Add author override, change crifx.toml format

crifx/cli.py

@@ -15,6 +15,7 @@
 
 
 def _dir_path_argparse_type(path):
+    """Check that the provided path is a directory."""
     if os.path.isdir(path):
         return path
     else:

crifx/config_parser.py

@@ -80,10 +80,13 @@ class LanguageGroupConfig:
     required_ac_count: int = 0
 
     @staticmethod
-    def from_toml_dict(
-        toml_dict: dict[str, Any], group_identifier: str
-    ) -> "LanguageGroupConfig":
+    def from_toml_dict(toml_dict: dict[str, Any]) -> "LanguageGroupConfig":
         """Initialize a LanguageGroupConfig from a toml dict."""
+        group_identifier = toml_dict.get("name")
+        if group_identifier is None:
+            raise ValueError(
+                "Language group in the `crifx.toml` file is missing a 'name'"
+            )
         language_names = toml_dict.get("languages", [])
         languages = []
         for language_name in language_names:
@@ -110,13 +113,16 @@ class AliasGroup:
     aliases: list[str]
 
     @staticmethod
-    def from_toml_dict(
-        toml_dict: dict[str, Any], group_identifier: str
-    ) -> "AliasGroup":
+    def from_toml_dict(toml_dict: dict[str, Any]) -> "AliasGroup":
         """Parse an AliasGroup from a toml dictionary."""
-        aliases = toml_dict.get("aliases", [])
+        primary_name = toml_dict.get("primary_name")
+        if primary_name is None:
+            raise ValueError(
+                "The `crifx.toml` file is missing a 'primary_name' for one or more 'judge' tables."
+            )
         git_name = toml_dict.get("git_name")
-        return AliasGroup(group_identifier, git_name, aliases)
+        aliases = toml_dict.get("aliases", [])
+        return AliasGroup(primary_name, git_name, aliases)
 
 
 class Config:
@@ -129,18 +135,18 @@ def __init__(self, toml_dict):
         )
         self.language_group_configs = []
         self.alias_groups = []
-        language_groups = toml_dict.get("language_groups", {})
-        for group_identifier, language_group_dict in language_groups.items():
+        language_groups = toml_dict.get("language_group", [])
+        for language_group_dict in language_groups:
             language_group_config = LanguageGroupConfig.from_toml_dict(
-                language_group_dict, group_identifier
+                language_group_dict,
             )
             if not language_group_config.language_group.languages:
                 # No languages parsed from the language group.
                 continue
             self.language_group_configs.append(language_group_config)
-        alias_groups = toml_dict.get("aliases", {})
-        for identifier, alias_group_dict in alias_groups.items():
-            alias_group = AliasGroup.from_toml_dict(alias_group_dict, identifier)
+        alias_groups = toml_dict.get("judge", [])
+        for alias_group_dict in alias_groups:
+            alias_group = AliasGroup.from_toml_dict(alias_group_dict)
             self.alias_groups.append(alias_group)
         for alias_group_1, alias_group_2 in itertools.product(
             self.alias_groups, self.alias_groups

crifx/problemset_parser.py

@@ -2,6 +2,7 @@
 
 import logging
 import os
+import re
 import tomllib
 from typing import Any
 
@@ -26,6 +27,7 @@
 
 TEST_CASE_IMAGE_EXTENSIONS = ["png", "jpg", "jpeg"]
 PROBLEM_REVIEW_STATUS_FILENAME = "crifx-problem-status.toml"
+CRIFX_AUTHOR_PATTERN = "crifx!\(author=([a-zA-Z0-9_ ]+)\)"
 
 
 class ProblemSetParser:
@@ -199,26 +201,45 @@ def _parse_submissions_dir(
             if language is None:
                 continue
             submission_path = os.path.join(submissions_dir, filename)
-            git_user_guess = self.git_manager.guess_file_author(submission_path)
-            filename_guess = self.guess_author_by_filename(filename)
-            if filename_guess is None:
-                judge = (
-                    self.judges_by_name.get(getattr(git_user_guess, "name"))
-                    or UNKNOWN_JUDGE
-                )
-            else:
-                judge = filename_guess
             lines_of_code = 0
             file_bytes = 0
+            author_name_override = None
             try:
                 with open(submission_path, "r") as submission_file:
+                    submission_lines = submission_file.readlines()
+                    for line_number, line in enumerate(submission_lines):
+                        author_match = re.search(CRIFX_AUTHOR_PATTERN, line)
+                        if author_match is not None:
+                            logging.debug(
+                                "Found author override for file %s on line %d",
+                                submission_path,
+                                line_number + 1,
+                            )
+                            author_name_override = author_match.group(1)
+                            break
                     lines_of_code = len(submission_file.readlines())
                 file_bytes = os.stat(submission_path).st_size
             except (FileExistsError, FileNotFoundError, PermissionError):
                 logging.warning(
                     "Could not determine size of submission at path '%s'",
                     submission_path,
                 )
+            git_user_guess = self.git_manager.guess_file_author(submission_path)
+            filename_guess = self.guess_author_by_filename(filename)
+            if author_name_override is not None:
+                judge = UNKNOWN_JUDGE
+                for candidate_judge in self.judges_by_name.values():
+                    if candidate_judge.has_alias(author_name_override):
+                        judge = candidate_judge
+                        break
+            elif filename_guess is not None:
+                judge = filename_guess
+
+            else:
+                judge = (
+                    self.judges_by_name.get(getattr(git_user_guess, "name"))
+                    or UNKNOWN_JUDGE
+                )
             submission = Submission(
                 judge,
                 filename,

examples/example_problemset/addtwonumbers/submissions/accepted/add_two.py → examples/example_problemset/addtwonumbers/submissions/accepted/add_two_jd.py

@@ -1,5 +1,11 @@
 #!/bin/python3
 
+"""
+This file uses 'jd' in the filename.
+
+This is an alias defined for Jane Doe in the crifx.toml file.
+"""
+
 from sys import stdin
 
 tokens = stdin.readline().split(" ")

examples/example_problemset/crifx.toml

@@ -22,21 +22,36 @@ validator_reviewers = 2
 # There must be at least 2 judges who have reviewed the test data.
 data_reviewers = 2
 
-# Section for defining language groups.
-[language_groups]
-# Non alphanumeric characters can be used in the language group name crifx
-# the language group name is written in double quotes "".
-[language_groups."c/c++"]
+# Define judge names in `[[judge]]` tables.
+[[judge]]
+# Primary name is required.
+primary_name = "Jane Doe"
+# git_name is not required.
+# git_name = "janedoe"
+# Aliases is a list of zero or more alternative names.
+aliases = ["jd", "jane"]
+
+[[judge]]
+primary_name = "Homer Simpson"
+git_name = "homers123"
+aliases = []
+
+# Tables for defining language groups.
+[[language_group]]
+# Non alphanumeric characters can be used in the language group name.
+name = "c/c++"
 # The names of the languages in the group. Language names are cast to
 # lowercase before comparison to language names known to crifx.
 languages = ["C", "C++"]
 # The number of AC submissions required for this group.
 required_ac_count = 0
 
-[language_groups."java/kotlin"]
+[[language_group]]
+name = "java/kotlin"
 languages = ["Java", "Kotlin"]
 required_ac_count = 0
 
-[language_groups.python]
+[[language_group]]
+name = "python"
 languages = ["Python"]
 required_ac_count = 0

examples/example_problemset/helloworld/submissions/accepted/hello.py

@@ -4,7 +4,7 @@
 
 # Use a crifx command to override the author for this submission instead
 # of relying on git blame information.
-# crifx!(author="Homer Simpson")
+# crifx!(author=Homer Simpson)
 
 from sys import stdin
 

tests/scenarios/sample_contest_1/crifx.toml

@@ -0,0 +1,3 @@
+[[judge]]
+primary_name = "Jane Doe"
+aliases = []

tests/scenarios/sample_contest_1/problem_a/submissions/accepted/hello_world.java

@@ -1,3 +1,9 @@
+/**
+ * Example solution with a name override.
+ *
+ * crifx!(author=Jane Doe)
+ */
+
 public class hello_world {
     public static void main(String[] args) {
         System.out.println("Hello world!");

tests/test_problemset_parsing.py

@@ -2,6 +2,8 @@
 
 import os
 
+from crifx.config_parser import parse_config
+from crifx.contest_objects import ProgrammingLanguage
 from crifx.git_manager import GitManager
 from crifx.problemset_parser import ProblemSetParser
 
@@ -10,7 +12,8 @@ def test_scenario_1(scenarios_path):
     """Test the basic scenario 1."""
     path = os.path.join(scenarios_path, "sample_contest_1")
     git_manager = GitManager(scenarios_path)
-    parser = ProblemSetParser(path, git_manager, [])
+    config = parse_config(path)
+    parser = ProblemSetParser(path, git_manager, config.alias_groups)
     problemset = parser.parse_problemset()
     problem_a = next(
         problem for problem in problemset.problems if problem.name == "problem_a"
@@ -32,3 +35,10 @@ def test_scenario_1(scenarios_path):
     assert len(problem_a.ac_submissions) == 2
     assert len(problem_a.wa_submissions) == 1
     assert len(problem_a.tle_submissions) == 0
+
+    problem_a_java_ac = next(
+        sub
+        for sub in problem_a.ac_submissions
+        if sub.language is ProgrammingLanguage.JAVA
+    )
+    assert problem_a_java_ac.author.primary_name == "Jane Doe"