docs

AQUARIUS Time-Series Developer Guide: Field Data Plug-In Framework

About This Guide

A Field Data Plug-In Framework has been added to AQUARIUS Time-Series.

A Field Data Plug-In Framework SDK is now included with the AQUARIUS Time-Series Server installation package. Integrators can use this SDK to extend AQUARIUS Time-Series by creating custom field data plug-ins to parse custom field data files.

After installing the plug-in on the AQUARIUS Time-Series Server and registering it with the Field Data Plug-In Framework, users can import their proprietary/custom field data files to the AQUARIUS Time-Series system using the available drag-and-drop upload method from Springboard, Location Manager, and the Field Data Editor, or by using the AQUARIUS Acquisition API.

This guide provides guidance for developing custom field data plug-ins for AQUARIUS Time-Series.

If you would like to see more examples of field data plug-ins, please visit our Examples Repository on GitHub.

Change Log

AQTS 2023.4 - Framework version 2.14

• Updated ServiceStack framework libraries to v6

AQTS 2021.4 Update 1 - Framework version 2.14

• Added GageHeightReferencePointName to DischargeActivity to set a Reference Point for the Mean Gage Height of a Discharge Activity.

AQTS 2021.4 - Framework version 2.13

• Added WaterQualityCrossSection flag to CompletedVisitActivities to indicate if a WaterQualityCrossSection activity was performed during the field visit.

AQTS 2021.3 Update 1 - Framework version 2.12

• Added GageHeightAdjustmentAmount and GageHeightComments to DischargeActivity to record an optional adjustment amount (and reason for the adjustment) for Mean Gage Height. MeanGageHeight and GageHeightAdjustmentAmount values are combined by the framework and should not be pre-calculated by the plugin.
• Added AdjustmentAmount to Reading to record an optional adjustment amount. Adjusted Reading and AdjustmentAmount values are combined by the framework and should not be pre-calculated by the plugin.

Note - Framework version 2.11 was not publicly released

AQTS 2020.3 - Framework version 2.10

• Added support for GageZeroFlowActivity activities to record the gage height at zero flow.
• Added support for OtherDischargeSection channel measurements which can be used to record channel measurements measured against any discharge monitoring method.
• Added support for EngineeredStructureDischarge channel measurements which can be used to record the discharge from a physical structure such as a weir or flume.
• Added support for VolumetricDischarge channel measurements which can be used to record the discharge based on an observation of the amount of time it takes for the stream to fill a container of known volume.
• ManualGaugingDischargeSection.DischargeMethod is now optional and will no longer default to MidSection if omitted.

AQTS 2020.2 - Framework version 2.9

• Added optional NumberOfVerticals property to ManualGaugingDischargeSection. This property can be used to specify the number of verticals a manual gauging activity has when there are no verticals to be imported. The NumberOfVerticals must be null or match the vertical count when the activity has verticals.
• Added optional MeasurementTime property to DischargeActivity. MeasurementTime is set to half way between the Start and End of the MeasurementPeriod when null.
• Added the Dictionary<string, string> GetPluginConfigurations() method to the IFieldDataResultsAppender interface. This method can be used to retrieve configuration settings for each installed plugin. The Settings page of the System Config app can be used to change plugin settings in the FieldDataPluginConfig-{PluginName} group.

AQTS 2019.4 Update 1 - Framework version 2.7

• Added optional Grade property to Reading. See Grade Data Type for more details about working with Grades.

AQTS 2019.4 - Framework version 2.6

• Added optional UseLocationDatumAsReference property to Reading. Set this property to indicate that the Reading is measured against the local assumed datum of the Reading's location. Cannot be used in combination with a Reference Point.

AQTS 2019.3 - Framework version 2.5

• Added optional SensorUniqueId property to Reading and Calibration. Set the UniqueId of an existing Sensor to associate the Reading / Calibration with the physical device used to retrieve the measurement. The SensorUnique Id can be retrieved from the Sensors/Gauges tab in Springboard or by using the Publish or Provisioning APIs.
• Added optional ReadingQualifiers property to Reading. Allows adding multiple qualifiers to a Reading.
• Deprecated ReadingQualifier property on Reading. Replaced with ReadingQualifiers.

AQTS 2019.2 - Framework version 2.3

• Added optional MeasurementGrade property to DischargeActivity. See Grade Data Type for more details about working with Grades.
• Added ActiveUncertaintyType enumeration and optional property on DischargeActivity. Can be one of None, Quantitative or Qualitative and is used to indicate which uncertainty property the system will use for Field Visit Readings in the Rating Development Toolbox.
• Added QualitativeUncertaintyType enumeration and property on DischargeActivity. This property is required when ActiveUncertaintyType is Qualitative otherwise, optional.
• Added QuantitativeUncertainty property to DischargeActivity. This property is required when ActiveUncertaintyType is Quantitative otherwise, optional.
• Added optional QualityAssuranceComments property to DischargeActivity.
• Added support for Inspections and Calibrations.

AQTS 2019.1 - Framework version 2.1

• Added optional MeterCalibration property to ManualGaugingDischargeSection.

AQTS 2018.4 - Framework version 2.0

• Field data plug-ins do not need to be re-installed following an upgrade to AQUARIUS Time-Series
• Changed location where field data plug-ins are installed to %ProgramData%\Aquatic Informatics\AQUARIUS Server\FieldDataPlugins
• Modified Provisioning API POST /fielddataplugins to upload and install field data plug-ins packaged in a *.plugin file
• Breaking change: Changed the parameter type of ControlCondition.ControlCondition from the ConditionType Enum to ControlConditionPickList. See PickList Data Type for more details about working with PickLists.
• Breaking change: Added a required PointOrder property to CrossSectionPoint. This property controls the order in which the points are drawn. Previously, points would always be drawn in order of Distance. Valid values for the property are 1, 2, ..., N where N is the total number of points in the cross-section. Values cannot be repeated. See the examples page for Cross-Section Surveys for more detail.
• Added support for vertical segments and overhangs in cross-section profiles.

AQTS 2018.3 - Framework version 1.3

• Added AdcpDischargeSection channel measurement activity
• Added ValidationChecks.CannotBeNegative() helper method

AQTS 2018.2 - Framework version 1.2

• Added support to set reference points on Readings
• Added Weather, CollectionAgency and CompletedVisitActivities properties on FieldVisitDetails
• Added AdjustmentType, AdjustmentAmount, and ReasonForAdjustment properties on DischargeActivity
• Expanded GageHeightCalculationType to support ManuallyCalculated and SimpleAverage options on DischargeActivity
• Added support for ControlConditions
• Enhanced the field data plug-in framework to support uploading attachments with a field data file

AQTS 2018.1 - Framework version 1.1

• Changed the framework guide from PDF to Markdown
• Added support for Level Surveys

AQTS 2017.4 - Framework version 1.0

• First supported release of the plug-in framework
• The PDF of the 2017.4 framework guide can be found here.

Overview

Figure 1 illustrates how the Field Data Plug-In Framework fits into the overall architecture of AQUARIUS Time-Series.

The Framework is called when:

• A user utilizes the available drag-and-drop upload method from Springboard, Location Manager, or the Field Data Editor

OR

• A program calls the PostFieldDataAttachment operation from the AQUARIUS Acquisition API via a POST http://<yourserver>/AQUARIUS/Acquisition/v2/locations/{LocationUniqueId}/visits/uploads/plugins operation.

Figure 1: Architecture Overview

Field data plug-ins are dynamically loaded at runtime and run within its own AppDomain. Each plug-in is sand-boxed, running locally and protected from other areas of the system.

Control Flow

Figure 2 illustrates the control flow after a user uploads a field data file to the Framework.

Figure 2: Control Flow

Plug-ins are registered with the Framework and given a priority. The priority determines the order that the Framework runs the plug-ins to parse the field data file. When a field data file is submitted to the Framework, the file is passed to each parser, in ascending priority order. Each parser returns one of the following statuses:

• CannotParse. The plug-in could not parse the file. The Framework will pass the file to the next highest priority plug-in. If the file is submitted through the drag-and-drop interface in Springboard, Location Manager, or the Field Data Editor, and there are no more registered plug-ins, then the Framework passes the file to the built-in plug-ins (e.g., SWAMI, FlowTracker *.DIS, AquaCalc) to parse. If the file is submitted through the AQUARIUS Acquisition API, an HTTP error is returned.
• SuccessfullyParsedButDataInvalid. The plug-in parsed the file but failed to process the data. The framework will exit with an error. The error will be logged to a log file (FieldDataPluginFramework.log) and a "Failed" status will be reported in Springboard or an HTTP error is returned in the AQUARIUS Acquisition API.
• SuccessfullyParsedAndDataValid. The plug-in parsed the file and processed the data without error. The framework will try to save the field data to the database and the save status will be reported below the drag-and-drop area in Springboard, Location Manager or the Field Data Editor (FDE) or an HTTP created status (201) is returned in the AQUARIUS Acquisition API.

Data Model

The Framework SDK provides a number of data objects. Your plug-in is responsible for mapping the contents of a field data file to one or more of these data objects. The Framework uses these data objects to save field data to the AQUARIUS Time-Series Server.

Figure 3 illustrates the data objects and their relationships, as currently supported by the Framework.

Figure 3: Data Object Diagram

LocationInfo is an immutable object. When a plug-in is called without location context, as is the case when data files are submitted through Springboard drag-and-drop, the plug-in can use the IFieldDataResultsAppender to look-up a location by its identifier or unique ID. We recommend that plug-ins and their field data file reference locations by its unique ID for the following reasons:

• The unique ID is a string identifier assigned by AQUARIUS Time-Series. It is guaranteed to be unique.
• The location identifier can be changed from Springboard or by using the Provisioning API. When this happens, any field data that references the location identifier needs to be updated or AQUARIUS Time-Series will throw a LocationNotFound exception.
• If the AQUARIUS Time-Series Server has an Oracle database, the location identifier is case sensitive. In this instance, locations with different capitalization can co-exist as separate locations (e.g., "My Location" vs. “my location”). This introduces fragility in the field data because it must be precise when identifying its location context.

All Framework SDK data objects require timestamps specified as DateTimeOffset. However, some plug-ins may process field data containing timestamps that do not include an UTC-offset. Plug-ins can use these timestamps to construct a DateTimeOffset object by combining it with the UtcOffset property in LocationInfo.

Similarly, FieldVisitInfo is an immutable object. It is created when the plug-in uses the IFieldDataResultsAppender to add an instance of a FieldVisitDetails object to a location.

There are factory methods to create an instance of CrossSectionSurvey, DischargeActivity, and ManualGaugingDischargeSection object. Plug-ins should also use IFieldDataResultsAppender to add field data objects to IFieldVisitInfo.

LevelSurveyMeasurement elevation values (MeasuredElevation) use the same units as the location elevation.

Some enumerated types in the Framework SDK default to "Unknown" value. When creating data objects that have these enumerated types as properties, the properties should be set to “Unspecified” value if there is no other appropriate value for it so that the data displays correctly in the Field Data Editor. For example, when creating an instance of ManualGaugingDischargeSection,

//These enums fail if you use "Unknown".  “Unspecified” is what should be used
//when nothing else fits;
var factory = new ManualGaugingDischargeSectionFactory(UnitSystem);
var manualGauging = factory.CreateManualGaugingDischargeSection(measurementPeriod, 
                                                                results.TotalDischarge.Value);

manualGauging.MeterSuspension = MeterSuspensionType.Unspecified;
manualGauging.DeploymentMethod = DeploymentMethodType.Unspecified;

If these properties are left as the default "Unknown" value, the Framework will save the field data, but when the data is viewed in the Field Data Editor, it will show an invalid icon (orange exclamation mark) by the discharge value. Unfortunately, this value cannot be corrected in the Field Data Editor.

PickList Data Type

In AQUARIUS Time-Series, certain field visit activity properties are pick lists, a customizable collection of key-value pairs, known as a PickListItem. A PickListItem has an identifier and a display name, which is shown in the Field Data Editor.

In the Framework, pick list properties are defined as data type, PickList. The PickList data type is similar to an enumeration, but the possible values are only known at run time. A PickList instance is constructed by specifying either the identifier or the display name of a PickListItem that belongs to the pick list. When the Framework saves the field data to the AQUARIUS Time-Series Server, it will validate the PickList. If the PickList specifies an invalid PickListItem, the Framework will return an error message that lists all of the PickListItems that belong to the pick list.

Grade Data Type

Similar to pick lists, Time-Series has a customizable collection of Grades. A Grade has a numeric code, display name, and other properties which can be viewed and edited with the Provisioning API or System Config page.

In the Framework, grades are defined as the Grade data type. Grades can be specified using either the numeric code or display name (the "Name" property in Provisioning), using the static methods Grade.FromCode or Grade.FromDisplayName. The two options are provided for convenience. In either case, the Framework looks up and assigns the matching Grade to the graded object. If it cannot find a matching Grade, the Framework will return an error message.

Setting up a Development Environment

Plug-ins are 64-bit libraries written using .NET Framework 4.7.2. If the .NET Framework 4.7.2 developer pack is not already installed, it can be downloaded here.

Install the Field Data Framework in your project via NuGet.

PM> Install-Package Aquarius.FieldDataFramework

This will install:

• the lib\FieldDataPluginFramework.dll assembly, which your plug-in must consume and implement.
• the tools\PluginTester.exe tool, for quickly testing your plug-in without needing an AQTS server.
• the tools\PluginPackager.exe tool, for packaging your plug-in into a single *.plugin file for easy deployment.

Writing a Field Data Plug-In

A plug-in implements the interface, IFieldDataPlugin:

namespace FieldDataPluginFramework
{
    public interface IFieldDataPlugin
    {
        ParseFileResult ParseFile(Stream fileStream,
                                  IFieldDataResultsAppender fieldDataResultsAppender, 
                                  ILog logger);
        ParseFileResult ParseFile(Stream fileStream, 
                                  LocationInfo targetLocation, 
                                  IFieldDataResultsAppender fieldDataResultsAppender, 
                                  ILog logger);
    }
}

In IFieldDataPlugin, there are two ParseFile method signatures:

1. Global (i.e. no location context)
   1. Called by the Framework when Springboard is used to import a data file;
   2. Data file must contain a reference to the location that owns the field data;
2. With location context (i.e. targetLocation)
   1. Called by the Framework when LocationManager, the Field Data Editor or AQUARIUS Acquisition API is used to import a data file;
   2. Field data will be created in the targetLocation. If the data file references a location, it must match the targetLocation or the field data will not be imported.

Each ParseFile method returns a ParseFileResult object. ParseFileResult has static constructors to create an instance to represent the plug-in status (SuccessullyParsedAndDataValid, SuccessfullyParsedButDataInvalid, CannotParse). If your plug-in returns a SuccessfullyParsedButDataInvalid or CannotParse status, we recommend that it includes an error message or exception, which will be displayed as an error message to the user by the drag-and-drop interfaces.

Tips

1. Plug-ins should not access the network. The Framework runs plug-ins locally in a sand-boxed environment.
2. Plug-ins should be thread safe. In the future, the Framework will call plug-ins concurrently from multiple threads or processes.
3. Plug-ins should only contain managed code. Aquatic Informatics will not provide developer support for plug-ins containing unmanaged code.
4. Plug-ins should be able to process field data files in less than one second. In the future, the Framework will terminate long-running plug-ins.
5. Plug-ins process any file that can be uploaded to the AQUARIUS Time-Series Server (through drag-and-drop in Springboard, Location Manager and the Field Data Editor), including images and video files. Plug-ins should try to quickly inspect the start of the file stream and return the CannotParse status if the content does not make sense.

Parsing the Field Data File

A plug-in is responsible for reading the field data file and mapping its contents to one or more of the data objects provided by the Framework SDK (FieldDataPluginFramework.dll). The ParseFile method provides a Stream as one of its inputs. The Stream is opened by the Framework and contains the memory stream byte representation of the field data file. Below is an example of how the contents of the stream are read (in C#):

public void ReadFile(Stream stream)
{
    using (var reader = CreateStreamReader(stream, Encoding.UTF8))
    {
        //Read data from file.
    }
}

private static StreamReader CreateStreamReader(Stream stream, Encoding fileEncoding)
{
    const int defaultByteBufferSize = 1024;
    //NOTE: Make sure to set leaveOpen property on StreamReader so the Stream’s
    //Dispose() method is not called when StreamReader.Dispose() is called.
    //Framework will take care of closing the Stream.

    return new StreamReader(fileStream, fileEncoding,
	                        detectEncodingFromByteOrderMarks: true,
	                        bufferSize: defaultByteBufferSize, leaveOpen: true);
}

Logging

The Field Data Plug-In Framework log messages are written to the log file, FieldDataPluginFramework.log, found on the server at:

%ProgramData%\Aquatic Informatics\AQUARIUS\Logs

Each plug-in is provided a reference to an ILog object when its ParseFile method is called. The ILog object writes log messages from the plug-in to the FieldDataPluginFramework.log. It is important to note that since both the Framework and plug-in log to FieldDataPluginFramework.log, error messages from a plug-in are logged as warnings because plug-in errors are not the same severity as errors from the Framework (Framework errors indicate problems saving field data to the AQUARIUS Time-Series Server).

The log messages written to FieldDataPluginFramework.log are designed to be easily parsed by log analytic tools, such as LogStash.

Packaging your Plug-In for Easy Deployment

See the PluginPacker project for details on adding a build step that will automatically create a *.plugin file every time your project builds in Visual Studio. This file can be deployed using the System Config page on your AQTS app server.

Tips

1. To update the registration of your plug-in (e.g., change the plug-in priority), you need to un-register and re-register your plug-in.
2. We recommend selecting plug-in priorities so that there is a gap between each priority. This makes it easier to insert new plug-ins without having to re-order the priorities of every registered plug-in.

For example, your AQUARIUS Time-Series Server has two plug-ins registered so that PluginA is assigned priority = 1 and PluginC is assigned priority = 3.

Now, you would like to install a new plug-in, PluginB, so that it is the second plug-in that will be run by the Framework. To do so, you need to un-register and re-register PluginC with priority = 3 so that you can register PluginB with priority = 2.

However, if PluginA and PluginB were registered so that PluginA is assigned priority = 100 and PluginC is assigned priority = 300, you do not need to change the priority of PluginC; you only need to assign PluginC a priority between 100 and 300.

Unregistering your Plug-In

A plug-in can be unregistered from the Framework using the DELETE /fielddataplugins endpoint in Provisioning API. The delete endpoint only removes the Framework’s awareness of the plug-in (i.e. when a field data file is submitted, the Framework will not run the plug-in), it does not delete the plug-in’s libraries from the AQUARIUS Time-Series Server.

To unregister a plug-in, you need to refer to the plug-in by its unique ID. A plug-in’s unique id can be found by using the GET /fielddataplugins endpoint in Provisioning API to retrieve a list of all plug-ins registered with the Framework.

Debugging your Plug-In

After installing and registering your plug-in on the AQUARIUS Time-Series Server, you can test and debug your plug-in by using the drag-and-drop upload method, (available from Springboard, Location Manager or the Field Data Editor), to upload your test field data files. We recommend that your plug-in specifies an error message or exception when it returns a SuccessfullyParsedButDataInvalid or CannotParse status. The error message or exception is displayed as feedback to the user by the drag-and-drop interfaces. Otherwise, the feedback from these drag-and-drop interfaces may not be very informative:

Figure 4: Drag-and-drop Error Message

To help debug problems with your plug-in, look in FieldDataPluginFramework.log to find more details about why the data failed to be saved to the AQUARIUS Time-Series Server.

Another approach to testing and debugging your plug-in is to use the AQUARIUS Acquisition API, which provides more details in its response to help you quickly diagnose problems:

Figure 5: AQUARIUS Acquisition API Response

However, we still recommend that you refer to FieldDataPluginFramework.log to find more details about your plug-in failure.

Additionally, the Aquatic Informatics Field Data Framework repo on GitHub provides the PluginTester.exe as a tool to help debug your plug-in outside of the AQUARIUS Time-Series Server environment. The Plug-in Tester can be found here. This tool is also available in tools folder of the NuGet package.

In particular, when the /Json=AppendedResults.json command line option is used, the JSON written to disk will include both the Framework data models written by your plug-in and the correct AssemblyQualifiedTypeName to use during plug-in registration.

If you make changes to your plug-in, it is not necessary to unregister and re-register your plug-in with AQUARIUS Time-Series Server. Simply replace the existing plug-in library on the server with its newer version.

Tips

1. Debugging your plug-in can be difficult if it throws any unhandled exceptions because these exception messages are logged to the FieldDataPluginFramework.log at the DEBUG log level. By default, all AQUARIUS Time-Series Server logs are set to the INFO log level. To avoid this problem, we recommend that you write your own top-level exception handler that will log all exceptions to FieldDataPluginFramework.log.
2. The DEBUG log level is the most verbose log level supported. You can increase the logging verbosity for all AQUARIUS Time-Series Server logs by changing the log level from <level value="INFO”> to <level value=”DEBUG”> in %ProgramData%\Aquatic Informatics\AQUARIUS\Logs\RemotingAppender.config.

This change will take effect immediately, without needing to re-start the server. While changing the log level to DEBUG can help with debugging, it will also fill up the log file with a large number of logging messages.

This will make reading the log files much more difficult, so you may want to search for specific terms instead of scanning the logs for errors.

When you have finished debugging your plug-in, we recommend that you restore the log level in RemotingAppender.config to the INFO log level.

Uploading Field Data with Attachments

The Field Data Plug-In Framework will attach files to field visits when the custom field data file and its attachments are bundled into a single zip file. The zip file is organized as follows:

• A single custom field data file that is located at the root of the zip file;
• A sub-folder that is located at the root of the zip file containing any field visit attachments (for example, images, videos, files). The sub-folder can be given any name; and
• If there are multiple files located at the root level of the zip folder, the Field Data Plug-In Framework will fail to parse the zip file for attachments.

When the Field Data Plug-In Framework processes the zip file, it will:

• Pass the field data file to each registered parser. If the field data file is successfully parsed by a plug-in, it will create the field visits;
• Attach all of the files contained in the sub-folder to each field visit; and
• Not attach the zip file to the field visits.