Custom code quality rules custom-code-quality-rules

This page describes the custom code quality rules executed by Cloud Manager as part of code quality testing. They are based on best practices from Experience Manager Engineering.

SonarQube rules sonarqube-rules

The following section details SonarQube rules executed by Cloud Manager.

Do not use potentially dangerous functions do-not-use-potentially-dangerous-functions

The methods Thread.stop() and Thread.interrupt() can produce hard-to-reproduce issues and, sometimes, security vulnerabilities. Their usage should be tightly monitored and validated. In general, message passing is a safer way to accomplish similar goals.

Non-compliant code non-compliant-code

public class DontDoThis implements Runnable {
  private Thread thread;

  public void start() {
    thread = new Thread(this);
    thread.start();
  }

  public void stop() {
    thread.stop();  // UNSAFE!
  }

  public void run() {
    while (true) {
        somethingWhichTakesAWhileToDo();
    }
  }
}

Compliant code compliant-code

public class DoThis implements Runnable {
  private Thread thread;
  private boolean keepGoing = true;

  public void start() {
    thread = new Thread(this);
    thread.start();
  }

  public void stop() {
    keepGoing = false;
  }

  public void run() {
    while (this.keepGoing) {
        somethingWhichTakesAWhileToDo();
    }
  }
}

Do not use format strings which may be externally controlled do-not-use-format-strings-which-may-be-externally-controlled

Using a format string from an external source (such a request parameter or user-generated content) can produce can expose an application to denial of service attacks. There are circumstances where a format string may be externally controlled, but is only allowed from trusted sources.

Non-compliant code non-compliant-code-1

protected void doPost(SlingHttpServletRequest request, SlingHttpServletResponse response) {
  String messageFormat = request.getParameter("messageFormat");
  request.getResource().getValueMap().put("some property", String.format(messageFormat, "some text"));
  response.sendStatus(HttpServletResponse.SC_OK);
}

HTTP requests should always have socket and connect timeouts http-requests-should-always-have-socket-and-connect-timeouts

When executing HTTP requests from inside an Experience Manager application, it is critical to ensure that proper timeouts are configured to avoid unnecessary thread consumption. Unfortunately, the default behavior of both Java™’s default HTTP Client (java.net.HttpUrlConnection) and the commonly used Apache HTTP Components client is to never time out, so timeouts must be explicitly set. Further, as a best practice, these timeouts should be no more than 60 seconds.

Non-compliant code non-compliant-code-2

@Reference
private HttpClientBuilderFactory httpClientBuilderFactory;

public void dontDoThis() {
  HttpClientBuilder builder = httpClientBuilderFactory.newBuilder();
  HttpClient httpClient = builder.build();

  // do something with the client
}

public void dontDoThisEither() {
  URL url = new URL("http://www.google.com");
  URLConnection urlConnection = url.openConnection();

  BufferedReader in = new BufferedReader(new InputStreamReader(
    urlConnection.getInputStream()));

  String inputLine;
  while ((inputLine = in.readLine()) != null) {
    logger.info(inputLine);
  }

  in.close();
}

Compliant code compliant-code-1

@Reference
private HttpClientBuilderFactory httpClientBuilderFactory;

public void doThis() {
  HttpClientBuilder builder = httpClientBuilderFactory.newBuilder();
  RequestConfig requestConfig = RequestConfig.custom()
    .setConnectTimeout(5000)
    .setSocketTimeout(5000)
    .build();
  builder.setDefaultRequestConfig(requestConfig);

  HttpClient httpClient = builder.build();

  // do something with the client
}

public void orDoThis() {
  URL url = new URL("http://www.google.com");
  URLConnection urlConnection = url.openConnection();
  urlConnection.setConnectTimeout(5000);
  urlConnection.setReadTimeout(5000);

  BufferedReader in = new BufferedReader(new InputStreamReader(
    urlConnection.getInputStream()));

  String inputLine;
  while ((inputLine = in.readLine()) != null) {
    logger.info(inputLine);
  }

  in.close();
}

Always close ResourceResolver objects resourceresolver-objects-should-always-be-closed

ResourceResolver objects obtained from the ResourceResolverFactory consume system resources. Although there are measures in place to reclaim these resources when a ResourceResolver is no longer in use, it is more efficient to explicitly close any opened ResourceResolver objects by calling the close() method.

One relatively common misconception is that ResourceResolver objects created using an existing JCR session should not be explicitly closed or that doing so closes the underlying JCR session. This is not the case. Regardless of how a ResourceResolver is opened, it should be closed when no longer used. Since ResourceResolver implements the Closeable interface, it is also possible to use the try-with-resources syntax instead of explicitly invoking close().

Non-compliant code non-compliant-code-4

public void dontDoThis(Session session) throws Exception {
  ResourceResolver resolver = factory.getResourceResolver(Collections.singletonMap("user.jcr.session", (Object)session));
  // do some stuff with the resolver
}

Compliant code compliant-code-2

public void doThis(Session session) throws Exception {
  ResourceResolver resolver = null;
  try {
    resolver = factory.getResourceResolver(Collections.singletonMap("user.jcr.session", (Object)session));
    // do something with the resolver
  } finally {
    if (resolver != null) {
      resolver.close();
    }
  }
}

public void orDoThis(Session session) throws Exception {
  try (ResourceResolver resolver = factory.getResourceResolver(Collections.singletonMap("user.jcr.session", (Object) session))){
    // do something with the resolver
  }
}

Do not use Sling servlet paths to register servlets do-not-use-sling-servlet-paths-to-register-servlets

As described in the Sling documentation, bindings servlets by paths is discouraged. Path-bound servlets cannot use standard JCR access controls and, as a result, require additional security rigor. Rather than using path-bound servlets, it is recommended to create nodes in the repository and register servlets by resource type.

Non-compliant code non-compliant-code-5

@Component(property = {
  "sling.servlet.paths=/apps/myco/endpoint"
})
public class DontDoThis extends SlingAllMethodsServlet {
 // implementation
}

Caught exceptions should be logged or thrown, not both caught-exceptions-should-be-logged-or-thrown-but-not-both

In general, an exception should be logged exactly one time. Logging exceptions multiple times can cause confusion as it is unclear how many times an exception occurred. The most common pattern which leads to this is logging and throwing a caught exception.

Non-compliant code non-compliant-code-6

public void dontDoThis() throws Exception {
  try {
    someOperation();
  } catch (Exception e) {
    logger.error("something went wrong", e);
    throw e;
  }
}

Compliant code compliant-code-3

public void doThis() {
  try {
    someOperation();
  } catch (Exception e) {
    logger.error("something went wrong", e);
  }
}

public void orDoThis() throws MyCustomException {
  try {
    someOperation();
  } catch (Exception e) {
    throw new MyCustomException(e);
  }
}

Avoid log statements immediately followed by a throw statement avoid-having-a-log-statement-immediately-followed-by-a-throw-statement

Another common pattern to avoid is to log a message and then immediately throw an exception. This practice generally indicates that the exception message ends up duplicated in log files.

Non-compliant code non-compliant-code-7

public void dontDoThis() throws Exception {
  logger.error("something went wrong");
  throw new RuntimeException("something went wrong");
}

Compliant code compliant-code-4

public void doThis() throws Exception {
  throw new RuntimeException("something went wrong");
}

Avoid logging at INFO when handling GET or HEAD requests avoid-logging-at-info-when-handling-get-or-head-requests

In general, the INFO log level should be used to demarcate important actions and, by default, Experience Manager is configured to log at the INFO level or above. GET and HEAD methods should only ever be read-only operations and thus do not constitute important actions. Logging at the INFO level in response to GET or HEAD requests is likely to create significant log noise, making it harder to identify useful information in log files. Logging when handling GET or HEAD requests should be either at the WARN or ERROR levels when something has gone wrong or at the DEBUG or TRACE levels if deeper troubleshooting information would be helpful.

Non-compliant code non-compliant-code-8

public void doGet() throws Exception {
  logger.info("handling a request from the user");
}

Compliant code compliant-code-5

public void doGet() throws Exception {
  logger.debug("handling a request from the user.");
}

Do not use Exception.getMessage() as the first parameter of a logging statement do-not-use-exception-getmessage-as-the-first-parameter-of-a-logging-statement

As a best practice, log messages should provide contextual information about where in the application an exception has occurred. While context can also be determined by using stack traces, in general the log message is going to be easier to read and understand. As a result, when logging an exception, it is a bad practice to use the exception’s message as the log message. The exception message contains what went wrong whereas the log message should be used to tell a log reader what the application was doing when the exception happened. The exception message is still logged. By specifying your own message, the logs are easier to understand.

Non-compliant code non-compliant-code-9

public void dontDoThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    logger.error(e.getMessage(), e);
  }
}

Compliant code compliant-code-6

public void doThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    logger.error("Unable to do something", e);
  }
}

Logging in catch blocks should be at the WARN or ERROR level logging-in-catch-blocks-should-be-at-the-warn-or-error-level

As the name suggests, Java™ exceptions should always be used in exceptional circumstances. As a result, when an exception is caught, it is important to ensure that log messages are logged at the appropriate level, either WARN or ERROR. This ensures that those messages appear correctly in the logs.

Non-compliant code non-compliant-code-10

public void dontDoThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    logger.debug(e.getMessage(), e);
  }
}

Compliant code compliant-code-7

public void doThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    logger.error("Unable to do something", e);
  }
}

Do not print stack traces to the console do-not-print-stack-traces-to-the-console

As mentioned, context is critical when understanding log messages. Using Exception.printStackTrace() causes only the stack trace to be output to the standard error stream, losing all context. Further, in a multi-threaded application like Experience Manager, if multiple exceptions are printed using this method in parallel, their stack traces may overlap, which produces significant confusion. Exceptions should be logged through the logging framework only.

Non-compliant code non-compliant-code-11

public void dontDoThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    e.printStackTrace();
  }
}

Compliant code compliant-code-8

public void doThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    logger.error("Unable to do something", e);
  }
}

Do not output to standard output or standard error do-not-output-to-standard-output-or-standard-error

Logging in Experience Manager should always be done through the logging framework (SLF4J). Outputting directly to the standard output or standard error streams loses the structural and contextual information provided by the logging framework. Sometimes, it may cause performance issues.

Non-compliant code non-compliant-code-12

public void dontDoThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    System.err.println("Unable to do something");
  }
}

Compliant code compliant-code-9

public void doThis() {
  try {
    someMethodThrowingAnException();
  } catch (Exception e) {
    logger.error("Unable to do something", e);
  }
}

Avoid hardcoded /apps and /libs paths avoid-hardcoded-apps-and-libs-paths

In general, paths which start with /libs and /apps should not be hardcoded as the paths they refer to are most commonly stored as paths relative to the Sling search path, which is set to /libs,/apps by default. Using the absolute path may introduce subtle defects which would only appear later in the project lifecycle.

Non-compliant code non-compliant-code-13

public boolean dontDoThis(Resource resource) {
  return resource.isResourceType("/libs/foundation/components/text");
}

Compliant code compliant-code-10

public void doThis(Resource resource) {
  return resource.isResourceType("foundation/components/text");
}

Do not use Sling scheduler sonarqube-sling-scheduler

Do not use the Sling Scheduler for tasks that require a guaranteed execution. Sling Scheduled Jobs guarantee execution and better suited for both clustered and non-clustered environments.

See Apache Sling Eventing and Job Handling to learn more about how Sling Jobs are handled in clustered environments.

Do not use Experience Manager deprecated APIs sonarqube-aem-deprecated

The Experience Manager API surface is under constant revision to identify APIs for which usage is discouraged and thus considered deprecated.

Often, these APIs are deprecated using the standard Java™ @Deprecated annotation and, as such, as identified by squid:CallToDeprecatedMethod.

However, there are cases where an API is deprecated in the context of Experience Manager but may not be deprecated in other contexts. This rule identifies this second class.

Do not use @Inject annotation with @Optional in Sling Models sonarqube-slingmodels-inject-optional

The Apache Sling project discourages the use of the @Inject annotation in the context of Sling Models, as it can lead to bad performance when combined with the DefaultInjectionStrategy.OPTIONAL (either at field or class level). Instead more specific injections (like the @ValueMapValue or @OsgiInjector annotations) should be used.

Check the Apache Sling documentation for more information about the recommended annotations and why this recommendation was made in the first place.

Reuse instances of a HTTPClient sonarqube-reuse-httpclient

AEM applications often reach out to other applications using the HTTP protocol, and the Apache HttpClient is an often used library to achieve this. But the creation of such an HttpClient object comes with some overhead, so these objects should be reused as much as possible.

This rule checks that such a HttpClient object is not private within a method, but global on a class level, so it can be reused. In this case the httpClient field should be set in the constructor of the class or the activate() method (if this class is an OSGi component/service).

Check the Optimization Guide of the HttpClient for some best practices regarding the use of the HttpClient.

Non-compliant code non-compliant-code-14

public void doHttpCall() {
  HttpClient httpclient = HttpClients.createDefault();
  // do something with the httpclient
}

Compliant code compliant-code-11

public class myClass {

  HttpClient httpclient;

  public void doHttpCall() {
    // do something with the httpclient
  }

}

OakPAL content rules oakpal-rules

The following section details the OakPAL checks executed by Cloud Manager.

Product APIs annotated with @ProviderType should not be implemented or extended by customers product-apis-annotated-with-providertype-should-not-be-implemented-or-extended-by-customers

The Experience Manager API contains Java™ interfaces and classes which are only meant to be used – but not implemented – by custom code. For example, the interface com.day.cq.wcm.api.Page should be implemented by Experience Manager only.

When new methods are added to these interfaces, those additional methods do not impact existing code which uses these interfaces. As a result, the addition of new methods to these interfaces is considered to be backwards-compatible. However, if custom code implements one of these interfaces, that custom code has introduced a backwards-compatibility risk for the customer.

Interfaces and classes – as implemented by Experience Manager – are annotated with org.osgi.annotation.versioning.ProviderType or sometimes a similar legacy annotation aQute.bnd.annotation.ProviderType. This rule identifies the cases where such an interface is implemented or a class is extended by custom code.

Non-compliant code non-compliant-code-3

import com.day.cq.wcm.api.Page;

public class DontDoThis implements Page {
// implementation here
}

Custom Lucene Oak indexes must have a Tika configuration oakpal-indextikanode

Multiple out-of-the-box Experience Manager Oak indexes include a Tika configuration and customizations of these indexes must include a Tika configuration. This rule checks for customizations of the damAssetLucene, lucene, and graphqlConfig indexes and raises an issue if either the tika  node is missing or if the tika node is missing a child node named config.xml.

See indexing documentation for more information on customizing index definitions.

Non-compliant code non-compliant-code-indextikanode

+ oak:index
    + damAssetLucene-1-custom
      - async: [async]
      - evaluatePathRestrictions: true
      - includedPaths: /content/dam
      - tags: [visualSimilaritySearch]
      - type: lucene

Compliant code compliant-code-indextikanode

+ oak:index
    + damAssetLucene-1-custom-2
      - async: [async]
      - evaluatePathRestrictions: true
      - includedPaths: /content/dam
      - tags: [visualSimilaritySearch]
      - type: lucene
      + tika
        + config.xml

Custom Lucene Oak Indexes must not be synchronous oakpal-indexasync

Oak indexes of type lucene must always be asynchronously indexed. Failure to do this may result in system instability. More information on the structure of Lucene indexes can be found in the Oak documentation.

Non-compliant code non-compliant-code-indexasync

+ oak:index
    + damAssetLucene-1-custom
      - evaluatePathRestrictions: true
      - includedPaths: /content/dam
      - type: lucene
      - tags: [visualSimilaritySearch]
      + tika
        + config.xml

Compliant code compliant-code-indexasync

+ oak:index
    + damAssetLucene-1-custom-2
      - async: [async]
      - evaluatePathRestrictions: true
      - includedPaths: /content/dam
      - tags: [visualSimilaritySearch]
      - type: lucene
      + tika
        + config.xml

Custom DAM Asset Lucene Oak indexes are properly structured oakpal-damAssetLucene-sanity-check

For asset search to work correctly in Experience Manager Assets, customizations of the damAssetLucene Oak index must follow a set of guidelines which are specific to this index. This rule checks that the index definition must have a multi-valued property named tags which contains the value visualSimilaritySearch.

Non-compliant code non-compliant-code-damAssetLucene

+ oak:index
    + damAssetLucene-1-custom
      - async: [async, nrt]
      - evaluatePathRestrictions: true
      - includedPaths: /content/dam
      - type: lucene
      + tika
        + config.xml

Compliant code compliant-code-damAssetLucene

+ oak:index
    + damAssetLucene-1-custom-2
      - async: [async, nrt]
      - evaluatePathRestrictions: true
      - includedPaths: /content/dam
      - tags: [visualSimilaritySearch]
      - type: lucene
      + tika
        + config.xml

Customer packages should not create or modify nodes under /libs oakpal-customer-package

It has been a long-standing best practice that the /libs content tree in the Experience Manager content repository should be considered read-only by customers. Modifying nodes and properties under /libs creates significant risk for major and minor updates. Modifications to /libs must be done by Adobe through official channels.

Packages should not contain duplicate OSGi configurations oakpal-package-osgi

A common problem that occurs on complex projects is where the same OSGi component is configured multiple times. This issue creates an ambiguity as to which configuration is applicable. This rule is “runmode-aware” in that it only identifies issues where the same component is configured multiple times in the same run mode or combination of run modes.

Non-compliant code non-compliant-code-osgi

+ apps
  + projectA
    + config
      + com.day.cq.commons.impl.ExternalizerImpl
  + projectB
    + config
      + com.day.cq.commons.impl.ExternalizerImpl

Compliant code compliant-code-osgi

+ apps
  + shared-config
    + config
      + com.day.cq.commons.impl.ExternalizerImpl

Config and install folders should only contain OSGi nodes oakpal-config-install

For security reasons, paths containing /config/ and /install/ are only readable by administrative users in Experience Manager and should be used only for OSGi configuration and OSGi bundles. Placing other types of content under paths which contain these segments results in application behavior which unintentionally varies between administrative and non-administrative users.

A common problem is use of nodes named config within component dialogs or when specifying the rich text editor configuration for inline editing. To resolve this issue, the offending node should be renamed to a compliant name. For the rich text editor configuration, use the configPath property on the cq:inplaceEditing node to specify the new location.

Non-compliant code non-compliant-code-config-install

+ cq:editConfig [cq:EditConfig]
  + cq:inplaceEditing [cq:InplaceEditConfig]
    + config [nt:unstructured]
      + rtePlugins [nt:unstructured]

Compliant code compliant-code-config-install

+ cq:editConfig [cq:EditConfig]
  + cq:inplaceEditing [cq:InplaceEditConfig]
    ./configPath = inplaceEditingConfig (String)
    + inplaceEditingConfig [nt:unstructured]
      + rtePlugins [nt:unstructured]

Packages should not overlap oakpal-no-overlap

Similar to the Packages Should Not Contain Duplicate OSGi Configurations rule, this is a common problem on complex projects where the same node path is written to by multiple separate content packages. While using content package dependencies can be used to ensure a consistent result, it is better to avoid overlaps entirely.

Default authoring mode should not be classic UI oakpal-default-authoring

The OSGi configuration com.day.cq.wcm.core.impl.AuthoringUIModeServiceImpl defines the default authoring mode within Experience Manager. Because the Classic UI has been deprecated since Experience Manager 6.4, an issue is now raised when the default authoring mode is configured to Classic UI.

Components with dialogs should have touch UI dialogs oakpal-components-dialogs

Experience Manager components which have a Classic UI dialog should always have a corresponding Touch UI dialog. Both provide an optimal authoring experience and to be compatible with the Cloud Service deployment model, where Classic UI is not supported. This rule verifies the following scenarios:

The Experience Manager Modernization Tools documentation provides documentation and tooling for how to convert components from Classic UI to Touch UI. See the Experience Manager Modernization Tools documentation for more details.

Packages should not mix mutable and immutable content oakpal-packages-immutable

To be compatible with the Cloud Service deployment model, individual content packages must contain either content for the immutable areas of the repository (/apps and /libs), or the mutable area (everything not in /apps or /libs), but not both. For example, a package which includes both /apps/myco/components/text and /etc/clientlibs/myco is not compatible with Cloud Service and cause an issue to be reported.

See Experience Manager Project Structure for more details.

Do not use reverse replication agents oakpal-reverse-replication

Support for reverse replication is not available in Cloud Service deployments, as described as part of Experience Manager as a Cloud Service’s release notes.

Customers using reverse replication should contact Adobe for alternative solutions.

Resources contained in proxy-enabled client libraries should be in a folder named resources oakpal-resources-proxy

Experience Manager client libraries may contain static resources like images and fonts. As described in the document Using Preprocessors, when using proxied client libraries these static resources must be contained in a child folder named resources to be effectively referenced on the publish instances.

Non-compliant code non-compliant-proxy-enabled

+ apps
  + projectA
    + clientlib
      - allowProxy=true
      + images
        + myimage.jpg

Compliant code compliant-proxy-enabled

+ apps
  + projectA
    + clientlib
      - allowProxy=true
      + resources
        + myimage.jpg

Usage of Cloud Service incompatible workflow processes oakpal-usage-cloud-service

With the move to asset micro-services for asset processing on Experience Manager as a Cloud Service, several workflow processes which were used in on-premise and AMS versions of Experience Manager have become either unsupported or unnecessary.

The migration tool in the Experience Manager as a Cloud Service Assets GitHub repository can be used to update workflow models during migration to Experience Manager as a Cloud Service.

Usage of static templates is discouraged in favor of editable templates oakpal-static-template

While the use of static templates is historically common in Experience Manager projects, Adobe recommends editable templates because they provide the most flexibility and support additional features not present in static templates. More information can be found in the document Page Templates.

Migration from static to editable templates can be largely automated using the Experience Manager Modernization Tools.

Usage of legacy foundation components is discouraged oakpal-usage-legacy

The legacy Foundation Components (that is, components under /libs/foundation) have been deprecated for several Experience Manager releases in favor of the Core Components. Usage of the Foundation Components as the basis for custom components (whether by overlay or inheritance) is discouraged and should be converted to the corresponding Core Components.

This conversion can be facilitated by the Experience Manager Modernization Tools.

Only use supported run mode names and ordering oakpal-supported-runmodes

Experience Manager as a Cloud Service enforces a strict naming policy for run mode names and a strict ordering for those run modes. The list of supported run modes can be found in the document Deploying to Experience Manager as a Cloud Service and any deviation from this is identified as an issue.

Custom search index definition nodes must be direct children of /oak:index oakpal-custom-search

Experience Manager as a Cloud Service requires that custom search index definitions (that is, nodes of type oak:QueryIndexDefinition) be direct child nodes of /oak:index. Indexes in other locations must be moved to be compatible with Experience Manager as a Cloud Service. More information on search indexes can be found in the document Content Search and Indexing.

Custom search index definition nodes must have a compatVersion of 2 oakpal-custom-search-compatVersion

Experience Manager as a Cloud Service requires that custom search index definitions (such as nodes of type oak:QueryIndexDefinition) must have the compatVersion property set to 2. Any other value is not supported by Experience Manager as a Cloud Service. More information on search indexes can be found on Content Search and Indexing.

Descendent nodes of custom search index definition nodes must be of type nt:unstructured oakpal-descendent-nodes

Hard to troubleshoot issues can occur when a custom search index definition node has unordered child nodes. To avoid this situation, it is recommended that all descendent nodes of an oak:QueryIndexDefinition node be of type nt:unstructured.

Custom search index definition nodes must contain a child node named indexRules that has children oakpal-custom-search-index

A properly defined custom search index definition node must contain a child node named indexRules which, in turn must have at least one child. More information can be found in the Oak documentation.

Custom search index definition nodes must follow naming conventions oakpal-custom-search-definitions

Experience Manager as a Cloud Service requires that custom search index definitions (that is, nodes of type oak:QueryIndexDefinition) must be named following a specific pattern described in the document Content Search and Indexing.

Custom search index definition nodes must use the index type Lucene oakpal-index-type-lucene

Experience Manager as a Cloud Service requires that custom search index definitions (that is, nodes of type oak:QueryIndexDefinition) have a type property with the value set to lucene. Indexing using legacy index types must be updated before migration to Experience Manager as a Cloud Service. See Content Search and Indexing for more information.

Custom search index definition nodes must not contain a property named seed oakpal-property-name-seed

Experience Manager as a Cloud Service prohibits custom search index definitions (that is, nodes of type oak:QueryIndexDefinition) from containing a property named seed. Indexing using this property must be updated before migration to Experience Manager as a Cloud Service. See the document Content Search and Indexing for more information.

Custom search index definition nodes must not contain a property named reindex oakpal-reindex-property

Experience Manager as a Cloud Service prohibits custom search index definitions (that is, nodes of type oak:QueryIndexDefinition) from containing a property named reindex. Indexing using this property must be updated before migration to Experience Manager as a
Cloud Service. See the document Content Search and Indexing for more information.

Custom DAM asset lucene nodes must not specify ‘queryPaths’ oakpal-damAssetLucene-queryPaths

Non-compliant code non-compliant-code-damAssetLucene-queryPaths

+ oak:index
    + damAssetLucene-1-custom-1
      - async: [async, nrt]
      - evaluatePathRestrictions: true
      - includedPaths: [/content/dam]
      - queryPaths: [/content/dam]
      - type: lucene
      + tika
        + config.xml

Compliant code compliant-code-damAssetLucene-queryPaths

+ oak:index
    + damAssetLucene-1-custom-2
      - async: [async, nrt]
      - evaluatePathRestrictions: true
      - includedPaths: [/content/dam]
      - tags: [visualSimilaritySearch]
      - type: lucene
      + tika
        + config.xml

If custom search index definition contains compatVersion, it must be set to 2 oakpal-compatVersion

Index node specifying ‘includedPaths’ should also specify ‘queryPaths’ with the same values oakpal-included-paths-without-query-paths

For custom indexes, both includedPaths and queryPaths should be configured with identical values. If one is specified, the other must match it. However, there’s a special case for indexes of damAssetLucene, including its custom versions. For these, you should only provide includedPaths.

Index node specifying nodeScopeIndex on generic node type should also specify includedPaths and queryPaths oakpal-full-text-on-generic-node-type

When setting the nodeScopeIndex property on a “generic” node type like nt:unstructured or nt:base, you must also specify the includedPaths and queryPaths properties.
nt:base can be considered “generic”, since all node types inherits from it. So setting a nodeScopeIndex on nt:base will make it index all nodes in the repository. Similarly, nt:unstructured is also considered “generic” as there are many nodes in repositories that are of this type.

Non-compliant code non-compliant-code-full-text-on-generic-node-type

+ oak:index/acme.someIndex-custom-1
  - async: [async, nrt]
  - evaluatePathRestrictions: true
  - tags: [visualSimilaritySearch]
  - type: lucene
    + indexRules
      - jcr:primaryType: nt:unstructured
      + nt:base
        - jcr:primaryType: nt:unstructured
        + properties
          + acme.someIndex-custom-1
            - nodeScopeIndex: true

Compliant code compliant-code-full-text-on-generic-node-type

+ oak:index/acme.someIndex-custom-1
  - async: [async, nrt]
  - evaluatePathRestrictions: true
  - tags: [visualSimilaritySearch]
  - type: lucene
  - includedPaths: ["/content/dam/"]
  - queryPaths: ["/content/dam/"]
    + indexRules
      - jcr:primaryType: nt:unstructured
      + nt:base
        - jcr:primaryType: nt:unstructured
        + properties
          + acme.someIndex-custom-1
            - nodeScopeIndex: true

The queryLimitReads property of the query engine should not be overridden oakpal-query-limit-reads

Overriding the default value can lead to very slow page reads, particularly when more content is added.

Multiple active versions of the same inde oakpal-multiple-active-versions

Non-compliant code non-compliant-code-multiple-active-versions

+ oak:index
  + damAssetLucene-1-custom-1
    ...
  + damAssetLucene-1-custom-2
    ...
  + damAssetLucene-1-custom-3
    ...

Compliant code compliant-code-multiple-active-versions

+ damAssetLucene-1-custom-3
    ...

The name of fully custom index definitions should conform to the official guidelines oakpal-fully-custom-index-name

The expected pattern for fully custom index names is: [prefix].[indexName]-custom-[version]. More information can be found in the document Content Search and Indexing.

Same property with different analyzed values in the same index definition oakpal-same-property-different-analyzed-values

Non-compliant code non-compliant-code-same-property-different-analyzed-values

+ indexRules
  + dam:Asset
    + properties
      + status
        - name: status
        - analyzed: true
  + dam:cfVariationNode
    + properties
      + status
        - name: status

Compliant code compliant-code-same-property-different-analyzed-values

Example:

+ indexRules
  + dam:Asset
    + properties
      + status
        - name: status
        - analyzed: true
  + dam:cfVariationNode
    + properties
      + status
        - name: status
        - analyzed: true

Example:

+ indexRules
  + dam:Asset
    + properties
      + status
        - name: status
  + dam:cfVariationNode
    + properties
      + status
        - name: status
        - analyzed: true

If the analyzed property has not been explicitly set, its default value will be false.

Tags property

For specific indexes, ensure you retain the tags property and its current values. While adding new values to the tags property is permissible, deleting any existing ones (or the property altogether) can lead to unexpected results.