Classify zero-family stock alias clusters

crates/rrt-runtime/src/building.rs

@@ -96,6 +96,7 @@ pub struct BuildingTypeRecoveredTableSummary {
     pub nonzero_bty_header_name_0x40_summaries: Vec<BuildingTypeBtyHeaderNameSummary>,
     pub nonzero_bty_header_name_0x5e_summaries: Vec<BuildingTypeBtyHeaderNameSummary>,
     pub nonzero_bty_header_name_0x7c_summaries: Vec<BuildingTypeBtyHeaderNameSummary>,
+    pub bty_header_name_0x5e_dword_summaries: Vec<BuildingTypeBtyHeaderNameDwordSummary>,
 }
 
 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
@@ -114,6 +115,16 @@ pub struct BuildingTypeBtyHeaderNameSummary {
     pub sample_file_names: Vec<String>,
 }
 
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct BuildingTypeBtyHeaderNameDwordSummary {
+    pub header_offset_hex: String,
+    pub header_value: String,
+    pub dword_0xbb: u32,
+    pub dword_0xbb_hex: String,
+    pub file_count: usize,
+    pub sample_file_names: Vec<String>,
+}
+
 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
 pub struct BuildingTypeSourceReport {
     pub directory_path: String,
@@ -483,6 +494,8 @@ fn summarize_recovered_table_families(
         summarize_nonzero_bty_header_name_lane(files, 0x5e, |probe| &probe.name_0x5e);
     let nonzero_bty_header_name_0x7c_summaries =
         summarize_nonzero_bty_header_name_lane(files, 0x7c, |probe| &probe.name_0x7c);
+    let bty_header_name_0x5e_dword_summaries =
+        summarize_bty_header_name_lane_by_dword(files, 0x5e, |probe| &probe.name_0x5e);
 
     BuildingTypeRecoveredTableSummary {
         recovered_style_themes: RECOVERED_STYLE_THEMES
@@ -500,6 +513,7 @@ fn summarize_recovered_table_families(
         nonzero_bty_header_name_0x40_summaries,
         nonzero_bty_header_name_0x5e_summaries,
         nonzero_bty_header_name_0x7c_summaries,
+        bty_header_name_0x5e_dword_summaries,
     }
 }
 
@@ -549,6 +563,52 @@ fn summarize_nonzero_bty_header_name_lane(
     summaries
 }
 
+fn summarize_bty_header_name_lane_by_dword(
+    files: &[BuildingTypeSourceFile],
+    offset: u32,
+    selector: impl Fn(&BuildingTypeBtyHeaderProbe) -> &String,
+) -> Vec<BuildingTypeBtyHeaderNameDwordSummary> {
+    let mut groups = BTreeMap::<(String, u32), Vec<String>>::new();
+    for file in files {
+        let Some(probe) = &file.bty_header_probe else {
+            continue;
+        };
+        let header_value = selector(probe).trim();
+        if header_value.is_empty() {
+            continue;
+        }
+        groups
+            .entry((header_value.to_string(), probe.dword_0xbb))
+            .or_default()
+            .push(file.file_name.clone());
+    }
+
+    let mut summaries = groups
+        .into_iter()
+        .map(|((header_value, dword_0xbb), mut file_names)| {
+            file_names.sort();
+            file_names.dedup();
+            BuildingTypeBtyHeaderNameDwordSummary {
+                header_offset_hex: format!("0x{offset:02x}"),
+                header_value,
+                dword_0xbb,
+                dword_0xbb_hex: format!("0x{dword_0xbb:08x}"),
+                file_count: file_names.len(),
+                sample_file_names: file_names.into_iter().take(24).collect(),
+            }
+        })
+        .collect::<Vec<_>>();
+    summaries.sort_by(|left, right| {
+        right
+            .file_count
+            .cmp(&left.file_count)
+            .then_with(|| left.dword_0xbb.cmp(&right.dword_0xbb))
+            .then_with(|| left.header_offset_hex.cmp(&right.header_offset_hex))
+            .then_with(|| left.header_value.cmp(&right.header_value))
+    });
+    summaries
+}
+
 #[derive(Debug, Clone, PartialEq, Eq, Deserialize)]
 struct BuildingBindingArtifact {
     bindings: Vec<BuildingBindingRow>,
@@ -731,5 +791,16 @@ mod tests {
                 sample_file_names: vec!["Port.bty".to_string()],
             }]
         );
+        assert_eq!(
+            summary.bty_header_name_0x5e_dword_summaries,
+            vec![BuildingTypeBtyHeaderNameDwordSummary {
+                header_offset_hex: "0x5e".to_string(),
+                header_value: "TextileMill".to_string(),
+                dword_0xbb: 0x01f4,
+                dword_0xbb_hex: "0x000001f4".to_string(),
+                file_count: 1,
+                sample_file_names: vec!["Port.bty".to_string()],
+            }]
+        );
     }
 }