DataFrame hash_rows function with detemine seed

* python polars 0.8.12

* DataFrame hash_rows function with detemine seed

* Fix docstring

* Switch to using black to fix formatting

Co-authored-by: Ritchie Vink <ritchie46@gmail.com>
Co-authored-by: thanhtm1 <thanhtm1@msb.com.vn>

polars/polars-core/src/frame/mod.rs

@@ -1644,9 +1644,9 @@ impl DataFrame {
 
     /// Hash and combine the row values
     #[cfg(feature = "row_hash")]
-    pub fn hash_rows(&self) -> Result<UInt64Chunked> {
+    pub fn hash_rows(&self, hasher_builder: Option<RandomState>) -> Result<UInt64Chunked> {
         let dfs = split_df(self, POOL.current_num_threads())?;
-        let (cas, _) = df_rows_to_hashes_threaded(&dfs, None);
+        let (cas, _) = df_rows_to_hashes_threaded(&dfs, hasher_builder);
 
         let mut iter = cas.into_iter();
         let mut acc_ca = iter.next().unwrap();

py-polars/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "py-polars"
-version = "0.8.11-beta.1"
+version = "0.8.12"
 authors = ["ritchie46 <ritchie46@gmail.com>"]
 edition = "2018"
 readme = "README.md"

py-polars/polars/eager/frame.py

@@ -2035,13 +2035,26 @@ def shrink_to_fit(self, in_place: bool = False) -> Optional["DataFrame"]:
             df._df.shrink_to_fit()
             return df
 
-    def hash_rows(self) -> "pl.Series":
+    def hash_rows(
+        self, k0: int = 0, k1: int = 1, k2: int = 2, k3: int = 3
+    ) -> "pl.Series":
         """
         Hash and combine the rows in this DataFrame.
 
         Hash value is UInt64
-        """
-        return pl.wrap_s(self._df.hash_rows())
+
+        Parameters
+        ----------
+        k0
+            seed parameter
+        k1
+            seed parameter
+        k2
+            seed parameter
+        k3
+            seed parameter
+        """
+        return pl.wrap_s(self._df.hash_rows(k0, k1, k2, k3))
 
 
 class GroupBy:

py-polars/polars/eager/series.py

@@ -1558,7 +1558,7 @@ def hash(self, k0: int = 0, k1: int = 1, k2: int = 2, k3: int = 3) -> "pl.Series
         """
         Hash the Series.
 
-        The hash value is of type `Date64`
+        The hash value is of type `UInt64`
 
         Parameters
         ----------

py-polars/src/dataframe.rs

@@ -886,8 +886,9 @@ impl PyDataFrame {
         self.df.shrink_to_fit();
     }
 
-    pub fn hash_rows(&self) -> PyResult<PySeries> {
-        let hash = self.df.hash_rows().map_err(PyPolarsEr::from)?;
+    pub fn hash_rows(&self, k0: u64, k1: u64, k2: u64, k3: u64) -> PyResult<PySeries> {
+        let hb = ahash::RandomState::with_seeds(k0, k1, k2, k3);
+        let hash = self.df.hash_rows(Some(hb)).map_err(PyPolarsEr::from)?;
         Ok(hash.into_series().into())
     }
 }