Refactor: Consolidated pronom modules into pronom package and created a utils package

fido/fido.py

@@ -15,15 +15,15 @@
 import tempfile
 import zipfile
 from contextlib import closing
-from time import perf_counter
 from typing import Optional
 from xml.etree import cElementTree as ET
 
 from fido import CONFIG_DIR, __version__
-from fido.char_handler import escape
 from fido.cli_args import parse_cli_args
 from fido.package import OlePackage, ZipPackage
-from fido.versions import get_local_versions, sig_file_actions
+from fido.pronom.versions import get_local_versions, sig_file_actions
+from fido.utils.char_handler import escape
+from fido.utils.timer import PerfTimer
 
 defaults = {
     "config_dir": CONFIG_DIR,
@@ -50,22 +50,6 @@
 }
 
 
-class PerfTimer:
-    """Utility class that carries out simple process timings."""
-
-    def __init__(self):
-        """New instance with start time running."""
-        self.start_time = perf_counter()
-
-    def start(self):
-        """Start new timer."""
-        self.start_time = perf_counter()
-
-    def duration(self):
-        """Return the duration since instantiation or start() was last called."""
-        return perf_counter() - self.start_time
-
-
 class Fido:
     """Main FIDO application class."""
 

fido/update_signatures.py → fido/pronom/update_signatures.py

@@ -21,15 +21,16 @@
 from shutil import rmtree
 from xml.etree import ElementTree as CET
 
-from . import CONFIG_DIR, __version__
-from .prepare import run as prepare_pronom_to_fido
-from .pronom.soap import (
+from pronom.prepare import run as prepare_pronom_to_fido
+
+from fido import CONFIG_DIR, __version__
+from fido.pronom.soap import (
     NS,
     get_droid_signatures,
     get_pronom_sig_version,
     get_sig_xml_for_puid,
 )
-from .versions import get_local_versions
+from fido.pronom.versions import get_local_versions
 
 ABORT_MSG = "Aborting update..."
 
@@ -117,9 +118,7 @@ def sig_version_check(version="latest"):
         print("Getting latest version number from PRONOM...")
         version = get_pronom_sig_version()
         if not version:
-            sys.exit(
-                "Failed to obtain PRONOM signature file version number, please try again."
-            )
+            sys.exit("Failed to obtain PRONOM signature file version number, please try again.")
 
     print("Querying PRONOM for signaturefile version {}.".format(version))
     sig_file_name = _sig_file_name(version)
@@ -159,9 +158,7 @@ def init_sig_download(defaults):
     resume = False
     if os.path.isdir(tmpdir):
         print("Found previously created temporary folder for download:", tmpdir)
-        resume = query_yes_no(
-            "Do you want to resume download (yes) or start over (no)?"
-        )
+        resume = query_yes_no("Do you want to resume download (yes) or start over (no)?")
         if resume:
             print("Resuming download...")
     else:
@@ -171,9 +168,7 @@ def init_sig_download(defaults):
         except OSError:
             pass
     if not os.path.isdir(tmpdir):
-        sys.stderr.write(
-            "Failed to create temporary folder for PUID's, using: " + tmpdir
-        )
+        sys.stderr.write("Failed to create temporary folder for PUID's, using: " + tmpdir)
     return tmpdir, resume
 
 
@@ -187,9 +182,7 @@ def download_signatures(defaults, format_eles, resume, tmpdir):
         download_sig(format_ele, tmpdir, resume, defaults)
         numfiles += 1
         print(
-            r"Downloaded {}/{} files [{}%]".format(
-                numfiles, puid_count, int(float(numfiles) / one_percent)
-            ),
+            r"Downloaded {}/{} files [{}%]".format(numfiles, puid_count, int(float(numfiles) / one_percent)),
             end="\r",
         )
     print("100%")
@@ -258,9 +251,7 @@ def update_versions_xml(version):
 
 def main():
     """Main CLI entrypoint."""
-    parser = ArgumentParser(
-        description="Download and convert the latest PRONOM signatures"
-    )
+    parser = ArgumentParser(description="Download and convert the latest PRONOM signatures")
     parser.add_argument(
         "-tmpdir",
         default=OPTIONS["tmp_dir"],

fido/toxml.py

@@ -22,8 +22,9 @@
 import csv
 import sys
 
+from fido.pronom.versions import get_local_versions
+
 from . import __version__
-from .versions import get_local_versions
 
 
 def main():
@@ -34,9 +35,7 @@ def main():
     <versions>
         <fido_version>{0}</fido_version>
         <signature_version>{1}</signature_version>
-    </versions>""".format(
-            __version__, get_local_versions().pronom_version
-        )
+    </versions>""".format(__version__, get_local_versions().pronom_version)
     )
 
     reader = csv.reader(sys.stdin)
@@ -54,9 +53,7 @@ def main():
         <formatname>{6}</formatname>
         <signaturename>{7}</signaturename>
         <filesize>{8}</filesize>
-    </file>""".format(
-                row[6], row[0], row[8], row[1], row[2], row[7], row[3], row[4], row[5]
-            )
+    </file>""".format(row[6], row[0], row[8], row[1], row[2], row[7], row[3], row[4], row[5])
         )
 
     sys.stdout.write("\n</fido_output>\n")

fido/utils/timer.py

@@ -0,0 +1,17 @@
+from time import perf_counter
+
+
+class PerfTimer:
+    """Utility class that carries out simple process timings."""
+
+    def __init__(self):
+        """New instance with start time running."""
+        self.start_time = perf_counter()
+
+    def start(self):
+        """Start new timer."""
+        self.start_time = perf_counter()
+
+    def duration(self):
+        """Return the duration since instantiation or start() was last called."""
+        return perf_counter() - self.start_time

pyproject.toml

@@ -45,9 +45,9 @@ testing = [
 
 [project.scripts]
 fido = "fido.fido:main"
-fido-prepare = "fido.prepare:main"
+fido-prepare = "fido.pronom.prepare:main"
 fido-toxml = "fido.toxml:main"
-
+fido-update-signatures = "fido.pronom.update_signatures:run"
 
 [tool.setuptools.package-data]
 "fido" = ["*.*", "conf/*.*", "pronom/*.*"]

tests/test_fido.py

@@ -7,7 +7,8 @@
 
 import pytest
 
-from fido.fido import Fido, PerfTimer
+from fido.fido import Fido
+from fido.utils.timer import PerfTimer
 
 
 def test_perf_timer():

tests/test_prepare.py

@@ -2,7 +2,7 @@
 
 import pytest
 
-from fido.prepare import convert_to_regex
+from fido.pronom.prepare import convert_to_regex
 
 
 def binrep_convert(byt):
@@ -64,17 +64,17 @@ def test_bitmasks(pronom_bytesequence, matches_predicate):
     ("pronom_bytesequence", "input_", "matches_bool"),
     (
         # These are good:
-        ("ab{3}cd(01|02|03)~07ff", "\xAB\xDD\xDD\xDD\xCD\x02\x11\xFF", True),
-        ("ab{3}cd(01|02|03)~07ff", "\xAB\xDD\xDD\xDD\xCD\x03\x11\xFF", True),
-        ("ab{3}cd(01|02|03)~07ff", "\xAB\xDD\xDD\xDD\xCD\x02\xFE\xFF", True),
+        ("ab{3}cd(01|02|03)~07ff", "\xab\xdd\xdd\xdd\xcd\x02\x11\xff", True),
+        ("ab{3}cd(01|02|03)~07ff", "\xab\xdd\xdd\xdd\xcd\x03\x11\xff", True),
+        ("ab{3}cd(01|02|03)~07ff", "\xab\xdd\xdd\xdd\xcd\x02\xfe\xff", True),
         # Bad because missing three anythings between AB and CD
-        ("ab{3}cd(01|02|03)~07ff", "\xAB\xDD\xDD\xCD\x02\x11\xFF", False),
+        ("ab{3}cd(01|02|03)~07ff", "\xab\xdd\xdd\xcd\x02\x11\xff", False),
         # Bad because not at start of string
-        ("ab{3}cd(01|02|03)~07ff", "\xDA\xAB\xDD\xDD\xDD\xCD\x02\x11\xFF", False),
+        ("ab{3}cd(01|02|03)~07ff", "\xda\xab\xdd\xdd\xdd\xcd\x02\x11\xff", False),
         # Bad because 04 is not in (01|02|03)
-        ("ab{3}cd(01|02|03)~07ff", "\xAB\xDD\xDD\xDD\xCD\x04\x11\xFF", False),
+        ("ab{3}cd(01|02|03)~07ff", "\xab\xdd\xdd\xdd\xcd\x04\x11\xff", False),
         # Bad because 18 is not in ~07
-        ("ab{3}cd(01|02|03)~07ff", "\xAB\xDD\xDD\xDD\xCD\x02\x18\xFF", False),
+        ("ab{3}cd(01|02|03)~07ff", "\xab\xdd\xdd\xdd\xcd\x02\x18\xff", False),
     ),
 )
 def test_heterogenous_sequences(pronom_bytesequence, input_, matches_bool):