Fix executable caching bug

- Make index non-unique to resolve FileExists errors of gridFs.
- Rename binary_cache to executable_cache for consistency.
- Ruff changed double quotes to single quotes in some files.
- Remove REPORT_LEAK command.
- Fix in simulation.

Author: GJFR

bughog/configuration.py

@@ -53,17 +53,17 @@ def get_database_params() -> DatabaseParameters:
 
         required_database_params = ['BCI_MONGO_HOST', 'BCI_MONGO_USERNAME', 'BCI_MONGO_DATABASE', 'BCI_MONGO_PASSWORD']
         missing_database_params = [param for param in required_database_params if os.getenv(param) in ['', None]]
-        binary_cache_limit = int(os.getenv('BCI_BINARY_CACHE_LIMIT', 0))
+        executable_cache_limit = int(os.getenv('BCI_EXECUTABLE_CACHE_LIMIT', 0))
         if missing_database_params:
             logger.info(f'Could not find database parameters {missing_database_params}, using database container...')
-            Global.__database_params = container.run(binary_cache_limit)
+            Global.__database_params = container.run(executable_cache_limit)
         else:
             Global.__database_params = DatabaseParameters(
                 os.getenv('BCI_MONGO_HOST'),
                 os.getenv('BCI_MONGO_USERNAME'),
                 os.getenv('BCI_MONGO_PASSWORD'),
                 os.getenv('BCI_MONGO_DATABASE'),
-                binary_cache_limit,
+                executable_cache_limit,
             )
             logger.info(f"Found database environment variables '{Global.__database_params}'")
         return Global.__database_params

bughog/database/mongo/container.py

@@ -20,7 +20,7 @@
 DEFAULT_HOST = 'bh_db'
 
 
-def run(binary_cache_limit: int) -> DatabaseParameters:
+def run(executable_cache_limit: int) -> DatabaseParameters:
     docker_client = docker.from_env()
     try:
         mongo_container = docker_client.containers.get(DEFAULT_HOST)
@@ -33,7 +33,7 @@ def run(binary_cache_limit: int) -> DatabaseParameters:
         LOGGER.debug('MongoDB container not found, creating a new one...')
         __create_new_container(DEFAULT_USER, DEFAULT_PW, DEFAULT_DB_NAME, DEFAULT_HOST)
     LOGGER.debug('MongoDB container has started!')
-    return DatabaseParameters(DEFAULT_HOST, DEFAULT_USER, DEFAULT_PW, DEFAULT_DB_NAME, binary_cache_limit)
+    return DatabaseParameters(DEFAULT_HOST, DEFAULT_USER, DEFAULT_PW, DEFAULT_DB_NAME, executable_cache_limit)
 
 
 def stop():

bughog/database/mongo/executable_cache.py

@@ -129,16 +129,12 @@ def store_file(file_path: str) -> None:
                     ),
                     exc_info=futures_with_exception[0].exception(),
                 )
-                ExecutableCache.__remove_commit_executable_files(state_name)
+                ExecutableCache.__remove_commit_executable_files(subject_config.subject_type, subject_config.subject_name, state_name)
                 logger.debug(f'Removed possibly incomplete cached executable files for {state_name}.')
             else:
                 elapsed_time = time.time() - start_time
                 logger.debug(f'Stored executable in {elapsed_time:.2f}s')
 
-    @staticmethod
-    def remove_executable_files(state_name) -> None:
-        ExecutableCache.__remove_commit_executable_files(state_name)
-
     @staticmethod
     def __count_cached_executables(state_type: Optional[str] = None) -> int:
         """
@@ -167,16 +163,24 @@ def __remove_least_used_commit_executable_files() -> None:
         )
         for state_doc in grid_cursor:
             state_name = state_doc['state_name']
-            ExecutableCache.__remove_commit_executable_files(state_name)
+            subject_type = state_doc['subject_type']
+            subject_name = state_doc['subject_name']
+            ExecutableCache.__remove_commit_executable_files(subject_type, subject_name, state_name)
             break
 
     @staticmethod
-    def __remove_commit_executable_files(state_name: str) -> None:
+    def __remove_commit_executable_files(subject_type: str, subject_name: str, state_name: str) -> None:
         """
         Removes the executable files associated with the parameters.
         """
         fs = MongoDB().gridfs
         files_collection = MongoDB().get_collection('fs.files')
 
-        for grid_doc in files_collection.find({'state_name': state_name}):
+        query = {
+            'file_type': 'executable',
+            'state_name': state_name,
+            'subject_type': subject_type,
+            'subject_name': subject_name,
+        }
+        for grid_doc in files_collection.find(query):
             fs.delete(grid_doc['_id'])

bughog/database/mongo/mongodb.py

@@ -22,7 +22,7 @@ class EvaluationParameters:
 
     def serialize(self) -> str:
         pickled_bytes = pickle.dumps(self, pickle.HIGHEST_PROTOCOL)
-        return base64.b64encode(pickled_bytes).decode("ascii")
+        return base64.b64encode(pickled_bytes).decode('ascii')
 
     @staticmethod
     def deserialize(pickled_str: str) -> EvaluationParameters:
@@ -53,7 +53,7 @@ def to_dict(self) -> dict:
 
     @staticmethod
     def from_dict(data: dict) -> SubjectConfiguration:
-        return SubjectConfiguration(data["subject_type"], data["subject_name"], data["subject_setting"], data["cli_options"], data["extensions"])
+        return SubjectConfiguration(data['subject_type'], data['subject_name'], data['subject_setting'], data['cli_options'], data['extensions'])
 
 
 @dataclass(frozen=True)
@@ -70,7 +70,7 @@ def __post_init__(self):
         elif self.commit_nb_range:
             assert self.commit_nb_range[0] <= self.commit_nb_range[1]
         else:
-            raise AttributeError("Evaluation ranges require either major versions or commit numbers")
+            raise AttributeError('Evaluation ranges require either major versions or commit numbers')
 
 
 @dataclass(frozen=True)
@@ -86,26 +86,26 @@ class DatabaseParameters:
     username: str
     password: str
     database_name: str
-    binary_cache_limit: int
+    executable_cache_limit: int
 
     def to_dict(self) -> dict:
         return asdict(self)
 
     @staticmethod
     def from_dict(data: dict) -> DatabaseParameters:
         return DatabaseParameters(
-            data["host"],
-            data["username"],
-            data["password"],
-            data["database_name"],
-            data["binary_cache_limit"],
+            data['host'],
+            data['username'],
+            data['password'],
+            data['database_name'],
+            data['executable_cache_limit'],
         )
 
     def __str__(self) -> str:
-        return f"{self.username}@{self.host}:27017/{self.database_name}"
+        return f'{self.username}@{self.host}:27017/{self.database_name}'
 
     def __repr__(self) -> str:
-        return f"{self.username}@{self.host}:27017/{self.database_name}"
+        return f'{self.username}@{self.host}:27017/{self.database_name}'
 
 
 @dataclass(frozen=True)
@@ -146,29 +146,28 @@ class PlotParameters(EvaluationParameters):
     #    )
 
 
-
 @staticmethod
 def evaluation_factory(kwargs: dict, database_params: DatabaseParameters, only_to_plot=False) -> list[EvaluationParameters]:
-    experiments = set(x for x in kwargs.get("experiments", []) + [kwargs.get("experiment_to_plot")] if x is not None)
+    experiments = set(x for x in kwargs.get('experiments', []) + [kwargs.get('experiment_to_plot')] if x is not None)
     if len(experiments) == 0:
         raise MissingParametersException()
 
     subject_configuration = SubjectConfiguration.from_dict(kwargs)
     sequence_configuration = SequenceConfiguration(
-        int(kwargs.get("nb_of_containers", 1)),
-        int(kwargs.get("sequence_limit", 50)),
-        kwargs.get("search_strategy"),
+        int(kwargs.get('nb_of_containers', 1)),
+        int(kwargs.get('sequence_limit', 50)),
+        kwargs.get('search_strategy'),
     )
     evaluation_params_list = []
     for experiment in sorted(experiments):
         if only_to_plot and experiment != kwargs.get('experiment_to_plot'):
             continue
         evaluation_range = EvaluationRange(
-            kwargs["project_name"],
+            kwargs['project_name'],
             experiment,
             __get_version_range(kwargs),
             __get_commit_nb_range(kwargs),
-            kwargs.get("only_release_commits", False),
+            kwargs.get('only_release_commits', False),
         )
         evaluation_params = EvaluationParameters(
             subject_configuration,
@@ -182,11 +181,11 @@ def evaluation_factory(kwargs: dict, database_params: DatabaseParameters, only_t
 
 @staticmethod
 def __get_cookie_name(form_data: dict[str, str]) -> str | None:
-    if form_data["check_for"] == "request":
+    if form_data['check_for'] == 'request':
         return None
-    if "cookie_name" in form_data:
-        return form_data["cookie_name"]
-    return "generic"
+    if 'cookie_name' in form_data:
+        return form_data['cookie_name']
+    return 'generic'
 
 
 @staticmethod
@@ -199,8 +198,8 @@ def __get_version_range(form_data: dict[str, str]) -> tuple[int, int] | None:
 
 @staticmethod
 def __get_commit_nb_range(form_data: dict[str, str]) -> tuple[int, int] | None:
-    lower_rev_number = form_data.get("lower_commit_nb", None)
-    upper_rev_number = form_data.get("upper_commit_nb", None)
+    lower_rev_number = form_data.get('lower_commit_nb', None)
+    upper_rev_number = form_data.get('upper_commit_nb', None)
     lower_rev_number = int(lower_rev_number) if lower_rev_number else None
     upper_rev_number = int(upper_rev_number) if upper_rev_number else None
     if lower_rev_number is None or upper_rev_number is None:

bughog/parameters.py

@@ -6,13 +6,13 @@
 
 from bughog.evaluation.file_structure import Folder
 from bughog.parameters import EvaluationParameters
-from bughog.subject.executable import Executable
 from bughog.subject.simulation import Simulation
+from bughog.subject.web_browser.executable import BrowserExecutable
 
 # TODO: all pyautogui are imported inside functions because the import needs DISPLAY var, while not all containers need and have that.
 
 class BrowserSimulation(Simulation):
-    def __init__(self, executable: Executable, folder: Folder, params: EvaluationParameters):
+    def __init__(self, executable: BrowserExecutable, folder: Folder, params: EvaluationParameters):
         import pyautogui as gui
         super().__init__(executable, folder, params)
         disp = Display(visible=True, size=(1920, 1080), backend='xvfb', use_xauth=True)
@@ -36,7 +36,7 @@ def supported_commands(self) -> list[str]:
             'hotkey',
             'sleep',
             'screenshot',
-            'report_leak',
+            'reproduced',
             'assert_file_contains',
             'open_file',
             'open_console',
@@ -111,7 +111,7 @@ def screenshot(self, filename: str):
         file_path = os.path.join('/app/logs/screenshots/', f'{project_name}-{experiment_name}-{self.executable.state.name}-{executable_name}.jpg')
         gui.screenshot(file_path)
 
-    def report_leak(self):
+    def reproduced(self):
         self.navigate('https://a.test/report/?bughog_reproduced=ok')
 
     def assert_file_contains(self, filename: str, content: str):
@@ -128,7 +128,7 @@ def open_file(self, filename: str):
         self.navigate(f'file:///root/Downloads/{filename}')
 
     def open_console(self):
-        self.hotkey(*self.executable.get_open_console_hotkey())
+        self.hotkey(*self.executable.open_console_hotkey)
         self.sleep('1.5')
 
 

bughog/subject/web_browser/interaction/simulation.py

@@ -143,15 +143,15 @@ def python_evaluation(project: str, experiment: str, file_name: str):
     sys.modules[module_name] = module
     spec.loader.exec_module(module)
 
-    def report_leak() -> None:
+    def reproduced() -> None:
         remote_ip = request.headers.get("X-Real-IP")
         response_data = {
-            "url": url_for("experiments.report_endpoint", leak=experiment),
+            "url": url_for("experiments.report_endpoint", bughog_reproduced='OK'),
             "method": request.method,
             "headers": dict(request.headers),
             "content": request.data.decode("utf-8"),
         }
 
         requests.post(f"http://{remote_ip}:5001/report/", json=response_data, timeout=5)
 
-    return module.main(request, report_leak)
+    return module.main(request, reproduced)

bughog/web/blueprints/experiments.py

@@ -1,4 +1,4 @@
-const KEYWORDS = "NAVIGATE|NEW_TAB|CLICK_POSITION|CLICK|WRITE|PRESS|HOLD|RELEASE|HOTKEY|SLEEP|SCREENSHOT|REPORT_LEAK|ASSERT_FILE_CONTAINS|OPEN_FILE|OPEN_CONSOLE";
+const KEYWORDS = "NAVIGATE|NEW_TAB|CLICK_POSITION|CLICK|WRITE|PRESS|HOLD|RELEASE|HOTKEY|SLEEP|SCREENSHOT|ASSERT_FILE_CONTAINS|OPEN_FILE|OPEN_CONSOLE";
 
 ace.define("ace/mode/interaction_script_highlight_rules",["require","exports","module","ace/lib/oop","ace/mode/text_highlight_rules"], function(require, exports, module){"use strict";
     const oop = require("../lib/oop");
@@ -44,4 +44,4 @@ const getMode = () => new Promise((resolve) => ace.require(["ace/mode/interactio
 
 export {
     getMode,
-};
+};

bughog/web/vue/src/interaction_script_mode.js

@@ -6,7 +6,7 @@ GITHUB_TOKEN=
 
 # Cache parameters
 # All binaries will be cached in the active MongoDB (either a local Docker container, or the one configured below).
-BCI_BINARY_CACHE_LIMIT=
+BCI_EXECUTABLE_CACHE_LIMIT=
 
 # Database parameters
 BCI_MONGO_HOST=

config/.env.example

@@ -11,11 +11,10 @@
 # RELEASE key
 # HOTKEY key1 key2 ...
 # SLEEP seconds         where seconds is a float or an int
-# REPORT_LEAK
 # ASSERT_FILE_CONTAINS file content     if the downloaded file exists and contains the given content as a substring, the evaluation continues
 #                                       otherwise the evaluation terminates and the exact reason is reported
 
 ### Debugging commands
 # SCREENSHOT file_name
 # OPEN_FILE file
-# OPEN_CONSOLE
+# OPEN_CONSOLE