Merge pull request #41 from jupyterlab/n-dim-support-w-parsing

n-dimensional support via parse_index

.gitignore

@@ -1,3 +1,4 @@
+.autoversion
 lib/
 node_modules/
 
@@ -18,6 +19,7 @@ MANIFEST
 /scratch/*.cutie
 /scratch/*.h5
 /scratch/*.hdf5
+/scratch/Untitled*
 
 # temp/system files
 *~
@@ -35,3 +37,7 @@ MANIFEST
 *.code-workspace
 .history
 .vscode
+
+# tmp dev stuff
+json.py
+json.ts

jupyterlab_hdf/baseHandler.py

@@ -4,8 +4,8 @@
 # Distributed under the terms of the Modified BSD License.
 
 import h5py
-import json
 import os
+import simplejson
 import traceback
 from tornado import gen
 from tornado.httpclient import HTTPError
@@ -14,6 +14,7 @@
 from notebook.utils import url_path_join
 
 # from .config import HdfConfig
+from .exception import JhdfError
 
 __all__ = ['HdfBaseManager', 'HdfFileManager', 'HdfBaseHandler']
 
@@ -22,31 +23,38 @@
 class HdfBaseManager:
     """Base class for implementing HDF5 handling
     """
-    def __init__(self, notebook_dir):
+    def __init__(self, log, notebook_dir):
+        self.log = log
         self.notebook_dir = notebook_dir
 
-    def _get(self, f, uri, row, col):
+    def _get(self, f, uri, **kwargs):
         raise NotImplementedError
 
-    def get(self, relfpath, uri, row, col):
+    def get(self, relfpath, uri, **kwargs):
         def _handleErr(code, msg):
-            raise HTTPError(code, '\n'.join((
-                msg,
-                f'relfpath: {relfpath}, uri: {uri}, row: {row}, col: {col}'
-            )))
+            extra = dict((
+                ('relfpath', relfpath),
+                ('uri', uri),
+                *kwargs.items(),
+            ))
+
+            if isinstance(msg, dict):
+                # encode msg as json
+                msg['debugVars'] = {**msg.get('debugVars', {}), **extra}
+                msg = simplejson.dumps(msg, ignore_nan=True)
+            else:
+                msg = '\n'.join((
+                    msg,
+                    ', '.join(f'{key}: {val}' for key,val in extra.items())
+                ))
+
+            self.log.error(msg)
+            raise HTTPError(code, msg)
 
         if not relfpath:
             msg = f'The request was malformed; fpath should not be empty.'
             _handleErr(400, msg)
 
-        if row and not col:
-            msg = f'The request was malformed; row slice was specified, but col slice was empty.'
-            _handleErr(400, msg)
-
-        if col and not row:
-            msg = f'The request was malformed; col slice was specified, but row slice was empty.'
-            _handleErr(400, msg)
-
         fpath = url_path_join(self.notebook_dir, relfpath)
 
         if not os.path.exists(fpath):
@@ -61,7 +69,12 @@ def _handleErr(code, msg):
                        f'Error: {traceback.format_exc()}')
                 _handleErr(401, msg)
             try:
-                out = self._get(fpath, uri, row, col)
+                out = self._get(fpath, uri, **kwargs)
+            except JhdfError as e:
+                msg = e.args[0]
+                msg['traceback'] = traceback.format_exc()
+                msg['type'] = 'JhdfError'
+                _handleErr(400, msg)
             except Exception as e:
                 msg = (f'Found and opened file, error getting contents from object specified by the uri.\n'
                        f'Error: {traceback.format_exc()}')
@@ -72,11 +85,11 @@ def _handleErr(code, msg):
 class HdfFileManager(HdfBaseManager):
     """Implements base HDF5 file handling
     """
-    def _get(self, fpath, uri, row, col):
+    def _get(self, fpath, uri, **kwargs):
         with h5py.File(fpath, 'r') as f:
-            return self._getFromFile(f, uri, row, col)
+            return self._getFromFile(f, uri, **kwargs)
 
-    def _getFromFile(self, f, uri, row, col):
+    def _getFromFile(self, f, uri, **kwargs):
         raise NotImplementedError
 
 ## handler
@@ -90,20 +103,21 @@ def initialize(self, notebook_dir):
             raise NotImplementedError
 
         self.notebook_dir = notebook_dir
-        self.manager = self.managerClass(notebook_dir=notebook_dir)
+        self.manager = self.managerClass(log=self.log, notebook_dir=notebook_dir)
 
     @gen.coroutine
     def get(self, path):
         """Based on an api request, get either the contents of a group or a
         slice of a dataset and return it as serialized JSON.
         """
         uri = '/' + self.get_query_argument('uri').lstrip('/')
-        row = self.getQueryArguments('row', int)
-        col = self.getQueryArguments('col', int)
 
-        try:
-            self.finish(json.dumps(self.manager.get(path, uri, row, col)))
+        _kws = ('atleast_2d', 'ixstr', 'subixstr')
+        _vals = (self.get_query_argument(kw, default=None) for kw in _kws)
+        kwargs = {kw:(val if val else None) for kw,val in zip(_kws, _vals)}
 
+        try:
+            self.finish(simplejson.dumps(self.manager.get(path, uri, **kwargs), ignore_nan=True))
         except HTTPError as err:
             self.set_status(err.code)
             response = err.response.body if err.response else str(err.code)
@@ -112,8 +126,8 @@ def get(self, path):
                 err.message
             )))
 
-    def getQueryArguments(self, key, func=None):
-        if func is not None:
-            return [func(x) for x in self.get_query_argument(key).split(',')] if key in self.request.query_arguments else None
-        else:
-            return [x for x in self.get_query_argument(key).split(',')] if key in self.request.query_arguments else None
+    # def getQueryArguments(self, key, func=None):
+    #     if func is not None:
+    #         return [func(x) for x in self.get_query_argument(key).split(',')] if key in self.request.query_arguments else None
+    #     else:
+    #         return [x for x in self.get_query_argument(key).split(',')] if key in self.request.query_arguments else None

jupyterlab_hdf/contents.py

@@ -6,27 +6,29 @@
 import h5py
 
 from .baseHandler import HdfFileManager, HdfBaseHandler
-from .util import dsetContentDict, dsetDict, groupDict, uriJoin, uriName
+from .util import dsetContentDict, dsetDict, groupDict, jsonize, uriJoin, uriName
 
 __all__ = ['HdfContentsManager', 'HdfContentsHandler']
 
 ## manager
 class HdfContentsManager(HdfFileManager):
     """Implements HDF5 contents handling
     """
-    def _getFromFile(self, f, uri, row, col):
+    def _getFromFile(self, f, uri, ixstr, **kwargs):
         obj = f[uri]
 
         if isinstance(obj, h5py.Group):
             return [(groupDict if isinstance(val, h5py.Group) else dsetDict)
                         (name=name, uri=uriJoin(uri, name))
                     for name,val in obj.items()]
-        else:
+        elif isinstance(obj, h5py.Dataset):
             return dsetDict(
                 name=uriName(uri),
                 uri=uri,
-                content=dsetContentDict(obj, row, col),
+                content=jsonize(dsetContentDict(obj, ixstr)),
             )
+        else:
+            raise ValueError("unknown h5py obj: %s" % obj)
 
 
 ## handler

jupyterlab_hdf/data.py

@@ -3,8 +3,6 @@
 # Copyright (c) Jupyter Development Team.
 # Distributed under the terms of the Modified BSD License.
 
-import h5py
-
 from .baseHandler import HdfFileManager, HdfBaseHandler
 from .util import dsetChunk
 
@@ -15,8 +13,16 @@
 class HdfDataManager(HdfFileManager):
     """Implements HDF5 data handling
     """
-    def _getFromFile(self, f, uri, row, col):
-        return dsetChunk(f[uri], row, col)
+    def _getFromFile(self, f, uri, ixstr, subixstr=None, atleast_2d=False, **kwargs):
+        # # DEBUG: uncomment for logging
+        # from .util import dsetContentDict, parseSubindex
+        # logd = dsetContentDict(f[uri], ixstr=ixstr)
+        # logd['subixstr'] = subixstr
+        # if subixstr is not None:
+        #     logd['ixcompound'] = parseSubindex(ixstr, subixstr, f[uri].shape)
+        # self.log.info('{}'.format(logd))
+
+        return dsetChunk(f[uri], ixstr, subixstr=subixstr, atleast_2d=atleast_2d).tolist()
 
 
 ## handler

jupyterlab_hdf/exception.py

@@ -0,0 +1,9 @@
+# -*- coding: utf-8 -*-
+
+# Copyright (c) Jupyter Development Team.
+# Distributed under the terms of the Modified BSD License.
+
+__all__ = ['JhdfError']
+
+class JhdfError(Exception):
+    pass

jupyterlab_hdf/snippet.py

@@ -18,7 +18,7 @@
 class HdfSnippetManager(HdfBaseManager):
     """Implements HDF5 contents handling
     """
-    def _get(self, fpath, uri, row, col):
+    def _get(self, fpath, uri, ixstr, **kwargs):
         return dsetTemp.format(fpath=fpath, uri=uri)