feat: check that container-relative URLs have no query

rdeltour · rdeltour · commit a4fed675b684 · 2022-11-28T10:54:48.000+01:00
EPUB relative URLs must be valid-relative-ocf-URL-with-fragment strings,
which must not have query components.

This commit adds a new check that reports `RSC-033` (error) when a query
is found in a container-relative URL.

It also slightly improves the URL checks in single-file validation mode,
to ensure a URL is not always considered remote in that mode.
diff --git a/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java b/src/main/java/com/adobe/epubcheck/messages/DefaultSeverities.java
@@ -349,6 +349,7 @@ private void initialize()
     severities.put(MessageId.RSC_030, Severity.ERROR);
     severities.put(MessageId.RSC_031, Severity.WARNING);
     severities.put(MessageId.RSC_032, Severity.ERROR);
+    severities.put(MessageId.RSC_033, Severity.ERROR);
 
     // Scripting
     severities.put(MessageId.SCP_001, Severity.SUPPRESSED); // checking scripts is out of scope
diff --git a/src/main/java/com/adobe/epubcheck/messages/MessageId.java b/src/main/java/com/adobe/epubcheck/messages/MessageId.java
@@ -343,6 +343,7 @@ public enum MessageId implements Comparable<MessageId>
   RSC_030("RSC-030"),
   RSC_031("RSC-031"),
   RSC_032("RSC-032"),
+  RSC_033("RSC-033"),
 
   // Messages relating to scripting
   SCP_001("SCP-001"),
diff --git a/src/main/java/com/adobe/epubcheck/opf/ValidationContext.java b/src/main/java/com/adobe/epubcheck/opf/ValidationContext.java
@@ -205,7 +205,7 @@ public String relativize(URL url)
     }
     else
     {
-      return url.toString();
+      return this.url.relativize(url);
     }
   }
 
diff --git a/src/main/java/org/w3c/epubcheck/core/references/URLChecker.java b/src/main/java/org/w3c/epubcheck/core/references/URLChecker.java
@@ -29,6 +29,7 @@ public class URLChecker
   private URL baseURLTestB;
   private boolean isRemoteBase;
   private final Report report;
+  private final ValidationContext context;
 
   public URLChecker(ValidationContext context)
   {
@@ -37,7 +38,8 @@ public URLChecker(ValidationContext context)
 
   public URLChecker(ValidationContext context, URL baseURL)
   {
-    this.report = Preconditions.checkNotNull(context).report;
+    this.context = Preconditions.checkNotNull(context);
+    this.report = context.report;
     this.baseURL = Preconditions.checkNotNull(baseURL);
     this.isRemoteBase = false;
     try
@@ -117,12 +119,22 @@ private URL resolveURL(String string, boolean isBase, EPUBLocation location)
         isRemoteBase = true;
         return url;
       }
-      // if relative URL "leaks" outside the container, report and continue
-      else if (!isBase && !testA.toString().startsWith(TEST_BASE_A_FULL)
-          || !testB.toString().startsWith(TEST_BASE_B_FULL))
+      else
       {
-        // FIXME !!! this is broken, base s/b taken into account
-        report.message(MessageId.RSC_026, location, string);
+        // if URL has a query string, report and continue
+        if (url.query() != null)
+        {
+          report.message(MessageId.RSC_033, location, string);
+          url = url.withQuery(null);
+        }
+        // if relative URL "leaks" outside the container, report and continue
+        // this check only make sense when the container is present
+        if (context.container.isPresent() && !isBase
+            && (!testA.toString().startsWith(TEST_BASE_A_FULL)
+                || !testB.toString().startsWith(TEST_BASE_B_FULL)))
+        {
+          report.message(MessageId.RSC_026, location, string);
+        }
       }
       return url;
     } catch (GalimatiasParseException e)
diff --git a/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties b/src/main/resources/com/adobe/epubcheck/messages/MessageBundle.properties
@@ -260,7 +260,7 @@ RSC_015=A fragment identifier is required for svg use tag references.
 RSC_016=Fatal Error while parsing file: %1$s
 RSC_017=Warning while parsing file: %1$s
 RSC_019=EPUBs with Multiple Renditions should contain a META-INF/metadata.xml file.
-RSC_020="%1$s" is not a valid URI (%2$s)
+RSC_020="%1$s" is not a valid URL (%2$s)
 RSC_021=A Search Key Map Document must point to Content Documents ("%1$s" was not found in the spine).
 RSC_022=Cannot check image details (requires Java version 7 or higher).
 RSC_024=Informative parsing warning: %1$s
@@ -272,3 +272,4 @@ RSC_029=Data URL is not allowed in this context.
 RSC_030=File URLs are not allowed in EPUB, but found "%1$s".
 RSC_031=Remote resource references should use HTTPS, but found "%1$s".
 RSC_032=Fallback must be provided for foreign resources, but found none for resource "%1$s" of type "%2$s".
+RSC_033=Relative URL strings must not have a query component, but found one in "%1$s".
diff --git a/src/test/resources/epub3/04-ocf/files/url-in-xhtml-valid.xhtml b/src/test/resources/epub3/04-ocf/files/url-in-xhtml-valid.xhtml
@@ -0,0 +1,15 @@
+<!DOCTYPE html>
+<html xmlns:epub="https://www.idpf.org/2007/ops" xmlns="https://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
+	<head>
+		<meta charset="utf-8"/>
+		<title>Minimal EPUB</title>
+	</head>
+	<body>
+		<h1>Loomings</h1>
+		<a href="https://example.org">asbolute URL</a>
+		<a href="content_002.xhtml">relative URL</a>
+		<a href="../content_002.xhtml">relative URL, one directory up</a>
+		<a href="content_002.xhtml#id">fragment URL</a>
+		<a href="#id">fragment-only URL</a>
+	</body>
+</html>
diff --git a/src/test/resources/epub3/04-ocf/files/url-query-in-package-item-error.opf b/src/test/resources/epub3/04-ocf/files/url-query-in-package-item-error.opf
@@ -0,0 +1,16 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<package xmlns="https://www.idpf.org/2007/opf" version="3.0" xml:lang="en" unique-identifier="q">
+  <metadata xmlns:dc="https://purl.org/dc/elements/1.1/">
+    <dc:title id="title">Minimal EPUB 3.0</dc:title>
+    <dc:language>en</dc:language>
+    <dc:identifier id="q">NOID</dc:identifier>
+    <meta property="dcterms:modified">2017-06-14T00:00:01Z</meta>
+  </metadata>
+  <manifest>
+    <item id="content_001"  href="content_001.xhtml?value=value" media-type="application/xhtml+xml"/>
+    <item id="nav" href="nav.xhtml" media-type="application/xhtml+xml" properties="nav"/>
+  </manifest>
+  <spine>
+    <itemref idref="content_001"/>
+  </spine>
+</package>
diff --git a/src/test/resources/epub3/04-ocf/files/url-query-in-package-link-error.opf b/src/test/resources/epub3/04-ocf/files/url-query-in-package-link-error.opf
@@ -0,0 +1,17 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<package xmlns="https://www.idpf.org/2007/opf" version="3.0" xml:lang="en" unique-identifier="q">
+<metadata xmlns:dc="https://purl.org/dc/elements/1.1/">
+  <dc:title id="title">Minimal EPUB 3.0</dc:title>
+  <dc:language>en</dc:language>
+  <dc:identifier id="q">NOID</dc:identifier>
+  <meta property="dcterms:modified">2017-06-14T00:00:01Z</meta>
+  <link rel="record" href="atom.xml?name=value" media-type="application/atom+xml"/>
+</metadata>
+<manifest>
+  <item id="content_001"  href="content_001.xhtml" media-type="application/xhtml+xml"/>
+  <item id="nav"  href="nav.xhtml" media-type="application/xhtml+xml" properties="nav"/>
+</manifest>
+<spine>
+  <itemref idref="content_001" />
+</spine>
+</package>
diff --git a/src/test/resources/epub3/04-ocf/files/url-query-in-xhtml-a-error.xhtml b/src/test/resources/epub3/04-ocf/files/url-query-in-xhtml-a-error.xhtml
@@ -0,0 +1,11 @@
+<!DOCTYPE html>
+<html xmlns:epub="https://www.idpf.org/2007/ops" xmlns="https://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
+	<head>
+		<meta charset="utf-8"/>
+		<title>Minimal EPUB</title>
+	</head>
+	<body>
+		<h1>Loomings</h1>
+		<a href="content_002.xhtml?name=value">Call me Ishmael.</a>
+	</body>
+</html>
diff --git a/src/test/resources/epub3/04-ocf/ocf.feature b/src/test/resources/epub3/04-ocf/ocf.feature
@@ -105,6 +105,15 @@ Feature: EPUB 3 — Open Container Format
     
   ###  4.1.5 URLs in the OCF abstract container
 
+  #### Valid container URLs
+
+  @spec @xref:sec-container-iri
+  Scenario: Allow valid container URLs in XHTML
+    When checking EPUB 'url-in-xhtml-valid.xhtml'
+    And no errors or warnings are reported
+
+  #### Invalid container URLs
+
   @spec @xref:sec-container-iri
   Scenario: Report leaking URLs in the package document
     When checking EPUB 'ocf-url-leaking-in-opf-error'
@@ -117,6 +126,28 @@ Feature: EPUB 3 — Open Container Format
     Then error RSC-026 is reported
     And no other errors or warnings are reported
 
+  #### URL query checks:
+
+  @spec @xref:sec-container-iri
+  Scenario: Report a URL query string found in a manifest item
+    When checking EPUB 'url-query-in-package-item-error.opf'
+    Then error RSC-033 is reported
+    And no other errors or warnings are reported
+
+  @spec @xref:sec-container-iri
+  Scenario: Report a URL query string found in a package link
+    When checking EPUB 'url-query-in-package-link-error.opf'
+    Then error RSC-033 is reported
+    And no other errors or warnings are reported
+
+  @spec @xref:sec-container-iri
+  Scenario: Report a URL query string found in a manifest item
+    When checking EPUB 'url-query-in-xhtml-a-error.xhtml'
+    Then error RSC-033 is reported
+    And no other errors or warnings are reported
+
+  #### resource existence checks:
+
   @spec @xref:sec-container-iri
   Scenario: Report a reference from an XHTML `cite` attribute not declared in the manifest
     When checking EPUB 'url-xhtml-cite-missing-resource-error'

Original file line number	Diff line number	Diff line change
`@@ -205,7 +205,7 @@ public String relativize(URL url)`
`205`	`205`	`}`
`206`	`206`	`else`
`207`	`207`	`{`
`208`		`- return url.toString();`
	`208`	`+ return this.url.relativize(url);`
`209`	`209`	`}`
`210`	`210`	`}`
`211`	`211`