h5tab 1.0.0__tar.gz

h5tab-1.0.0/PKG-INFO

@@ -0,0 +1,7 @@
+Metadata-Version: 2.4
+Name: h5tab
+Version: 1.0.0
+Summary: Simple HDF5 file viewer.
+Author-email: John Helly <j.c.helly@durham.ac.uk>
+Requires-Python: >=3.7
+Requires-Dist: h5py

h5tab-1.0.0/README.md

@@ -0,0 +1,35 @@
+# h5tab - view tabular data stored as HDF5 datasets
+
+This is a simple utility to display multiple HDF5 datasets as columns in a
+table. This is useful when you need to look up elements with the same index
+between different datasets.
+
+## Usage
+
+Run the script with the names of one or more HDF5 files as arguments.
+```
+h5tab.py filename ...
+```
+This brings up a tree view where you can select datasets to open. Clicking
+a dataset toggles it between selected and not selected.
+
+Once you've selected all of the datasets to open click the "Open
+selected datasets" button at the bottom of the window. This opens a new
+window with a table where the columns correspond to the selected datasets
+and the rows are the dataset element index (in the first dimension in case
+of multidimensional datasets).
+
+In the case of compound data types, individual fields can be designated
+as table columns.
+
+Only the displayed portions of the datasets are read in, so very large
+datasets can be viewed.
+
+## Limitations
+
+This is not a general purpose HDF5 file viewer.
+
+  * Table rows always correspond to the first dimension so datasets with
+    large dimensions other than the first can't be displayed
+  * The program has no good way to display variable length datasets
+  * The tree view can be slow for large numbers of datasets

h5tab-1.0.0/h5tab.egg-info/PKG-INFO

@@ -0,0 +1,7 @@
+Metadata-Version: 2.4
+Name: h5tab
+Version: 1.0.0
+Summary: Simple HDF5 file viewer.
+Author-email: John Helly <j.c.helly@durham.ac.uk>
+Requires-Python: >=3.7
+Requires-Dist: h5py

h5tab-1.0.0/h5tab.egg-info/SOURCES.txt

@@ -0,0 +1,9 @@
+README.md
+h5tab.py
+pyproject.toml
+h5tab.egg-info/PKG-INFO
+h5tab.egg-info/SOURCES.txt
+h5tab.egg-info/dependency_links.txt
+h5tab.egg-info/entry_points.txt
+h5tab.egg-info/requires.txt
+h5tab.egg-info/top_level.txt

h5tab-1.0.0/h5tab.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

h5tab-1.0.0/h5tab.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+h5tab = h5tab:main

h5tab-1.0.0/h5tab.egg-info/requires.txt

@@ -0,0 +1 @@
+h5py

h5tab-1.0.0/h5tab.egg-info/top_level.txt

@@ -0,0 +1 @@
+h5tab

h5tab-1.0.0/h5tab.py

@@ -0,0 +1,724 @@
+#!/usr/bin/env python
+
+#
+# TODO:
+#
+# More sensible way to determine no. of rows
+# Add attribute/info icons
+#
+
+# Avoid warnings if h5py has MPI support
+try:
+    import mpi4py
+except ImportError:
+    pass
+else:
+    mpi4py.rc.initialize = False
+
+import h5py
+
+import os.path
+import sys
+
+from tkinter import *
+from idlelib.tree import TreeItem, TreeNode
+
+import tkinter.font as tkFont
+
+ITEM_TOP     = 0
+ITEM_GROUP   = 1
+ITEM_DATASET = 2
+ITEM_ATTR    = 3
+ITEM_DESCR   = 4
+
+icon_images = None
+
+# Nodes which are highlighted
+selected_nodes = {}
+
+# Format for displaying values
+float_fmt = "%14.6e"
+int_fmt   = "%d"
+
+# Minimum column width
+min_width = 16
+
+# Maximum elements per row
+max_elements = 6
+
+# Maximum column width
+max_width = max_elements * min_width
+
+def get_shape(data):
+    """Return shape of array or [1] if its a scalar"""
+    if len(data.shape) > 0:
+        return data.shape
+    else:
+        return [1]
+
+def get_format(dtype):
+    # Check for variable length data types
+    if h5py.check_string_dtype(dtype) is not None:
+        return "%s"
+    elif h5py.check_vlen_dtype(dtype) is not None:
+        dtype = h5py.check_vlen_dtype(dtype)
+    # Choose format string depending on base type
+    if dtype.kind == "i" or dtype.kind == "u":
+        fmt = int_fmt
+    elif dtype.kind == "f":
+        fmt = float_fmt
+    elif dtype.kind == "S" or dtype.kind == "U" or dtype.kind == "a":
+        fmt = "%s"
+    else:
+        fmt = None
+    return fmt
+
+def get_width(dtype):
+    if h5py.check_string_dtype(dtype) is not None or h5py.check_vlen_dtype(dtype) is not None:
+        return max_width
+    elif dtype.kind == "S" or dtype.kind == "U" or dtype.kind == "a":
+        return min(max_width, max((dtype.itemsize, min_width)))
+    else:
+        return min_width
+
+def is_too_big(shape):
+    if len(shape) < 2:
+        # 1D arrays are always ok
+        return False
+    elif len(shape) == 2:
+        # 2D is ok if second dim is small
+        return (shape[1] > max_elements)
+    else:
+        # Multidimensional array - check no. of elements
+        i = shape[1]
+        for j in shape[2:]:
+            i *= j
+        return (i > max_elements)
+
+def argsort(seq):
+    return sorted(range(len(seq)), key=seq.__getitem__)
+
+def get_icons():
+    # idlelib doesn't have appropriate icons for datasets/attributes
+    # so encode some here (data strings are base64 encoded .gif files)
+    dataset = PhotoImage(data=r'R0lGODlhEAANAPIEADAwMAD//8PDw9zc3P///6usr'+
+                         r'QAAAAAAACH5BAEAAAUALAAAAAAQAA0AAAMqOLDcrSHKORe9'+
+                         r'0uktuv/fIhBkaZKCCK6deL5ECrCsC5syvaq6BygbzSABADs=')
+    node    = PhotoImage(data=r'R0lGODlhCwALAPECAAAAAH9/f////6usrSH5BAEAAA'+
+                         r'MALAAAAAALAAsAAAIanI8Wy6wCowhPQjqCvdVivX3b1Elf0y'+
+                         r'TqUAAAOw==')
+    return {"data":dataset, "attr":node}
+
+def value_to_text(value, fmt):
+    """Write out a multidimensional array as text"""
+    if isinstance(value, str):
+        return value
+    elif value.dtype.kind == "O" and value.shape == ():
+        # Value is a single object reference
+        return value_to_text(value[()], fmt)
+    elif len(value.shape) > 0:
+        # Value is an array, possibly of objects
+        s = []
+        for i in range(value.shape[0]):
+            s.append(value_to_text(value[i,...], fmt))
+        return "[" + (",".join(s)) + "]"
+    else:
+        # Value is a non-object scalar
+        if fmt is None:
+            return str(value)
+        else:
+            return fmt % value
+
+def type_string(dtype):
+    """Return a description of a HDF5 data type"""
+    if h5py.check_string_dtype(dtype) is not None:
+        return "vlen string"
+    elif h5py.check_vlen_dtype(dtype) is not None:
+        return "vlen "+str(h5py.check_vlen_dtype(dtype))
+    else:
+        return str(dtype)
+
+class MyTreeNode(TreeNode):
+    def __init__(self, canvas, parent, item):
+        TreeNode.__init__(self, canvas, parent, item)
+        self.item = item
+    def select(self, event=None, value=None):
+        if self.item.item_type == ITEM_DATASET:
+            if value is None:
+                self.selected = not(self.selected)
+            else:
+                self.selected = value
+            self.canvas.delete(self.image_id)
+            self.drawicon()
+            self.drawtext()
+            global selected_nodes
+            my_id = (self.item.file.filename, self.item.path, self.item.field)
+            if self.selected:
+                selected_nodes[my_id] = self.item
+            else:
+                if my_id in selected_nodes:
+                    del selected_nodes[my_id]
+                
+    def geticonimage(self, name):
+        global icon_images
+        if icon_images is None:
+            icon_images = get_icons()
+        if name in icon_images:
+            return icon_images[name]
+        else:
+            return TreeNode.geticonimage(self, name)
+
+class TopItem(TreeItem):
+    """Top level tree widget item with one sub-item for each file"""
+    def __init__(self):
+        self.nodes = []
+        self.item_type = ITEM_TOP
+    def GetText(self):
+        return "HDF5 files"
+    def IsExpandable(self):
+        return True
+    def GetSubList(self):
+        return self.nodes
+    def add_child(self, child):
+        self.nodes.append(child)
+    
+class AttrItem(TreeItem):
+    """Tree widget item which represents a HDF5 attribute"""
+    def __init__(self, name, value):
+        self.name  = name
+        self.value = value_to_text(value, None)
+        self.item_type = ITEM_ATTR
+    def GetText(self):
+        return self.name + " = " + self.value
+    def IsExpandable(self):
+        return False
+    def GetSubList(self):
+        return []
+    def GetIconName(self):
+        return "attr"
+
+class DescrItem(TreeItem):
+    """Tree widget item which represents a description of a dataset"""
+    def __init__(self, id, dtype, field):
+        self.item_type = ITEM_DESCR
+        self.text = "shape="+str(id.shape)+", dtype="+type_string(dtype)
+        if field is None:
+            # Don't show layout and chunks for dataset sub-fields
+            plist = id.get_create_plist()
+            if plist.get_layout() == h5py.h5d.CHUNKED:
+                chunk = plist.get_chunk()
+                self.text += ", chunks="+str(chunk)
+            for i in range(plist.get_nfilters()):
+                (code, flags, value, name) = plist.get_filter(i)
+                self.text += ", "+str(name.decode())
+    def GetText(self):
+        return self.text
+    def IsExpandable(self):
+        return False
+    def GetSubList(self):
+        return []
+    def GetIconName(self):
+        return "attr"
+
+#
+# Use low level interface instead of instantiating h5py objects for every
+# tree item in here to avoid performance issue with datasets with
+# many chunks if using h5py < 2.0.
+#
+class HDF5Item(TreeItem):
+    """Tree widget item which represents a HDF5 file, group or dataset"""
+    def __init__(self, id, name="", file=None, is_file=False, parent_name="", field=None):
+        self.id      = id
+        self.file    = file
+        self.is_file = is_file
+        self.name    = name
+        if field is None:
+            self.path    = (parent_name+"/"+name).lstrip("/")
+        else:
+            self.path = parent_name
+        self.field   = field if field is not None else ()
+        if isinstance(self.id, h5py.h5d.DatasetID):
+            self.item_type = ITEM_DATASET
+        else:
+            self.item_type = ITEM_GROUP
+    def GetText(self):
+        if self.is_file:
+            return self.file.filename
+        else:
+            return self.name
+    def IsExpandable(self):
+        return True
+    def GetSubList(self):
+        sublist = []
+        if isinstance(self.id, h5py.h5d.DatasetID):
+            # This is a dataset or a sub-field of a dataset. Find its data type.
+            dtype = self.id.get_type().dtype
+            if self.field is not None:
+                for f in self.field:
+                    dtype = dtype[f]
+            # Add its description to the tree.
+            sublist.append(DescrItem(self.id, dtype, field=self.field))
+            # Check for compound type and add sub fields
+            if dtype.fields is not None:
+                for field in dtype.fields:
+                    sublist.append(HDF5Item(self.id, field, self.file,
+                                            parent_name=self.path, field=self.field+(field,)))
+        if isinstance(self.id, h5py.h5g.GroupID):
+            # This is a group. Add its members to the tree.
+            n = self.id.get_num_objs()
+            for i in range(n):
+                name = self.id.get_objname_by_idx(i)
+                ot   = self.id.get_objtype_by_idx(i)
+                if ot == h5py.h5g.DATASET:
+                    subid = h5py.h5d.open(self.id, name)
+                    sublist.append(HDF5Item(subid, name.decode(), self.file,
+                                            parent_name=self.path)) 
+                elif ot == h5py.h5g.GROUP:
+                    subid = h5py.h5g.open(self.id, name)
+                    sublist.append(HDF5Item(subid, name.decode(), self.file,
+                                            parent_name=self.path))
+                else:
+                    # Unsupported object type
+                    pass
+        # Get number of attributes on this object
+        try:
+            nr_attrs = h5py.h5a.get_num_attrs(self.id)
+        except ValueError:
+            # Workaround for old HDF5 versions: need to open root group
+            if self.is_file:
+                nr_attrs = h5py.h5a.get_num_attrs(self.id["/"])
+            else:
+                raise
+        # Add attributes of this object to the tree
+        if nr_attrs > 0: 
+            for name in self.get_data().attrs.keys():
+                try:
+                    value = self.get_data().attrs[name]
+                except IOError as e:
+                    pass
+                else:
+                    sublist.append(AttrItem(name, value))
+
+        return sublist
+    def GetIconName(self):
+        if self.item_type == ITEM_DATASET:
+            return "data"
+        else:
+            return "folder"
+    def get_data(self):
+        if len(self.path) > 0:
+            return self.file[self.path]
+        else:
+            return self.file
+
+class AutoScrollbar(Scrollbar):
+    def set(self, lo, hi):
+        if float(lo) <= 0.0 and float(hi) >= 1.0:
+            self.tk.call("grid", "remove", self)
+        else:
+            self.grid()
+        Scrollbar.set(self, lo, hi)
+
+class DataWindow():
+    def __init__(self, data):
+
+        global windows_open
+        windows_open += 1
+        
+        self.ids = data.keys()
+        self.filenames = [id[0] for id in self.ids]
+        self.names     = [id[1] for id in self.ids]
+        self.fields    = [id[2] for id in self.ids]
+        self.datasets  = [data[k].get_data() for k in self.ids]
+        self.nrows_tot = max([get_shape(arr)[0] for arr in self.datasets])
+
+        # Sort by file name
+        #idx = argsort(self.filenames)
+        sort_key = [fn + n for (fn,n) in zip(self.filenames, self.names)]
+        idx = argsort(sort_key)
+        self.names     = [self.names[i]     for i in idx]
+        self.filenames = [self.filenames[i] for i in idx]
+        self.datasets  = [self.datasets[i]  for i in idx]
+        self.fields    = [self.fields[i]    for i in idx]
+
+        # Figure out which columns need to show file names
+        self.nspan = [1 for _ in range(len(self.names))]
+        for i in range(len(self.names)-1):
+            for j in range(i+1,len(self.names)):
+                if self.filenames[i] == self.filenames[j]:
+                    self.nspan[i] += 1
+        for i in range(1, len(self.names)):
+            if self.filenames[i] == self.filenames[i-1]:
+                self.nspan[i] = 0
+                    
+        # Make top level window
+        self.toplevel = Toplevel()
+        self.toplevel.title("HDF5 data table")
+        self.toplevel.grid_rowconfigure(0,weight=1)
+        self.toplevel.grid_columnconfigure(0,weight=1)
+
+        # Add label, entry and button for jumping to row
+        lframe = Frame(self.toplevel)
+        lframe.grid(row=2, column=0, sticky=E+W)
+        lframe.columnconfigure(0, weight=0)
+        lframe.columnconfigure(1, weight=0)
+        lframe.columnconfigure(2, weight=1)
+        self.label = Label(lframe, text="Find index: ")
+        self.label.grid(row=2, column=0)
+        self.entry = Entry(lframe)
+        self.entry.grid(row=2, column=1)
+        self.entry.bind("<Return>", self.on_return)
+
+        # Add row count
+        self.rowlabel = Label(lframe, text="Total rows: "+str(self.nrows_tot))
+        self.rowlabel.grid(row=2, column=2, sticky=E)
+
+        # Make vertical scroll bar
+        self.vsbar = AutoScrollbar(self.toplevel, orient=VERTICAL,
+                                   command=self.scrollbar_moved)
+        self.vsbar.grid(row=0, column=1, sticky=N+S)
+
+        # Make horizontal scroll bar
+        hsbar = AutoScrollbar(self.toplevel, orient=HORIZONTAL)
+        hsbar.grid(row=1, column=0, sticky=E+W)
+
+        # Put a canvas in the window
+        self.canvas = Canvas(self.toplevel, xscrollcommand=hsbar.set)
+        self.canvas.grid(row=0, column=0, sticky=N+S+E+W)
+        hsbar.config(command=self.canvas.xview)                   
+
+        # Put a frame in the canvas
+        self.frame = Frame(self.canvas, bd=3, relief=SUNKEN)
+        self.frame.rowconfigure(2,weight=1)
+
+        # Calculate column widths
+        self.widths = []
+        for (name,dset,fname,fields) in zip(self.names,self.datasets,self.filenames,self.fields):
+            dtype = dset.dtype
+            for f in fields:
+                dtype = dtype[f]
+            width = get_width(dtype)
+            if len(dset.shape) > 1:
+                width += 3
+                for s in dset.shape[1:]:
+                    width *= s
+            if is_too_big(dset.shape):
+                width = min_width
+            width = min((max_width, max(width, len(name))))
+            self.widths.append(width)
+
+        # Calculate file name widths
+        self.fname_widths = []
+        icol = 0
+        for (width, nspan) in zip(self.widths, self.nspan):
+            w = 0
+            for i in range(nspan):
+                w += self.widths[icol+i]
+            self.fname_widths.append(w)
+            icol += 1
+            
+        # Add text widgets with column headings and data columns
+        font = tkFont.Font(family="courier", size=-16)
+        self.title = []
+        self.column = []
+        self.fname  = []
+        icol = 1
+        self.index  = Text(self.frame, font=font, wrap=NONE,
+                           width=12, bg="light cyan")
+        self.index.grid(row=2, column=0, sticky=W+N+S)
+        for (name,dset,fname,width,fw,fields) in zip(self.names,self.datasets,
+                                                     self.filenames, self.widths,
+                                                     self.fname_widths, self.fields):
+            if len(fields) == 0:
+                # Basic type - just display the dataset name
+                display_name = name
+            else:
+                # Compound type - need to display the field name too
+                display_name = "|".join((name,)+fields)
+
+            self.title.append(Text(self.frame, font=font, wrap=NONE,
+                                   width=width, height=1, bg="light yellow"))
+            self.title[-1].insert("end", display_name+" ")
+            self.title[-1].grid(row=1, column=icol,sticky=N+W)
+            self.title[-1].config(state=DISABLED)
+            self.column.append(Text(self.frame, font=font, wrap=NONE,
+                                    width=width, bg="white", setgrid=1))
+            self.column[-1].grid(row=2, column=icol, sticky=W+N+S)
+            self.column[-1].config(state=DISABLED)
+
+            # Add file name spanning columns from same file
+            if self.nspan[icol-1] > 0:
+                self.fname.append(Text(self.frame, font=font, wrap=NONE,
+                                       height=1, width=fw))
+                self.fname[-1].grid(row=0, column=icol, sticky=E+W,
+                                    columnspan=self.nspan[icol-1])
+                self.fname[-1].insert("end", fname)
+                self.fname[-1].config(state=DISABLED)
+
+            icol += 1
+
+        self.win = self.canvas.create_window(0, 0, anchor=NW,
+                                             window=self.frame)
+        self.frame.update_idletasks()
+        self.canvas.config(scrollregion=self.canvas.bbox("all"))
+
+        # Resize window if size of canvas changes
+        self.canvas.bind( '<Configure>', self.on_resize)
+
+        # Scrolling with keys/mouse wheel
+        # TODO - see if there's a way to get events passed to the parent
+        # frame instead of binding everything!
+        self.bind_scrolling(self.canvas)
+        self.bind_scrolling(self.frame)
+        self.bind_scrolling(self.index)
+        self.bind_scrolling(self.vsbar)
+        self.bind_scrolling(self.entry)
+        self.bind_scrolling(self.label)
+        for c in self.column:
+            self.bind_scrolling(c)
+        for t in self.title:
+            self.bind_scrolling(t)
+        for f in self.fname:
+            self.bind_scrolling(f)
+            
+        # Store height of one row of text
+        for i in range(1000):
+            self.index.insert("end", str(i)+"\n")
+        height = self.index.winfo_height()
+        nrows = int(self.index.index("@0,%d" % height).split(".")[0])
+        self.row_height = int(height//nrows)
+
+        # Set up initial content
+        self.update_nrows()
+        self.move_to_row(0)
+        self.frame.focus_set()
+
+        self.toplevel.protocol("WM_DELETE_WINDOW", self.destroy)
+
+    def bind_scrolling(self, widget):
+        widget.bind("<Prior>", self.pageup)
+        widget.bind("<Next>",  self.pagedown)
+        widget.bind("<Button-4>", self.pageup)
+        widget.bind("<Button-5>", self.pagedown)
+
+    def on_resize(self, event):
+        # Resize the data window
+        self.canvas.itemconfig(self.win, height=event.height)
+        self.canvas.update_idletasks()
+        self.update_nrows()
+        # Resize text widgets
+        #self.index.configure(height=self.nrows_display)
+        #for c in self.column:
+        #    c.configure(height=self.nrows_display)
+        # Redraw values
+        self.update_scroll_bar()
+        self.update_data()
+
+    def update_data(self):
+        # Redraw array index
+        self.index.config(state=NORMAL)
+        self.index.delete(1.0, END)
+        self.ilast = min(self.ifirst + self.nrows_display, self.nrows_tot)
+        for i in range(self.ifirst, self.ilast):
+            if i < self.ilast -1:
+                self.index.insert("end", str(i)+"\n")
+            else:
+                self.index.insert("end", str(i))
+        self.index.config(state=DISABLED)
+        # Redraw columns
+        for col, dat, fields in zip(self.column, self.datasets, self.fields):
+            # Find data type of this column
+            dtype = dat.dtype
+            for f in fields:
+                dtype = dtype[f]
+            # Clear this column
+            col.config(state=NORMAL)
+            col.delete(1.0, END)
+            # Check the data is something we can sensibly display
+            if is_too_big(dat.shape):
+                col.insert("end", "Unable to display!")
+            else:
+                # Check at least one value is to be displayed
+                if len(dat.shape) == 0:
+                    # Its a scalar, so only visible if scrolled up to the top
+                    if self.ifirst == 0:
+                        # Decide format to use
+                        fmt = get_format(dat.dtype)
+                        col.insert("end", value_to_text(dat[...], fmt)+"\n")
+                else:
+                    # Its an array
+                    if self.ifirst < dat.shape[0] and self.ilast > self.ifirst:
+                        ilast = min(self.ilast, dat.shape[0])
+                        # Read the data
+                        d = dat[self.ifirst:ilast, ...]
+                        for f in fields:
+                            d = d[f]
+                        # Decide format to use
+                        fmt = get_format(d.dtype)
+                        # Add values to display
+                        for i in range(d.shape[0]):
+                            if i < d.shape[0] - 1:
+                                col.insert("end", value_to_text(d[i,...],fmt)+
+                                           "\n")
+                            else:
+                                col.insert("end", value_to_text(d[i,...],fmt))
+  
+            col.config(state=DISABLED)
+
+
+    def update_scroll_bar(self):
+        lo = float(self.ifirst) / float(self.nrows_tot)
+        hi = float(self.ilast)  / float(self.nrows_tot)
+        self.vsbar.set(lo, hi)
+
+    def scrollbar_moved(self, action, *args):
+        if action == "moveto":
+            offset = float(args[0])
+            i = int(offset * self.nrows_tot)
+        elif action == "scroll":
+            step = int(args[0])
+            what = args[1]
+            if what == "units":
+                i = self.ifirst + step
+            else:
+                i = self.ifirst + step * self.nrows_display
+        self.move_to_row(i)
+    
+    def on_return(self, event):
+        try:
+            i = int(self.entry.get())
+        except ValueError:
+            pass
+        else:
+            self.move_to_row(i)
+
+    def move_to_row(self, i):
+        self.ifirst = i
+        if self.ifirst >= self.nrows_tot - self.nrows_display:
+            self.ifirst = self.nrows_tot - self.nrows_display
+        if self.ifirst < 0:
+            self.ifirst = 0
+        self.ilast = min((self.ifirst + self.nrows_display,
+                          self.nrows_tot))
+        self.update_scroll_bar()
+        self.update_data()
+
+    def pageup(self, event):
+        self.move_to_row(self.ifirst - self.nrows_display)
+        return "break"
+
+    def pagedown(self, event):
+        self.move_to_row(self.ifirst + self.nrows_display)
+        return "break"
+
+    def destroy(self, *args):
+        global windows_open
+        global root
+        self.toplevel.destroy()
+        windows_open -= 1
+        if windows_open == 0:
+            root.destroy()
+
+    def update_nrows(self):
+        n = ((self.index.winfo_height()-0.5*self.row_height)/
+             self.row_height)
+        n = max((0, n))
+        self.nrows_display = int(n)
+        
+class BrowserWindow():
+    def __init__(self, filenames):
+
+        global windows_open
+        windows_open += 1
+
+        nloaded = 0
+
+        self.windows = []
+
+        self.toplevel = Toplevel()
+        self.toplevel.title("HDF5 Browser")
+        self.toplevel.config(bg="white")
+
+        canvas = Canvas(self.toplevel)
+        canvas.config(bg='white')
+        canvas.grid(sticky=N+S+E+W)
+        top_item = TopItem()
+        top_node = MyTreeNode(canvas, None, top_item)
+
+        vsbar = AutoScrollbar(self.toplevel)
+        vsbar.config(command=canvas.yview)                   
+        canvas.config(yscrollcommand=vsbar.set)              
+        vsbar.grid(row=0, column=1, sticky=N+S)
+
+        hsbar = AutoScrollbar(self.toplevel, orient=HORIZONTAL)
+        hsbar.config(command=canvas.xview)                   
+        canvas.config(xscrollcommand=hsbar.set)              
+        hsbar.grid(row=1, column=0, sticky=E+W)
+        canvas.bind('<4>', lambda event : canvas.yview('scroll', -1, 'units'))
+        canvas.bind('<5>', lambda event : canvas.yview('scroll', 1, 'units'))
+
+        self.toplevel.grid_rowconfigure(0, weight=1)
+        self.toplevel.grid_columnconfigure(0, weight=1)
+        
+        for fname in sys.argv[1:]:
+            try:
+                f = h5py.File(fname, "r")
+            except IOError:
+                print("Unable to read file: ", fname)
+                continue
+            top_item.add_child(HDF5Item(f.id, file=f, is_file=True))
+            nloaded += 1
+
+        if nloaded == 0:
+            raise IOError("No files loaded!")
+
+        bframe = Frame(self.toplevel, relief="raised", borderwidth=2)
+        bframe.grid(row=2, column=0, sticky=N+S+E+W)
+        b1 = Button(bframe, text="Open selected datasets as table",
+                    command=self.open_datasets)
+        b1.grid(row=0, column=0)
+
+        b2 = Button(bframe, text="Deselect all",
+                    command=self.deselect_all)
+        b2.grid(row=0, column=1)
+        
+        self.canvas   = canvas
+        self.top_node = top_node
+        self.top_item = top_item
+        
+        top_node.update()
+        top_node.expand()
+
+        self.toplevel.protocol("WM_DELETE_WINDOW", self.destroy)
+
+    def open_datasets(self):
+        if len(selected_nodes.keys()) > 0:
+            self.windows.append(DataWindow(selected_nodes))
+
+    def deselect_all(self):
+        global selected_nodes
+        self.top_node.deselecttree()
+        selected_nodes = {}
+
+    def destroy(self, *args):
+        global windows_open
+        global root
+        self.toplevel.destroy()
+        windows_open -= 1
+        if windows_open == 0:
+            root.destroy()
+
+def main():
+            
+    global root
+    global windows_open
+
+    root = Tk()
+    root.withdraw()
+    
+    windows_open = 0
+
+    filenames = sys.argv[1:]
+    mainwin = BrowserWindow(filenames)
+
+    root.mainloop()

h5tab-1.0.0/pyproject.toml

@@ -0,0 +1,18 @@
+[build-system]
+requires = ["setuptools>=61.0"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "h5tab"
+version = "1.0.0"
+authors = [
+  { name="John Helly", email="j.c.helly@durham.ac.uk" },
+]
+description = "Simple HDF5 file viewer."
+requires-python = ">=3.7"
+dependencies = [
+    "h5py",
+]
+
+[project.scripts]
+h5tab = "h5tab:main"

h5tab-1.0.0/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+