Accept '_' as alias/typo for '-' in lang strings. Replace on parsing.

Fixes Issue #0000574
http://bugs.librdf.org/mantis/view.php?id=574

5 files changed, 24 insertions, 6 deletions

diff --git a/src/raptor_ntriples.c b/src/raptor_ntriples.c
index 50154ecc..e23debd9 100644
--- a/src/raptor_ntriples.c
+++ b/src/raptor_ntriples.c
@@ -87,10 +87,12 @@ raptor_ntriples_term_valid(unsigned char c, int position,
       break;
 
     case RAPTOR_TERM_CLASS_LANGUAGE:
-      /* ends on first non [a-zA-Z]+ ('-' [a-zA-Z0-9]+ )? */
+      /* ends on first non [a-zA-Z]+ ('-' [a-zA-Z0-9]+ )?
+       * Accept _ as synonym / typo for -.
+       */
       result = IS_ASCII_ALPHA(c);
       if(position)
-        result = (result || IS_ASCII_DIGIT(c) || c == '-');
+        result = (result || IS_ASCII_DIGIT(c) || c == '-' || c == '_');
       break;
 
     default:
@@ -575,10 +577,13 @@ raptor_ntriples_parse_term(raptor_world* world, raptor_locator* locator,
 
           /* Normalize language to lowercase
            * http://www.w3.org/TR/rdf-concepts/#dfn-language-identifier
+           * Also convert _ to - as synonym / typo.
            */
           for(q = object_literal_language; *q; q++) {
             if(IS_ASCII_UPPER(*q))
               *q = TO_ASCII_LOWER(*q);
+            if(*q == '_')
+              *q = '-';
           }
 
         }
diff --git a/src/raptor_term.c b/src/raptor_term.c
index d64da8fc..d6fa52b2 100644
--- a/src/raptor_term.c
+++ b/src/raptor_term.c
@@ -199,13 +199,22 @@ raptor_new_term_from_counted_literal(raptor_world* world,
     *new_literal = '\0';
 
   if(language) {
+    unsigned char c;
+    unsigned char* l;
+    
     new_language = RAPTOR_MALLOC(unsigned char*, language_len + 1);
     if(!new_language) {
       RAPTOR_FREE(char*, new_literal);
       return NULL;
     }
-    memcpy(new_language, language, language_len);
-    new_language[language_len] = '\0';
+
+    l = new_language;
+    while((c = *language++)) {
+      if(c == '_')
+        c = '-';
+      *l++ = c;
+    }
+    *l = '\0';
   } else
     language_len = 0;
 
diff --git a/tests/ntriples/Makefile.am b/tests/ntriples/Makefile.am
index 18f025cb..15b34cf8 100644
--- a/tests/ntriples/Makefile.am
+++ b/tests/ntriples/Makefile.am
@@ -26,14 +26,16 @@ nt2013-ex1.nt \
 nt2013-ex2.nt \
 nt2013-ex3.nt \
 nt2013-ex4.nt \
-bug-562.nt
+bug-562.nt \
+bug-574.nt
 
 NT_OUT_FILES = test.out \
 nt2013-ex1.out \
 nt2013-ex2.out \
 nt2013-ex3.out \
 nt2013-ex4.out \
-bug-562.out
+bug-562.out \
+bug-574.out
 
 NT_BAD_TEST_FILES=bad-00.nt bad-02.nt bad-03.nt bad-04.nt \
 bad-05.nt bad-06.nt bad-07.nt
diff --git a/tests/ntriples/bug-574.nt b/tests/ntriples/bug-574.nt
new file mode 100644
index 00000000..9f204971
--- /dev/null
+++ b/tests/ntriples/bug-574.nt
@@ -0,0 +1 @@
+<http://example.org/resource> <http://www.w3.org/2000/01/rdf-schema#label> "SOMETHING"@zh_pinyin .
diff --git a/tests/ntriples/bug-574.out b/tests/ntriples/bug-574.out
new file mode 100644
index 00000000..09aa702e
--- /dev/null
+++ b/tests/ntriples/bug-574.out
@@ -0,0 +1 @@
+<http://example.org/resource> <http://www.w3.org/2000/01/rdf-schema#label> "SOMETHING"@zh-pinyin .