fix pd.read_csv inferred type is Unknown but go to definition works. #1819

pyrefly/lib/module/bundled.rs

@@ -34,6 +34,17 @@ pub fn set_readonly(path: &Path, value: bool) -> anyhow::Result<()> {
     Ok(())
 }
 
+/// Prefer a package initializer when a bundle contains both `foo.pyi` and
+/// `foo/__init__.pyi`, matching Python's filesystem import precedence.
+pub(crate) fn bundled_module_path_is_preferred(candidate: &Path, existing: &Path) -> bool {
+    candidate
+        .file_name()
+        .is_some_and(|name| name == "__init__.pyi")
+        && existing
+            .file_name()
+            .is_none_or(|name| name != "__init__.pyi")
+}
+
 /// Creates a base config file for bundled stubs with common settings.
 ///
 /// This helper function encapsulates the common configuration logic shared across

pyrefly/lib/module/third_party.rs

@@ -21,6 +21,7 @@ use pyrefly_util::arc_id::ArcId;
 use starlark_map::small_map::SmallMap;
 
 use crate::module::bundled::BundledStub;
+use crate::module::bundled::bundled_module_path_is_preferred;
 use crate::module::bundled::create_bundled_stub_config;
 
 #[derive(Debug, Clone)]
@@ -60,7 +61,13 @@ impl BundledStub for BundledThirdParty {
         for (relative_path, contents) in contents {
             let adjusted_path = strip_stubs_suffix_from_path(&relative_path);
             let module_name = ModuleName::from_relative_path(&adjusted_path)?;
-            res.find.insert(module_name, relative_path.clone());
+            if res
+                .find
+                .get(&module_name)
+                .is_none_or(|existing| bundled_module_path_is_preferred(&relative_path, existing))
+            {
+                res.find.insert(module_name, relative_path.clone());
+            }
             res.load.insert(relative_path, Arc::new(contents));
         }
         Ok(res)
@@ -140,6 +147,16 @@ mod tests {
         }
     }
 
+    #[test]
+    fn test_bundled_third_party_prefers_package_over_module() {
+        let stubs = get_bundled_third_party().unwrap();
+        let path = stubs
+            .find
+            .get(&ModuleName::from_str("pandas.io.parsers"))
+            .expect("pandas.io.parsers should be bundled");
+        assert_eq!(path.file_name().unwrap(), "__init__.pyi");
+    }
+
     #[test]
     fn test_bundled_third_party_load_works() {
         let stub = get_bundled_third_party().unwrap();

pyrefly/lib/module/typeshed.rs

@@ -24,6 +24,7 @@ use starlark_map::small_map::SmallMap;
 
 use crate::config::config::ConfigFile;
 use crate::module::bundled::BundledStub;
+use crate::module::bundled::bundled_module_path_is_preferred;
 use crate::module::bundled::create_bundled_stub_config;
 
 #[derive(Debug, Clone)]
@@ -41,7 +42,13 @@ impl BundledStub for BundledTypeshedStdlib {
         };
         for (relative_path, contents) in contents {
             let module_name = ModuleName::from_relative_path(&relative_path)?;
-            res.find.insert(module_name, relative_path.clone());
+            if res
+                .find
+                .get(&module_name)
+                .is_none_or(|existing| bundled_module_path_is_preferred(&relative_path, existing))
+            {
+                res.find.insert(module_name, relative_path.clone());
+            }
             res.load.insert(relative_path, Arc::new(contents));
         }
         Ok(res)

pyrefly/lib/module/typeshed_third_party.rs

@@ -20,6 +20,7 @@ use pyrefly_util::arc_id::ArcId;
 use starlark_map::small_map::SmallMap;
 
 use crate::module::bundled::BundledStub;
+use crate::module::bundled::bundled_module_path_is_preferred;
 use crate::module::bundled::create_bundled_stub_config;
 
 #[derive(Debug, Clone)]
@@ -45,8 +46,12 @@ impl BundledStub for BundledTypeshedThirdParty {
                 .get(&relative_path)
                 .map(|s| ModuleName::from_str(s))
                 .unwrap_or_else(|| ModuleName::from_name(&module_name.first_component()));
-            res.find
-                .insert(module_name, (relative_path.clone(), package_name));
+            if res.find.get(&module_name).is_none_or(|(existing, _)| {
+                bundled_module_path_is_preferred(&relative_path, existing)
+            }) {
+                res.find
+                    .insert(module_name, (relative_path.clone(), package_name));
+            }
             res.load.insert(relative_path, Arc::new(contents));
         }
         Ok(res)

pyrefly/lib/test/lsp/lsp_interaction/hover.rs

@@ -115,6 +115,43 @@ fn hover_shows_third_party_function_name() {
     interaction.shutdown().unwrap();
 }
 
+#[test]
+fn hover_shows_bundled_pandas_function_type() {
+    let root = get_test_files_root();
+    let mut interaction = LspInteraction::new();
+    interaction.set_root(root.path().join("pandas_read_csv_hover"));
+    interaction
+        .initialize(InitializeSettings {
+            configuration: Some(None),
+            ..Default::default()
+        })
+        .unwrap();
+
+    interaction.client.did_open("test.py");
+    interaction
+        .client
+        .definition("test.py", 2, 5)
+        .expect_definition_response_from_root(
+            "site_packages/pandas/io/parsers/readers.py",
+            0,
+            4,
+            0,
+            12,
+        )
+        .unwrap();
+    interaction
+        .client
+        .hover("test.py", 2, 5)
+        .expect_hover_response_with_markup(|value| {
+            value.is_some_and(|text| {
+                text.contains("read_csv:") && !text.contains("read_csv: Unknown")
+            })
+        })
+        .unwrap();
+
+    interaction.shutdown().unwrap();
+}
+
 #[test]
 fn test_hover_import() {
     let root = get_test_files_root();

pyrefly/lib/test/lsp/lsp_interaction/test_files/pandas_read_csv_hover/pyrefly.toml

@@ -0,0 +1,2 @@
+search-path = ["."]
+site-package-path = ["site_packages"]

pyrefly/lib/test/lsp/lsp_interaction/test_files/pandas_read_csv_hover/site_packages/pandas/__init__.py

@@ -0,0 +1 @@
+from pandas.io.api import read_csv as read_csv

pyrefly/lib/test/lsp/lsp_interaction/test_files/pandas_read_csv_hover/site_packages/pandas/io/api.py

@@ -0,0 +1 @@
+from pandas.io.parsers.readers import read_csv as read_csv

pyrefly/lib/test/lsp/lsp_interaction/test_files/pandas_read_csv_hover/site_packages/pandas/io/parsers/readers.py

@@ -0,0 +1,2 @@
+def read_csv(path: str) -> object:
+    return object()

pyrefly/lib/test/lsp/lsp_interaction/test_files/pandas_read_csv_hover/test.py

@@ -0,0 +1,3 @@
+import pandas as pd
+
+pd.read_csv("data.csv")