.gitignore

@@ -2,4 +2,3 @@ pyerrors_corrlib.egg-info
 __pycache__
 *.egg-info
 test.ipynb
-test_ds

corrlib/__init__.py

@@ -19,6 +19,5 @@ from .main import *
 from .import input as input
 from .initialization import *
 from .meas_io import *
-from .cache_io import *
 from .find import *
 from .version import __version__

corrlib/cache_io.py

@@ -1,19 +1,6 @@
 from typing import Union, Optional
 import os
 import shutil
-from .tools import record2name_key
-from pyerrors import dump_object
-import datalad.api as dl
-import sqlite3
-
-
-def get_version_hash(path, record):
-    db = os.path.join(path, "backlogger.db")
-    dl.get(db, dataset=path)
-    conn = sqlite3.connect(db)
-    c = conn.cursor()
-    c.execute(f"SELECT current_version FROM 'backlogs' WHERE path = '{record}'")
-    return c.fetchall()[0][0]
 
 
 def drop_cache_files(path: str, fs: Optional[list[str]]=None):
@@ -32,27 +19,15 @@ def cache_dir(path, file):
     return cache_path
 
 
-def cache_path(path, file, sha_hash, key):
+def cache_path(path, file, hash, key):
-    cache_path = os.path.join(cache_dir(path, file), key + "_" +  sha_hash)
+    cache_path = os.path.join(cache_dir(path, file), hash, key)
     return cache_path
 
+def is_in_cache(path, record, hash):
     
-def is_old_version(path, record):
+    if os.file.exists(cache_path(path, file, hash, key)):
-    version_hash = get_version_hash(path, record)
-    file, key = record2name_key(record)
-    meas_cache_path = os.path.join(cache_dir(path, file))
-    ls = []
-    for p, ds, fs in os.walk(meas_cache_path):
-        ls.extend(fs)
-    for filename in ls:
-        if key == filename.split("_")[0]:
-            if not version_hash == filename.split("_")[1][:-2]:
         return True
     else:
         return False
     
     
-def is_in_cache(path, record):
-    version_hash = get_version_hash(path, record)
-    file, key = record2name_key(record)
-    return os.path.exists(cache_path(path, file, version_hash, key) + ".p")

corrlib/cli.py

@@ -6,7 +6,7 @@ from .toml import import_tomls, update_project, reimport_project
 from .find import find_record, list_projects
 from .tools import str2list
 from .main import update_aliases
-from .cache_io import drop_cache_files as cio_drop_cache_files
+from .meas_io import drop_cache as mio_drop_cache
 import os
 
 
@@ -171,7 +171,7 @@ def drop_cache(
     """
     Drop the currect cache directory of the dataset.
     """
-    cio_drop_cache_files(path)
+    mio_drop_cache(path)
     return
 
 

corrlib/initialization.py

@@ -21,8 +21,7 @@ def _create_db(db):
                   parameters TEXT,
                   parameter_file TEXT,
                   created_at TEXT,
-                  updated_at TEXT,
+                  updated_at TEXT)''')
-                  current_version TEXT)''')
     c.execute('''CREATE TABLE IF NOT EXISTS projects
                  (id TEXT PRIMARY KEY,
                   aliases TEXT,

corrlib/meas_io.py

@@ -5,10 +5,11 @@ import sqlite3
 from .input import sfcf,openQCD
 import json
 from typing import Union, Optional
-from pyerrors import Obs, Corr, load_object, dump_object
+from pyerrors import Obs, Corr, dump_object, load_object
-from hashlib import sha256
+from hashlib import sha256, sha1
-from .tools import cached, record2name_key, make_version_hash
+from .tools import cached, record2name_key
-from .cache_io import is_in_cache, cache_path, cache_dir, get_version_hash
+import shutil
+from .caching import cache_path, cache_dir
 
 
 def write_measurement(path, ensemble, measurement, uuid, code, parameter_file: Optional[str]=None):
@@ -78,13 +79,11 @@ def write_measurement(path, ensemble, measurement, uuid, code, parameter_file: O
                 subkey = "/".join(par_list)
                 subkeys = [subkey]
                 pars[subkey] = json.dumps(parameters)
-
-        meas_paths = []
         for subkey in subkeys:                
-            par_hash = sha256(str(pars[subkey]).encode('UTF-8')).hexdigest()
+            parHash = sha256(str(pars[subkey]).encode('UTF-8')).hexdigest()
-            meas_path = file_in_archive + "::" + par_hash
+            meas_path = file_in_archive + "::" + parHash
-            meas_paths.append(meas_path)
+
-            known_meas[par_hash] = measurement[corr][subkey]
+            known_meas[parHash] = measurement[corr][subkey]
             
             if c.execute("SELECT * FROM backlogs WHERE path = ?", (meas_path,)).fetchone() is not None:
                 c.execute("UPDATE backlogs SET updated_at = datetime('now') WHERE path = ?", (meas_path, ))
@@ -93,12 +92,7 @@ def write_measurement(path, ensemble, measurement, uuid, code, parameter_file: O
                                                 (corr, ensemble, code, meas_path, uuid, pars[subkey], parameter_file))
         conn.commit()
         pj.dump_dict_to_json(known_meas, file)
-        for meas_path in meas_paths:
+    files.append(path + '/backlogger.db')
-            version_hash = make_version_hash(path, meas_path)
-            print(version_hash)
-            c.execute("UPDATE backlogs SET current_version = ? WHERE project = ? AND code = ? and name = ?", (version_hash, uuid, code, corr))
-        conn.commit()
-    files.append(db)
     conn.close()
     dl.save(files, message="Add measurements to database", dataset=path)
 
@@ -146,21 +140,16 @@ def load_records(path: str, record_paths: list[str], preloaded = {}) -> list[Uni
     returned_data: list = []
     for file in needed_data.keys():
         for key in list(needed_data[file]):
-            record = file + "::" + key
+            if os.path.exists(cache_path(path, file, key) + ".p"):
-            current_version = get_version_hash(path, record)
+                returned_data.append(load_object(cache_path(path, file, key) + ".p"))
-            if is_in_cache(path, record):
-                returned_data.append(load_object(cache_path(path, file, current_version, key) + ".p"))
             else:
                 if file not in preloaded:
                     preloaded[file] = preload(path, file)
                 returned_data.append(preloaded[file][key])
                 if cached:
-                    if not is_in_cache(path, record):
-                        file, key = record2name_key(record)
                     if not os.path.exists(cache_dir(path, file)):
                         os.makedirs(cache_dir(path, file))
-                        current_version = get_version_hash(path, record)
+                    dump_object(preloaded[file][key], cache_path(path, file, key))
-                        dump_object(preloaded[file][key], cache_path(path, file, current_version, key))
     return returned_data
 
 

corrlib/tools.py

@@ -1,5 +1,5 @@
-import hashlib
+
-import os
+
 
 def str2list(string):
     return string.split(",")
@@ -22,10 +22,3 @@ def record2name_key(record_path: str):
     file = record_path.split("::")[0]
     key = record_path.split("::")[1]
     return file, key
-
-
-def make_version_hash(path, record):
-    file, key = record2name_key(record)
-    with open(os.path.join(path, file), 'rb') as fp:
-        file_hash = hashlib.file_digest(fp, 'sha1').hexdigest()
-    return file_hash

setup.py

@@ -10,7 +10,7 @@ setup(name='pycorrlib',
       version=version['__version__'],
       author='Justus Kuhlmann',
       author_email='j_kuhl19@uni-muenster.de',
-      install_requires=['pyerrors>=2.11.1', 'datalad>=1.1.0', 'typer>=0.12.5', 'gitpython>=3.1.45'],
+      install_requires=['pyerrors>=2.11.1', 'datalad>=1.1.0', 'typer>=0.12.5'],
       entry_points = {
         'console_scripts': ['pcl=corrlib.cli:app'],
         },