xedocs 0.2.37__py3-none-any.whl → 0.2.39__py3-none-any.whl

xedocs/__init__.py

@@ -2,7 +2,7 @@
 
 __author__ = """Yossi Mosbacher"""
 __email__ = "joe.mosbacher@gmail.com"
-__version__ = "0.2.37"
+__version__ = "0.2.39"
 
 
 import logging

xedocs/dispatchers.py

@@ -54,8 +54,8 @@ def read_csv_files(path, **kwargs) -> List[dict]:
     docs =[]
     
     with fsspec.open_files(path, **kwargs) as fs:
-         kwargs = rframe.utils.filter_kwargs(pd.read_csv, kwargs)
-         for f in fs:
+        kwargs = rframe.utils.filter_kwargs(pd.read_csv, kwargs)
+        for f in fs:
             ds = pd.read_csv(f, **kwargs).to_dict(orient='records')
             docs.extend(ds)
     return docs

xedocs/utils.py

@@ -134,20 +134,38 @@ class LazyFileAccessor(DataAccessor):
     def iter_path_records(self, ignore_paths=(), **labels):
         for path in self.urlpaths:
             glob_patttern = self.format_to_glob(path)
-            fs, _, fpaths = fsspec.get_fs_token_paths(glob_patttern,
-                                                   storage_options=self.storage_options)
-            pattern = path.replace(f"{fs.protocol}://", "")
+            fs, _, fpaths = fsspec.get_fs_token_paths(glob_patttern, storage_options=self.storage_options)
+
+            # Always define original_protocol
+            original_protocol = fs.protocol if fs.protocol else 'file'
+            if isinstance(original_protocol, tuple):
+                original_protocol = original_protocol[0]
+            else:
+                original_protocol = original_protocol
+
+            # Debug prints
+            print(f"\n[DEBUG]")
+            print(f"Processing path: {path}")
+            print(f"fs.protocol: {fs.protocol}")
+            print(f"Original protocol: {original_protocol}")
+            print(f"Glob pattern: {glob_patttern}")
+            print(f"File paths: {fpaths}")
+    
+            pattern = path.replace(f"{original_protocol}://", "")
             pattern = parse.compile(self.glob_to_format(pattern))
+    
+            # More debug prints
+            print(f"Parse pattern: {pattern}")
+    
             loaded = set(ignore_paths)
-
+    
             for fpath in fpaths:
                 if fpath in loaded:
                     continue
                 r = pattern.parse(fpath)
                 if r is None:
                     continue
-                for k,vs in labels.items():
-                    
+                for k, vs in labels.items():
                     if vs is None:
                         continue
                     if not isinstance(vs, list):
@@ -164,10 +182,11 @@ class LazyFileAccessor(DataAccessor):
                     if label not in vs:
                         break
                 else:
-                    records = read_files(fpath, protocol=fs.protocol, **fs.storage_options)
+                    records = read_files(fpath, protocol=original_protocol, **fs.storage_options)
                     yield fpath, records
                     self.loaded.add(fpath)
 
+
     def load_files(self, **labels):
         index_fields = list(self.schema.get_index_fields())
         if len(index_fields) == 1:

{xedocs-0.2.37.dist-info → xedocs-0.2.39.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: xedocs
-Version: 0.2.37
+Version: 0.2.39
 Summary: Top-level package for xedocs.
 Author: Yossi Mosbacher
 Author-email: joe.mosbacher@gmail.com

{xedocs-0.2.37.dist-info → xedocs-0.2.39.dist-info}/RECORD

@@ -1,4 +1,4 @@
-xedocs/__init__.py,sha256=MZpxJ9AELOsLt2GtU8QL0weUt8ocdvBhQuaOGfv68xs,1054
+xedocs/__init__.py,sha256=wH2b2ynVVq-ZQwqENUSAJhPA_Z7fD8mBQ_9woNgkbOI,1054
 xedocs/_settings.py,sha256=_Fjj5XJBHBrN6N1Dqs_JzovbEcF6--5BGnAggK1ksck,4199
 xedocs/_straxen_plugin.py,sha256=bxKWkVDNCS0Mah_GGNEvE8TCk-ibrY8xJbO0UMvfajM,1904
 xedocs/cli.py,sha256=1oZXa4Iqs-ANRTABZP8K6e3oSIatKnJNhQ1DZ4rsNrc,3493
@@ -10,7 +10,7 @@ xedocs/data_locations/data_folder.py,sha256=reCX7CgZMjXJYWdsuMI3JLmSgwd2RTMjPO7i
 xedocs/data_locations/github.py,sha256=Tg3Mwo1vUE7Z3gzhjgEtSDlEtUx7F0gY8U-Uj89GIZI,2683
 xedocs/data_locations/mongodb.py,sha256=2tVlHgSlw7mVQ9zRtLz2DPZgyJ7d5JNYMZm0VO_4VDo,3203
 xedocs/databases.py,sha256=t8WIZt0zf_DHvie2NG-q0CNi7zsXgOVyvrKunCz0lhc,1799
-xedocs/dispatchers.py,sha256=GJ5F3mJYMf95wNyFrrsrNevYqfuEOEShSdsPo9gkBD4,2664
+xedocs/dispatchers.py,sha256=zY64OzMXwcJyuvM9pgsciMcmpaML7nPOIRS1r_lG2p8,2662
 xedocs/entrypoints.py,sha256=1y61KiT0gaTWGlM-PHawXSSzI3ayEhjOucNAkgH_Aps,2098
 xedocs/json_records.py,sha256=lUPUkaDkRvM_KSJ1nNQzB0HgAvKxWPxMUFgtQoFJ1pE,1768
 xedocs/regex_dispatcher.py,sha256=LXUN4nKRCnVX5b7_KOTMuUh_jCPrVm1W3UusRCjnrac,2481
@@ -78,12 +78,12 @@ xedocs/schemas/pmt_data/pmt_installs.py,sha256=zVz04RbVvS9DpnM6kp8Ql14g1aOtRsqWA
 xedocs/schemas/pmt_data/pmts.py,sha256=VTDpDPz0HdTlfkPypN6CFjmuxEUNJj0-cC6uqRRuXZs,271
 xedocs/schemas/pmt_data/voltage_changes.py,sha256=ieTNCSoE4A3VHrWZ3Hk5HN_q5Nrwu04jwqji7bgbJto,428
 xedocs/schemas/pmt_data/voltage_settings.py,sha256=L2cScDQNviWgAlWvMRjnIEgsP0urVg0gYWAxK3FrFPc,428
-xedocs/utils.py,sha256=3OmrYetj83uhy0x5GHzk-_z_vSQeFR50X_9R51B3oWg,8398
+xedocs/utils.py,sha256=KYdyye-avgteZiqwDgIB4fx4P3Kg6scENbxoD7MLcXw,9049
 xedocs/widgets.py,sha256=l2pgQaWqycZQF5pF6eEeLuHdqjVBkJ8OB-o-mcivMl4,28810
 xedocs/xedocs.py,sha256=Dk-ld5X6qB6hrp40svDkJX_9K0MKkZTxal-fLVQpIz8,6949
 xedocs/xenon_config.py,sha256=QQlg2icofZWq_UAh5kDSuViai0uP_3ClozNNv6mpx3s,2674
-xedocs-0.2.37.dist-info/AUTHORS.rst,sha256=7XNtvXT_irnC5nzau-Al_VVmmhk8EgW811pPs8OXgBo,164
-xedocs-0.2.37.dist-info/METADATA,sha256=yKAXo7x0J8rmOb47S-clqvOXHpTc2qaN0jAIF76ZR-k,5827
-xedocs-0.2.37.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
-xedocs-0.2.37.dist-info/entry_points.txt,sha256=AfKuejqaCB-nA4rYbuWtMcR0d-t4nNzrM3s8V_e7YSk,92
-xedocs-0.2.37.dist-info/RECORD,,
+xedocs-0.2.39.dist-info/AUTHORS.rst,sha256=7XNtvXT_irnC5nzau-Al_VVmmhk8EgW811pPs8OXgBo,164
+xedocs-0.2.39.dist-info/METADATA,sha256=rS02k5-w_QZz3FPZTg5jJaO0N0gTF8kXZJYj8TfG7HU,5827
+xedocs-0.2.39.dist-info/WHEEL,sha256=fGIA9gx4Qxk2KDKeNJCbOEwSrmLtjWCwzBz351GyrPQ,88
+xedocs-0.2.39.dist-info/entry_points.txt,sha256=AfKuejqaCB-nA4rYbuWtMcR0d-t4nNzrM3s8V_e7YSk,92
+xedocs-0.2.39.dist-info/RECORD,,

{xedocs-0.2.37.dist-info → xedocs-0.2.39.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: poetry-core 2.0.1
+Generator: poetry-core 2.1.2
 Root-Is-Purelib: true
 Tag: py3-none-any