Merge branch 'master' into issue/58-bugfix-turtle-bool-parser

lib/Parser/Turtle.php

@@ -554,6 +554,8 @@ protected function parseQuotedLiteral()
     {
         $label = $this->parseQuotedString();
 
+        $this->skipWSC();
+
         // Check for presence of a language tag or datatype
         $c = $this->peek();
 
@@ -906,7 +908,7 @@ protected function parseQNameOrBoolean()
                 if ('true' == $value || 'false' == $value) {
                     // Unread last character
                     $this->unread($c);
-                    
+
                     return [
                         'type' => 'literal',
                         'value' => $value,
@@ -943,6 +945,11 @@ protected function parseQNameOrBoolean()
                 }
                 $c = $this->read();
             }
+
+            // Last char of name must not be a dot
+            if (mb_substr($localName, -1) === '.') {
+                throw new Exception("Turtle Parse Error: last character of QName must not be a dot", $this->line, $this->column - 1);
+            }
         }
 
         // Unread last character
@@ -1220,6 +1227,7 @@ public static function isNameChar($c)
             self::isNameStartChar($c)
             || $o >= 0x30 && $o <= 0x39     // 0-9
             || '-' == $c
+            || '.' == $c                    // dots are allowed in the middle of a name, not as start char
             || 0x00B7 == $o;
     }
 

test/fixtures/turtle/bad-07.ttl

@@ -1,4 +1,3 @@
 # paths are not in turtle
 @prefix : <http://example.org/stuff/1.0/> .
-:a.:b.:c .
 :a^:b^:c .

test/fixtures/turtle/gh51-sweetrdf-dot-in-name-bad.ttl

@@ -0,0 +1,9 @@
+@prefix : <http://example.org/#> .
+
+# A dot can't be the first nor the last character of the "after the semicolon part of a prefixed name" but is allowed in the middle
+# See https://www.w3.org/TR/turtle/#grammar-production-PN_LOCAL
+# See https://github.com/sweetrdf/easyrdf/issues/51
+# The triples below should not parse!
+:SubjectWithEndDot. :predicate :Object .
+:Subject :predicateWithEndDot. :Object .
+:Subject :predicate :ObjectWithEndDot. .

test/fixtures/turtle/gh51-sweetrdf-dot-in-name.out

@@ -0,0 +1,2 @@
+<http://example.org/#Subject.WithADot> <http://example.org/#predicate.withADot> <http://example.org/#Object.WithADot> .
+<http://example.org/#Subject.With.Dots> <http://example.org/#predicate.with.dots> <http://example.org/#Object.With.Dots> .

test/fixtures/turtle/gh51-sweetrdf-dot-in-name.ttl

@@ -0,0 +1,7 @@
+@prefix : <http://example.org/#> .
+
+# A dot can't be the first nor the last character of the "after the semicolon part of a prefixed name" but is allowed in the middle
+# See https://www.w3.org/TR/turtle/#grammar-production-PN_LOCAL
+# See https://github.com/sweetrdf/easyrdf/issues/51
+:Subject.WithADot :predicate.withADot :Object.WithADot .
+:Subject.With.Dots :predicate.with.dots :Object.With.Dots .

test/fixtures/turtle/gh52-sweetrdf-whitespace-langtag.out

@@ -0,0 +1,2 @@
+<http://example.org/#a> <http://www.w3.org/2000/01/rdf-schema#label> "Test"@en .
+<http://example.org/#a> <http://www.w3.org/2000/01/rdf-schema#label> "Test"^^<http://www.w3.org/2001/XMLSchema#string> .

test/fixtures/turtle/gh52-sweetrdf-whitespace-langtag.ttl

@@ -0,0 +1,9 @@
+@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
+@prefix xs: <http://www.w3.org/2001/XMLSchema#> .
+@prefix : <http://example.org/#> .
+
+# Allow whitespace between literal and language tag
+:a rdfs:label "Test"      @en .
+
+# Allow whitespace between literal and datatype IRI
+:a rdfs:label "Test"      ^^xs:string .

tests/EasyRdf/Parser/TurtleTest.php

@@ -434,7 +434,7 @@ public function testBad07()
         // paths are not in turtle
         $this->expectException('EasyRdf\Parser\Exception');
         $this->expectExceptionMessage(
-            'Turtle Parse Error: object for statement missing on line 3, column 5'
+            "Turtle Parse Error: expected an RDF value here, found '^' on line 3, column 3"
         );
         $this->parseTurtle('turtle/bad-07.ttl');
     }
@@ -580,14 +580,45 @@ public function testIssue140()
     }
 
     /**
+     * Allow usage of dot in middle of name
+     *
+     * @see https://github.com/sweetrdf/easyrdf/issues/51
+     */
+    public function testIssue51()
+    {
+        $this->turtleTestCase('gh51-sweetrdf-dot-in-name');
+    }
+
+    public function testIssue51Bad()
+    {
+        // Test long literals with missing end
+        $this->expectException('EasyRdf\Parser\Exception');
+        $this->expectExceptionMessage(
+            'Turtle Parse Error: last character of QName must not be a dot on line 7, column 20'
+        );
+        $this->parseTurtle('turtle/gh51-sweetrdf-dot-in-name-bad.ttl');
+    }
+
+    /**
+     * Allow whitespace(s) between quoted string literal and langtag or datatype
+     *
+     * @see https://github.com/sweetrdf/easyrdf/issues/52
+     */
+    public function testIssue52()
+    {
+        $this->turtleTestCase('gh52-sweetrdf-whitespace-langtag');
+    }
+
+    /**
+     * Bug in Turtle parser for boolean value that is directly followed by a semicolon
+     *
      * @see https://github.com/sweetrdf/easyrdf/issues/58
-     * Notice this is an issue reported in the sweetrdf/easyrdf fork
      */
     public function testIssue58()
     {
         // Test file should parse without exceptions
         $this->expectNotToPerformAssertions();
 
-        $this->parseTurtle("turtle/gh58-sweetrdf-bool-parser.ttl");
+        $this->parseTurtle('turtle/gh58-sweetrdf-bool-parser.ttl');
     }
 }