first version of caching

TODO.md

@@ -3,7 +3,13 @@
 ## Features
 - [ ] implement import of non-datalad projects
 - [ ] implement a way to use another backlog repo as a project
+
+- [ ] find a way to convey the mathematical structure of what EXACTLY is the form of the correlator in a specific project
+    - this could e.g. be done along the lines of mandatory documentation
+- [ ] keep better track of the versions the code, that was used for a specific measurement.
+    - maybe let this be an input in the project file?
+    - git repo and commit hash/version tag
 - [ ] implement local caching with pickle files
 
 ## Bugfixes
-- [ ] revisit the reimport function for single files
+- [ ] revisit the reimport function for single files

corrlib/initialization.py

@@ -46,4 +46,7 @@ def create(path):
     os.makedirs(path + '/archive')
     os.makedirs(path + '/toml_imports')
     os.makedirs(path + '/import_scripts/template.py')
+    with open(path + "/.gitignore", "w") as fp:
+        fp.write(".cache")
+        fp.close()
     dl.save(path, dataset=path, message="Initialize backlogger directory.")

corrlib/meas_io.py

@@ -5,8 +5,9 @@ import sqlite3
 from .input import sfcf,openQCD
 import json
 from typing import Union
-from pyerrors import Obs, Corr
+from pyerrors import Obs, Corr, dump_obj, load_obj
 from hashlib import sha256
+from .tools import cached
 
 
 def write_measurement(path, ensemble, measurement, uuid, code, parameter_file=None):
@@ -135,16 +136,35 @@ def load_records(path: str, meas_paths: list[str], preloaded = {}) -> list[Union
             needed_data[file] = []
         key = mpath.split("::")[1]
         needed_data[file].append(key)
-    for filename in needed_data.keys():
-        if not filename in preloaded:
-            preloaded[filename] = preload(path, filename)
     returned_data: list = []
-    for filename in needed_data.keys():
-        for key in list(needed_data[filename]):
-            returned_data.append(preloaded[filename][key])
+    for file in needed_data.keys():
+        for key in list(needed_data[file]):
+            if os.path.exists(cache_path(path, file, key)):
+                returned_data.append(load_obj(cache_path(path, file, key)))
+            else:
+                if file not in preloaded:
+                    preloaded[file] = preload(path, file)
+                returned_data.append(preloaded[file][key])
+                if cached:
+                    if not os.path.exists(cache_dir(path, file)):
+                        os.makedirs(cache_dir(path, file))
+                    dump_obj(preloaded[file][key], cache_path(path, file, key))
     return returned_data
 
 
+def cache_dir(path, file):
+    cache_path_list = [path]
+    cache_path_list.append(".cache")
+    cache_path_list.extend(file.split("/")[1:])
+    cache_path = os.path.join(cache_path_list)
+    return cache_path
+
+
+def cache_path(path, file, key):
+    cache_path = os.path.join(cache_dir(path, file), key)
+    return cache_path + ".p"
+
+
 def preload(path: str, file: str):
     dl.get(os.path.join(path, file), dataset=path)
     filedict = pj.load_json_dict(os.path.join(path, file))

corrlib/toml.py

@@ -104,12 +104,14 @@ def import_toml(path: str, file: str, copy_file: bool=True) -> None:
                     for rwp in ["integrator", "eps", "ntot", "dnms"]:
                         param[rwp] = "Unknown"
                 param['type'] = 't0'
-                measurement = openQCD.extract_t0(path, uuid, md['path'], param, md["prefix"], md["dtr_read"], md["xmin"], md["spatial_extent"], fit_range=md.get('fit_range', 5), postfix=md.get('postfix', None), names=md.get('names', None))
+                measurement = openQCD.extract_t0(path, uuid, md['path'], param, md["prefix"], md["dtr_read"], md["xmin"], md["spatial_extent"],
+                                                 fit_range=md.get('fit_range', 5), postfix=md.get('postfix', None), names=md.get('names', None), files=md.get('files', None))
             elif md['measurement'] == 't1':
                 if 'param_file' in md:
                     param = openQCD.read_ms3_param(path, uuid, md['param_file'])
                 param['type'] = 't1'
-                measurement = openQCD.extract_t1(path, uuid, md['path'], param, md["prefix"], md["dtr_read"], md["xmin"], md["spatial_extent"], fit_range=md.get('fit_range', 5), postfix=md.get('postfix', None), names=md.get('names', None))
+                measurement = openQCD.extract_t1(path, uuid, md['path'], param, md["prefix"], md["dtr_read"], md["xmin"], md["spatial_extent"],
+                                                 fit_range=md.get('fit_range', 5), postfix=md.get('postfix', None), names=md.get('names', None), files=md.get('files', None))
 
         write_measurement(path, ensemble, measurement, uuid, project['code'], (md['param_file'] if 'param_file' in md else None))
 

corrlib/tools.py

@@ -6,4 +6,6 @@ def str2list(string):
 
 def list2str(mylist):
     s = ",".join(mylist)
-    return s
+    return s
+
+cached = True