erl_interface: Enable decode of unicode atoms

No API changes or additions. Just the ability for erl_interface to decode
unicode atoms and convert them into latin1 strings to preserve backward
compatibility for the existing API.

5 files changed, 78 insertions, 73 deletions

diff --git a/lib/erl_interface/src/decode/decode_atom.c b/lib/erl_interface/src/decode/decode_atom.c
index c2e6a0426e..84edf1766a 100644
--- a/lib/erl_interface/src/decode/decode_atom.c
+++ b/lib/erl_interface/src/decode/decode_atom.c
@@ -21,24 +21,76 @@
 #include "eiext.h"
 #include "putget.h"
 
+static int utf8_to_latin1(char* dest, const char* source, unsigned len);
+
 int ei_decode_atom(const char *buf, int *index, char *p)
 {
   const char *s = buf + *index;
   const char *s0 = s;
   int len;
 
-  if (get8(s) != ERL_ATOM_EXT) return -1;
+  switch (get8(s)) {
+  case ERL_ATOM_EXT:
+      len = get16be(s);
+      if (len > MAXATOMLEN) return -1;
+      if (p) {
+	  memmove(p,s,len); 
+	  p[len] = (char)0;
+      }
+      break;
+
+  case ERL_SMALL_ATOM_EXT:
+      len = get8(s);
+      if (p) {
+	  memmove(p,s,len); 
+	  p[len] = (char)0;
+      }
+      break;
+
+  case ERL_UNICODE_ATOM_EXT:
+      len = get16be(s);
 
-  len = get16be(s);
+      if (len > 2*MAXATOMLEN) return -1;
 
-  if (len > MAXATOMLEN) return -1;
+      if (p && utf8_to_latin1(p, s, len) < 0) return -1;
+      break;
 
-  if (p) {
-    memmove(p,s,len); 
-    p[len] = (char)0;
+  default:
+      return -1;
   }
+
   s += len;
   *index += s-s0;
-  
-  return 0; 
+  return 0;
+}
+
+int ei_internal_get_atom(const char** bufp, char* p)
+{
+    int ix = 0;
+    if (ei_decode_atom(*bufp, &ix, p) < 0) return -1;
+    *bufp += ix;
+    return 0;
+}
+
+static int utf8_to_latin1(char* dest, const char* source, unsigned slen)
+{
+    const char* dest_end = dest + MAXATOMLEN - 1;
+
+    while (slen > 0 && dest < dest_end) {
+	if ((source[0] & 0x80) == 0) {
+	    *dest++ = *source++;
+	    --slen;
+	}
+	else if (slen > 1 &&
+		 (source[0] & 0xFE) == 0xC2 &&
+		 (source[1] & 0xC0) == 0x80) {
+	    *dest++ = (char) ((source[0] << 6) | (source[1] & 0x3F));
+	    source += 2;
+	    slen -= 2;
+	}
+	else return -1;
+    }
+    *dest = 0;
+    return 0;
 }
+
diff --git a/lib/erl_interface/src/decode/decode_boolean.c b/lib/erl_interface/src/decode/decode_boolean.c
index 9fd09c63f1..0a7a06f1d4 100644
--- a/lib/erl_interface/src/decode/decode_boolean.c
+++ b/lib/erl_interface/src/decode/decode_boolean.c
@@ -26,32 +26,20 @@ int ei_decode_boolean(const char *buf, int *index, int *p)
 {
   const char *s = buf + *index;
   const char *s0 = s;
-  int len;
+  char tbuf[MAXATOMLEN+1];
   int t;
 
-  if (get8(s) != ERL_ATOM_EXT) return -1;
+  if (get_atom(&s, tbuf) < 0) return -1;
 
-  len = get16be(s);
-
-  switch (len) {
-  case 4:
-    /* typecast makes ansi happy */
-    if (strncmp((char*)s,"true",4)) return -1;
-    t = 1;
-    break;
-
-  case 5:
-    if (strncmp((char*)s,"false",5)) return -1;
-    t = 0;
-    break;
-    
-  default:
-    return -1;
-  }
-  
-  s += len;
+  if (memcmp(tbuf, "true", 5) == 0)
+      t = 1;
+  else if (memcmp(tbuf, "false", 6) == 0)
+      t = 0;
+  else
+      return -1;
+      
   if (p) *p = t;
   *index += s-s0;
-
   return 0; 
 }
+
diff --git a/lib/erl_interface/src/decode/decode_pid.c b/lib/erl_interface/src/decode/decode_pid.c
index 9ed1c36db6..a762ae499e 100644
--- a/lib/erl_interface/src/decode/decode_pid.c
+++ b/lib/erl_interface/src/decode/decode_pid.c
@@ -21,6 +21,7 @@
 #include "eiext.h"
 #include "putget.h"
 
+
 int ei_decode_pid(const char *buf, int *index, erlang_pid *p)
 {
   const char *s = buf + *index;
@@ -30,17 +31,7 @@ int ei_decode_pid(const char *buf, int *index, erlang_pid *p)
   if (get8(s) != ERL_PID_EXT) return -1;
 
   /* first the nodename */
-  if (get8(s) != ERL_ATOM_EXT) return -1;
-
-  len = get16be(s);
-
-  if (len > MAXATOMLEN) return -1;
-  
-  if (p) {
-    memmove(p->node, s, len);
-    p->node[len] = (char)0;
-  }
-  s += len;
+  if (get_atom(&s, p->node) < 0) return -1;
   
   /* now the numbers: num (4), serial (4), creation (1) */
   if (p) {
diff --git a/lib/erl_interface/src/decode/decode_port.c b/lib/erl_interface/src/decode/decode_port.c
index 28abed801a..6eb2bc9197 100644
--- a/lib/erl_interface/src/decode/decode_port.c
+++ b/lib/erl_interface/src/decode/decode_port.c
@@ -30,17 +30,7 @@ int ei_decode_port(const char *buf, int *index, erlang_port *p)
   if (get8(s) != ERL_PORT_EXT) return -1;
 
   /* first the nodename */
-  if (get8(s) != ERL_ATOM_EXT) return -1;
-
-  len = get16be(s);
-
-  if (len > MAXATOMLEN) return -1;
-
-  if (p) {
-    memmove(p->node, s, len);
-    p->node[len] = (char)0;
-  }
-  s += len;
+  if (get_atom(&s, p->node) < 0) return -1;
   
   /* now the numbers: num (4), creation (1) */
   if (p) {
diff --git a/lib/erl_interface/src/decode/decode_ref.c b/lib/erl_interface/src/decode/decode_ref.c
index 7b15808bc5..df3c30777b 100644
--- a/lib/erl_interface/src/decode/decode_ref.c
+++ b/lib/erl_interface/src/decode/decode_ref.c
@@ -21,6 +21,7 @@
 #include "eiext.h"
 #include "putget.h"
 
+
 int ei_decode_ref(const char *buf, int *index, erlang_ref *p)
 {
   const char *s = buf + *index;
@@ -30,18 +31,8 @@ int ei_decode_ref(const char *buf, int *index, erlang_ref *p)
   switch (get8(s)) {
     case ERL_REFERENCE_EXT:
 
-      /* first the nodename */
-      if (get8(s) != ERL_ATOM_EXT) return -1;
-
-      len = get16be(s);
-
-      if (len > MAXATOMLEN) return -1;
-
-      if (p) {
-	  memmove(p->node, s, len);
-	  p->node[len] = (char)0;
-      }
-      s += len;
+      /* nodename */
+      if (get_atom(&s, p->node) < 0) return -1;
   
       /* now the numbers: num (4), creation (1) */
       if (p) {
@@ -62,15 +53,7 @@ int ei_decode_ref(const char *buf, int *index, erlang_ref *p)
       if (p) p->len = count;
 
       /* then the nodename */
-      if (get8(s) != ERL_ATOM_EXT) return -1;
-      len = get16be(s);
-      if (len > MAXATOMLEN) return -1;
-
-      if (p) {
-	  memmove(p->node, s, len);
-	  p->node[len] = (char)0;
-      }
-      s += len;
+      if (get_atom(&s, p->node) < 0) return -1;
 
       /* creation */
       if (p) {
@@ -95,3 +78,4 @@ int ei_decode_ref(const char *buf, int *index, erlang_ref *p)
       return -1;
   }
 }
+