fix: correctly handle case when heads from cache not equal heads from works. (#243)

Author: Artanias

Makefile

@@ -1,4 +1,4 @@
-UTIL_VERSION            := 0.5.29
+UTIL_VERSION            := 0.5.30
 UTIL_NAME               := codeplag
 PWD                     := $(shell pwd)
 

src/codeplag/algorithms/featurebased.py

@@ -94,7 +94,7 @@ def get_children_indexes(
         count_of_nodes (int): count of elements in the tree.
 
     Complexity:
-        O(len(tree))
+        O(n)
     """
     indexes = []
     count_of_children = 0
@@ -118,9 +118,7 @@ def find_max_index(array: np.ndarray) -> np.ndarray:
         array (np.ndarray): matrix of compliance (np.ndarray object).
 
     Complexity:
-        rows = array.shape[0]
-        columns = array.shape[1]
-        O(rows * columns)
+        O(n^2)
 
     """
     maximum = 0

src/codeplag/handlers/check.py

@@ -320,6 +320,15 @@ def _do_step(
         metrics = None
         if self.reporter is not None:
             metrics = self.reporter.get_result(work1, work2)
+            if isinstance(metrics, FullCompareInfo) and (
+                metrics.first_heads != work1.head_nodes or metrics.second_heads != work2.head_nodes
+            ):
+                logger.warning(
+                    "Invalid data for the '%s' and '%s' found in cache.",
+                    work1.filepath,
+                    work2.filepath,
+                )
+                metrics = None
         if metrics is None:
             future = self._create_future_compare(executor, work1, work2)
             future.id = len(processing)  # type: ignore