No format analysis for ongoing ABC contests (#286)

* Enable the functionality to skip input format analysis for ongoing ABC contests

* Fix format

* Fix unit tests

Author: kyuridenamida

atcodertools/atcoder_tools.py

@@ -43,6 +43,8 @@ def main():
     if len(sys.argv) < 2 or sys.argv[1] not in ("gen", "test", "submit", "codegen", "compile", "set", "version"):
         print("Usage:")
         print("{} gen -- to generate workspace".format(sys.argv[0]))
+        print(
+            "{} codegen -- to generate code for one problem".format(sys.argv[0]))
         print(
             "{} compile -- to compile codes in your workspace".format(sys.argv[0]))
         print("{} test -- to test codes in your workspace".format(sys.argv[0]))

atcodertools/common/language.py

@@ -123,7 +123,8 @@ def from_name(cls, name: str):
     name="python",
     display_name="Python",
     extension="py",
-    submission_lang_pattern=re.compile(".*Python \\(3.*|.*Python \\(CPython 3.*"),
+    submission_lang_pattern=re.compile(
+        ".*Python \\(3.*|.*Python \\(CPython 3.*"),
     default_code_generator=python.main,
     default_template_path=get_default_template_path('py'),
     compile_command="python3 -mpy_compile {filename}.py",

atcodertools/constprediction/constants_prediction.py

@@ -36,6 +36,11 @@ def __init__(self, cands):
         self.cands = cands
 
 
+class FailingToKnowFormatAnalysisAllowedByRuleError(Exception):
+    def __init__(self, reason):
+        self.reason = reason
+
+
 MOD_ANCHORS = ["余り", "あまり", "mod", "割っ", "modulo"]
 DECIMAL_ANCHORS = ["誤差", " error "]
 LIMIT_ANCHORS = ["時間制限", "Time Limit"]
@@ -206,6 +211,26 @@ def normalize(sentence):
     raise MultipleModCandidatesError(limit_cands)
 
 
+def predict_is_format_analysis_allowed_by_rule(html: str) -> bool:
+    soup = BeautifulSoup(html, "html.parser")
+    url_meta_tag = soup.find("meta", property="og:url")
+    if url_meta_tag is None:
+        raise FailingToKnowFormatAnalysisAllowedByRuleError(
+            "No meta tag detected, which is important as a clue to know if the input analysis is allowed.")
+
+    # ABC is the target of "No LLM rules" by AtCoder (See https://info.atcoder.jp/entry/llm-abc-rules-ja or https://info.atcoder.jp/entry/llm-abc-rules-en)
+    is_target_contest_of_no_llm_rule = "/contests/abc" in url_meta_tag["content"]
+
+    # If there is no virtual standings link, assume it's ongoing.
+    is_ongoing_contest = len([tag for tag in soup.find_all(
+        "a") if "/standings/virtual" in tag.get("href", "")]) == 0
+
+    if is_target_contest_of_no_llm_rule and is_ongoing_contest:
+        return False
+
+    return True
+
+
 def predict_constants(html: str) -> ProblemConstantSet:
     try:
         yes_str, no_str = predict_yes_no(html)
@@ -233,4 +258,19 @@ def predict_constants(html: str) -> ProblemConstantSet:
                        "two or more candidates {} are detected as limit".format(e.cands))
         timeout = None
 
-    return ProblemConstantSet(mod=mod, yes_str=yes_str, no_str=no_str, judge_method=judge, timeout=timeout)
+    try:
+        is_format_analysis_allowed_by_rule = predict_is_format_analysis_allowed_by_rule(
+            html)
+    except FailingToKnowFormatAnalysisAllowedByRuleError as e:
+        logger.warning(
+            "Failed to know whether format analysis is allowed by the contest rules -- ", e.reason)
+        is_format_analysis_allowed_by_rule = None
+
+    return ProblemConstantSet(
+        mod=mod,
+        yes_str=yes_str,
+        no_str=no_str,
+        judge_method=judge,
+        timeout=timeout,
+        is_format_analysis_allowed_by_rule=is_format_analysis_allowed_by_rule,
+    )

atcodertools/constprediction/models/problem_constant_set.py

@@ -8,10 +8,12 @@ def __init__(self,
                  yes_str: str = None,
                  no_str: str = None,
                  judge_method: Judge = None,
-                 timeout: float = None
+                 timeout: float = None,
+                 is_format_analysis_allowed_by_rule: bool = None,
                  ):
         self.mod = mod
         self.yes_str = yes_str
         self.no_str = no_str
         self.judge_method = judge_method
         self.timeout = timeout
+        self.is_format_analysis_allowed_by_rule = is_format_analysis_allowed_by_rule

atcodertools/fmtprediction/predict_format.py

@@ -16,7 +16,14 @@ def __init__(self, cands):
         self.cands = cands
 
 
-def predict_format(content: ProblemContent) -> FormatPredictionResult:
+class PredictionNotAllowedError(Exception):
+    pass
+
+
+def predict_format(content: ProblemContent, is_format_analysis_allowed_by_rule: bool) -> FormatPredictionResult:
+    if not is_format_analysis_allowed_by_rule:
+        raise PredictionNotAllowedError
+
     input_format = content.get_input_format()
     samples = content.get_samples()
 

atcodertools/tools/codegen.py

@@ -19,7 +19,7 @@
 from atcodertools.config.config import Config
 from atcodertools.constprediction.constants_prediction import predict_constants
 from atcodertools.fmtprediction.models.format_prediction_result import FormatPredictionResult
-from atcodertools.fmtprediction.predict_format import MultiplePredictionResultsError, NoPredictionResultError, predict_format
+from atcodertools.fmtprediction.predict_format import MultiplePredictionResultsError, NoPredictionResultError, PredictionNotAllowedError, predict_format
 from atcodertools.tools import get_default_config_path
 from atcodertools.tools.envgen import USER_CONFIG_PATH, get_config, output_splitter
 from atcodertools.tools.utils import with_color
@@ -89,19 +89,23 @@ def emit_info(text):
         emit_error("Failed to download samples.")
         raise e
 
+    constants = predict_constants(content.original_html)
+
     try:
-        prediction_result = predict_format(content)
+        prediction_result = predict_format(
+            content, constants.is_format_analysis_allowed_by_rule or False)
         emit_info(
             with_color("Format prediction succeeded", Fore.LIGHTGREEN_EX))
-    except (NoPredictionResultError, MultiplePredictionResultsError) as e:
+    except (NoPredictionResultError, MultiplePredictionResultsError, PredictionNotAllowedError) as e:
         prediction_result = FormatPredictionResult.empty_result()
         if isinstance(e, NoPredictionResultError):
             msg = "No prediction -- Failed to understand the input format"
-        else:
+        elif isinstance(e, MultiplePredictionResultsError):
             msg = "Too many prediction -- Failed to understand the input format"
+        elif isinstance(e, PredictionNotAllowedError):
+            msg = "Format prediction is skipped because it's not allowed by AtCoder rules (At least not allowed in ongoing ABC contests) -- See https://info.atcoder.jp/entry/llm-abc-rules-en"
         emit_warning(with_color(msg, Fore.LIGHTRED_EX))
 
-    constants = predict_constants(content.original_html)
     code_generator = config.code_style_config.code_generator
     with open(template_code_path, "r") as f:
         template = f.read()

atcodertools/tools/envgen.py

@@ -24,7 +24,7 @@
     create_code
 from atcodertools.fmtprediction.models.format_prediction_result import FormatPredictionResult
 from atcodertools.fmtprediction.predict_format import NoPredictionResultError, \
-    MultiplePredictionResultsError, predict_format
+    MultiplePredictionResultsError, PredictionNotAllowedError, predict_format
 from atcodertools.tools import get_default_config_path
 from atcodertools.tools.models.metadata import Metadata
 from atcodertools.tools.utils import with_color
@@ -117,19 +117,23 @@ def emit_info(text):
                 code_file_path,
                 new_path))
 
+    constants = predict_constants(content.original_html)
+
     try:
-        prediction_result = predict_format(content)
+        prediction_result = predict_format(
+            content, constants.is_format_analysis_allowed_by_rule or False)
         emit_info(
             with_color("Format prediction succeeded", Fore.LIGHTGREEN_EX))
-    except (NoPredictionResultError, MultiplePredictionResultsError) as e:
+    except (NoPredictionResultError, MultiplePredictionResultsError, PredictionNotAllowedError) as e:
         prediction_result = FormatPredictionResult.empty_result()
         if isinstance(e, NoPredictionResultError):
             msg = "No prediction -- Failed to understand the input format"
-        else:
+        elif isinstance(e, MultiplePredictionResultsError):
             msg = "Too many prediction -- Failed to understand the input format"
+        elif isinstance(e, PredictionNotAllowedError):
+            msg = "Format prediction is skipped because it's not allowed by AtCoder rules (At least not allowed in ongoing ABC contests) -- See https://info.atcoder.jp/entry/llm-abc-rules-en"
         emit_warning(with_color(msg, Fore.LIGHTRED_EX))
 
-    constants = predict_constants(content.original_html)
     code_generator = config.code_style_config.code_generator
     with open(template_code_path, "r") as f:
         template = f.read()