code-coverage: handle functions in __init__.py files (#1005)

* code-coverage: handle functions in __init__.py files

Fixes: ossf/fuzz-introspector#1003

Signed-off-by: David Korczynski <[email protected]>

* utils: fix link-handling of functions declared in __init__.py files

Signed-off-by: David Korczynski <[email protected]>

* fix formatting

Signed-off-by: David Korczynski <[email protected]>

* formatting nit

Signed-off-by: David Korczynski <[email protected]>

---------

Signed-off-by: David Korczynski <[email protected]>

src/fuzz_introspector/code_coverage.py

@@ -180,9 +180,11 @@ def _python_ast_funcname_to_cov_file(self, function_name) -> Optional[str]:
         # Resolve name if required. This is needed to normalise filenames.
         logger.debug("Resolving name")
         potential_paths = []
+        init_paths = []
         current_path = ""
         for module_name in function_name.split("."):
             current_path += module_name
+            init_paths.append(current_path + "/__init__.py")
             potential_paths.append(current_path + ".py")
             current_path += "/"
 
@@ -193,6 +195,24 @@ def _python_ast_funcname_to_cov_file(self, function_name) -> Optional[str]:
                 if potential_key.endswith(potential_path):
                     logger.debug(f"Found key: {str(potential_key)}")
                     return potential_key
+        # We found no matches when filenames exclude __init__.py. Try to
+        # include these now.
+        init_matches = []
+        # Iterate based in init paths since we want to have the longest match
+        # at the end of __init__matches list.
+        logger.info("Scanning for init paths")
+        for potential_init_path in init_paths:
+            logger.info("Trying %s" % (potential_init_path))
+            for potential_key in self.file_map:
+                logger.debug(f"Scanning {str(potential_key)}")
+                if potential_key.endswith(potential_init_path):
+                    logger.debug(f"Found __init__ match: {str(potential_key)}")
+                    init_matches.append(potential_key)
+
+        # Return the last match, as this signals the path with most precise
+        # matching.
+        if len(init_matches) > 0:
+            return init_matches[-1]
 
         # If this is reached then no match was found. Return None.
         logger.debug("Could not find key")
@@ -639,7 +659,6 @@ def load_python_json_coverage(json_file: str,
             'executed_lines']
         cp.dual_file_map[cov_entry]['missing_lines'] = data['files'][entry][
             'missing_lines']
-
     return cp
 
 

src/fuzz_introspector/utils.py

@@ -200,11 +200,13 @@ def approximate_python_coverage_files(src1: str, src2: str) -> bool:
 
     # Generate list of potential candidates
     possible_candidates = []
+    possible_init_candidates = []
     splits = src1.split(".")
     curr_str = ""
     for s2 in splits:
         curr_str = curr_str + s2
         possible_candidates.append(curr_str + ".py")
+        possible_init_candidates.append("/__init__.py")
         curr_str = curr_str + "/"
 
     # Start from backwards to find te longest possible candidate
@@ -218,6 +220,16 @@ def approximate_python_coverage_files(src1: str, src2: str) -> bool:
             target = candidate
             break
 
+    if target is None:
+        for init_candidate in reversed(possible_init_candidates):
+            if src2.endswith(init_candidate):
+                # ensure the entire filename is matched in the event of not slashes
+                if "/" not in init_candidate:
+                    if not src2.split("/")[-1] == init_candidate:
+                        continue
+                target = init_candidate
+                break
+
     if target is not None:
         logger.debug(f"Found target {target}")
         return True