Remove werkzeug requirement, replace generate_filename

docs/requirements.txt

@@ -7,6 +7,3 @@ portalocker>=2.4.0
 semantic_version>=2.9.0
 Sphinx>=5.2.2
 sphinx_rtd_theme>=1.0.0
-
-# This is just needed for secure_filename and should probably be removed
-Werkzeug>=2.1.2

hed/tools/__init__.py

@@ -40,7 +40,7 @@
 
 from .util.hed_logger import HedLogger
 from .util.data_util import get_new_dataframe, get_value_dict, replace_values, reorder_columns
-from .util.io_util import check_filename, generate_filename, extract_suffix_path, get_file_list, make_path
+from .util.io_util import check_filename, clean_filename, extract_suffix_path, get_file_list, make_path
 from .util.io_util import get_dir_dictionary, get_file_list, get_path_components, parse_bids_filename
 
 from .analysis import annotation_util

hed/tools/remodeling/dispatcher.py

@@ -8,7 +8,7 @@
 from hed.schema.hed_schema_io import get_schema
 from hed.tools.remodeling.backup_manager import BackupManager
 from hed.tools.remodeling.operations.valid_operations import valid_operations
-from hed.tools.util.io_util import generate_filename, extract_suffix_path, get_timestamp
+from hed.tools.util.io_util import clean_filename, extract_suffix_path, get_timestamp
 
 
 class Dispatcher:
@@ -66,7 +66,7 @@ def get_summaries(self, file_formats=['.txt', '.json']):
             file_base = context_item.context_filename
             if self.data_root:
                 file_base = extract_suffix_path(self.data_root, file_base)
-            file_base = generate_filename(file_base)
+            file_base = clean_filename(file_base)
             for file_format in file_formats:
                 if file_format == '.txt':
                     summary = context_item.get_text_summary(individual_summaries="consolidated")

hed/tools/util/io_util.py

@@ -1,8 +1,8 @@
 """Utilities for generating and handling file names."""
 
 import os
+import re
 from datetime import datetime
-from werkzeug.utils import secure_filename
 from hed.errors.exceptions import HedFileError
 
 TIME_FORMAT = '%Y_%m_%d_T_%H_%M_%S_%f'
@@ -92,39 +92,19 @@ def extract_suffix_path(path, prefix_path):
     return return_path
 
 
-def generate_filename(base_name, name_prefix=None, name_suffix=None, extension=None, append_datetime=False):
-    """ Generate a filename for the attachment.
+def clean_filename(filename):
+    """ Replaces invalid characters with under-bars
 
     Parameters:
-        base_name (str):   Name of the base, usually the name of the file that the issues were generated from.
-        name_prefix (str): Prefix prepended to the front of the base name.
-        name_suffix (str): Suffix appended to the end of the base name.
-        extension (str):   Extension to use.
-        append_datetime (bool): If True, append the current date-time to the base output filename.
+        filename (str):   source filename
 
     Returns:
-        str:  Name of the attachment other containing the issues.
-
-    Notes:
-        - The form prefix_basename_suffix + extension.
-
+        str:  The filename with anything but alphanumeric, period, hyphens, and under-bars removed.
     """
-
-    pieces = []
-    if name_prefix:
-        pieces = pieces + [name_prefix]
-    if base_name:
-        pieces.append(os.path.splitext(base_name)[0])
-    if name_suffix:
-        pieces = pieces + [name_suffix]
-    filename = "".join(pieces)
-    if append_datetime:
-        now = datetime.now()
-        filename = filename + '_' + now.strftime(TIME_FORMAT)[:-3]
-    if filename and extension:
-        filename = filename + extension
-
-    return secure_filename(filename)
+    if not filename:
+        return ""
+    out_name = re.sub(r'[^a-zA-Z0-9._-]+', '_', filename)
+    return out_name
 
 
 def get_dir_dictionary(dir_path, name_prefix=None, name_suffix=None, extensions=None, skip_empty=True,

requirements.txt

@@ -4,6 +4,3 @@ openpyxl>=3.0.9
 pandas>=1.3.5
 portalocker>=2.4.0
 semantic_version>=2.9.0
-
-# This is just needed for secure_filename and should probably be removed
-Werkzeug>=2.1.2

setup.cfg

@@ -33,7 +33,6 @@ install_requires =
     pytz
     semantic-version
     six
-    Werkzeug
 
 
 [options.packages.find]

tests/tools/util/test_io_util.py

@@ -1,7 +1,7 @@
 import os
 import unittest
 from hed.errors.exceptions import HedFileError
-from hed.tools.util.io_util import check_filename, extract_suffix_path, generate_filename, \
+from hed.tools.util.io_util import check_filename, extract_suffix_path, clean_filename, \
     get_dir_dictionary, get_file_list, get_path_components, parse_bids_filename, \
     _split_entity, get_allowed, get_filtered_by_element
 
@@ -46,77 +46,39 @@ def test_extract_suffix_path(self):
         suffix_path = extract_suffix_path('c:/myroot/temp.tsv', 'c:')
         self.assertTrue(suffix_path.endswith('temp.tsv'), "extract_suffix_path has the right path")
 
-    def test_generate_file_name(self):
-        file1 = generate_filename('mybase')
+    def test_clean_file_name(self):
+        file1 = clean_filename('mybase')
         self.assertEqual(file1, "mybase", "generate_file_name should return the base when other arguments not set")
-        file2 = generate_filename('mybase', name_prefix="prefix")
-        self.assertEqual(file2, "prefixmybase", "generate_file_name should return correct name when prefix set")
-        file3 = generate_filename('mybase', name_prefix="prefix", extension=".json")
-        self.assertEqual(file3, "prefixmybase.json", "generate_file_name should return correct name for extension")
-        file4 = generate_filename('mybase', name_suffix="suffix")
-        self.assertEqual(file4, "mybasesuffix", "generate_file_name should return correct name when suffix set")
-        file5 = generate_filename('mybase', name_suffix="suffix", extension=".json")
-        self.assertEqual(file5, "mybasesuffix.json", "generate_file_name should return correct name for extension")
-        file6 = generate_filename('mybase', name_prefix="prefix", name_suffix="suffix", extension=".json")
-        self.assertEqual(file6, "prefixmybasesuffix.json",
-                         "generate_file_name should return correct name for all set")
-        filename = generate_filename(None, name_prefix=None, name_suffix=None, extension=None)
-        self.assertEqual('', filename, "Return empty when all arguments are none")
-        filename = generate_filename(None, name_prefix=None, name_suffix=None, extension='.txt')
-        self.assertEqual('', filename,
-                         "Return empty when base_name, prefix, and suffix are None, but extension is not")
-        filename = generate_filename('c:/temp.json', name_prefix=None, name_suffix=None, extension='.txt')
-        self.assertEqual('c_temp.txt', filename,
-                         "Returns stripped base_name + extension when prefix, and suffix are None")
-        filename = generate_filename('temp.json', name_prefix='prefix_', name_suffix='_suffix', extension='.txt')
-        self.assertEqual('prefix_temp_suffix.txt', filename,
-                         "Return stripped base_name + extension when prefix, and suffix are None")
-        filename = generate_filename(None, name_prefix='prefix_', name_suffix='suffix', extension='.txt')
-        self.assertEqual('prefix_suffix.txt', filename,
-                         "Returns correct string when no base_name")
-        filename = generate_filename('event-strategy-v3_task-matchingpennies_events.json',
-                                     name_suffix='_blech', extension='.txt')
-        self.assertEqual('event-strategy-v3_task-matchingpennies_events_blech.txt', filename,
-                         "Returns correct string when base_name with hyphens")
-        filename = generate_filename('HED7.2.0.xml', name_suffix='_blech', extension='.txt')
-        self.assertEqual('HED7.2.0_blech.txt', filename, "Returns correct string when base_name has periods")
-
-    def test_generate_file_name_with_date(self):
-        file1 = generate_filename('mybase')
-        file1t = generate_filename('mybase', append_datetime=True)
-        self.assertGreater(len(file1t), len(file1), "generate_file_name generates a longer file when datetime is used.")
-        # TODO convert more of these tests.
-        # self.assertEqual(file1, "mybase", "generate_file_name should return the base when other arguments not set")
-        # file2 = generate_filename('mybase', name_prefix="prefix")
+        # file2 = clean_filename('mybase', name_prefix="prefix")
         # self.assertEqual(file2, "prefixmybase", "generate_file_name should return correct name when prefix set")
-        # file3 = generate_filename('mybase', name_prefix="prefix", extension=".json")
+        # file3 = clean_filename('mybase', name_prefix="prefix", extension=".json")
         # self.assertEqual(file3, "prefixmybase.json", "generate_file_name should return correct name for extension")
-        # file4 = generate_filename('mybase', name_suffix="suffix")
+        # file4 = clean_filename('mybase', name_suffix="suffix")
         # self.assertEqual(file4, "mybasesuffix", "generate_file_name should return correct name when suffix set")
-        # file5 = generate_filename('mybase', name_suffix="suffix", extension=".json")
+        # file5 = clean_filename('mybase', name_suffix="suffix", extension=".json")
         # self.assertEqual(file5, "mybasesuffix.json", "generate_file_name should return correct name for extension")
-        # file6 = generate_filename('mybase', name_prefix="prefix", name_suffix="suffix", extension=".json")
+        # file6 = clean_filename('mybase', name_prefix="prefix", name_suffix="suffix", extension=".json")
         # self.assertEqual(file6, "prefixmybasesuffix.json",
         #                  "generate_file_name should return correct name for all set")
-        # filename = generate_filename(None, name_prefix=None, name_suffix=None, extension=None)
-        # self.assertEqual('', filename, "Return empty when all arguments are none")
-        # filename = generate_filename(None, name_prefix=None, name_suffix=None, extension='.txt')
-        # self.assertEqual('', filename,
-        #                  "Return empty when base_name, prefix, and suffix are None, but extension is not")
-        # filename = generate_filename('c:/temp.json', name_prefix=None, name_suffix=None, extension='.txt')
-        # self.assertEqual('c_temp.txt', filename,
-        #                  "Returns stripped base_name + extension when prefix, and suffix are None")
-        # filename = generate_filename('temp.json', name_prefix='prefix_', name_suffix='_suffix', extension='.txt')
+        filename = clean_filename("")
+        self.assertEqual('', filename, "Return empty when all arguments are none")
+        filename = clean_filename(None)
+        self.assertEqual('', filename,
+                         "Return empty when base_name, prefix, and suffix are None, but extension is not")
+        filename = clean_filename('c:/temp.json')
+        self.assertEqual('c_temp.json', filename,
+                         "Returns stripped base_name + extension when prefix, and suffix are None")
+        # filename = clean_filename('temp.json', name_prefix='prefix_', name_suffix='_suffix', extension='.txt')
         # self.assertEqual('prefix_temp_suffix.txt', filename,
         #                  "Return stripped base_name + extension when prefix, and suffix are None")
-        # filename = generate_filename(None, name_prefix='prefix_', name_suffix='suffix', extension='.txt')
+        # filename = clean_filename(None, name_prefix='prefix_', name_suffix='suffix', extension='.txt')
         # self.assertEqual('prefix_suffix.txt', filename,
         #                  "Returns correct string when no base_name")
-        # filename = generate_filename('event-strategy-v3_task-matchingpennies_events.json',
-        #                              name_suffix='_blech', extension='.txt')
+        # filename = clean_filename('event-strategy-v3_task-matchingpennies_events.json',
+        #                           name_suffix='_blech', extension='.txt')
         # self.assertEqual('event-strategy-v3_task-matchingpennies_events_blech.txt', filename,
         #                  "Returns correct string when base_name with hyphens")
-        # filename = generate_filename('HED7.2.0.xml', name_suffix='_blech', extension='.txt')
+        # filename = clean_filename('HED7.2.0.xml', name_suffix='_blech', extension='.txt')
         # self.assertEqual('HED7.2.0_blech.txt', filename, "Returns correct string when base_name has periods")
 
     def test_get_dir_dictionary(self):