consider .ans and .out for deduplication of testcases

bin/config.py

@@ -32,10 +32,13 @@
 FILE_NAME_REGEX = '[a-zA-Z0-9][a-zA-Z0-9_.-]*[a-zA-Z0-9]'
 COMPILED_FILE_NAME_REGEX = re.compile(FILE_NAME_REGEX)
 
-KNOWN_DATA_EXTENSIONS = [
+KNOWN_TESTCASE_EXTENSIONS = [
     '.in',
     '.ans',
     '.out',
+]
+
+KNOWN_DATA_EXTENSIONS = KNOWN_TESTCASE_EXTENSIONS + [
     '.interaction',
     '.hint',
     '.desc',
@@ -46,15 +49,19 @@
     '.gif',
 ]
 
-KNOWN_TEXT_DATA_EXTENSIONS = [
-    '.in',
-    '.ans',
-    '.out',
+KNOWN_TEXT_DATA_EXTENSIONS = KNOWN_TESTCASE_EXTENSIONS + [
     '.interaction',
     '.hint',
     '.desc',
 ]
 
+INVALID_CASE_DIRECTORIES = [
+    'invalid_inputs',
+    'invalid_answers',
+    'invalid_outputs',
+    'bad',
+]
+
 
 SEED_DEPENDENCY_RETRIES = 10
 

bin/generate.py

@@ -423,7 +423,7 @@ def __init__(self, problem, generator_config, key, name: str, yaml, parent, list
 
         # Hash of testcase for caching.
         self.hash = None
-        hashes = {}
+
         # Filled during generate(), since `self.config.solution` will only be set later for the default solution.
         self.cache_data = {}
 
@@ -438,6 +438,14 @@ def __init__(self, problem, generator_config, key, name: str, yaml, parent, list
         # root in /data
         self.root = self.path.parts[0]
 
+        # files to consider for hashing
+        hashes = {}
+        extensions = config.KNOWN_TESTCASE_EXTENSIONS.copy()
+        if self.root not in config.INVALID_CASE_DIRECTORIES[1:]:
+            extensions.remove('.ans')
+        if self.root not in config.INVALID_CASE_DIRECTORIES[2:]:
+            extensions.remove('.out')
+
         if yaml is None:
             self.inline = True
             yaml = dict()
@@ -491,7 +499,7 @@ def __init__(self, problem, generator_config, key, name: str, yaml, parent, list
                 if self.copy.is_file():
                     self.in_is_generated = False
                 self.rule['copy'] = str(self.copy)
-                for ext in ['.in', '.ans', '.out']:
+                for ext in extensions:
                     if self.copy.with_suffix(ext).is_file():
                         hashes[ext] = hash_file(self.copy.with_suffix(ext))
 
@@ -507,7 +515,7 @@ def __init__(self, problem, generator_config, key, name: str, yaml, parent, list
             if '.in' in self.hardcoded:
                 self.in_is_generated = False
                 self.rule['in'] = self.hardcoded['.in']
-            for ext in ['.in', '.ans', '.out']:
+            for ext in extensions:
                 if ext in self.hardcoded:
                     hashes[ext] = hash_string(self.hardcoded[ext])
 
@@ -523,17 +531,8 @@ def __init__(self, problem, generator_config, key, name: str, yaml, parent, list
             # An error is shown during generate.
             return
 
-        # remove filed that should not be considered for hashing
-        if '.ans' in hashes and self.root not in ['invalid_answers', 'invalid_outputs']:
-            hashes.pop('.ans')
-        if '.out' in hashes and self.root not in ['invalid_outputs']:
-            hashes.pop('.out')
-
         # build ordered list of hashes we want to consider
-        self.hash = []
-        for ext in ['.in', '.ans', '.out']:
-            if ext in hashes:
-                self.hash.append(hashes[ext])
+        self.hash = [hashes[ext] for ext in config.KNOWN_TESTCASE_EXTENSIONS if ext in hashes]
 
         # combine hashes
         if len(self.hash) == 1:
@@ -785,7 +784,29 @@ def move_generated():
 
         def add_testdata_to_cache():
             # Store the generated testdata for deduplication test cases.
-            test_hash = hash_file(target_infile)
+            hashes = {}
+
+            # remove files that should not be considered for this testcase
+            extensions = config.KNOWN_TESTCASE_EXTENSIONS.copy()
+            if t.root not in config.INVALID_CASE_DIRECTORIES[1:]:
+                extensions.remove('.ans')
+            if t.root not in config.INVALID_CASE_DIRECTORIES[2:]:
+                extensions.remove('.out')
+
+            for ext in extensions:
+                if target_infile.with_suffix(ext).is_file():
+                    hashes[ext] = hash_file(target_infile.with_suffix(ext))
+
+            # build ordered list of hashes we want to consider
+            test_hash = [hashes[ext] for ext in extensions if ext in hashes]
+
+            # combine hashes
+            if len(test_hash) == 1:
+                test_hash = test_hash[0]
+            else:
+                test_hash = combine_hashes(test_hash)
+
+            # check for duplicates
             if test_hash not in generator_config.generated_testdata:
                 generator_config.generated_testdata[test_hash] = t
             else:
@@ -823,7 +844,7 @@ def add_testdata_to_cache():
 
                 # Step 3: Write hardcoded files.
                 for ext, contents in t.hardcoded.items():
-                    if contents == '' and t.root in ['bad', 'invalid_inputs', 'invalid_answers', 'invalid_outputs']:
+                    if contents == '' and t.root in config.INVALID_CASE_DIRECTORIES:
                         bar.error(f'Hardcoded {ext} data must not be empty!')
                         return
                     else:
@@ -871,7 +892,7 @@ def add_testdata_to_cache():
 
             if not generator_up_to_date:
                 # Generate .ans and .interaction if needed.
-                if not config.args.no_solution and testcase.root not in ["invalid_inputs", "invalid_answers"]:
+                if not config.args.no_solution and testcase.root not in config.INVALID_CASE_DIRECTORIES:
                     if not problem.interactive:
                         # Generate a .ans if not already generated by earlier steps.
                         if not testcase.ans_path.is_file():
@@ -1316,7 +1337,7 @@ def parse(key, name, yaml, parent, listed=True):
             nonlocal testcase_id
             # Skip unlisted `data/bad` directory: we should not generate .ans files there.
             if (
-                name in ['bad', 'invalid_inputs', 'invalid_answers']
+                name in config.INVALID_CASE_DIRECTORIES
                 and parent.path == Path('.')
                 and listed is False
             ):
@@ -1799,7 +1820,7 @@ def clean_testcase(t):
             if (
                 not process_testcase(self.problem, t.path)
                 or t.listed
-                or (len(t.path.parts) > 0 and t.path.parts[0] in ['bad', 'invalid_inputs', 'invalid_answers'])
+                or (len(t.path.parts) > 0 and t.path.parts[0] in config.INVALID_CASE_DIRECTORIES)
             ):
                 bar.done()
                 return