if the SGF file contains BOM, q5Go can't read it properly.

BOM explanation: <a href="https://en.wikipedia.org/wiki/Byte_order_mark" rel="nofollow

Here's a better version. <div class="snippet-clipboard-content notranslate positio

reading SGF file with BOM failed about q5go HOT 7 CLOSED

bernds commented on August 28, 2024

reading SGF file with BOM failed

from q5go.

Comments (7)

bernds commented on August 28, 2024

What is BOM? And how does it fail?

from q5go.

bernds commented on August 28, 2024

Could you attach the SGF file in question?

from q5go.

missdeer commented on August 28, 2024

BOM explanation: https://en.wikipedia.org/wiki/Byte_order_mark

You may download a SGF file contains BOM at https://www.101weiqi.com/file/sgf/96/962980_1_时越_VS_杨鼎新.sgf

from q5go.

bernds commented on August 28, 2024

Hmm. That really doesn't seem valid according to the SGF spec (which really implies that SGF is a binary format, not a text format), but I guess if it's fairly common, we should try to special-case it.

from q5go.

bernds commented on August 28, 2024

Here's a patch that just ignores such a BOM. Does this work for the files you have? If not, could you attach more test cases?

diff --git a/src/sgfload.cc b/src/sgfload.cc
index a7cd3ae..ac48a8d 100644
--- a/src/sgfload.cc
+++ b/src/sgfload.cc
@@ -93,6 +93,20 @@ sgf *load_sgf (std::istream &in)
 {
     char nextch;
 
+    if (!in.get (nextch))
+	throw premature_eof ();
+    /* Look for (and discard) a UTF-8 BOM.  Bogus, since SGF is a
+       binary file format, but it occurs in the wild.  */
+    if (nextch == (char)0xEF) {
+	if (!in.get (nextch))
+	    throw premature_eof ();
+	if (nextch != (char)0xBB)
+	    throw broken_sgf ();
+	if (!in.get (nextch))
+	    throw premature_eof ();
+	if (nextch != (char)0xBF)
+	    throw broken_sgf ();
+    }
     nextch = skip_whitespace (in);
     if (nextch != '(')
 	throw broken_sgf ();

from q5go.

bernds commented on August 28, 2024

Here's a better version.

diff --git a/src/sgfload.cc b/src/sgfload.cc
index a7cd3ae..cdcacb3 100644
--- a/src/sgfload.cc
+++ b/src/sgfload.cc
@@ -93,7 +93,24 @@ sgf *load_sgf (std::istream &in)
 {
     char nextch;
 
-    nextch = skip_whitespace (in);
+    if (!in.get (nextch))
+       throw premature_eof ();
+    /* Look for (and discard) a UTF-8 BOM.  Bogus, since SGF is a
+       binary file format, but it occurs in the wild.  */
+    if (nextch == (char)0xEF) {
+       if (!in.get (nextch))
+           throw premature_eof ();
+       if (nextch != (char)0xBB)
+           throw broken_sgf ();
+       if (!in.get (nextch))
+           throw premature_eof ();
+       if (nextch != (char)0xBF)
+           throw broken_sgf ();
+       if (!in.get (nextch))
+           throw premature_eof ();
+    }
+    if (isspace (nextch))
+       nextch = skip_whitespace (in);
     if (nextch != '(')
        throw broken_sgf ();
     sgf *s = new sgf (parse_gametree (in));

from q5go.

missdeer commented on August 28, 2024

so far, I only found UTF-8 encoded SGF files may contains BOM.
the patch works for me.

from q5go.

Recommend Projects