dcicutils 8.16.6.1b1__py3-none-any.whl → 8.16.6.1b3__py3-none-any.whl

dcicutils/data_readers.py

@@ -31,15 +31,22 @@ class RowReader(abc.ABC):
         self.open()
 
     def __iter__(self) -> Iterator:
+        header = self._iter_header()
         for row in self.rows:
             self.row_number += 1
             if self.is_comment_row(row):
                 continue
             if self.is_terminating_row(row):
                 break
-            if len(self.header) < len(row):  # Row values beyond what there are headers for are ignored.
+            if len(header) < len(row):  # Row values beyond what there are headers for are ignored.
                 self._warning_extra_values.append(self.row_number)
-            yield {column: self.cell_value(value) for column, value in zip(self.header, row)}
+            yield self._iter_mapper({column: self.cell_value(value) for column, value in zip(header, row)})
+
+    def _iter_header(self) -> List[str]:
+        return self.header
+
+    def _iter_mapper(self, row: dict) -> List[str]:
+        return row
 
     def _define_header(self, header: List[Optional[Any]]) -> None:
         self.header = []

dcicutils/structured_data.py

@@ -59,7 +59,7 @@ class StructuredDataSet:
                  progress: Optional[Callable] = None,
                  validator_hook: Optional[Callable] = None,
                  validator_sheet_hook: Optional[Callable] = None,
-                 row_reader_mapper: Optional[Callable] = None,
+                 excel_class: Optional[Excel] = None,
                  debug_sleep: Optional[str] = None) -> None:
         self._progress = progress if callable(progress) else None
         self._data = {}
@@ -80,7 +80,7 @@ class StructuredDataSet:
         self._merge = True if merge is True else False  # New merge functionality (2024-05-25)
         self._validator_hook = validator_hook if callable(validator_hook) else None
         self._validator_sheet_hook = validator_sheet_hook if callable(validator_sheet_hook) else None
-        self._row_reader_mapper = row_reader_mapper if callable(row_reader_mapper) else None
+        self._excel_class = excel_class if excel_class is not None and issubclass(excel_class, Excel) else Excel
         self._debug_sleep = None
         if debug_sleep:
             try:
@@ -306,7 +306,7 @@ class StructuredDataSet:
     def _load_excel_file(self, file: str) -> None:
         def get_counts() -> Tuple[int, int]:
             nonlocal file
-            excel = Excel(file)
+            excel = self._excel_class(file)
             nrows = 0
             for sheet_name in excel.sheet_names:
                 for row in excel.sheet_reader(sheet_name):
@@ -316,7 +316,8 @@ class StructuredDataSet:
             nrows, nsheets = get_counts()
             self._progress({PROGRESS.LOAD_START: PROGRESS.NOW(),
                             PROGRESS.LOAD_COUNT_SHEETS: nsheets, PROGRESS.LOAD_COUNT_ROWS: nrows})
-        excel = Excel(file)  # Order the sheet names by any specified ordering (e.g. ala snovault.loadxl).
+        excel = self._excel_class(file)
+        # Order the sheet names by any specified ordering (e.g. ala snovault.loadxl).
         order = {Schema.type_name(key): index for index, key in enumerate(self._order)} if self._order else {}
         for sheet_name in sorted(excel.sheet_names, key=lambda key: order.get(Schema.type_name(key), sys.maxsize)):
             self._load_reader(excel.sheet_reader(sheet_name), type_name=Schema.type_name(sheet_name))
@@ -376,10 +377,7 @@ class StructuredDataSet:
         schema = None
         noschema = False
         structured_row_template = None
-        sheet_name = reader.sheet_name if self._row_reader_mapper and hasattr(reader, "sheet_name") else None
         for row in reader:
-            if self._row_reader_mapper and (not isinstance(row := self._row_reader_mapper(row, sheet_name), dict)):
-                continue
             self._nrows += 1
             if self._debug_sleep:
                 time.sleep(float(self._debug_sleep))

{dcicutils-8.16.6.1b1.dist-info → dcicutils-8.16.6.1b3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: dcicutils
-Version: 8.16.6.1b1
+Version: 8.16.6.1b3
 Summary: Utility package for interacting with the 4DN Data Portal and other 4DN resources
 Home-page: https://github.com/4dn-dcic/utils
 License: MIT

{dcicutils-8.16.6.1b1.dist-info → dcicutils-8.16.6.1b3.dist-info}/RECORD

@@ -10,7 +10,7 @@ dcicutils/common.py,sha256=U6tedhp6FFl9EUGoMqiMAqON_WqGJ_317JE_Tr4dXDs,3926
 dcicutils/contribution_scripts.py,sha256=0k5Gw1TumcD5SAcXVkDd6-yvuMEw-jUp5Kfb7FJH6XQ,2015
 dcicutils/contribution_utils.py,sha256=vYLS1JUB3sKd24BUxZ29qUBqYeQBLK9cwo8x3k64uPg,25653
 dcicutils/creds_utils.py,sha256=64BbIfS90T1eJmmQJrDyfrRa3V2F1x7T8lOrEeFfqJY,11127
-dcicutils/data_readers.py,sha256=6EMrY7TjDE8H7bA_TCWtpLQP7slJ0YTL77_dNh6e7sg,7626
+dcicutils/data_readers.py,sha256=CgxWPB2hJzFfKqVUWL_0fERe4CZRMvt-5Afo9lqEt74,7813
 dcicutils/data_utils.py,sha256=k2OxOlsx7AJ6jF-YNlMyGus_JqSUBe4_n1s65Mv1gQQ,3098
 dcicutils/datetime_utils.py,sha256=xaJ9rzSsk4uBPidwV2sByrl2xcQiFWXTnx6HLpt8EcY,13650
 dcicutils/deployment_utils.py,sha256=6Sv1cM9T944d_9CTJt8O2RPtpw9MiTIPKw_DgYLOnCU,69916
@@ -66,7 +66,7 @@ dcicutils/secrets_utils.py,sha256=8dppXAsiHhJzI6NmOcvJV5ldvKkQZzh3Fl-cb8Wm7MI,19
 dcicutils/sheet_utils.py,sha256=VlmzteONW5VF_Q4vo0yA5vesz1ViUah1MZ_yA1rwZ0M,33629
 dcicutils/snapshot_utils.py,sha256=YDeI3vD-MhAtHwKDzfEm2q-n3l-da2yRpRR3xp0Ah1M,23021
 dcicutils/ssl_certificate_utils.py,sha256=F0ifz_wnRRN9dfrfsz7aCp4UDLgHEY8LaK7PjnNvrAQ,9707
-dcicutils/structured_data.py,sha256=AnX8opkCw-lfAnd4KPH-OIneh2ZB2oUYQCo7GWmASh4,66789
+dcicutils/structured_data.py,sha256=fnqrB5d59EoD6AcilUUmR2THGdjYeNK17eLoAhBCT4w,66580
 dcicutils/submitr/progress_constants.py,sha256=5bxyX77ql8qEJearfHEvsvXl7D0GuUODW0T65mbRmnE,2895
 dcicutils/submitr/ref_lookup_strategy.py,sha256=VJN-Oo0LLna6Vo2cu47eC-eU-yUC9NFlQP29xajejVU,4741
 dcicutils/task_utils.py,sha256=MF8ujmTD6-O2AC2gRGPHyGdUrVKgtr8epT5XU8WtNjk,8082
@@ -75,8 +75,8 @@ dcicutils/trace_utils.py,sha256=g8kwV4ebEy5kXW6oOrEAUsurBcCROvwtZqz9fczsGRE,1769
 dcicutils/validation_utils.py,sha256=cMZIU2cY98FYtzK52z5WUYck7urH6JcqOuz9jkXpqzg,14797
 dcicutils/variant_utils.py,sha256=2H9azNx3xAj-MySg-uZ2SFqbWs4kZvf61JnK6b-h4Qw,4343
 dcicutils/zip_utils.py,sha256=_Y9EmL3D2dUZhxucxHvrtmmlbZmK4FpSsHEb7rGSJLU,3265
-dcicutils-8.16.6.1b1.dist-info/LICENSE.txt,sha256=IrWHOFtMb2inTgr2ZqwQriRC8H3uMQfQVO6lniaaoE0,1102
-dcicutils-8.16.6.1b1.dist-info/METADATA,sha256=sGAPNyW77aQY-rWnX0ycgt9Ya-qDir36Dm-UNjiku1w,3493
-dcicutils-8.16.6.1b1.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-dcicutils-8.16.6.1b1.dist-info/entry_points.txt,sha256=W6kEWdUJk9tQ4myAgpehPdebcwvCAZ7UgB-wyPgDUMg,335
-dcicutils-8.16.6.1b1.dist-info/RECORD,,
+dcicutils-8.16.6.1b3.dist-info/LICENSE.txt,sha256=IrWHOFtMb2inTgr2ZqwQriRC8H3uMQfQVO6lniaaoE0,1102
+dcicutils-8.16.6.1b3.dist-info/METADATA,sha256=NMooRRWHCGqrfzf8D0lOhEuFyf7jQSM_TmgC4o3CfTk,3493
+dcicutils-8.16.6.1b3.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+dcicutils-8.16.6.1b3.dist-info/entry_points.txt,sha256=W6kEWdUJk9tQ4myAgpehPdebcwvCAZ7UgB-wyPgDUMg,335
+dcicutils-8.16.6.1b3.dist-info/RECORD,,