From df3442b0eb1ec4c6f8f4547cc20e5ad0824e7cd5 Mon Sep 17 00:00:00 2001
From: Sverker Eriksson <sverker@erlang.org>
Date: Wed, 5 Jul 2017 16:08:59 +0200
Subject: erts: Refactor erts_unicode_list_to_buf

to get bytes written when truncated.
---
 erts/emulator/beam/bif.c    | 17 +++++++------
 erts/emulator/beam/global.h |  2 +-
 erts/emulator/beam/utils.c  | 60 +++++++++++++++++++++++++--------------------
 3 files changed, 44 insertions(+), 35 deletions(-)

(limited to 'erts')

diff --git a/erts/emulator/beam/bif.c b/erts/emulator/beam/bif.c
index 40dd4129d2..8380efaf13 100644
--- a/erts/emulator/beam/bif.c
+++ b/erts/emulator/beam/bif.c
@@ -3021,17 +3021,17 @@ BIF_RETTYPE list_to_atom_1(BIF_ALIST_1)
 {
     Eterm res;
     byte *buf = (byte *) erts_alloc(ERTS_ALC_T_TMP, MAX_ATOM_SZ_LIMIT);
-    Sint i = erts_unicode_list_to_buf(BIF_ARG_1, buf, MAX_ATOM_CHARACTERS);
-
+    Sint written;
+    int i = erts_unicode_list_to_buf(BIF_ARG_1, buf, MAX_ATOM_CHARACTERS,
+                                     &written);
     if (i < 0) {
 	erts_free(ERTS_ALC_T_TMP, (void *) buf);
-	i = erts_list_length(BIF_ARG_1);
-	if (i > MAX_ATOM_CHARACTERS) {
+	if (i == -2) {
 	    BIF_ERROR(BIF_P, SYSTEM_LIMIT);
 	}
 	BIF_ERROR(BIF_P, BADARG);
     }
-    res = erts_atom_put(buf, i, ERTS_ATOM_ENC_UTF8, 1);
+    res = erts_atom_put(buf, written, ERTS_ATOM_ENC_UTF8, 1);
     ASSERT(is_atom(res));
     erts_free(ERTS_ALC_T_TMP, (void *) buf);
     BIF_RET(res);
@@ -3042,8 +3042,9 @@ BIF_RETTYPE list_to_atom_1(BIF_ALIST_1)
 BIF_RETTYPE list_to_existing_atom_1(BIF_ALIST_1)
 {
     byte *buf = (byte *) erts_alloc(ERTS_ALC_T_TMP, MAX_ATOM_SZ_LIMIT);
-    Sint i = erts_unicode_list_to_buf(BIF_ARG_1, buf, MAX_ATOM_CHARACTERS);
-
+    Sint written;
+    int i = erts_unicode_list_to_buf(BIF_ARG_1, buf, MAX_ATOM_CHARACTERS,
+                                     &written);
     if (i < 0) {
     error:
 	erts_free(ERTS_ALC_T_TMP, (void *) buf);
@@ -3051,7 +3052,7 @@ BIF_RETTYPE list_to_existing_atom_1(BIF_ALIST_1)
     } else {
 	Eterm a;
 	
-	if (erts_atom_get((char *) buf, i, &a, ERTS_ATOM_ENC_UTF8)) {
+	if (erts_atom_get((char *) buf, written, &a, ERTS_ATOM_ENC_UTF8)) {
 	    erts_free(ERTS_ALC_T_TMP, (void *) buf);
 	    BIF_RET(a);
 	} else {
diff --git a/erts/emulator/beam/global.h b/erts/emulator/beam/global.h
index fc95535ec3..fcb88712e9 100644
--- a/erts/emulator/beam/global.h
+++ b/erts/emulator/beam/global.h
@@ -1287,7 +1287,7 @@ int erts_utf8_to_latin1(byte* dest, const byte* source, int slen);
 
 void bin_write(fmtfn_t, void*, byte*, size_t);
 Sint intlist_to_buf(Eterm, char*, Sint); /* most callers pass plain char*'s */
-Sint erts_unicode_list_to_buf(Eterm list, byte *buf, Sint len);
+int erts_unicode_list_to_buf(Eterm list, byte *buf, Sint len, Sint* written);
 
 struct Sint_buf {
 #if defined(ARCH_64)
diff --git a/erts/emulator/beam/utils.c b/erts/emulator/beam/utils.c
index 457cada745..ba7596cde4 100644
--- a/erts/emulator/beam/utils.c
+++ b/erts/emulator/beam/utils.c
@@ -3634,30 +3634,40 @@ intlist_to_buf(Eterm list, char *buf, Sint len)
     return -2;			/* not enough space */
 }
 
-/* Fill buf with the contents of the unicode list.
- * Return the number of bytes in the buffer,
- * or -1 for type error,
- * or -2 for not enough buffer space (buffer contains truncated result).
+/** @brief Fill buf with the UTF8 contents of the unicode list
+ * @param len Max number of characters to write.
+ * @param written NULL or bytes written.
+ * @return 0 ok,
+ *        -1 type error,
+ *        -2 list too long, only \c len characters written
  */
-Sint
-erts_unicode_list_to_buf(Eterm list, byte *buf, Sint len)
+int
+erts_unicode_list_to_buf(Eterm list, byte *buf, Sint len, Sint* written)
 {
     Eterm* listptr;
     Sint sz = 0;
+    Sint val;
+    int res;
 
-    if (is_nil(list)) {
-	return 0;
-    }
-    if (is_not_list(list)) {
-	return -1;
-    }
-    listptr = list_val(list);
+    while (1) {
+        if (is_nil(list)) {
+            res = 0;
+            break;
+        }
+        if (is_not_list(list)) {
+            res = -1;
+            break;
+        }
+        listptr = list_val(list);
 
-    while (len-- > 0) {
-	Sint val;
+        if (len-- <= 0) {
+            res = -2;
+            break;
+        }
 
 	if (is_not_small(CAR(listptr))) {
-	    return -1;
+	    res = -1;
+            break;
 	}
 	val = signed_val(CAR(listptr));
 	if (0 <= val && val < 0x80) {
@@ -3669,7 +3679,8 @@ erts_unicode_list_to_buf(Eterm list, byte *buf, Sint len)
 	    sz += 2;
 	} else if (val < 0x10000UL) {
 	    if (0xD800 <= val && val <= 0xDFFF) {
-		return -1;
+		res = -1;
+                break;
 	    }
 	    buf[sz+0] = 0xE0 | (val >> 12);
 	    buf[sz+1] = 0x80 | ((val >> 6) & 0x3F);
@@ -3682,18 +3693,15 @@ erts_unicode_list_to_buf(Eterm list, byte *buf, Sint len)
 	    buf[sz+3] = 0x80 | (val & 0x3F);
 	    sz += 4;
 	} else {
-	    return -1;
+            res = -1;
+            break;
 	}
 	list = CDR(listptr);
-	if (is_nil(list)) {
-	    return sz;
-	}
-	if (is_not_list(list)) {
-	    return -1;
-	}
-	listptr = list_val(list);
     }
-    return -2;			/* not enough space */
+
+    if (written)
+        *written = sz;
+    return res;
 }
 
 /*
-- 
cgit v1.2.3