add check for links to files

corrlib/integrity.py

@@ -3,6 +3,8 @@ from pathlib import Path
 from .tools import get_db_file
 import pandas as pd
 import sqlite3
+from .tracker import get
+import pyerrors.input.json as pj
 
 
 def has_valid_times(result: pd.Series) -> bool:
@@ -38,10 +40,45 @@ def check_db_integrity(path: Path) -> None:
         if not has_valid_times(result):
             raise ValueError(f"Result with id {result[id]} has wrong time signatures.")
     print("DB:\t✅")
+    return
+
+
+def _check_db2paths(path: Path, meas_paths: list[str]) -> None:
+    needed_data: dict[str, list[str]] = {}
+    for mpath in meas_paths:
+        file = mpath.split("::")[0]
+        if file not in needed_data.keys():
+            needed_data[file] = []
+        key = mpath.split("::")[1]
+        needed_data[file].append(key)
+
+    for file in needed_data.keys():
+        get(path, Path(file))
+        filedict: dict[str, Any] = pj.load_json_dict(str(path / file))
+        if not set(filedict.keys()).issubset(needed_data[file]):
+            for key in filedict.keys():
+                if key not in needed_data[file]:
+                    raise ValueError(f"Found unintended key {key} in file {file}.")
+        elif not set(needed_data[file]).issubset(filedict.keys()):
+            for key in needed_data[file]:
+                if key not in filedict.keys():
+                    raise ValueError(f"Did not find data for key {key} that should be in file {file}.")
+    print("Links:\t✅")
+    return
+
+
+def check_db_file_links(path: Path) -> None:
+    db = get_db_file(path)
+    search_expr = "SELECT path FROM 'backlogs'"
+    conn = sqlite3.connect(path / db)
+    results = pd.read_sql(search_expr, conn)['path'].values
+    print(results)
+    _check_db2paths(path, results)
 
 
 def full_integrity_check(path: Path) -> None:
     check_db_integrity(path)
+    check_db_file_links(path)
     print("Full:\t✅")