Read the column names from the first existing step 3 result file

ledsa/core/file_handling.py

@@ -1,4 +1,5 @@
 import os
+import re
 from typing import List, Union
 
 import numpy as np
@@ -304,6 +305,32 @@ def create_binary_data(channel: int, path: str = '.') -> None:
 LEGACY_FIT_COLUMNS = BASE_COLUMNS + LED_CENTER_COLUMNS + FIT_COLUMNS[:-1]
 
 
+def _find_first_result_file(channel: int) -> str:
+    """
+    Find the step 3 result file with the lowest image ID of a channel.
+
+    An image that could not be read has no result file at all, so the lowest analysed image ID is not
+    necessarily 1 and the result file of image 1 does not have to exist.
+
+    :param channel: Channel number whose result files are searched.
+    :type channel: int
+    :return: Path of the result file with the lowest image ID.
+    :rtype: str
+    :raises FileNotFoundError: If the channel has no result file.
+    """
+    dir_path = os.path.join('analysis', f'channel{channel}')
+    img_ids = []
+    if os.path.isdir(dir_path):
+        for file_name in os.listdir(dir_path):
+            img = re.search(r"(\d+)_led_positions.csv", file_name)
+            if img is not None:
+                img_ids.append(int(img.group(1)))
+    if not img_ids:
+        raise FileNotFoundError(f'No analysis results in {dir_path}. Run the data extraction (-s3 or '
+                                f'-s3_fast) before the analysis.')
+    return os.path.join(dir_path, f'{min(img_ids)}_led_positions.csv')
+
+
 def _get_column_names(channel: int) -> List[str]:
     """
     Get the column names for the specified channel from the number of columns of the CSV files.
@@ -313,8 +340,9 @@ def _get_column_names(channel: int) -> List[str]:
     :return: List of column names.
     :rtype: List[str]
     :raises ValueError: If the file cannot be parsed or the number of columns matches no known layout.
+    :raises FileNotFoundError: If the channel has no result file.
     """
-    file_path = os.path.join('analysis', f'channel{channel}', '1_led_positions.csv')
+    file_path = _find_first_result_file(channel)
     try:
         parameters = ledsa.core.file_handling.read_table(file_path, delim=',', silent=True, atleast_2d=True)
     except ValueError as error:

ledsa/tests/AcceptanceTests/07_test_corrupted_images.robot

@@ -40,6 +40,19 @@ Analysis Skips Images Without Extraction Results
     # mismatch of the time axes and must not be reported as one
     Extco Time Axis Should Be Accepted
 
+Analysis Reads The Column Names From The First Existing Result File
+    [Documentation]     The column names of the analysis are read from one result file. An image that could
+    ...                 not be read has none at all, so an unreadable first image must not keep the analysis
+    ...                 from reading the names from the first file that does exist.
+    Corrupt Image   test_img_1.jpg  truncate
+    Execute Ledsa   -s3_fast
+    Corrupted Images Report Should List     test_img_1.jpg  extraction of the LED intensities
+    # image 1 is the default reference image and has no results here, so image 2 provides the reference
+    Create And Fill Config Analysis     linear      num_ref_images=2
+    Execute Ledsa   --analysis
+    Analysis Should Exclude Image   test_img_1.jpg
+    Extco Time Axis Should Be Accepted
+
 *** Keywords ***
 Restore Test Images
     Restore Corrupted Images

ledsa/tests/AcceptanceTests/LedsaATestLibrary.py

@@ -311,10 +311,11 @@ def create_and_fill_config(self, first=1, last=4):
         conf.save()
 
     @keyword
-    def create_and_fill_config_analysis(self, solver, reference_property='sum_col_val'):
+    def create_and_fill_config_analysis(self, solver, reference_property='sum_col_val', num_ref_images=1):
+        """Write config_analysis.ini. num_ref_images lets a test take the reference from another image than the first."""
         conf = ConfigDataAnalysis(load_config_file=False, camera_position=None, num_layers=20, domain_bounds=None,
-                                  led_array_indices=0, num_ref_images=1, camera_channels=0, num_cores=1,
-                                  reference_property=reference_property,
+                                  led_array_indices=0, num_ref_images=int(num_ref_images), camera_channels=0,
+                                  num_cores=1, reference_property=reference_property,
                                   average_images=False, solver=solver, weighting_preference=-6e-4,
                                   weighting_curvature=1e-7,
                                   num_iterations=2000, lambda_reg=1e-3)

ledsa/tests/UnitTests/test_file_handling.py

@@ -0,0 +1,56 @@
+import os
+
+import pytest
+
+from ledsa.core.file_handling import _find_first_result_file, _get_column_names
+
+
+class TestFindFirstResultFile:
+
+    def _write_result_files(self, tmp_path, img_ids):
+        channel_dir = tmp_path / 'analysis' / 'channel0'
+        channel_dir.mkdir(parents=True)
+        for img_id in img_ids:
+            (channel_dir / f'{img_id}_led_positions.csv').write_text('# header\n', encoding='utf-8')
+
+    def test_lowest_image_id_wins(self, tmp_path, monkeypatch):
+        # the IDs are compared as numbers, so 2 comes before 10
+        self._write_result_files(tmp_path, [10, 2, 3])
+        monkeypatch.chdir(tmp_path)
+        assert _find_first_result_file(0) == os.path.join('analysis', 'channel0', '2_led_positions.csv')
+
+    def test_image_without_results_is_skipped(self, tmp_path, monkeypatch):
+        # image 1 could not be read, so it has no result file at all
+        self._write_result_files(tmp_path, [2, 4])
+        monkeypatch.chdir(tmp_path)
+        assert _find_first_result_file(0) == os.path.join('analysis', 'channel0', '2_led_positions.csv')
+
+    def test_other_files_are_ignored(self, tmp_path, monkeypatch):
+        self._write_result_files(tmp_path, [3])
+        (tmp_path / 'analysis' / 'channel0' / 'all_parameters.h5').write_text('', encoding='utf-8')
+        monkeypatch.chdir(tmp_path)
+        assert _find_first_result_file(0) == os.path.join('analysis', 'channel0', '3_led_positions.csv')
+
+    def test_channel_without_results_raises(self, tmp_path, monkeypatch):
+        self._write_result_files(tmp_path, [])
+        monkeypatch.chdir(tmp_path)
+        with pytest.raises(FileNotFoundError, match='Run the data extraction'):
+            _find_first_result_file(0)
+
+    def test_missing_channel_directory_raises(self, tmp_path, monkeypatch):
+        monkeypatch.chdir(tmp_path)
+        with pytest.raises(FileNotFoundError, match='No analysis results'):
+            _find_first_result_file(0)
+
+
+class TestColumnNamesWithoutTheFirstImage:
+
+    def test_names_are_read_from_the_first_existing_file(self, tmp_path, monkeypatch):
+        # the reported case: image 1 could not be read, so the analysis has to fall back to image 2
+        channel_dir = tmp_path / 'analysis' / 'channel0'
+        channel_dir.mkdir(parents=True)
+        row = ','.join(str(float(i)) for i in range(6))
+        (channel_dir / '2_led_positions.csv').write_text(f'# header\n{row}\n{row}\n', encoding='utf-8')
+        monkeypatch.chdir(tmp_path)
+        assert _get_column_names(0) == ["img_id", "led_id", "led_array_id", "sum_col_val", "mean_col_val",
+                                        "max_col_val", "bgsub_sum_col_val", "width", "height"]