use pathlib.Path for directories and files

corrlib/meas_io.py

@@ -10,9 +10,13 @@ from .tools import get_db_file, cache_enabled
 from .tracker import get, save, unlock
 import shutil
 from typing import Any
+from pathlib import Path
 
 
-def write_measurement(path: str, ensemble: str, measurement: dict[str, dict[str, dict[str, Any]]], uuid: str, code: str, parameter_file: Union[str, None]) -> None:
+CACHE_DIR = ".cache"
+
+
+def write_measurement(path: Path, ensemble: str, measurement: dict[str, dict[str, dict[str, Any]]], uuid: str, code: str, parameter_file: Union[str, None]) -> None:
     """
     Write a measurement to the backlog.
     If the file for the measurement already exists, update the measurement.
@@ -33,7 +37,7 @@ def write_measurement(path: str, ensemble: str, measurement: dict[str, dict[str,
         The parameter file used for the measurement.
     """
     db_file = get_db_file(path)
-    db = os.path.join(path, db_file)
+    db = path / db_file
 
     files_to_save = []
 
@@ -44,11 +48,11 @@ def write_measurement(path: str, ensemble: str, measurement: dict[str, dict[str,
     conn = sqlite3.connect(db)
     c = conn.cursor()
     for corr in measurement.keys():
-        file_in_archive = os.path.join('.', 'archive', ensemble, corr,  uuid + '.json.gz')
-        file = os.path.join(path, file_in_archive)
+        file_in_archive = Path('.') / 'archive' / ensemble / corr / str(uuid + '.json.gz')
+        file = path / file_in_archive
         known_meas = {}
-        if not os.path.exists(os.path.join(path, '.', 'archive', ensemble, corr)):
-            os.makedirs(os.path.join(path, '.', 'archive', ensemble, corr))
+        if not os.path.exists(path / 'archive' / ensemble / corr):
+            os.makedirs(path / 'archive' / ensemble / corr)
             files_to_save.append(file_in_archive)
         else:
             if os.path.exists(file):
@@ -99,7 +103,7 @@ def write_measurement(path: str, ensemble: str, measurement: dict[str, dict[str,
                 pars[subkey] = json.dumps(parameters)
         for subkey in subkeys:
             parHash = sha256(str(pars[subkey]).encode('UTF-8')).hexdigest()
-            meas_path = file_in_archive + "::" + parHash
+            meas_path = str(file_in_archive) + "::" + parHash
 
             known_meas[parHash] = measurement[corr][subkey]
 
@@ -115,7 +119,7 @@ def write_measurement(path: str, ensemble: str, measurement: dict[str, dict[str,
     return
 
 
-def load_record(path: str, meas_path: str) -> Union[Corr, Obs]:
+def load_record(path: Path, meas_path: str) -> Union[Corr, Obs]:
     """
     Load a list of records by their paths.
 
@@ -134,7 +138,7 @@ def load_record(path: str, meas_path: str) -> Union[Corr, Obs]:
     return load_records(path, [meas_path])[0]
 
 
-def load_records(path: str, meas_paths: list[str], preloaded: dict[str, Any] = {}) -> list[Union[Corr, Obs]]:
+def load_records(path: Path, meas_paths: list[str], preloaded: dict[str, Any] = {}) -> list[Union[Corr, Obs]]:
     """
     Load a list of records by their paths.
 
@@ -162,11 +166,11 @@ def load_records(path: str, meas_paths: list[str], preloaded: dict[str, Any] = {
     returned_data: list[Any] = []
     for file in needed_data.keys():
         for key in list(needed_data[file]):
-            if os.path.exists(cache_path(path, file, key) + ".p"):
-                returned_data.append(load_object(cache_path(path, file, key) + ".p"))
+            if os.path.exists(str(cache_path(path, file, key)) + ".p"):
+                returned_data.append(load_object(str(cache_path(path, file, key)) + ".p"))
             else:
                 if file not in preloaded:
-                    preloaded[file] = preload(path, file)
+                    preloaded[file] = preload(path, Path(file))
                 returned_data.append(preloaded[file][key])
                 if cache_enabled(path):
                     if not os.path.exists(cache_dir(path, file)):
@@ -175,7 +179,7 @@ def load_records(path: str, meas_paths: list[str], preloaded: dict[str, Any] = {
     return returned_data
 
 
-def cache_dir(path: str, file: str) -> str:
+def cache_dir(path: Path, file: str) -> Path:
     """
     Returns the directory corresponding to the cache for the given file.
 
@@ -190,14 +194,14 @@ def cache_dir(path: str, file: str) -> str:
     cache_path: str
         The path holding the cached data for the given file.
     """
-    cache_path_list = [path]
-    cache_path_list.append(".cache")
-    cache_path_list.extend(file.split("/")[1:])
-    cache_path = "/".join(cache_path_list)
+    cache_path_list = file.split("/")[1:]
+    cache_path = path / CACHE_DIR
+    for directory in cache_path_list:
+        cache_path /= directory
     return cache_path
 
 
-def cache_path(path: str, file: str, key: str) -> str:
+def cache_path(path: Path, file: str, key: str) -> Path:
     """
     Parameters
     ----------
@@ -213,11 +217,11 @@ def cache_path(path: str, file: str, key: str) -> str:
     cache_path: str
         The path at which the measurement of the given file and key is cached.
     """
-    cache_path = os.path.join(cache_dir(path, file), key)
+    cache_path = cache_dir(path, file) / key
     return cache_path
 
 
-def preload(path: str, file: str) -> dict[str, Any]:
+def preload(path: Path, file: Path) -> dict[str, Any]:
     """
     Read the contents of a file into a json dictionary with the pyerrors.json.load_json_dict method.
 
@@ -234,12 +238,12 @@ def preload(path: str, file: str) -> dict[str, Any]:
         The data read from the file.
     """
     get(path, file)
-    filedict: dict[str, Any] = pj.load_json_dict(os.path.join(path, file))
+    filedict: dict[str, Any] = pj.load_json_dict(path / file)
     print("> read file")
     return filedict
 
 
-def drop_record(path: str, meas_path: str) -> None:
+def drop_record(path: Path, meas_path: str) -> None:
     """
     Drop a record by it's path.
 
@@ -251,9 +255,9 @@ def drop_record(path: str, meas_path: str) -> None:
         The measurement path as noted in the database.
     """
     file_in_archive = meas_path.split("::")[0]
-    file = os.path.join(path, file_in_archive)
+    file = path / file_in_archive
     db_file = get_db_file(path)
-    db = os.path.join(path, db_file)
+    db = path / db_file
     get(path, db_file)
     sub_key = meas_path.split("::")[1]
     unlock(path, db_file)
@@ -268,7 +272,7 @@ def drop_record(path: str, meas_path: str) -> None:
     known_meas = pj.load_json_dict(file)
     if sub_key in known_meas:
         del known_meas[sub_key]
-        unlock(path, file_in_archive)
+        unlock(path, Path(file_in_archive))
         pj.dump_dict_to_json(known_meas, file)
         save(path, message="Drop measurements to database", files=[db, file])
         return
@@ -276,7 +280,7 @@ def drop_record(path: str, meas_path: str) -> None:
         raise ValueError("This measurement does not exist as a file!")
 
 
-def drop_cache(path: str) -> None:
+def drop_cache(path: Path) -> None:
     """
     Drop the cache directory of the library.
 
@@ -285,7 +289,7 @@ def drop_cache(path: str) -> None:
     path: str
         The path of the library.
     """
-    cache_dir = os.path.join(path, ".cache")
+    cache_dir = path / ".cache"
     for f in os.listdir(cache_dir):
-        shutil.rmtree(os.path.join(cache_dir, f))
+        shutil.rmtree(cache_dir / f)
     return