Error Analysis

Tracking will start at the 1st screenview event, it is required to have screen tracking implemented. Make sure to follow the Android Track screens sections.

If you are in the process of implementing the SDK for the 1st time (or choose to take this update as an opportunity to review your Privacy related implementation), make sure to follow the Android Privacy sections.

A new Gradle dependency needs to be added to your Gradle build file.

implementation "com.contentsquare.android:error-analysis:4.23.1"

implementation "com.contentsquare.android:error-analysis:4.23.1"

You do not need to do anything to start the SDK, it will start itself with Contentsquare SDK.

When the SDK starts, you should see a log like this one:

// success case
Contentsquare Error Analysis [SDKVersionNumber] starting in app : [your.app.packageName]

// fail case
// << No public fail log

By using code instrumentation, network calls will be instrumented to collect network data.
This way, three largely used network stacks are supported, and all the libraries that use them:

• OkHttp ↗,
• HttpURLConnection ↗,
• HttpClient ↗ (deprecated ↗)

With the use of a Gradle plugin, an instrumentation of your code will be performed to collect the data about network errors.

In your root-level (project-level) Gradle file (<project>/build.gradle or <project>/build.gradle.kts), add the Contentsquare Network Gradle plugin to the plugins block:

plugins {
   // Add the dependency for the Contentsquare Network Gradle plugin
   id "com.contentsquare.error.analysis.network" version "1.1.0" apply false
}

buildscript {
   repositories {
      maven {
         url "https://plugins.gradle.org/m2"
      }
   }
   dependencies {
      classpath "com.contentsquare.gradle:error-analysis-network:1.1.0"
   }
}

plugins {
   // Add the dependency for the Contentsquare Network Gradle plugin
   id("com.contentsquare.error.analysis.network") version "1.1.0" apply false
}

buildscript {
   repositories {
      maven {
         url = uri("https://plugins.gradle.org/m2")
      }
   }
   dependencies {
      classpath("com.contentsquare.gradle:error-analysis-network:1.1.0")
   }
}

In your module (app-level) Gradle file (usually <project>/<app-module>/build.gradle or <project>/<app-module>/build.gradle.kts), add the Contentsquare Network Gradle plugin:

plugins {
   id "com.contentsquare.error.analysis.network"
}

apply plugin: "com.contentsquare.error.analysis.network"

plugins {
   id("com.contentsquare.error.analysis.network")
}

apply(plugin = "com.contentsquare.error.analysis.network")

The plugin adds new tasks for network code instrumentation: transformVariantClassesWithAsm

This task can be time consuming, and should be disabled for development build.

By default the plugin is enabled for each variant, there are three ways to disable the Gradle plugin:

android {
   buildTypes {
      debug {
         errorAnalysisNetwork {
               setInstrumentationEnabled(false)
         }
      }
   }
}

android {
   buildTypes {
      debug {
         configure<com.contentsquare.gradle.error.analysis.network.ErrorAnalysisNetworkExtension> {
               setInstrumentationEnabled(false)
         }
      }
   }
}

android {
   flavorDimensions "example"
   productFlavors {
      enable {
         dimension "example"
      }
      disable {
         dimension "example"
         errorAnalysisNetwork {
               setInstrumentationEnabled(false)
         }
      }
   }
}

android {
   flavorDimensions += "example"
   productFlavors {
      create("enable") {
         dimension = "example"
      }
      create("disable") {
         dimension = "example"
         configure<com.contentsquare.gradle.error.analysis.network.ErrorAnalysisNetworkExtension> {
               setInstrumentationEnabled(false)
         }
      }
   }
}

errorAnalysisNetworkInstrumentationEnabled=false

Command line:

./gradlew :app:assembleDebug -PerrorAnalysisNetworkInstrumentationEnabled=false

If your project is using OkHttp client, an interceptor (ErrorAnalysisInterceptor) can be added to collect all the network errors:

 import com.contentsquare.android.error.analysis.network.ErrorAnalysisInterceptor;

 new OkHttpClient().newBuilder()
    .addInterceptor(new ErrorAnalysisInterceptor())
    .build();

 import com.contentsquare.android.error.analysis.network.ErrorAnalysisInterceptor

 OkHttpClient().newBuilder()
    .addInterceptor(ErrorAnalysisInterceptor())
    .build()

If no other solution works, a custom implementation can be created via the newNetworkMetric method.

Multiple data will need to be gathered: URL and method at the creation of the metric, and status code after the request is done. Also calling start (before the request) and stop (after all the rest) methods for time metrics to be collect and to send the event.

import com.contentsquare.android.error.analysis.ErrorAnalysis;
import com.contentsquare.android.error.analysis.NetworkMetric;


NetworkMetric metric = ErrorAnalysis.getInstance().newNetworkMetric(url, ErrorAnalysis.HttpMethod.GET);
metric.start();

// do your request

metric.setStatusCode(code);
metric.stop();

import com.contentsquare.android.error.analysis.ErrorAnalysis
import com.contentsquare.android.error.analysis.trace

ErrorAnalysis.getInstance().newNetworkMetric(url, ErrorAnalysis.HttpMethod.GET).trace {

   // do your request

   setStatusCode(code)
}

By default, the API errors feature collects the URL path of the failed API requests. To prevent the collection of Personal Data in the URL path, you can rewrite the request URL path with the ErrorAnalysis.setUrlMaskingPatterns SDK API.

Replace a step of the path - meaning between two slashes (/) - containing Personal Data with a variable:

• :user_id becomes CS_ANONYMIZED_USER_ID
• :address becomes CS_ANONYMIZED_ADDRESS

ErrorAnalysis.setUrlMaskingPatterns(
   List.of("https://www.contentsquare.com/users/:user_id/address/:address")
);

ErrorAnalysis.setUrlMaskingPatterns(
   listOf("https://www.contentsquare.com/users/:user_id/address/:address")
)

Current version does not provide a log to confirm if API error collection is enabled.

If you have logging enabled, you will see API errors:

• Displayed in Contentsquare Log Visualizer.
• As an Info log in Logcat:

CSLIB: API Error - 401 GET https://api.client.com

The way our SDK works is by auto-starting with the application launch and collects the API errors through the instrumented code added by the Gradle plugin.

Once started, our SDK fetches its config from our servers. It will start collecting data from network events if the API errors setting is enabled in the config (this is handled by the Contentsquare team).

The SDK monitors the only API error with response code above 400, and generates analytics data. These events are then locally stored, and eventually sent to our servers in batches.

For each network error, a new event will be sent in analytics and Session Replay data. Check the following sections to learn more about how data is processed and sent:

• Analytics requests
• Session Replay requests

API errors troubleshooting details enables you to collect more information about API errors so you can troubleshoot errors faster.

With this feature you will be able to see three types of additional API error details in the Event Stream of Session Replay.

• The HTTP headers of the request and the response.
• The body (the data sent by the request or received in the response).
• The query parameters of the request endpoint (of the URL of the information you request for).

See API Troubleshooting Details ↗ for more details.

Only network calls in error (response code above 400) will be collected.
Here the exhaustive list of data collected:

• Response code
• URL (without query strings)
• HTTP method
• Timestamp of the request
• Timestamp of the response

Code instrumentation does not work well with Firebase Performance plugin.
Network events will not be logged for OkHttp when using Contentsquare Error Analysis network and Firebase performance plugin (only one of the two will log events).
If you want to use both libraries, prefer using interceptors than code instrumentation.

In order to produce readable crash reports from builds that have been obfuscated, Contentsquare requires you to upload the mapping file produced by your build to our servers. To simplify this process, we’ve provided a Gradle plugin that you can integrate into your build.

In your root-level (project-level) Gradle file (<project>/build.gradle or <project>/build.gradle.kts), add the Contentsquare Crash Gradle plugin to the plugins block:

plugins {
   // Add the dependency for the Contentsquare Crash Gradle plugin
   id "com.contentsquare.error.analysis.crash" version "1.2.0" apply false
}

buildscript {
   repositories {
      maven {
         url "https://plugins.gradle.org/m2"
      }
   }
   dependencies {
      classpath "com.contentsquare.gradle:error-analysis-crash:1.2.0"
   }
}

plugins {
   // Add the dependency for the Contentsquare Crash Gradle plugin
   id("com.contentsquare.error.analysis.crash") version "1.2.0" apply false
}

buildscript {
   repositories {
      maven {
         url = uri("https://plugins.gradle.org/m2")
      }
   }
   dependencies {
      classpath("com.contentsquare.gradle:error-analysis-crash:1.2.0")
   }
}

In your module (app-level) Gradle file (usually <project>/<app-module>/build.gradle or <project>/<app-module>/build.gradle.kts), add the Contentsquare Crash Gradle plugin:

plugins {
   id "com.contentsquare.error.analysis.crash"
}

apply plugin: "com.contentsquare.error.analysis.crash"

plugins {
   id("com.contentsquare.error.analysis.crash")
}

apply(plugin = "com.contentsquare.error.analysis.crash")

To use the plugin, you must provide it with the relevant credentials to be able to upload the mapping file.

1. Login to the Contentsquare platform on https://app.contentsquare.com ↗
2. Make sure to be on the right project
3. The project ID can be found in the URL query parameter project: https://app.contentsquare.com/#/{MODULE_NAME}?project={PROJECT_ID}&hash={HASH}

Follow the dedicated documentation from the Help Center to get the client ID and client Secret: How to create API credentials ↗.

You can do this by setting the uploadCredentials property of the errorAnalysisCrash extension in your build.gradle or build.gradle.kts file, as shown below. The errorAnalysisCrash extension is available on all build types and product flavours.

errorAnalysisCrash {
   uploadCredentials {
      projectId = <your project Id>
      clientId = "<your client Id>"
      clientSecret = "<your client secret>"
   }
}

configure<com.contentsquare.gradle.error.analysis.crash.extensions.ErrorAnalysisCrashExtension> {
    uploadCredentials {
        projectId = <your project Id>
        clientId = "<your client Id>"
        clientSecret = "<your client secret>"
    }
}

To ensure that we can provide line numbers and file names in the crash reports, add the following configuration to your ProGuard file:

-keepattributes SourceFile,LineNumberTable        # Keep file names and line numbers

The upload task is independent from any other tasks in build graph. Once your build is complete, and the mapping file has been generated, you can upload it follows:

./gradlew uploadContentsquareMappingFile<VariantName>

Replace <VariantName> with the name of the build variant you want to upload the mapping file for.

We strongly recommend to run this as a task in a CI/CD pipeline for the obfuscated builds.

If in-app features are enabled, an info log should appear when the Crash reporter has successfully been initialized (see Android Debugging and Logging section).:

CSLIB: Crash reporter is enabled

To test that the Crash Reporter is working properly, you can force a crash in your application. The next time you start your application, the crash will be sent to our servers. If you have logging enabled, you will see the upload reflected in the log:

I/CSLIB: Crash event detected and sent for userID: <user-id>, session: <session-number> on screen: <screen-number> crashID: <crash-id>

Crash Reporter is started automatically and begins to report basic information for crashes in your application when it is launched.

Once started, our SDK fetches its configuration from our servers. It will start reporting crashes if the Crash Reporter setting is enabled (this is handled by the Contentsquare team).

When a crash occurs, a report is created and stored locally in the device. Once the app is launched again, the report is sent to our servers and then removed from the local storage.

For each crash, a new event will be sent to Analytics and Session Replay data when the app is launched again after the crash occurred and the network conditions allow for the server to be reached.

The Crash Reporter will capture any crashes caused by an unhandled exception in applications written in Java or Kotlin. It will collect information about the application, system, process, threads, stack traces and some other event metadata you can find in collected data points

Native crashes from applications written using C or C++ are currently not supported.

It is possible that the Crash Reporter will work with other SDKs, but we cannot guarantee it. Android has a global error handling class and therefore we rely on other SDKs forwarding the error on to us in the event of a crash, should they have been registered first. If the Contentsquare Crash Reporter is registered first, it will forward on errors to any other registered handlers.

Based on our tests, we can confirm it is compatible with Firebase Crashlytics.

Errors occurring in web pages loaded in WebViews can be collected if Webview tracking is setup in your app.

See 📚 Webview Tracking implementation.

Once WebView tracking is implemented, errors collection can be set up.

See 📚 Errors (in WebView)

We always strive to be non-intrusive, and transparent to the developers of the client app. We apply this rule on the performance as well. These are the technical specifics we can share on performance, if you have any questions feel free to reach out to us.

The following performance results were obtained under the following conditions:

We used Android Microbenchmark ↗ to measure the time added by our implementation.

The instrumented code added is running in less than 1ms per request: our benchmarks show less 400µs.

About CPU and memory, the overhead of adding the collection of network error is negligible.