Return empty dataframe when reading an empty table (#34)

* fix loading `.cbox` files

* add test and empty check directly on TextBuffer

---------

Co-authored-by: alisterburt <[email protected]>

starfile/parser.py

@@ -54,7 +54,8 @@ def _parse_data_block(self):
                 self._parse_loop_block()
                 return
 
-            elif line.startswith('data_') or self.crawler.current_line_number == self.n_lines:
+            elif line.startswith(
+                'data_') or self.crawler.current_line_number == self.n_lines:
                 self._parse_simple_block_from_buffer()
                 return
 
@@ -74,6 +75,8 @@ def _parse_loop_block(self):
         self.crawler.increment_line_number()
         header = self._parse_loop_header()
         df = self._parse_loop_data()
+        if df is None:
+            df = pd.DataFrame({h: None for h in header}, index=[0])
         df.columns = header
         df.name = self._current_data_block_name
         self._add_dataframe(df)
@@ -157,7 +160,7 @@ def _parse_loop_header(self) -> List[str]:
 
         return self.text_buffer.buffer
 
-    def _parse_loop_data(self) -> pd.DataFrame:
+    def _parse_loop_data(self) -> Union[pd.DataFrame, None]:
         self.text_buffer.clear()
 
         while self.crawler.current_line_number <= self.n_lines:
@@ -167,8 +170,16 @@ def _parse_loop_data(self) -> pd.DataFrame:
             self.text_buffer.add_line(current_line)
             self.crawler.increment_line_number()
 
-        df = pd.read_csv(StringIO(self.text_buffer.as_str()), delim_whitespace=True, header=None,
-                         comment='#')
+        # check whether the buffer is empty
+        if self.text_buffer.is_empty:
+            return None
+
+        df = pd.read_csv(
+            StringIO(self.text_buffer.as_str()),
+            delim_whitespace=True,
+            header=None,
+            comment='#'
+        )
         return df
 
     def dataframes_to_numeric(self):
@@ -204,7 +215,3 @@ def dataframe_at_index(self, idx: int):
 
     def dataframes_as_list(self):
         return list(self.dataframes.values())
-
-
-
-

starfile/utils.py

@@ -12,6 +12,15 @@ class TextBuffer:
     def __init__(self):
         self.buffer = deque()
 
+    @property
+    def is_empty(self) -> bool:
+        if len(self.buffer) == 0:
+            return True
+        elif len(self.buffer) <= 100:  # arbitrary, avoid iterating large buffer
+            return all([item.strip() == '' for item in self.buffer])
+        else:
+            return False
+
     def clear(self):
         self.buffer = deque()
 

tests/constants.py

@@ -20,6 +20,7 @@
 non_existant_file = test_data_directory / 'non_existant_file.star'
 two_single_line_loop_blocks = test_data_directory / 'two_single_line_loop_blocks.star'
 two_basic_blocks = test_data_directory / 'two_basic_blocks.star'
+empty_loop = test_data_directory / 'empty_loop.star'
 
 # Example DataFrame for testing
 cars = {'Brand': ['Honda_Civic', 'Toyota_Corolla', 'Ford_Focus', 'Audi_A4'],

tests/data/empty_loop.star

@@ -0,0 +1,6 @@
+
+data_
+
+loop_
+_rlnCoordinateX #1
+

tests/test_parsing.py

@@ -5,10 +5,23 @@
 import pytest
 
 from starfile.parser import StarParser
-from .constants import loop_simple, postprocess, pipeline, rln31_style, optimiser_2d, optimiser_3d, \
-    sampling_2d, \
-    sampling_3d, single_line_middle_of_multiblock, single_line_end_of_multiblock, non_existant_file, \
-    loop_simple_columns, two_single_line_loop_blocks, two_basic_blocks
+from .constants import (
+    loop_simple,
+    postprocess,
+    pipeline,
+    rln31_style,
+    optimiser_2d,
+    optimiser_3d,
+    sampling_2d,
+    sampling_3d,
+    single_line_middle_of_multiblock,
+    single_line_end_of_multiblock,
+    non_existant_file,
+    loop_simple_columns,
+    two_single_line_loop_blocks,
+    two_basic_blocks,
+    empty_loop,
+)
 from .utils import generate_large_star_file, remove_large_star_file, million_row_file
 
 
@@ -59,9 +72,10 @@ def test_read_multiblock_file():
         assert isinstance(df, pd.DataFrame)
 
     assert s.dataframes['general'].shape == (1, 6)
-    assert all(['rlnFinalResolution', 'rlnBfactorUsedForSharpening', 'rlnUnfilteredMapHalf1',
-                'rlnUnfilteredMapHalf2', 'rlnMaskName', 'rlnRandomiseFrom']
-               == s.dataframes['general'].columns)
+    assert all(
+        ['rlnFinalResolution', 'rlnBfactorUsedForSharpening', 'rlnUnfilteredMapHalf1',
+         'rlnUnfilteredMapHalf2', 'rlnMaskName', 'rlnRandomiseFrom']
+        == s.dataframes['general'].columns)
     assert s.dataframes['fsc'].shape == (49, 7)
     assert s.dataframes['guinier'].shape == (49, 3)
 
@@ -197,3 +211,9 @@ def test_two_basic_blocks():
     assert len(parser.dataframes) == 2
     for df in parser.dataframes.values():
         assert df.shape == (1, 3)
+
+
+def test_empty_loop_block():
+    """Parsing an empty loop block should return an empty dataframe."""
+    parser = StarParser(empty_loop)
+    assert len(parser.dataframes) == 1