Purview DLP Share Point Alert Info

Query

let alert_id = "<<<>>>";
let query_period = 1d;
let _FilePaths =
    AlertEvidence
    | where Timestamp > ago(query_period)
    | where EntityType == "File" and AlertId == strcat("dl", alert_id)
    | extend AdditionalFields = todynamic(AdditionalFields)
    | distinct SourceFilePath = strcat(AdditionalFields["Directory"], "/", AdditionalFields["Name"]), ObjectName = url_decode(tostring(AdditionalFields["Name"]))
;
CloudAppEvents
| where Timestamp > ago(query_period)
| where ActionType in ("AlertTriggered", "AlertUpdated") and tostring(RawEventData["AlertId"]) == alert_id
| summarize arg_max(Timestamp, *)
| extend
    AlertId = tostring(RawEventData["AlertId"]),
    Data = todynamic(tostring(RawEventData["Data"]))
| extend
    UserPrincipalName = tolower(tostring(Data["f3u"])),
    Severity = tostring(Data["sev"]),
    Workload = tostring(Data["wl"]),
    SensitiveInformationContentType = split(tostring(Data["sict"]), ";"),
    SensitiveInformationTypeMatchInfo = split(tostring(Data["sitmi"]), ";"),
    PolicyId = tostring(Data["dpid"]),
    PolicyName = tostring(Data["dpn"]),
    ManagementRuleId = tostring(Data["drid"]),
    //RuleId = tostring(Data["drid"]),
    RuleName = tostring(Data["drn"]),
    ProtectionActions = split(tostring(Data["dact"]), ", "),
    ObjectName = tostring(Data["von"]),
    DMC = split(tostring(Data["dmc"]), ","),
    FileOwnerName = tolower(tostring(Data["flmb"])), // It might not the same as f3u - UserPrincipalName
    FileOwnerDisplayName = tostring(Data["fo"]),
    USI = tostring(Data["usi"])
| where Workload == "SharePoint" and tostring(RawEventData["Category"]) == "DataLossPrevention"
| lookup kind=leftouter _FilePaths on ObjectName
| project
    Timestamp,
    Application,
    ActionType,
    AlertId,
    UserPrincipalName,
    Severity,
    Workload,
    SensitiveInformationContentType,
    SensitiveInformationTypeMatchInfo,
    PolicyId,
    PolicyName,
    ManagementRuleId,
    //RuleId,
    RuleName,
    DMC,
    ProtectionActions,
    ObjectName,
    SourceFilePath,
    FileOwnerName,
    FileOwnerDisplayName,
    USI,
    AlertTriggered_RawEventData = RawEventData,
    AlertTriggered_ReportId = ReportId
// | summarize arg_max(CreationTime, *) by AlertId, Timestamp, SourceFilePath, ObjectName, ManagementRuleId
| join hint.strategy=shuffle kind=leftouter (
    CloudAppEvents
    | where Timestamp > ago(query_period)
    | where ActionType in ("DLPRuleMatch", "DLPRuleUndo") and Application == "Microsoft SharePoint Online"
    | mv-expand PolicyDetail = iff(array_length(RawEventData["PolicyDetails"]) > 0, RawEventData["PolicyDetails"], dynamic([""]))
    | mv-expand Rule = iff(array_length(PolicyDetail["Rules"]) > 0, PolicyDetail["Rules"], dynamic([""]))
    | mv-apply SensitiveInformationItem = Rule["ConditionsMatched"]["SensitiveInformation"] on (
        summarize SensitiveInformation = make_list(bag_remove_keys(SensitiveInformationItem, dynamic(["SensitiveInformationDetailedClassificationAttributes"])))
    )
    | project
        CreationTime = todatetime(RawEventData["CreationTime"]),
        ManagementRuleId = tostring(Rule["ManagementRuleId"]),
        ObjectName,
        //IncidentId = tostring(RawEventData["IncidentId"]),
        SensitiveInformation,
        ActionType,
        Reason = tostring(RawEventData["ExceptionInfo"]["Reason"]),
        //SharePointMetaData = RawEventData["SharePointMetaData"],
        SiteCollectionUrl = tostring(RawEventData["SharePointMetaData"]["SiteCollectionUrl"]),
        //IsViewableByExternalUsers = tostring(RawEventData["SharePointMetaData"]["IsViewableByExternalUsers"]),
        //ItemLastSharedTime = todatetime(RawEventData["SharePointMetaData"]["ItemLastSharedTime"]),
        //SharedBy = todatetime(RawEventData["SharePointMetaData"]["SharedBy"]),
        DLPRuleMatch_RawEventData = RawEventData,
        DLPRuleMatch_ReportId = ReportId
    // The same file can be updated, and suffer several "DLPRuleMatch" and "DLPRuleUndo" actions
    | summarize
        EvaluationEventCount = count(),
        arg_min(CreationTime, SensitiveInformation, SiteCollectionUrl, DLPRuleMatch_RawEventData, DLPRuleMatch_ReportId),
        (LastEvaluationTime, LastActionType, LastReason, LastSensitiveInformation) = arg_max(CreationTime, ActionType, Reason, SensitiveInformation)
        by ObjectName, ManagementRuleId
    | extend
        LastEvaluationTime = iff(EvaluationEventCount == 1, datetime(null), LastEvaluationTime),
        LastActionType = iff(EvaluationEventCount == 1, "", LastActionType),
        LastReason = iff(EvaluationEventCount == 1, "", LastReason),
        LastSensitiveInformation = iff(EvaluationEventCount == 1, dynamic(null), LastSensitiveInformation)
    ) on $left.SourceFilePath == $right.ObjectName, ManagementRuleId //, CreationTime does not always match
| project-away ObjectName1, ManagementRuleId1
// Add info about the sharing targets
// Did not have any example to work with, where a DLP alert happened and the file had ActionType in ("AddedToGroup", "AddedToSecureLink", "AddedToSharingLink")
| as hint.materialized=true _Events
| join kind=leftouter (
    CloudAppEvents
    | where Timestamp > ago(query_period)
    | where Application == "Microsoft SharePoint Online" and ActionType in ("AddedToGroup", "AddedToSecureLink", "AddedToSharingLink") // "GroupAdded", "SecureLinkCreated", "SharingLinkCreated", "SharingSet", "SharingInheritanceBroken"
    | extend
        UserId = tostring(RawEventData["UserId"]),
        CreationTime = todatetime(RawEventData["CreationTime"]),
        SharingOperation = tostring(RawEventData["Operation"]),
        SiteUrl = tostring(RawEventData["SiteUrl"]),
        ListId = tostring(RawEventData["ListId"]),
        ListItemUniqueId = tostring(RawEventData["ListItemUniqueId"]),
        ObjectName = tostring(RawEventData["ObjectId"]),
        ItemType = tostring(RawEventData["ItemType"]),
        Permission = tostring(RawEventData["Permission"]),
        SharingLinkScope = tostring(RawEventData["SharingLinkScope"]),
        EventData = tostring(RawEventData["EventData"]),
        ModifiedProperties = tostring(RawEventData["ModifiedProperties"]),
        TargetUserOrGroupName = tostring(RawEventData["TargetUserOrGroupName"]),
        TargetUserOrGroupType = tostring(RawEventData["TargetUserOrGroupType"]),
        UniqueSharingId = tostring(RawEventData["UniqueSharingId"]),
        CorrelationId = tostring(RawEventData["CorrelationId"]),
        SharingEvent_RawEventData = RawEventData,
        SharingEvent_ReportId = ReportId
    | where tostring(toscalar(_Events | summarize make_set_if(url_decode(SourceFilePath), isnotempty(SourceFilePath)))) has ObjectName
    | summarize
        CreationTime = min(CreationTime),
        TargetUserOrGroupName = array_sort_asc(make_set_if(TargetUserOrGroupName, isnotempty(TargetUserOrGroupName))) // You could exclude targets from the same tenant
        by CorrelationId, SiteUrl, UserId, SharingOperation, UniqueSharingId, ObjectName, ItemType, Permission, SharingLinkScope, EventData, TargetUserOrGroupType
    | summarize
        CreationTime = min(CreationTime),
        TargetUserOrGroup = make_bag(bag_pack(TargetUserOrGroupType, TargetUserOrGroupName))
        by CorrelationId, SiteUrl, UserId, SharingOperation, UniqueSharingId, ObjectName, ItemType, Permission, SharingLinkScope, EventData
    | extend EmptyColumns = array_concat(
        iff(isnotempty(Permission), dynamic(null), dynamic(["Permission"])),
        iff(isnotempty(SharingLinkScope), dynamic(null), dynamic(["SharingLinkScope"])),
        iff(isnotempty(EventData), dynamic(null), dynamic(["EventData"]))
        )
    | summarize
        CreationTime = min(CreationTime),
        Operations = make_bag(bag_pack(SharingOperation, bag_remove_keys(bag_pack_columns(Permission, SharingLinkScope, EventData), EmptyColumns))),
        take_any(TargetUserOrGroup)
        by CorrelationId, SiteUrl, UserId, UniqueSharingId, ObjectName, ItemType, Auxiliar = tostring(TargetUserOrGroup)
    | summarize
        CreationTime = min(CreationTime),
        Objects = make_list(bag_pack_columns(ObjectName, ItemType, Operations)),
        take_any(TargetUserOrGroup)
        by CorrelationId, SiteUrl, UserId, Auxiliar
    | summarize
        SharingTime = min(CreationTime),
        SharingTargets = make_list(bag_pack_columns(TargetUserOrGroup, Objects))
        by CorrelationId, SiteUrl, UserId
    | sort by SharingTime asc
    | summarize
        SharingEvents = make_list(bag_pack_columns(SharingTime, CorrelationId, SharingTargets))
        by SiteUrl, UserId
    | project
        SiteCollectionUrl = SiteUrl,
        UserPrincipalName = UserId,
        SharingEvents
    ) on SiteCollectionUrl, UserPrincipalName
| project-away SiteCollectionUrl1, UserPrincipalName1
| project-reorder
    Timestamp,
    Application,
    ActionType,
    AlertId,
    UserPrincipalName,
    Severity,
    Workload,
    SensitiveInformationContentType,
    SensitiveInformationTypeMatchInfo,
    PolicyId,
    PolicyName,
    ManagementRuleId,
    //RuleId,
    RuleName,
    DMC,
    ProtectionActions,
    ObjectName,
    SourceFilePath,
    FileOwnerName,
    FileOwnerDisplayName,
    USI,
    AlertTriggered_RawEventData,
    AlertTriggered_ReportId,
    CreationTime,
    SensitiveInformation,
    EvaluationEventCount,
    LastEvaluationTime,
    LastActionType,
    LastReason,
    LastSensitiveInformation,
    SiteCollectionUrl,
    DLPRuleMatch_RawEventData,
    DLPRuleMatch_ReportId,
    SharingEvents

Explanation

This KQL (Kusto Query Language) query is designed to analyze and correlate data loss prevention (DLP) alerts and related events in a Microsoft SharePoint Online environment. Here's a simplified breakdown of what the query does:

Define Variables:
- alert_id: A placeholder for a specific alert ID.
- query_period: The time range for the query, set to 1 day.
Extract File Paths:
- From the AlertEvidence table, it retrieves distinct file paths and names related to the specified alert ID within the last day.
Retrieve Alert Events:
- From the CloudAppEvents table, it fetches events where an alert was triggered or updated, matching the specified alert ID.
- It extracts various details such as user information, severity, workload, policy details, and file owner information.
Filter for SharePoint DLP Alerts:
- It filters the events to only include those related to SharePoint and categorized as data loss prevention.
Join with File Paths:
- It performs a left outer join with the previously extracted file paths to correlate alert events with specific files.
Analyze DLP Rule Matches:
- It joins with additional CloudAppEvents to analyze rule matches and undos related to DLP policies.
- It summarizes the evaluation events, capturing details like the last action type and sensitive information involved.
Identify Sharing Events:
- It further joins with events related to file sharing actions (e.g., added to a group or secure link) in SharePoint.
- It aggregates sharing events to identify targets and operations related to the files involved in the DLP alerts.
Project and Reorder Columns:
- Finally, it selects and orders the columns to present a comprehensive view of the alert, including timestamps, user details, policy information, file paths, sensitive information, and sharing events.

In summary, this query is a complex analysis that correlates DLP alerts with file activities and sharing events in SharePoint, providing insights into potential data loss incidents and how files were shared or accessed.