Improve dot support.

Extend test suite to cover more of the new Turtle draft.

git-svn-id: http://svn.drobilla.net/serd/trunk@433 490d8e77-9747-427b-9fa3-0b8f29cee8a0

16 files changed, 155 insertions, 36 deletions

diff --git a/src/reader.c b/src/reader.c
index 5dd9578e..3a245877 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -671,7 +671,7 @@ read_PN_LOCAL(SerdReader* reader, Ref dest)
 	default:
 		if ((st = read_PLX(reader, dest)) > SERD_FAILURE) {
 			return st;
-		} else if (st != SERD_SUCCESS && !read_PN_CHARS(reader, dest)) {
+		} else if (st != SERD_SUCCESS && !read_PN_CHARS_BASE(reader, dest)) {
 			return SERD_FAILURE;
 		}
 	}
@@ -679,12 +679,10 @@ read_PN_LOCAL(SerdReader* reader, Ref dest)
 	while ((c = peek_byte(reader))) {  // Middle: (PN_CHARS | '.' | ';')*
 		if (/*c == '.' || */c == ':') {
 			push_byte(reader, dest, eat_byte_safe(reader, c));
-		} else if (!read_PN_CHARS(reader, dest)) {
-			if ((st = read_PLX(reader, dest)) > SERD_FAILURE) {
-				return st;
-			} else if (st != SERD_SUCCESS) {
-				break;
-			}
+		} else if ((st = read_PLX(reader, dest)) > SERD_FAILURE) {
+			return st;
+		} else if (st != SERD_SUCCESS && !read_PN_CHARS(reader, dest)) {
+			break;
 		}
 	}
 
@@ -707,8 +705,9 @@ read_PN_PREFIX(SerdReader* reader, Ref dest)
 		}
 	}
 
-	if (c == '.' && !read_PN_CHARS(reader, dest)) {  // Last: PN_CHARS
-		r_err(reader, SERD_ERR_BAD_SYNTAX, "invalid prefix character\n");
+	const SerdNode* const n = deref(reader, dest);
+	if (n->buf[n->n_bytes - 1] == '.' && !read_PN_CHARS(reader, dest)) {
+		r_err(reader, SERD_ERR_BAD_SYNTAX, "prefix ends with `.'\n");
 		return SERD_ERR_BAD_SYNTAX;
 	}
 
@@ -960,7 +959,7 @@ read_BLANK_NODE_LABEL(SerdReader* reader)
 
 	uint8_t c = peek_byte(reader);  // First: (PN_CHARS | '_' | [0-9])
 	if (is_digit(c) || c == '_') {
-		push_byte(reader, ref, c);
+		push_byte(reader, ref, eat_byte_safe(reader, c));
 	} else if (!read_PN_CHARS(reader, ref)) {
 		r_err(reader, SERD_ERR_BAD_SYNTAX, "invalid name start character\n");
 		return pop_node(reader, ref);
@@ -974,18 +973,18 @@ read_BLANK_NODE_LABEL(SerdReader* reader)
 		}
 	}
 
-	if (c == '.' && !read_PN_CHARS(reader, ref)) {  // Last: PN_CHARS
-		r_err(reader, SERD_ERR_BAD_SYNTAX, "invalid name character\n");
+	const SerdNode* n = deref(reader, ref);
+	if (n->buf[n->n_bytes - 1] == '.' && !read_PN_CHARS(reader, ref)) {
+		r_err(reader, SERD_ERR_BAD_SYNTAX, "name ends with `.'\n");
 		return pop_node(reader, ref);
 	}
 
 	if (reader->syntax == SERD_TURTLE) {
-		const char* const buf = (const char*)deref(reader, ref)->buf;
-		if (is_digit(buf[1])) {
-			if (buf[0] == 'b') {
-				((char*)buf)[0] = 'B';  // Prevent clash
+		if (is_digit(n->buf[1])) {
+			if (n->buf[0] == 'b') {
+				((char*)n->buf)[0] = 'B';  // Prevent clash
 				reader->seen_genid = true;
-			} else if (reader->seen_genid && buf[0] == 'B') {
+			} else if (reader->seen_genid && n->buf[0] == 'B') {
 				r_err(reader, SERD_ERR_ID_CLASH,
 				      "found both `b' and `B' blank IDs, prefix required\n");
 				return pop_node(reader, ref);
@@ -1152,29 +1151,30 @@ read_objectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
 static bool
 read_predicateObjectList(SerdReader* reader, ReadContext ctx, bool* ate_dot)
 {
-	TRY_RET(read_verb(reader, &ctx.predicate));
-	read_ws_star(reader);
-	TRY_THROW(read_objectList(reader, ctx, ate_dot));
-	ctx.predicate = pop_node(reader, ctx.predicate);
-	if (*ate_dot) {
-		return true;
-	}
-	while (eat_delim(reader, ';')) {
-		switch (peek_byte(reader)) {
-		case ';':
-			continue;
-		case '.': case ']':
+	uint8_t c;
+	while (true) {
+		TRY_THROW(read_verb(reader, &ctx.predicate));
+		read_ws_star(reader);
+
+		TRY_THROW(read_objectList(reader, ctx, ate_dot));
+		ctx.predicate = pop_node(reader, ctx.predicate);
+		if (*ate_dot) {
 			return true;
-		default:
-			TRY_THROW(read_verb(reader, &ctx.predicate));
+		}
+
+		do {
 			read_ws_star(reader);
-			TRY_THROW(read_objectList(reader, ctx, ate_dot));
-			ctx.predicate = pop_node(reader, ctx.predicate);
-			if (*ate_dot) {
+			switch (c = peek_byte(reader)) {
+			case 0:
+				return false;
+			case '.': case ']':
 				return true;
+			case ';':
+				eat_byte_safe(reader, c);
 			}
-		}
+		} while (c == ';');
 	}
+
 	pop_node(reader, ctx.predicate);
 	return true;
 except:
diff --git a/tests/extra/manifest.ttl b/tests/extra/manifest.ttl
new file mode 100644
index 00000000..f373f3a2
--- /dev/null
+++ b/tests/extra/manifest.ttl
@@ -0,0 +1,86 @@
+@prefix mf: <http://www.w3.org/2001/sw/DataAccess/tests/test-manifest#> .
+@prefix qt: <http://www.w3.org/2001/sw/DataAccess/tests/test-query#> .
+@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
+@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+@prefix rdft: <http://www.w3.org/ns/rdftest#> .
+
+<>
+	rdf:type mf:Manifest ;
+	rdfs:comment "Extra Turtle tests from Serd" ;
+	mf:entries (
+		<#turtle-syntax-bad-blank-label-dot-end>
+		<#turtle-syntax-bad-ln-dash-start>
+		<#turtle-syntax-bad-ln-escape>
+		<#turtle-syntax-bad-ns-dot-end>
+		<#turtle-syntax-bad-ns-dot-start>
+		<#turtle-syntax-bad-ns-missing-dot-end>
+		<#turtle-syntax-bad-ns-missing-dot-start>
+		<#turtle-syntax-blank-label>
+		<#turtle-syntax-ns-dots>
+	) .
+
+<#turtle-syntax-bad-blank-label-dot-end>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Blank node label must not end in dot" ;
+	mf:name "turtle-syntax-bad-blank-label-dot-end" ;
+	mf:action <turtle-syntax-bad-blank-label-dot-end.ttl> .
+
+<#turtle-syntax-bad-number-dot-in-anon>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Dot delimeter may not appear in anonymous nodes" ;
+	mf:name "turtle-syntax-bad-number-dot-in-anon" ;
+	mf:action <turtle-syntax-bad-number-dot-in-anon.ttl> .
+
+<#turtle-syntax-bad-ln-dash-start>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Local name must not begin with dash" ;
+	mf:name "turtle-syntax-bad-ln-dash-start" ;
+	mf:action <turtle-syntax-bad-ln-dash-start.ttl> .
+
+<#turtle-syntax-bad-ln-escape>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Bad hex escape in local name" ;
+	mf:name "turtle-syntax-bad-ln-escape" ;
+	mf:action <turtle-syntax-bad-ln-escape.ttl> .
+
+<#turtle-syntax-bad-ln-escape-start>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Bad hex escape at start of local name" ;
+	mf:name "turtle-syntax-bad-ln-escape-start" ;
+	mf:action <turtle-syntax-bad-ln-escape-start.ttl> .
+
+<#turtle-syntax-bad-ns-dot-end>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Prefix must not end in dot" ;
+	mf:name "turtle-syntax-bad-ns-dot-end" ;
+	mf:action <turtle-syntax-bad-ns-dot-end.ttl> .
+
+<#turtle-syntax-bad-ns-dot-start>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Prefix must not start with dot" ;
+	mf:name "turtle-syntax-bad-ns-dot-start" ;
+	mf:action <turtle-syntax-bad-ns-dot-start.ttl> .
+
+<#turtle-syntax-bad-ns-missing-dot-end>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Prefix must not end in dot (error in triple, not prefix directive like turtle-syntax-bad-ns-dot-end)" ;
+	mf:name "turtle-syntax-bad-missing-ns-dot-end" ;
+	mf:action <turtle-syntax-bad-missing-ns-dot-end.ttl> .
+
+<#turtle-syntax-bad-ns-missing-dot-start>
+	rdf:type rdft:TestTurtleNegativeSyntax ;
+	rdfs:comment "Prefix must not start with dot (error in triple, not prefix directive like turtle-syntax-bad-ns-dot-end)" ;
+	mf:name "turtle-syntax-bad-missing-ns-dot-start" ;
+	mf:action <turtle-syntax-bad-missing-ns-dot-start.ttl> .
+
+<#turtle-syntax-ns-dots>
+	rdf:type rdft:TestTurtlePositiveSyntax ;
+	rdfs:comment "Dots in namespace names" ;
+	mf:name "turtle-syntax-ns-dots" ;
+	mf:action <turtle-syntax-ns-dots.ttl> .
+
+<#turtle-syntax-blank-label>
+	rdf:type rdft:TestTurtlePositiveSyntax ;
+	rdfs:comment "Characters allowed in blank node labels" ;
+	mf:name "turtle-syntax-blank-label" ;
+	mf:action <turtle-syntax-blank-label.ttl> .
diff --git a/tests/extra/turtle-syntax-bad-blank-label-dot-end.ttl b/tests/extra/turtle-syntax-bad-blank-label-dot-end.ttl
new file mode 100644
index 00000000..613e775f
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-blank-label-dot-end.ttl
@@ -0,0 +1,2 @@
+@prefix : <http://example/> .
+_:b1. :p :o .
diff --git a/tests/extra/turtle-syntax-bad-ln-dash-start.ttl b/tests/extra/turtle-syntax-bad-ln-dash-start.ttl
new file mode 100644
index 00000000..291bdda3
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-ln-dash-start.ttl
@@ -0,0 +1,2 @@
+@prefix : <http://example/> .
+:s :p :-o .
diff --git a/tests/extra/turtle-syntax-bad-ln-escape-start.ttl b/tests/extra/turtle-syntax-bad-ln-escape-start.ttl
new file mode 100644
index 00000000..d1661bd7
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-ln-escape-start.ttl
@@ -0,0 +1,2 @@
+@prefix : <http://example/> .
+:s :p :%2o .
diff --git a/tests/extra/turtle-syntax-bad-ln-escape.ttl b/tests/extra/turtle-syntax-bad-ln-escape.ttl
new file mode 100644
index 00000000..585e28b5
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-ln-escape.ttl
@@ -0,0 +1,2 @@
+@prefix : <http://example/> .
+:s :p :o%2 .
diff --git a/tests/extra/turtle-syntax-bad-missing-ns-dot-end.ttl b/tests/extra/turtle-syntax-bad-missing-ns-dot-end.ttl
new file mode 100644
index 00000000..4a593eb7
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-missing-ns-dot-end.ttl
@@ -0,0 +1 @@
+undefined.:s undefined.:p undefined.:o .
diff --git a/tests/extra/turtle-syntax-bad-missing-ns-dot-start.ttl b/tests/extra/turtle-syntax-bad-missing-ns-dot-start.ttl
new file mode 100644
index 00000000..23514bcc
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-missing-ns-dot-start.ttl
@@ -0,0 +1 @@
+.undefined:s .undefined:p .undefined:o .
diff --git a/tests/extra/turtle-syntax-bad-ns-dot-end.ttl b/tests/extra/turtle-syntax-bad-ns-dot-end.ttl
new file mode 100644
index 00000000..a99fc98d
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-ns-dot-end.ttl
@@ -0,0 +1,2 @@
+@prefix eg. : <http://example/> .
+eg.:s eg.:p eg.:o .
diff --git a/tests/extra/turtle-syntax-bad-ns-dot-start.ttl b/tests/extra/turtle-syntax-bad-ns-dot-start.ttl
new file mode 100644
index 00000000..b3993dd7
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-ns-dot-start.ttl
@@ -0,0 +1,2 @@
+@prefix .eg : <http://example/> .
+.eg:s .eg:p .eg:o .
diff --git a/tests/extra/turtle-syntax-bad-number-dot-in-anon.ttl b/tests/extra/turtle-syntax-bad-number-dot-in-anon.ttl
new file mode 100644
index 00000000..46d15f9a
--- /dev/null
+++ b/tests/extra/turtle-syntax-bad-number-dot-in-anon.ttl
@@ -0,0 +1,6 @@
+@prefix : <http://example/> .
+
+:s
+	:p [
+		:p1 27.
+	] .
diff --git a/tests/extra/turtle-syntax-blank-label.nt b/tests/extra/turtle-syntax-blank-label.nt
new file mode 100644
index 00000000..05dcc396
--- /dev/null
+++ b/tests/extra/turtle-syntax-blank-label.nt
@@ -0,0 +1,3 @@
+_:0b <http://example/p> <http://example/o> .
+_:_b <http://example/p> <http://example/o> .
+_:b.0 <http://example/p> <http://example/o> .
diff --git a/tests/extra/turtle-syntax-blank-label.ttl b/tests/extra/turtle-syntax-blank-label.ttl
new file mode 100644
index 00000000..3e3e516d
--- /dev/null
+++ b/tests/extra/turtle-syntax-blank-label.ttl
@@ -0,0 +1,4 @@
+@prefix : <http://example/> .
+_:0b :p :o . # Starts with digit
+_:_b :p :o . # Starts with underscore
+_:b.0 :p :o . # Contains dot, ends with digit
diff --git a/tests/extra/turtle-syntax-ns-dots.ttl b/tests/extra/turtle-syntax-ns-dots.ttl
new file mode 100644
index 00000000..78465870
--- /dev/null
+++ b/tests/extra/turtle-syntax-ns-dots.ttl
@@ -0,0 +1,2 @@
+@prefix e.g: <http://example/> .
+e.g:s e.g:p e.g:o .
diff --git a/tests/extra/turtle-syntax-pname-dots.ttl b/tests/extra/turtle-syntax-pname-dots.ttl
new file mode 100644
index 00000000..a708cf70
--- /dev/null
+++ b/tests/extra/turtle-syntax-pname-dots.ttl
@@ -0,0 +1,2 @@
+@prefix : <http://example/> .
+:s.ubj :p.r.ed :o.bject.
diff --git a/wscript b/wscript
index b787760f..8abfa67d 100644
--- a/wscript
+++ b/wscript
@@ -385,7 +385,7 @@ def test_manifest(ctx, srcdir, testdir, report, test_base, parse_base):
 
 def test(ctx):
     blddir = autowaf.build_dir(APPNAME, 'tests')
-    for i in ['', 'bad', 'good', 'new', 'tests-ttl']:
+    for i in ['', 'bad', 'good', 'new', 'tests-ttl', 'extra']:
         try:
             os.makedirs(os.path.join(blddir, i))
         except:
@@ -522,6 +522,8 @@ def test(ctx):
                       rdf_turtle + 'coverage/tests/', 'http://example/base/')
         test_manifest(ctx, srcdir, 'tests-ttl', report,
                       rdf_turtle + 'tests-ttl/', 'http://example/base/')
+        test_manifest(ctx, srcdir, 'extra', report,
+                      rdf_turtle + 'extra/', 'http://example/base/')
 
         report.close()