Multi-file batch analysis (+ Parquet, upload limit, Globus file/directory)

.gitignore

@@ -185,3 +185,14 @@ examples/*_vector_store/
 
 dump.rdb
 celerybeat-schedule.db
+
+# Runtime data (uploads, persisted file lists, scratch test data)
+data/
+
+# Runtime-generated custom metrics & remedy data (package keeps __init__.py + base_dr.py)
+aidrin/custom_metrics/customDR_*.py
+aidrin/custom_metrics/remedy_data/
+
+# session/local artifacts
+AIDRIN-gtag/
+claude.session

aidrin/__init__.py

@@ -334,8 +334,20 @@ def compute_entropy_risk(quasi_identifiers, file_info):
     return _fn(quasi_identifiers, file_info)
 
 
+# ---------------------------------------------------------------------------
+# Batch / Structural Overview
+# ---------------------------------------------------------------------------
+
+def summarize_files(file_infos):
+    """Per-file structural overview + totals for a batch of files (no metrics)."""
+    from aidrin.batch import summarize_files as _fn
+    return _fn(file_infos)
+
+
 __all__ = [
     "__version__",
+    # Batch / Structural Overview
+    "summarize_files",
     # Data Quality
     "calculate_completeness",
     "calculate_duplicates",

aidrin/batch.py

@@ -0,0 +1,71 @@
+"""Session-free batch helpers shared by the web UI, CLI, and library users."""
+
+import os
+
+import pandas as pd
+
+from aidrin.file_handling.file_parser import read_file
+
+
+def _summarize_one(file_info):
+    path, name, file_type = file_info
+    if not path:
+        return {
+            "name": name, "type": file_type,
+            "records": None, "features": None,
+            "numerical": None, "categorical": None,
+            "size_bytes": None, "status": "error",
+            "error": "No file path provided.",
+        }
+    size = None
+    try:
+        if path and os.path.exists(path):
+            size = os.path.getsize(path)
+    except OSError:
+        size = None
+
+    result = read_file(file_info)  # DataFrame | None | str
+    if isinstance(result, pd.DataFrame):
+        # Same dtype convention as the Data Overview panel: numeric vs string.
+        numerical = int(sum(pd.api.types.is_numeric_dtype(d) for d in result.dtypes))
+        categorical = int(sum(pd.api.types.is_string_dtype(d) for d in result.dtypes))
+        return {
+            "name": name, "type": file_type,
+            "records": int(len(result)),
+            "features": int(len(result.columns)),
+            "numerical": numerical, "categorical": categorical,
+            "size_bytes": size, "status": "ok", "error": None,
+        }
+
+    message = result if isinstance(result, str) else (
+        "Could not read the file. The format may be unsupported or the file "
+        "may be corrupted."
+    )
+    return {
+        "name": name, "type": file_type,
+        "records": None, "features": None,
+        "numerical": None, "categorical": None,
+        "size_bytes": size, "status": "error", "error": message,
+    }
+
+
+def summarize_files(file_infos):
+    """Return {"files": [per_file, ...], "totals": {...}} for a list of files.
+
+    Computes structural facts only (records, features, size, load status) — no
+    metrics. A file that fails to load becomes a status:"error" row and never
+    aborts the batch. ``file_infos`` is a list of (path, name, type) tuples.
+    """
+    files = [_summarize_one(fi) for fi in file_infos]
+    by_type = {}
+    for f in files:
+        by_type[f["type"]] = by_type.get(f["type"], 0) + 1
+    totals = {
+        "file_count": len(files),
+        "ok_count": sum(1 for f in files if f["status"] == "ok"),
+        "error_count": sum(1 for f in files if f["status"] == "error"),
+        "total_records": sum(f["records"] or 0 for f in files),
+        "total_size_bytes": sum(f["size_bytes"] or 0 for f in files),
+        "by_type": by_type,
+    }
+    return {"files": files, "totals": totals}

aidrin/file_handling/file_parser.py

@@ -38,6 +38,37 @@
     # (file_type,file_type_name)
 ]
 
+# Map a real file extension to the READER_MAP key that handles it.
+# Excel uses a single combined reader key, so all its extensions point at it.
+_EXCEL_KEY = ".xls, .xlsb, .xlsx, .xlsm"
+EXTENSION_MAP = {
+    ".csv": ".csv",
+    ".json": ".json",
+    ".npz": ".npz",
+    ".h5": ".h5",
+    ".parquet": ".parquet",   # resolves only once a parquet reader is registered in READER_MAP
+    ".xls": _EXCEL_KEY,
+    ".xlsb": _EXCEL_KEY,
+    ".xlsx": _EXCEL_KEY,
+    ".xlsm": _EXCEL_KEY,
+}
+
+
+def file_extension(filename):
+    """Return the lowercased real extension (e.g. ``.csv``), or ``""``."""
+    return os.path.splitext(filename or "")[1].lower()
+
+
+def infer_file_type(filename):
+    """Return the READER_MAP key for a filename's extension, or None.
+
+    Only returns a key that is actually registered in READER_MAP on this
+    install (e.g. ``.parquet`` resolves only if the parquet reader exists).
+    """
+    key = EXTENSION_MAP.get(file_extension(filename))
+    return key if key in READER_MAP else None
+
+
 # logger config
 file_upload_time_log = logging.getLogger("file_upload")
 

aidrin/structured_data_metrics/privacy_measure.py

@@ -517,6 +517,8 @@ def compute_k_anonymity(quasi_identifiers: List[str], file_info):
         "k-Anonymity Visualization": base64_str}``
         or ``{"Error": str}`` on validation failure.
     """
+    if not quasi_identifiers:
+        return {"Error": "Please select at least one quasi-identifier."}
     # Handle both DataFrame and tuple inputs
     if isinstance(file_info, tuple):
         data = read_file(file_info)
@@ -634,6 +636,8 @@ def compute_l_diversity(
         "l-Diversity Visualization": base64_str}``
         or ``{"Error": str}`` on validation failure.
     """
+    if not quasi_identifiers:
+        return {"Error": "Please select at least one quasi-identifier."}
     # Handle both DataFrame and tuple inputs
     if isinstance(file_info, tuple):
         data = read_file(file_info)
@@ -765,6 +769,8 @@ def compute_t_closeness(
         where ``t-Value`` is in ``[0, 1]``,
         or ``{"Error": str}`` on validation failure.
     """
+    if not quasi_identifiers:
+        return {"Error": "Please select at least one quasi-identifier."}
     # Handle both DataFrame and tuple inputs
     if isinstance(file_info, tuple):
         data = read_file(file_info)
@@ -895,6 +901,8 @@ def compute_entropy_risk(quasi_identifiers, file_info):
         where ``Entropy-Value >= 0``,
         or ``{"Error": str}`` on validation failure.
     """
+    if not quasi_identifiers:
+        return {"Error": "Please select at least one quasi-identifier."}
     # Handle both DataFrame and tuple inputs
     if isinstance(file_info, tuple):
         data = read_file(file_info)