest_base.digest_length -> Integer
+ *
+ * Return the length of the hash value in bytes.
+ */
 static VALUE
 rb_digest_base_digest_length(VALUE self)
 {
     rb_digest_metadata_t *algo;
 
-    algo = get_digest_base_metadata(rb_obj_class(self));
+    algo = get_digest_obj_metadata(self);
 
-    return INT2NUM(algo->digest_len);
+    return SIZET2NUM(algo->digest_len);
 }
 
-/* :nodoc: */
+/*
+ * call-seq: digest_base.block_length -> Integer
+ *
+ * Return the block length of the digest in bytes.
+ */
 static VALUE
 rb_digest_base_block_length(VALUE self)
 {
     rb_digest_metadata_t *algo;
 
-    algo = get_digest_base_metadata(rb_obj_class(self));
+    algo = get_digest_obj_metadata(self);
 
-    return INT2NUM(algo->block_len);
+    return SIZET2NUM(algo->block_len);
 }
 
 void
 Init_digest(void)
 {
+#undef rb_intern
     id_reset           = rb_intern("reset");
     id_update          = rb_intern("update");
     id_finish          = rb_intern("finish");
     id_digest          = rb_intern("digest");
     id_hexdigest       = rb_intern("hexdigest");
     id_digest_length   = rb_intern("digest_length");
+    id_metadata        = rb_id_metadata();
+    InitVM(digest);
+}
 
+void
+InitVM_digest(void)
+{
     /*
      * module Digest
      */
     rb_mDigest = rb_define_module("Digest");
 
+#ifdef HAVE_RB_EXT_RACTOR_SAFE
+    rb_ext_ractor_safe(true);
+#endif
+
     /* module functions */
     rb_define_module_function(rb_mDigest, "hexencode", rb_digest_s_hexencode, 1);
 
@@ -619,14 +827,13 @@ Init_digest(void)
      * class Digest::Class
      */
     rb_cDigest_Class = rb_define_class_under(rb_mDigest, "Class", rb_cObject);
+    rb_define_method(rb_cDigest_Class, "initialize",  rb_digest_class_init, 0);
     rb_include_module(rb_cDigest_Class, rb_mDigest_Instance);
 
     /* class methods */
     rb_define_singleton_method(rb_cDigest_Class, "digest", rb_digest_class_s_digest, -1);
     rb_define_singleton_method(rb_cDigest_Class, "hexdigest", rb_digest_class_s_hexdigest, -1);
 
-    id_metadata = rb_intern("metadata");
-
     /* class Digest::Base < Digest::Class */
     rb_cDigest_Base = rb_define_class_under(rb_mDigest, "Base", rb_cDigest_Class);