fix: classify CPC image errors and clarify intersection naming

CHANGELOG.md

@@ -23,6 +23,7 @@ All significant changes to this project will be documented in this file.
 ### Bug fixes
 
 * `FrequentItemsSketch::serialize` now writes the full 8-byte preamble for an empty sketch, matching the Java and C++ encoding. Empty sketches previously serialized to 6 bytes, which `FrequentItemsSketch::deserialize` rejected with an insufficient-data error.
+* `CpcSketch` and `CpcWrapper` now classify out-of-range fields in serialized images as `InvalidData` rather than `InvalidArgument`.
 
 ## v0.3.0 (2026-05-18)
 

datasketches/src/cpc/sketch.rs

@@ -630,13 +630,10 @@ impl CpcSketch {
             ErrorKind::InvalidData,
         )?;
         if !(MIN_LG_K..=MAX_LG_K).contains(&lg_k) {
-            return Err(Error::invalid_argument(format!(
-                "lg_k out of range; got {}",
-                lg_k
-            )));
+            return Err(Error::deserial(format!("lg_k out of range; got {}", lg_k)));
         }
         if first_interesting_column > 63 {
-            return Err(Error::invalid_argument(format!(
+            return Err(Error::deserial(format!(
                 "first_interesting_column out of range; got {}",
                 first_interesting_column
             )));

datasketches/src/cpc/wrapper.rs

@@ -62,13 +62,10 @@ impl CpcWrapper {
             .read_u8()
             .map_err(insufficient_data("first_interesting_column"))?;
         if !(MIN_LG_K..=MAX_LG_K).contains(&lg_k) {
-            return Err(Error::invalid_argument(format!(
-                "lg_k out of range; got {}",
-                lg_k
-            )));
+            return Err(Error::deserial(format!("lg_k out of range; got {}", lg_k)));
         }
         if first_interesting_column > 63 {
-            return Err(Error::invalid_argument(format!(
+            return Err(Error::deserial(format!(
                 "first_interesting_column out of range; got {}",
                 first_interesting_column
             )));

datasketches/src/thetafamily/common/intersection.rs

@@ -243,8 +243,8 @@ where
         self.table.estimated_size()
     }
 
-    /// Return the current intersection state as compact-sketch parts.
-    pub fn result(&self, ordered: bool) -> CompactSketchParts<E>
+    /// Returns the current intersection state as compact-sketch parts.
+    pub fn to_compact_parts(&self, ordered: bool) -> CompactSketchParts<E>
     where
         E: Clone,
     {

datasketches/src/thetafamily/common/jaccard_similarity.rs

@@ -194,7 +194,7 @@ where
     let mut intersection = IntersectionState::new(seed, NoopMergePolicy);
     intersection.update(KeyEntries(sketch_a))?;
     intersection.update(KeyEntries(sketch_b))?;
-    let intersection = intersection.result(false);
+    let intersection = intersection.to_compact_parts(false);
     let intersection_count = intersection
         .entries
         .iter()

datasketches/src/thetafamily/theta/intersection.rs

@@ -83,7 +83,7 @@ impl ThetaIntersection {
             self.state.has_result(),
             "ThetaIntersection::to_sketch() called before first update()"
         );
-        let parts = self.state.result(ordered);
+        let parts = self.state.to_compact_parts(ordered);
         CompactThetaSketch::from_parts(
             parts
                 .entries

datasketches/src/thetafamily/tuple/intersection.rs

@@ -153,7 +153,7 @@ where
             self.state.has_result(),
             "TupleIntersection::to_sketch() called before first update()"
         );
-        let parts = self.state.result(ordered);
+        let parts = self.state.to_compact_parts(ordered);
         CompactTupleSketch::from_parts(
             parts.entries,
             parts.theta,

datasketches/tests/cpc_test/wrapper.rs

@@ -19,6 +19,7 @@ use datasketches::common::NumStdDev;
 use datasketches::cpc::CpcSketch;
 use datasketches::cpc::CpcUnion;
 use datasketches::cpc::CpcWrapper;
+use datasketches::error::ErrorKind;
 use googletest::assert_that;
 use googletest::prelude::contains_substring;
 use googletest::prelude::eq;
@@ -90,3 +91,25 @@ fn test_is_compressed() {
         contains_substring("only compressed sketches are supported")
     );
 }
+
+#[test]
+fn test_invalid_image_fields_are_invalid_data() {
+    let original = CpcSketch::new(10).serialize();
+
+    for (index, value, field) in [
+        (3, 3, "lg_k"),
+        (3, 27, "lg_k"),
+        (4, 64, "first_interesting_column"),
+    ] {
+        let mut bytes = original.clone();
+        bytes[index] = value;
+
+        let sketch_err = CpcSketch::deserialize(&bytes).unwrap_err();
+        assert_that!(sketch_err.kind(), eq(ErrorKind::InvalidData));
+        assert_that!(sketch_err.message(), contains_substring(field));
+
+        let wrapper_err = CpcWrapper::new(&bytes).unwrap_err();
+        assert_that!(wrapper_err.kind(), eq(ErrorKind::InvalidData));
+        assert_that!(wrapper_err.message(), contains_substring(field));
+    }
+}