fix(program): disambiguate repeated Python symbols

crates/compass-languages/src/program/python.rs

@@ -1,4 +1,4 @@
-use std::collections::BTreeMap;
+use std::collections::{BTreeMap, HashMap};
 use std::path::Path;
 
 use compass_ir::{
@@ -17,6 +17,7 @@ pub(super) fn extract(
     let mut collector = Collector {
         descriptor,
         input,
+        function_occurrences: HashMap::new(),
         functions: Vec::new(),
         evidence: Vec::new(),
     };
@@ -27,6 +28,7 @@ pub(super) fn extract(
 struct Collector<'a> {
     descriptor: ProviderDescriptor,
     input: &'a FileInput<'a>,
+    function_occurrences: HashMap<String, u32>,
     functions: Vec<FunctionIr>,
     evidence: Vec<compass_ir::EvidenceRecord>,
 }
@@ -78,7 +80,7 @@ impl Collector<'_> {
             |owner| format!("{owner}.{short_name}"),
         );
         let signature = signature_bytes(self.input.source, node);
-        let symbol_id = hex_sha256(
+        let base_symbol_id = hex_sha256(
             format!(
                 "{}\0{}\0{}",
                 self.input.source_file,
@@ -87,6 +89,7 @@ impl Collector<'_> {
             )
             .as_bytes(),
         );
+        let symbol_id = unique_symbol_id(base_symbol_id, &mut self.function_occurrences);
         let definition = evidence_record(
             &self.descriptor.id,
             Some(self.input.source_file),
@@ -455,6 +458,17 @@ fn contains_token(source: &[u8], expected: &str) -> bool {
         .any(|token| token == expected)
 }
 
+fn unique_symbol_id(base: String, occurrences: &mut HashMap<String, u32>) -> String {
+    let occurrence = occurrences.entry(base.clone()).or_default();
+    let symbol = if *occurrence == 0 {
+        base.clone()
+    } else {
+        hex_sha256(format!("{base}\0{occurrence}").as_bytes())
+    };
+    *occurrence = occurrence.saturating_add(1);
+    symbol
+}
+
 fn signature_bytes<'a>(source: &'a [u8], node: Node<'_>) -> &'a [u8] {
     let end = node
         .child_by_field_name("body")

crates/compass-languages/tests/program_evidence.rs

@@ -132,6 +132,30 @@ fn repeated_signatures_in_distinct_lexical_scopes_have_unique_syntax_symbols()
     Ok(())
 }
 
+#[test]
+fn python_conditional_redefinitions_have_unique_syntax_symbols() -> Result<(), Box<dyn Error>> {
+    let source = b"try:\n    def mogrify(sql, params, connection):\n        return first(sql)\nexcept ImportError:\n    def mogrify(sql, params, connection):\n        return second(sql)\n";
+    let batch = TreeSitterSyntaxProvider::default()
+        .analyze_file(FileInput {
+            source_file: "django/db/backends/postgresql/psycopg_any.py",
+            language: "python",
+            source,
+        })?
+        .ok_or("missing Python batch")?;
+    let functions = &batch.modules[0].functions;
+    assert_eq!(functions.len(), 2);
+    assert_eq!(
+        functions
+            .iter()
+            .map(|function| function.symbol_id.as_str())
+            .collect::<BTreeSet<_>>()
+            .len(),
+        functions.len()
+    );
+    merge_evidence(vec![batch])?.validate()?;
+    Ok(())
+}
+
 #[test]
 fn syntax_merge_resolves_unique_local_calls() -> Result<(), Box<dyn Error>> {
     let batch = TreeSitterSyntaxProvider::default()