feat(scrubbing): Scrub span.data.http.query with default scrubbers

CHANGELOG.md

@@ -8,6 +8,7 @@
 - Protocol validation for source map image type. ([#1869](https://github.com/getsentry/relay/pull/1869))
 - Strip quotes from client hint values. ([#1874](https://github.com/getsentry/relay/pull/1874))
 - Add Dotnet, Javascript and PHP support for profiling. ([#1871](https://github.com/getsentry/relay/pull/1871), [#1876](https://github.com/getsentry/relay/pull/1876), [#1885](https://github.com/getsentry/relay/pull/1885))
+- Scrub `span.data.http.query` with default scrubbers. ([#1889](https://github.com/getsentry/relay/pull/1889))
 
 **Bug Fixes**:
 

relay-general/src/pii/processor.rs

@@ -363,11 +363,12 @@ fn insert_replacement_chunks(rule: &RuleRef, text: &str, output: &mut Vec<Chunk<
 
 #[cfg(test)]
 mod tests {
-    use crate::pii::{PiiConfig, ReplaceRedaction};
+
+    use crate::pii::{DataScrubbingConfig, PiiConfig, ReplaceRedaction};
     use crate::processor::process_value;
     use crate::protocol::{
-        Addr, DebugImage, DebugMeta, Event, ExtraValue, Headers, LogEntry, NativeDebugImage,
-        Request, TagEntry, Tags,
+        Addr, DataElement, DebugImage, DebugMeta, Event, ExtraValue, Headers, HttpElement,
+        LogEntry, NativeDebugImage, Request, Span, TagEntry, Tags,
     };
     use crate::testutils::assert_annotated_snapshot;
     use crate::types::{Annotated, Object, Value};
@@ -967,4 +968,54 @@ mod tests {
         );
         assert_eq!(chunks, res);
     }
+
+    #[test]
+    fn test_scrub_span_data_not_scrubbed() {
+        let mut span = Annotated::new(Span {
+            data: Annotated::new(DataElement {
+                http: Annotated::new(HttpElement {
+                    query: Annotated::new("dance=true".to_owned()),
+                    ..Default::default()
+                }),
+                ..Default::default()
+            }),
+            ..Default::default()
+        });
+
+        let ds_config = DataScrubbingConfig {
+            scrub_data: true,
+            scrub_defaults: true,
+            ..Default::default()
+        };
+        let pii_config = ds_config.pii_config().unwrap().as_ref().unwrap();
+        let mut pii_processor = PiiProcessor::new(pii_config.compiled());
+
+        process_value(&mut span, &mut pii_processor, ProcessingState::root()).unwrap();
+        assert_annotated_snapshot!(span);
+    }
+
+    #[test]
+    fn test_scrub_span_data_is_scrubbed() {
+        let mut span = Annotated::new(Span {
+            data: Annotated::new(DataElement {
+                http: Annotated::new(HttpElement {
+                    query: Annotated::new("ccnumber=5105105105105100&process_id=123".to_owned()),
+                    ..Default::default()
+                }),
+                ..Default::default()
+            }),
+            ..Default::default()
+        });
+
+        let ds_config = DataScrubbingConfig {
+            scrub_data: true,
+            scrub_defaults: true,
+            ..Default::default()
+        };
+        let pii_config = ds_config.pii_config().unwrap().as_ref().unwrap();
+        let mut pii_processor = PiiProcessor::new(pii_config.compiled());
+
+        process_value(&mut span, &mut pii_processor, ProcessingState::root()).unwrap();
+        assert_annotated_snapshot!(span);
+    }
 }

relay-general/src/pii/snapshots/relay_general__pii__processor__tests__scrub_span_data_is_scrubbed.snap

@@ -0,0 +1,30 @@
+---
+source: relay-general/src/pii/processor.rs
+expression: span
+---
+{
+  "data": {
+    "http": {
+      "query": "ccnumber=[Filtered]&process_id=123"
+    }
+  },
+  "_meta": {
+    "data": {
+      "http": {
+        "query": {
+          "": {
+            "rem": [
+              [
+                "@creditcard:filter",
+                "s",
+                9,
+                19
+              ]
+            ],
+            "len": 40
+          }
+        }
+      }
+    }
+  }
+}

relay-general/src/pii/snapshots/relay_general__pii__processor__tests__scrub_span_data_not_scrubbed.snap

@@ -0,0 +1,11 @@
+---
+source: relay-general/src/pii/processor.rs
+expression: span
+---
+{
+  "data": {
+    "http": {
+      "query": "dance=true"
+    }
+  }
+}

relay-general/src/protocol/span.rs

@@ -1,4 +1,6 @@
-use crate::protocol::{JsonLenientString, OperationType, SpanId, SpanStatus, Timestamp, TraceId};
+use crate::protocol::{
+    DataElement, JsonLenientString, OperationType, SpanId, SpanStatus, Timestamp, TraceId,
+};
 use crate::types::{Annotated, Object, Value};
 
 #[derive(Clone, Debug, Default, PartialEq, Empty, FromValue, IntoValue, ProcessValue)]
@@ -45,7 +47,7 @@ pub struct Span {
 
     /// Arbitrary additional data on a span, like `extra` on the top-level event.
     #[metastructure(pii = "maybe")]
-    pub data: Annotated<Object<Value>>,
+    pub data: Annotated<DataElement>,
 
     // TODO remove retain when the api stabilizes
     /// Additional arbitrary fields for forwards compatibility.
@@ -58,6 +60,8 @@ mod tests {
     use chrono::{TimeZone, Utc};
     use similar_asserts::assert_eq;
 
+    use crate::protocol::HttpElement;
+
     use super::*;
 
     #[test]
@@ -70,7 +74,12 @@ mod tests {
   "op": "operation",
   "span_id": "fa90fdead5f74052",
   "trace_id": "4c79f60c11214eb38604f4ae0781bfb2",
-  "status": "ok"
+  "status": "ok",
+  "data": {
+    "http": {
+      "query": "is_sample_query=true"
+    }
+  }
 }"#;
 
         let span = Annotated::new(Span {
@@ -84,6 +93,13 @@ mod tests {
             trace_id: Annotated::new(TraceId("4c79f60c11214eb38604f4ae0781bfb2".into())),
             span_id: Annotated::new(SpanId("fa90fdead5f74052".into())),
             status: Annotated::new(SpanStatus::Ok),
+            data: Annotated::new(DataElement {
+                http: Annotated::new(HttpElement {
+                    query: Annotated::new("is_sample_query=true".to_owned()),
+                    ..Default::default()
+                }),
+                ..Default::default()
+            }),
             ..Default::default()
         });
         assert_eq!(json, span.to_json_pretty().unwrap());

relay-general/src/protocol/types.rs

@@ -1117,6 +1117,26 @@ impl schemars::JsonSchema for Timestamp {
     }
 }
 
+#[derive(Clone, Debug, Default, PartialEq, Empty, FromValue, IntoValue, ProcessValue)]
+#[cfg_attr(feature = "jsonschema", derive(JsonSchema))]
+pub struct DataElement {
+    #[metastructure(pii = "maybe")]
+    pub http: Annotated<HttpElement>,
+
+    #[metastructure(additional_properties, retain = "true", pii = "maybe")]
+    pub other: Object<Value>,
+}
+
+#[derive(Clone, Debug, Default, PartialEq, Empty, FromValue, IntoValue, ProcessValue)]
+#[cfg_attr(feature = "jsonschema", derive(JsonSchema))]
+pub struct HttpElement {
+    #[metastructure(pii = "true")]
+    pub query: Annotated<String>,
+
+    #[metastructure(additional_properties, retain = "true", pii = "maybe")]
+    pub other: Object<Value>,
+}
+
 #[cfg(test)]
 mod tests {
     use similar_asserts::assert_eq;