Remove string-grouper

optimus/helpers/functions.py

@@ -283,78 +283,6 @@ def deep_sort(obj):
     return _sorted
 
 
-def infer_dataframes_keys(df_left: pd.DataFrame, df_right: pd.DataFrame):
-    """
-    Infer the possible key columns in two data frames
-    :param df_left:  
-    :param df_right: 
-    :return: 
-    """
-    from string_grouper import match_strings
-
-    result = []
-
-    df_left = df_left.dropna().astype(str)
-    df_right = df_right.dropna().astype(str)
-
-    # Search column names wiht *id* substring
-    def check_ids_columns(_df):
-        return [x for x in _df.columns if re.search(r"_id| id|id_| id ", x)]
-
-    ids_columns_left = check_ids_columns(df_left)
-    ids_columns_right = check_ids_columns(df_right)
-    if len(ids_columns_left) == len(ids_columns_right):
-        for i, j in zip(ids_columns_left, ids_columns_right):
-            result.append((i, j,))
-
-    # Numeric median len
-    def min_max_len(_df):
-
-        df_is_int = _df.applymap(lambda value: fastnumbers.isint(value)).sum()
-        df_is_int = df_is_int[df_is_int == len(_df)]
-        int_columns_names = df_is_int.index.values
-        int_columns_df = _df[int_columns_names]
-        string_len = int_columns_df.applymap(lambda value: len(value))
-        return (int_columns_names, string_len.min().values, string_len.max().values)
-
-    min_max_df_left = min_max_len(df_left)
-    min_max_df_right = min_max_len(df_right)
-
-    def median_len(arr, idx):
-        """
-        Calculate median len of the columns string
-        :param arr:
-        :param idx:
-        :return:
-        """
-        _min = arr[1][idx]
-        _max = arr[2][idx]
-        if _min != _max:
-            _median = _max - _min
-        else:
-            _median = _max
-        return _median
-
-    for i, col_l in enumerate(min_max_df_left[0]):
-        median_left = median_len(min_max_df_left, i)
-        for j, col_r in enumerate(min_max_df_right[0]):
-            median_right = median_len(min_max_df_right, j)
-            if median_left == median_right:
-                result.append((col_l, col_r,))
-
-    # String Clustering
-    for col_l in df_left:
-        for col_r in df_right:
-            try:
-                m = match_strings(df_left[col_l], df_right[col_r], min_similarity=0.05)
-                if len(m) > 0:
-                    result.append((col_l, col_r,))
-            except ValueError:
-                pass
-    # Count tuples
-    return [(count,) + item for item, count in Counter(result).items()]
-
-
 def update_dict(d, u):
     """
     Update only the given keys