update docstring for all writers

rltk/io/writer/block_array_writer.py

@@ -1,15 +1,24 @@
+import sys
+
 from rltk.io.writer import BlockWriter
 
 
 class BlockArrayWriter(BlockWriter):
-    def __init__(self, set_size=float('inf'), index_blacklist:set=None):
+    """
+    Block writer stores data in array (list).
+    
+    Args:
+        set_size (int, optional): Maximum size of a block. Defaults to `sys.maxsize`.
+        index_blacklist (set, optional): If a Record's id is in blacklist, it won't be in output. Default's to None.
+    """
+    def __init__(self, set_size: int = sys.maxsize, index_blacklist: set = None):
         self._array = list()
         self._dict = dict()
         self._is_dirty = False
         self._set_size = set_size
         self._blacklist = index_blacklist or set()
 
-    def write(self, id1, id2):
+    def write(self, id1: str, id2: str):
         if id1 in self._blacklist:
             return
 

rltk/io/writer/block_file_writer.py

@@ -6,7 +6,17 @@
 
 
 class BlockFileWriter(BlockWriter):
-    def __init__(self, filename, buffer_size=10000, set_size=sys.maxsize, index_blacklist:set=None):
+    """
+    Block writer stores data in file.
+    
+    Args:
+        filename (str): File name.
+        buffer_size (int, optional): Buffer size. Defaults to 10,000.
+        set_size (int, optional): Maximum size of a block. Defaults to `sys.maxsize`.
+        index_blacklist (set, optional): If a Record's id is in blacklist, it won't be in output. Default's to None.
+    """
+    def __init__(self, filename: str, buffer_size: int = 10000,
+                 set_size: int = sys.maxsize, index_blacklist: set = None):
         self._filename = filename
         self._temp_filename = filename + '.temp'
         self._buffer_size = buffer_size
@@ -18,7 +28,7 @@ def __init__(self, filename, buffer_size=10000, set_size=sys.maxsize, index_blac
         # clean up output file
         open(self._filename, 'w').close()
 
-    def write(self, id1, id2):
+    def write(self, id1: str, id2: str):
         # skip if id1 is in blacklist
         if id1 in self._blacklist:
             return

rltk/io/writer/block_writer.py

@@ -3,22 +3,40 @@
 
 class BlockWriter(Writer):
     """
-    index: sets
-    black list is on index
+    Super class of block writer
     """
 
-    def write(self, id1, id2):
+    def write(self, id1: str, id2: str):
+        """
+        Args:
+            id1 (str): Record 1 id.
+            id2 (str): Record 2 id.
+        """
         raise NotImplementedError
 
     def get_handler(self):
+        """
+        The handler which can be used as input of corresponding :meth:`BlockReader`.
+        """
         raise NotImplementedError
 
     def flush(self):
+        """
+        Force to flush buffer.
+        """
         raise NotImplementedError
 
     def get_blacklist(self):
-        return self._blacklist
+        """
+        Black list of indices. 
+        """
+        if getattr(self, '_blacklist'):
+            return self._blacklist
+        raise NotImplementedError
 
     def close(self):
+        """
+        Close handler.
+        """
         self.flush()
         super().close()

rltk/io/writer/ground_truth_writer.py

@@ -1,19 +1,34 @@
 import csv
 
 from rltk.io.writer import Writer
+from rltk.io.io_utils import get_file_handler
 
 
 class GroundTruthWriter(Writer):
+    """
+    Ground truth writer.
+    
+    Args:
+        file_handler (io.IOBase): It can be file name or file handler.
+    """
     def __init__(self, file_handler):
-        self._file_handler = Writer.get_file_handler(file_handler)
+        self._file_handler = get_file_handler(file_handler)
         fieldnames = ['id1', 'id2', 'label']
         self._csv_writer = csv.DictWriter(self._file_handler, fieldnames=fieldnames)
         self._csv_writer.writeheader()
 
-    def write(self, id1 : str, id2 : str, label : int):
+    def write(self, id1: str, id2: str, label: bool):
+        """
+        Writer to file.
+        
+        Args:
+            id1 (str): Id 1.
+            id2 (str): Id 2.
+            label (bool): Positive (True) or negative (False).
+        """
         self._csv_writer.writerow({'id1': id1, 'id2': id2, 'label': label})
 
-    def __del__(self):
+    def close(self):
         try:
             self._file_handler.close()
         except:

rltk/io/writer/writer.py

@@ -2,21 +2,26 @@
 
 
 class Writer(object):
+    """
+    Writer.
+    """
     def __init__(self):
         pass
 
     def write(self):
+        """
+        Write content.
+        """
         raise NotImplementedError
 
     def __del__(self):
-        pass
+        """
+        Same to :meth:`close`.
+        """
+        self.close()
 
     def close(self):
-        self.__del__()
-
-    @staticmethod
-    def get_file_handler(f):
-        if isinstance(f, io.IOBase):
-            return f
-
-        return open(f, 'w')
+        """
+        Close handler.
+        """
+        pass