From b5c9b7ffd95b2934e503e8524ac0f571ed2b9f91 Mon Sep 17 00:00:00 2001
From: Kaz Kylheku <kaz@kylheku.com>
Date: Thu, 29 Aug 2019 06:58:27 -0700
Subject: sha256/md5: allow characters and byte values.

* chksum.c (sha256_utf8_byte_callback, md5_utf8_byte_callback):
New static functions.
(sha256_hash, md5_hash): Support character and integer
objects.

* txr.1: Documented.
---
 chksum.c | 42 ++++++++++++++++++++++++++++++++++++++++++
 txr.1    | 10 ++++++----
 2 files changed, 48 insertions(+), 4 deletions(-)

diff --git a/chksum.c b/chksum.c
index 8934583b..4af08d47 100644
--- a/chksum.c
+++ b/chksum.c
@@ -183,6 +183,14 @@ val sha256_begin(void)
   return cobj(coerce(mem_t *, ps256), sha256_ctx_s, &sha256_ops);
 }
 
+static int sha256_utf8_byte_callback(int b, mem_t *ctx)
+{
+  SHA256_t *ps256 = coerce(SHA256_t *, ctx);
+  unsigned char uc = b;
+  SHA256_update(ps256, &uc, 1);
+  return 1;
+}
+
 val sha256_hash(val ctx, val obj)
 {
   val self = lit("sha256-hash");
@@ -201,6 +209,19 @@ val sha256_hash(val ctx, val obj)
   case BUF:
     sha256_szmax_upd(ps256, obj->b.data, c_unum(obj->b.len));
     break;
+  case CHR:
+    utf8_encode(c_chr(obj), sha256_utf8_byte_callback, coerce(mem_t *, ps256));
+    break;
+  case NUM:
+    {
+      cnum n = c_num(obj);
+      unsigned char uc = n;
+      if (n < 0 || n > 255)
+        uw_throwf(error_s, lit("~a: byte value ~s out of range"),
+                  self, obj, nao);
+      SHA256_update(ps256, &uc, 1);
+    }
+    break;
   default:
     uw_throwf(error_s, lit("~a: cannot hash ~s, only buffer and strings"),
               self, obj, nao);
@@ -424,6 +445,14 @@ val md5_begin(void)
   return cobj(coerce(mem_t *, pmd5), md5_ctx_s, &md5_ops);
 }
 
+static int md5_utf8_byte_callback(int b, mem_t *ctx)
+{
+  MD5_t *ps256 = coerce(MD5_t *, ctx);
+  unsigned char uc = b;
+  MD5_update(ps256, &uc, 1);
+  return 1;
+}
+
 val md5_hash(val ctx, val obj)
 {
   val self = lit("md5-hash");
@@ -442,6 +471,19 @@ val md5_hash(val ctx, val obj)
   case BUF:
     md5_szmax_upd(pmd5, obj->b.data, c_unum(obj->b.len));
     break;
+  case CHR:
+    utf8_encode(c_chr(obj), md5_utf8_byte_callback, coerce(mem_t *, pmd5));
+    break;
+  case NUM:
+    {
+      cnum n = c_num(obj);
+      unsigned char uc = n;
+      if (n < 0 || n > 255)
+        uw_throwf(error_s, lit("~a: byte value ~s out of range"),
+                  self, obj, nao);
+      MD5_update(pmd5, &uc, 1);
+    }
+    break;
   default:
     uw_throwf(error_s, lit("~a: cannot hash ~s, only buffer and strings"),
               self, obj, nao);
diff --git a/txr.1 b/txr.1
index b8b7c120..df165629 100644
--- a/txr.1
+++ b/txr.1
@@ -52411,8 +52411,9 @@ by including
 .meta obj
 into the digest calculation. The
 .meta obj
-argument may be a character string, whose UTF-8 representation is digested,
-or a buffer object, whose contents are digested.
+argument may be: a character or character string, whose UTF-8 representation is
+digested; a buffer object, whose contents are digested; or an integer,
+representing a byte value in the range 0 to 255 included in the digest.
 The
 .code sha256-hash
 function may be called multiple times to include any mixture of
@@ -52467,8 +52468,9 @@ by including
 .meta obj
 into the digest calculation. The
 .meta obj
-argument may be a character string, whose UTF-8 representation is digested,
-or a buffer object, whose contents are digested.
+argument may be: a character or character string, whose UTF-8 representation is
+digested; a buffer object, whose contents are digested; or an integer,
+representing a byte value in the range 0 to 255 included in the digest.
 The
 .code md5-hash
 function may be called multiple times to include any mixture of
-- 
cgit v1.2.3