rdflib 2.3.0-1c284ee8 → 2.3.0-344dec99

package/esm/serialize.js

@@ -40,7 +40,8 @@ contentType, callback, options) {
         return executeCallback(null, documentString);
       case TurtleContentType:
       case TurtleLegacyContentType:
-        sz.setFlags('si'); // Suppress = for sameAs and => for implies
+        // Suppress = for sameAs and => for implies; preserve any user-specified flags (e.g., 'o')
+        sz.setFlags('si' + (opts.flags ? ' ' + opts.flags : ''));
         documentString = sz.statementsToN3(newSts);
         return executeCallback(null, documentString);
       case NTriplesContentType:
@@ -48,7 +49,8 @@ contentType, callback, options) {
         documentString = sz.statementsToNTriples(newSts);
         return executeCallback(null, documentString);
       case JSONLDContentType:
-        sz.setFlags('si dr'); // turtle + dr (means no default, no relative prefix)
+        // turtle + dr (means no default, no relative prefix); preserve user flags
+        sz.setFlags('si dr' + (opts.flags ? ' ' + opts.flags : ''));
         documentString = sz.statementsToJsonld(newSts); // convert via turtle
         return executeCallback(null, documentString);
       case NQuadsContentType:

package/esm/serializer.js

@@ -20,7 +20,7 @@ export default function createSerializer(store) {
 ;
 export class Serializer {
   constructor(store) {
-    _defineProperty(this, "_notQNameChars", '\t\r\n !"#$%&\'()*.,+/;<=>?@[\\]^`{|}~');
+    _defineProperty(this, "_notQNameChars", '\t\r\n !"#$%&\'()*,+/;<=>?@[\\]^`{|}~');
     // issue#228
     _defineProperty(this, "_notNameChars", this._notQNameChars + ':');
     //  stringToN3:  String escaping for N3
@@ -54,6 +54,13 @@ export class Serializer {
     this.base = base;
     return this;
   }
+
+  /**
+   * Set serializer behavior flags. Letters can be combined with spaces.
+   * Examples: 'si', 'deinprstux', 'si dr', 'o'.
+   * Notable flags:
+   *  - 'o': do not abbreviate to a prefixed name when the local part contains a dot
+   */
   setFlags(flags) {
     this.flags = flags || '';
     return this;
@@ -244,6 +251,28 @@ export class Serializer {
   toN3(f) {
     return this.statementsToN3(f.statements);
   }
+  // Validate if a string is a valid PN_LOCAL per Turtle 1.1 spec
+  // Allows dots inside the local name but not as trailing character
+  // Also allows empty local names (for URIs ending in / or #)
+  isValidPNLocal(local) {
+    // Empty local name is valid (e.g., ex: for http://example.com/)
+    if (local.length === 0) return true;
+
+    // Cannot end with a dot
+    if (local[local.length - 1] === '.') return false;
+
+    // Check each character (allow dots mid-string)
+    for (var i = 0; i < local.length; i++) {
+      var ch = local[i];
+      // Dot is allowed unless it's the last character (checked above)
+      if (ch === '.') continue;
+      // Other characters must not be in the blacklist
+      if (this._notNameChars.indexOf(ch) >= 0) {
+        return false;
+      }
+    }
+    return true;
+  }
   explicitURI(uri) {
     if (this.flags.indexOf('r') < 0 && this.base) {
       uri = Uri.refTo(this.base, uri);
@@ -606,13 +635,17 @@ export class Serializer {
     if (j >= 0 && this.flags.indexOf('p') < 0 && (
     // Can split at namespace but only if http[s]: URI or file: or ws[s] (why not others?)
     uri.indexOf('http') === 0 || uri.indexOf('ws') === 0 || uri.indexOf('file') === 0)) {
-      var canSplit = true;
-      for (var k = j + 1; k < uri.length; k++) {
-        if (this._notNameChars.indexOf(uri[k]) >= 0) {
-          canSplit = false;
-          break;
-        }
-      }
+      var localid = uri.slice(j + 1);
+      var namesp = uri.slice(0, j + 1);
+      // Don't split if namespace is just the protocol (e.g., https://)
+      // A valid namespace should have content after the protocol
+      var minNamespaceLength = uri.indexOf('://') + 4; // e.g., "http://x" minimum
+      // Also don't split if namespace is the base directory (would serialize as relative URI)
+      var baseDir = this.base ? this.base.slice(0, Math.max(this.base.lastIndexOf('/'), this.base.lastIndexOf('#')) + 1) : null;
+      var namespaceIsBaseDir = baseDir && namesp === baseDir;
+      // If flag 'o' is present, forbid dots in local part when abbreviating
+      var forbidDotLocal = this.flags.indexOf('o') >= 0 && localid.indexOf('.') >= 0;
+      var canSplit = !namespaceIsBaseDir && !forbidDotLocal && namesp.length > minNamespaceLength && this.isValidPNLocal(localid);
       /*
             if (uri.slice(0, j + 1) === this.base + '#') { // base-relative
               if (canSplit) {
@@ -623,8 +656,6 @@ export class Serializer {
             }
       */
       if (canSplit) {
-        var localid = uri.slice(j + 1);
-        var namesp = uri.slice(0, j + 1);
         if (this.defaultNamespace && this.defaultNamespace === namesp && this.flags.indexOf('d') < 0) {
           // d -> suppress default
           if (this.flags.indexOf('k') >= 0 && this.keyords.indexOf(localid) < 0) {

package/lib/serialize.js

@@ -47,7 +47,8 @@ contentType, callback, options) {
         return executeCallback(null, documentString);
       case _types.TurtleContentType:
       case _types.TurtleLegacyContentType:
-        sz.setFlags('si'); // Suppress = for sameAs and => for implies
+        // Suppress = for sameAs and => for implies; preserve any user-specified flags (e.g., 'o')
+        sz.setFlags('si' + (opts.flags ? ' ' + opts.flags : ''));
         documentString = sz.statementsToN3(newSts);
         return executeCallback(null, documentString);
       case _types.NTriplesContentType:
@@ -55,7 +56,8 @@ contentType, callback, options) {
         documentString = sz.statementsToNTriples(newSts);
         return executeCallback(null, documentString);
       case _types.JSONLDContentType:
-        sz.setFlags('si dr'); // turtle + dr (means no default, no relative prefix)
+        // turtle + dr (means no default, no relative prefix); preserve user flags
+        sz.setFlags('si dr' + (opts.flags ? ' ' + opts.flags : ''));
         documentString = sz.statementsToJsonld(newSts); // convert via turtle
         return executeCallback(null, documentString);
       case _types.NQuadsContentType:

package/lib/serializer.d.ts

@@ -22,6 +22,12 @@ export class Serializer {
         string: NamedNode;
     };
     setBase(base: any): Serializer;
+    /**
+     * Set serializer behavior flags. Letters can be combined with spaces.
+     * Examples: 'si', 'deinprstux', 'si dr', 'o'.
+     * Notable flags:
+     *  - 'o': do not abbreviate to a prefixed name when the local part contains a dot
+     */
     setFlags(flags: any): Serializer;
     toStr(x: any): any;
     fromStr(s: any): any;
@@ -51,6 +57,7 @@ export class Serializer {
     toN3(f: any): string;
     _notQNameChars: string;
     _notNameChars: string;
+    isValidPNLocal(local: any): boolean;
     explicitURI(uri: any): string;
     statementsToNTriples(sts: any): string;
     statementsToN3(sts: any): string;

package/lib/serializer.js

@@ -30,7 +30,7 @@ function createSerializer(store) {
 ;
 class Serializer {
   constructor(store) {
-    (0, _defineProperty2.default)(this, "_notQNameChars", '\t\r\n !"#$%&\'()*.,+/;<=>?@[\\]^`{|}~');
+    (0, _defineProperty2.default)(this, "_notQNameChars", '\t\r\n !"#$%&\'()*,+/;<=>?@[\\]^`{|}~');
     // issue#228
     (0, _defineProperty2.default)(this, "_notNameChars", this._notQNameChars + ':');
     //  stringToN3:  String escaping for N3
@@ -64,6 +64,13 @@ class Serializer {
     this.base = base;
     return this;
   }
+
+  /**
+   * Set serializer behavior flags. Letters can be combined with spaces.
+   * Examples: 'si', 'deinprstux', 'si dr', 'o'.
+   * Notable flags:
+   *  - 'o': do not abbreviate to a prefixed name when the local part contains a dot
+   */
   setFlags(flags) {
     this.flags = flags || '';
     return this;
@@ -254,6 +261,28 @@ class Serializer {
   toN3(f) {
     return this.statementsToN3(f.statements);
   }
+  // Validate if a string is a valid PN_LOCAL per Turtle 1.1 spec
+  // Allows dots inside the local name but not as trailing character
+  // Also allows empty local names (for URIs ending in / or #)
+  isValidPNLocal(local) {
+    // Empty local name is valid (e.g., ex: for http://example.com/)
+    if (local.length === 0) return true;
+
+    // Cannot end with a dot
+    if (local[local.length - 1] === '.') return false;
+
+    // Check each character (allow dots mid-string)
+    for (var i = 0; i < local.length; i++) {
+      var ch = local[i];
+      // Dot is allowed unless it's the last character (checked above)
+      if (ch === '.') continue;
+      // Other characters must not be in the blacklist
+      if (this._notNameChars.indexOf(ch) >= 0) {
+        return false;
+      }
+    }
+    return true;
+  }
   explicitURI(uri) {
     if (this.flags.indexOf('r') < 0 && this.base) {
       uri = Uri.refTo(this.base, uri);
@@ -616,13 +645,17 @@ class Serializer {
     if (j >= 0 && this.flags.indexOf('p') < 0 && (
     // Can split at namespace but only if http[s]: URI or file: or ws[s] (why not others?)
     uri.indexOf('http') === 0 || uri.indexOf('ws') === 0 || uri.indexOf('file') === 0)) {
-      var canSplit = true;
-      for (var k = j + 1; k < uri.length; k++) {
-        if (this._notNameChars.indexOf(uri[k]) >= 0) {
-          canSplit = false;
-          break;
-        }
-      }
+      var localid = uri.slice(j + 1);
+      var namesp = uri.slice(0, j + 1);
+      // Don't split if namespace is just the protocol (e.g., https://)
+      // A valid namespace should have content after the protocol
+      var minNamespaceLength = uri.indexOf('://') + 4; // e.g., "http://x" minimum
+      // Also don't split if namespace is the base directory (would serialize as relative URI)
+      var baseDir = this.base ? this.base.slice(0, Math.max(this.base.lastIndexOf('/'), this.base.lastIndexOf('#')) + 1) : null;
+      var namespaceIsBaseDir = baseDir && namesp === baseDir;
+      // If flag 'o' is present, forbid dots in local part when abbreviating
+      var forbidDotLocal = this.flags.indexOf('o') >= 0 && localid.indexOf('.') >= 0;
+      var canSplit = !namespaceIsBaseDir && !forbidDotLocal && namesp.length > minNamespaceLength && this.isValidPNLocal(localid);
       /*
             if (uri.slice(0, j + 1) === this.base + '#') { // base-relative
               if (canSplit) {
@@ -633,8 +666,6 @@ class Serializer {
             }
       */
       if (canSplit) {
-        var localid = uri.slice(j + 1);
-        var namesp = uri.slice(0, j + 1);
         if (this.defaultNamespace && this.defaultNamespace === namesp && this.flags.indexOf('d') < 0) {
           // d -> suppress default
           if (this.flags.indexOf('k') >= 0 && this.keyords.indexOf(localid) < 0) {

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "rdflib",
   "description": "an RDF library for node.js. Suitable for client and server side.",
-  "version": "2.3.0-1c284ee8",
+  "version": "2.3.0-344dec99",
   "private": false,
   "browserslist": [
     "> 0.5%"
@@ -117,6 +117,7 @@
     "test:serialize:14": "cd ./tests/serialize && node ./data.js -in=t14.html -format=text/turtle -out=,t14.ttl && node diff ,t14.ttl t14-ref.ttl",
     "test:serialize:15": "cd ./tests/serialize && node ./data.js -in=t15.html -format=text/turtle -out=,t15.ttl && node diff ,t15.ttl t15-ref.ttl",
     "test:serialize:16": "cd ./tests/serialize && node ./data.js -in=t1.ttl -format=application/ld+json -out=,t1.jsonld && node diff ,t1.jsonld t16-ref.jsonld",
+    "test:serialize:17": "cd ./tests/serialize && node ./data.js -in=t17.ttl -format=application/rdf+xml  -out=,t17.xml && node diff ,t17.xml t17-ref.xml",
     "test:types": "tsc --noEmit --target es2019 --moduleResolution node tests/types/*.ts",
     "test:unit": "mocha --growl --require  ./tests/babel-register.js tests/unit/**-test.*",
     "test:unit:egp": "mocha --require ./tests/babel-register.js tests/unit/fetcher-egp-test.js",

package/src/serialize.ts

@@ -72,7 +72,8 @@ export default function serialize (
         return executeCallback(null, documentString)
       case TurtleContentType:
       case TurtleLegacyContentType:
-        sz.setFlags('si') // Suppress = for sameAs and => for implies
+        // Suppress = for sameAs and => for implies; preserve any user-specified flags (e.g., 'o')
+        sz.setFlags('si' + (opts.flags ? (' ' + opts.flags) : ''))
         documentString = sz.statementsToN3(newSts)
         return executeCallback(null, documentString)
       case NTriplesContentType:
@@ -80,7 +81,8 @@ export default function serialize (
         documentString = sz.statementsToNTriples(newSts)
         return executeCallback(null, documentString)
       case JSONLDContentType:
-        sz.setFlags('si dr') // turtle + dr (means no default, no relative prefix)
+        // turtle + dr (means no default, no relative prefix); preserve user flags
+        sz.setFlags('si dr' + (opts.flags ? (' ' + opts.flags) : ''))
         documentString = sz.statementsToJsonld(newSts) // convert via turtle
         return executeCallback(null, documentString)
       case NQuadsContentType:

package/src/serializer.js

@@ -52,6 +52,12 @@ export class Serializer {
     return this
   }
 
+  /**
+   * Set serializer behavior flags. Letters can be combined with spaces.
+   * Examples: 'si', 'deinprstux', 'si dr', 'o'.
+   * Notable flags:
+   *  - 'o': do not abbreviate to a prefixed name when the local part contains a dot
+   */
   setFlags(flags) {
     this.flags = flags || '';
     return this
@@ -251,10 +257,33 @@ export class Serializer {
     return this.statementsToN3(f.statements)
   }
 
-  _notQNameChars = '\t\r\n !"#$%&\'()*.,+/;<=>?@[\\]^`{|}~' // issue#228
+  _notQNameChars = '\t\r\n !"#$%&\'()*,+/;<=>?@[\\]^`{|}~' // issue#228
   _notNameChars =
     (this._notQNameChars + ':')
 
+  // Validate if a string is a valid PN_LOCAL per Turtle 1.1 spec
+  // Allows dots inside the local name but not as trailing character
+  // Also allows empty local names (for URIs ending in / or #)
+  isValidPNLocal(local) {
+    // Empty local name is valid (e.g., ex: for http://example.com/)
+    if (local.length === 0) return true
+    
+    // Cannot end with a dot
+    if (local[local.length - 1] === '.') return false
+    
+    // Check each character (allow dots mid-string)
+    for (var i = 0; i < local.length; i++) {
+      var ch = local[i]
+      // Dot is allowed unless it's the last character (checked above)
+      if (ch === '.') continue
+      // Other characters must not be in the blacklist
+      if (this._notNameChars.indexOf(ch) >= 0) {
+        return false
+      }
+    }
+    return true
+  }
+
   explicitURI(uri) {
     if (this.flags.indexOf('r') < 0 && this.base) {
       uri = Uri.refTo(this.base, uri)
@@ -628,13 +657,17 @@ export class Serializer {
     if (j >= 0 && this.flags.indexOf('p') < 0 &&
       // Can split at namespace but only if http[s]: URI or file: or ws[s] (why not others?)
       (uri.indexOf('http') === 0 || uri.indexOf('ws') === 0 || uri.indexOf('file') === 0)) {
-      var canSplit = true
-      for (var k = j + 1; k < uri.length; k++) {
-        if (this._notNameChars.indexOf(uri[k]) >= 0) {
-          canSplit = false
-          break
-        }
-      }
+      var localid = uri.slice(j + 1)
+      var namesp = uri.slice(0, j + 1)
+      // Don't split if namespace is just the protocol (e.g., https://)
+      // A valid namespace should have content after the protocol
+  var minNamespaceLength = uri.indexOf('://') + 4 // e.g., "http://x" minimum
+  // Also don't split if namespace is the base directory (would serialize as relative URI)
+  var baseDir = this.base ? this.base.slice(0, Math.max(this.base.lastIndexOf('/'), this.base.lastIndexOf('#')) + 1) : null
+  var namespaceIsBaseDir = baseDir && namesp === baseDir
+  // If flag 'o' is present, forbid dots in local part when abbreviating
+  var forbidDotLocal = this.flags.indexOf('o') >= 0 && localid.indexOf('.') >= 0
+  var canSplit = !namespaceIsBaseDir && !forbidDotLocal && namesp.length > minNamespaceLength && this.isValidPNLocal(localid)
 /*
       if (uri.slice(0, j + 1) === this.base + '#') { // base-relative
         if (canSplit) {
@@ -645,8 +678,6 @@ export class Serializer {
       }
 */
       if (canSplit) {
-        var localid = uri.slice(j + 1)
-        var namesp = uri.slice(0, j + 1)
         if (this.defaultNamespace && this.defaultNamespace === namesp &&
             this.flags.indexOf('d') < 0) { // d -> suppress default
           if (this.flags.indexOf('k') >= 0 &&