added HDF5 read capability. Closes #27
Build Debian Packages / build (trixie, debian13) (pull_request) Successful in 14m37s
Build Debian Packages / build (bookworm, debian12) (pull_request) Successful in 14m50s
Build Debian Packages / build (bullseye, debian11) (pull_request) Has been cancelled

This commit is contained in:
2026-07-06 20:42:04 +02:00
parent c181e15aa8
commit 7ba5c7d56a
3 changed files with 419 additions and 32 deletions
+244 -4
View File
@@ -7,6 +7,7 @@ import collections
import threading
import traceback
import io
import numpy
class DataPool(collections.abc.MutableMapping):
"""
@@ -107,7 +108,7 @@ class DataPool(collections.abc.MutableMapping):
self.__registered_listeners=None
def write_hdf5(self,hdffile,where="/",name="data_pool", complib=None, complevel=None):
if type(hdffile) is bytes:
if isinstance(hdffile, (bytes, str)):
dump_file=tables.open_file(hdffile, mode="a")
elif isinstance(hdffile,tables.File):
dump_file=hdffile
@@ -156,7 +157,7 @@ class DataPool(collections.abc.MutableMapping):
value=self.__mydict[key]
self.__dictlock.release()
# now write data, assuming, the object is constant during write operation
if "write_to_hdf" in dir(value):
if hasattr(value, "write_to_hdf"):
try:
value.write_to_hdf(hdffile=dump_file,
where=dump_dir,
@@ -170,16 +171,255 @@ class DataPool(collections.abc.MutableMapping):
traceback.print_tb(sys.exc_info()[2], None, traceback_file)
print("detailed traceback: %s\n"%str(e)+traceback_file.getvalue())
traceback_file=None
elif isinstance(value, (numpy.ndarray, int, float, str, bytes)):
try:
obj_to_save = value
if isinstance(value, str):
obj_to_save = value.encode('utf-8')
dump_file.create_array(dump_dir,
name=group_keyname,
obj=obj_to_save,
title=key)
except Exception as e:
print("failed to write data_pool[\"%s\"]: %s"%(key,str(e)))
else:
print("don't know how to store data_pool[\"%s\"]"%key)
print("don't know how to store data_pool[\"%s\"] (type: %s)"%(key, type(value)))
value=None
finally:
dump_group=None
if type(hdffile) is bytes:
if isinstance(hdffile, (bytes, str)):
dump_file.close()
dump_file=None
def read_from_hdf(self, hdffile, where="/data_pool"):
"""
Read data from HDF5 file and populate the DataPool.
Parameters:
- hdffile: HDF5 file object or filename (bytes/str)
- where: Location within HDF5 file to read from (default: "/data_pool")
"""
# Handle both filename and file object
if type(hdffile) is bytes or isinstance(hdffile, str):
hdf_file = tables.open_file(hdffile, mode="r")
close_file = True
elif isinstance(hdffile, tables.File):
hdf_file = hdffile
close_file = False
else:
raise Exception("expecting hdffile or string")
try:
# Navigate to the data pool group
if where not in hdf_file:
raise Exception(f"HDF5 location '{where}' not found")
data_pool_group = hdf_file.get_node(where)
# Recursively process the group structure
self._read_hdf_group(data_pool_group, "")
finally:
if close_file:
hdf_file.close()
def _read_hdf_group(self, group, path_prefix):
"""
Recursively read HDF5 group and populate DataPool.
Parameters:
- group: HDF5 group object
- path_prefix: Current path in DataPool hierarchy
"""
# Process all nodes in this group
for node in group:
# Get the node name (last part of the path)
node_name = node._v_name
# Try to get the original name from title
title = node._v_title
if title:
# For DAMARIS objects, the title is usually the full path
# For intermediate groups, it's just the part name
if "/" in title:
full_path = title
else:
if path_prefix:
full_path = f"{path_prefix}/{title}"
else:
full_path = title
else:
# Strip dir_ and dict_ prefixes from node name
clean_name = self._clean_hdf_name(node_name)
# Build the full path in DataPool hierarchy
if path_prefix:
full_path = f"{path_prefix}/{clean_name}"
else:
full_path = clean_name
full_path = full_path.replace("//", "/")
if isinstance(node, (tables.Group, tables.Table)):
# Check if this is a known DAMARIS object type
if hasattr(node._v_attrs, 'damaris_type'):
damaris_type = node._v_attrs.damaris_type
# Handle different DAMARIS object types
if damaris_type == "Accumulation":
from damaris.data.Accumulation import read_from_hdf
obj = read_from_hdf(node)
if obj is not None:
self[full_path] = obj
elif damaris_type == "ADC_Result":
from damaris.data.ADC_Result import read_from_hdf
obj = read_from_hdf(node)
if obj is not None:
self[full_path] = obj
elif damaris_type == "MeasurementResult":
from damaris.data.MeasurementResult import read_from_hdf
obj = read_from_hdf(node)
if obj is not None:
self[full_path] = obj
# Skip further processing of this group since we handled it
continue
# Recursively process sub-groups
if isinstance(node, tables.Group):
self._read_hdf_group(node, full_path)
elif isinstance(node, (tables.Array, tables.CArray, tables.EArray, tables.Table)):
# Handle simple array data
try:
data = node.read()
# If it's a scalar array of bytes, decode it to string
if hasattr(data, 'decode'):
try:
data = data.decode('utf-8')
except:
pass
elif isinstance(data, numpy.ndarray) and data.dtype.kind in ('S', 'V'):
# For numpy byte arrays, try to decode if they are scalar or small
if data.size == 1:
try:
decoded = data.item().decode('utf-8')
data = decoded
except:
pass
self[full_path] = data
except Exception as e:
print(f"Warning: Could not read array {full_path}: {e}")
# Note: Other node types (Table, etc.) would be handled here as needed
def _clean_hdf_name(self, name):
"""
Clean HDF5 node name by removing dir_ and dict_ prefixes.
Parameters:
- name: HDF5 node name (may have dir_ or dict_ prefix)
Returns:
- Cleaned name with prefixes removed
"""
if name.startswith("dir_"):
return name[4:] # Remove "dir_" prefix
elif name.startswith("dict_"):
return name[5:] # Remove "dict_" prefix
else:
return name
@classmethod
def load_hdf5(cls, filename):
"""
Load a DAMARIS HDF5 file and return DataPool and metadata.
Parameters:
- filename: Path to HDF5 file
Returns:
- tuple: (DataPool instance, metadata dict)
metadata dict contains:
- 'scripts': dict with experiment scripts and spool info
- 'log': log data if available
- 'timeline': timeline data if available
"""
# Validate file
if not tables.is_pytables_file(filename):
raise Exception(f"File {filename} is not a valid PyTables HDF5 file")
# Create DataPool and load data
data_pool = cls()
try:
hdf_file = tables.open_file(filename, mode="r")
# Load main data pool
if "/data_pool" in hdf_file:
data_pool.read_from_hdf(hdf_file, where="/data_pool")
# Extract metadata
metadata = {}
# Load scripts
if "/scripts" in hdf_file:
scripts_group = hdf_file.get_node("/scripts")
scripts_data = {}
# Use _v_children to get node names safely
if hasattr(scripts_group, '_v_children'):
for node_name, node in scripts_group._v_children.items():
try:
if hasattr(node, 'read'):
data = node.read()
# Handle numpy string arrays and other types
if hasattr(data, 'decode'):
# bytes or numpy bytes
scripts_data[node_name] = data.decode('utf-8', errors='replace')
elif hasattr(data, 'tolist'):
# numpy array
scripts_data[node_name] = str(data.tolist())
elif isinstance(data, (str, bytes)):
# string or bytes
if isinstance(data, bytes):
scripts_data[node_name] = data.decode('utf-8', errors='replace')
else:
scripts_data[node_name] = data
else:
scripts_data[node_name] = str(data)
except Exception as e:
print(f"Warning: Could not read script {node_name}: {e}")
metadata['scripts'] = scripts_data
# Load log
if "/log" in hdf_file:
try:
log_node = hdf_file.get_node("/log")
metadata['log'] = log_node.read()
except Exception as e:
print(f"Warning: Could not read log: {e}")
# Load timeline
if "/timeline" in hdf_file:
try:
timeline_node = hdf_file.get_node("/timeline")
metadata['timeline'] = timeline_node.read()
except Exception as e:
print(f"Warning: Could not read timeline: {e}")
return data_pool, metadata
except Exception as e:
raise Exception(f"Failed to load HDF5 file {filename}: {e}")
finally:
if 'hdf_file' in locals():
hdf_file.close()
def register_listener(self, listening_function):
self.__registered_listeners.append(listening_function)