Tidy up, and improve docs for, external libraries

cardinal_pythonlib/datetimefunc.py

@@ -46,35 +46,19 @@
 from string import Formatter
 from typing import Any, Optional, Union
 
-try:
-    # noinspection PyPackageRequirements
-    from arrow import Arrow
-except ImportError:
-    Arrow = None
-
-try:
-    import dateutil.parser
-except ImportError:
-    dateutil = None
-
+from arrow import Arrow
+import dateutil.parser
 from isodate.isoduration import parse_duration, Duration as IsodateDuration
 import pendulum
 from pendulum import Date, DateTime, Duration, Time
 from pendulum.tz import local_timezone
 from pendulum.tz.timezone import Timezone
 
-if Arrow is not None:
-    PotentialDatetimeType = Union[
-        None, datetime.datetime, datetime.date, DateTime, str, Arrow
-    ]
-    DateTimeLikeType = Union[datetime.datetime, DateTime, Arrow]
-    DateLikeType = Union[datetime.date, DateTime, Arrow]
-else:
-    PotentialDatetimeType = Union[
-        None, datetime.datetime, datetime.date, DateTime, str
-    ]
-    DateTimeLikeType = Union[datetime.datetime, DateTime]
-    DateLikeType = Union[datetime.date, DateTime]
+PotentialDatetimeType = Union[
+    None, datetime.datetime, datetime.date, DateTime, str, Arrow
+]
+DateTimeLikeType = Union[datetime.datetime, DateTime, Arrow]
+DateLikeType = Union[datetime.date, DateTime, Arrow]
 
 log = logging.getLogger(__name__)
 

cardinal_pythonlib/extract_text.py

@@ -96,46 +96,21 @@
     Optional,
 )
 from xml.etree import ElementTree as ElementTree
-
-# ... cElementTree used to be the fast implementation; now ElementTree is fast
-# and cElementTree is deprecated; see
-# https://docs.python.org/3.4/library/xml.etree.elementtree.html
 import zipfile
 
 import bs4
+import chardet
+from chardet.universaldetector import UniversalDetector
+import pdfminer  # pip install pdfminer.six
+import pdfminer.pdfinterp
+import pdfminer.converter
+import pdfminer.layout
+import pdfminer.pdfpage
 import prettytable
 from semantic_version import Version
 
 from cardinal_pythonlib.logs import get_brace_style_log_with_null_handler
 
-try:
-    import chardet
-    from chardet.universaldetector import UniversalDetector
-except ImportError:
-    chardet = None
-    UniversalDetector = None
-
-try:
-    # noinspection PyPackageRequirements
-    import pdfminer  # pip install pdfminer
-
-    assert (
-        int(pdfminer.__version__) > 20191010
-    ), "pdfminer installed but too old"  # version string is e.g. '20191125'
-
-    # noinspection PyPackageRequirements
-    import pdfminer.pdfinterp
-
-    # noinspection PyPackageRequirements
-    import pdfminer.converter
-
-    # noinspection PyPackageRequirements
-    import pdfminer.layout
-
-    # noinspection PyPackageRequirements
-    import pdfminer.pdfpage
-except ImportError:
-    pdfminer = None
 
 log = get_brace_style_log_with_null_handler(__name__)
 

cardinal_pythonlib/getch.py

@@ -34,13 +34,13 @@
 import sys
 
 try:
-    import msvcrt  # Windows only
+    import msvcrt  # Windows only, but part of core Python under Windows
 
     termios = None
     tty = None
 except ImportError:
     msvcrt = None
-    import termios  # Unix only
+    import termios  # Unix only; tty/termios are part of core Python under Unix
     import tty  # requires termios, so Unix only  # noqa: F401
 
 

cardinal_pythonlib/hash.py

@@ -64,14 +64,6 @@
 except ImportError:
     mmh3 = None
 
-# try:
-#     import xxhash
-#     pyhashxx = None
-# except ImportError:
-#     xxhash = None
-#     import pyhashxx
-
-
 # https://docs.python.org/3/library/platform.html#platform.architecture
 IS_64_BIT = sys.maxsize > 2**32
 TIMING_HASH = "hash"

cardinal_pythonlib/module_version.py

@@ -35,23 +35,13 @@
 # Imports
 # =============================================================================
 
-import sys
+from importlib.metadata import version  # structure for Python 3.8 or higher
 
 from semantic_version import Version
 
 # noinspection PyUnresolvedReferences
 import cardinal_pythonlib.ensure_test_executed_correctly  # noqa: F401
 
-if sys.version_info > (3, 8):
-    # Python 3.8 or higher
-    # noinspection PyCompatibility
-    from importlib.metadata import version
-else:
-    try:
-        from importlib_metadata import version
-    except ImportError:
-        raise
-
 
 # =============================================================================
 # Report Python module versions

cardinal_pythonlib/spreadsheets.py

@@ -38,25 +38,17 @@
 import logging
 from typing import Any, Dict, Iterable, List, Optional, Sequence, Tuple, Union
 
+import xlrd
+from xlrd import Book
+from xlrd.sheet import Cell, Sheet
+
 from cardinal_pythonlib.datetimefunc import (
     coerce_to_pendulum,
     pendulum_to_datetime_stripping_tz,
 )
 from cardinal_pythonlib.progress import ActivityCounter
 from cardinal_pythonlib.reprfunc import simple_repr
 
-try:
-    # noinspection PyPackageRequirements
-    import xlrd
-
-    # noinspection PyPackageRequirements
-    from xlrd import Book
-
-    # noinspection PyPackageRequirements
-    from xlrd.sheet import Cell, Sheet
-except ImportError:
-    raise ImportError("You must install the 'xlrd' package.")
-
 log = logging.getLogger(__name__)
 
 

cardinal_pythonlib/sqlalchemy/arrow_types.py

@@ -30,7 +30,6 @@
 import datetime
 from typing import Any, Iterable, Optional
 
-# noinspection PyUnresolvedReferences
 import arrow
 import sqlalchemy.dialects.mssql
 import sqlalchemy.dialects.mysql

docs/source/external_dependencies.rst

@@ -18,17 +18,21 @@
 External libraries
 ------------------
 
-
-This package also installs (and uses or extends):
+This package also installs (and uses or extends) the following packages, which
+are generally "pure Python", meaning that they can easily be installed e.g. on
+a Windows computer with no C compiler system installed.
 
 - ``alembic``: http://alembic.zzzcomputing.com/
 - ``appdirs``: https://pypi.org/project/appdirs/
+- ``arrow``: https://arrow.readthedocs.io/
 - ``beautifulsoup4``: https://www.crummy.com/software/BeautifulSoup/bs4/doc/
+- ``chardet``: https://chardet.readthedocs.io/en/latest/
 - ``colorlog``: https://pypi.org/project/colorlog/
 - ``isodate``: https://pypi.org/project/isodate/
 - ``numpy``: http://www.numpy.org/
 - ``openpyxl``: https://openpyxl.readthedocs.io/
 - ``pandas``: https://pandas.pydata.org/
+- ``pdfminer.six``: https://pdfminersix.readthedocs.io/en/latest/
 - ``pendulum``: https://pendulum.eustace.io/
 - ``prettytable``: https://pypi.org/project/PrettyTable/
 - ``psutil``: https://pypi.org/project/psutil/
@@ -40,26 +44,40 @@ This package also installs (and uses or extends):
 - ``semantic_version``: https://pypi.org/project/semantic_version/
 - ``SQLAlchemy``: https://www.sqlalchemy.org/
 - ``sqlparse``: https://sqlparse.readthedocs.io/
+- ``xlrd``: https://pypi.org/project/xlrd/
 
-The following will be used, if present (and an exception raised if you use
-library code that requires one of these packages without it being installed):
+The following packages will be used, if present, and an exception raised if you
+use library code that requires one of these packages without it being
+installed. They include large packages (e.g. Django), some other "less core"
+aspects, and packages that require a C compiler and so may be harder to install
+in some contexts.
 
-- ``arrow``: https://arrow.readthedocs.io/
-- ``bcrypt``: https://pypi.org/project/bcrypt/
+- ``bcrypt``: https://pypi.org/project/bcrypt/ (C-based)
 - ``colander``: https://docs.pylonsproject.org/projects/colander/
+- ``cryptography``: https://cryptography.io/
 - ``deform``: https://docs.pylonsproject.org/projects/deform/
-- ``Django``: https://www.djangoproject.com/
+- ``Django`` >= 4.2: https://www.djangoproject.com/
 - ``dogpile.cache``: https://dogpilecache.readthedocs.io/
+- ``libChEBIpy``: https://pypi.org/project/libChEBIpy/ (Python 2 only?)
 - ``pyramid``: https://trypyramid.com/
-- ``webob``: https://webob.org/ (used by Pyramid)
+- ``webob``: https://webob.org/ (used and installed by Pyramid)
 
-The following will be used, but the library code won't complain if not:
+The following packages will be used sometimes, but the library code won't
+complain much if they are absent. They include some other C-based packages, one
+that is specific to Windows and won't install on other platforms, and a
+selection of PDF-handling libraries where there is not a clear best choice.
 
-- ``mmh3``: https://pypi.org/project/mmh3/
+- ``brotlipy``: https://pypi.org/project/brotlipy/ (C-based)
+- ``mmh3``: https://pypi.org/project/mmh3/ (C-based)
+- ``matplotlib``: https://matplotlib.org/
 - ``pdfkit``: https://pypi.org/project/pdfkit/
-- ``pdfminer``: https://pypi.org/project/pdfminer/
-- ``pypiwin32``: https://pypi.org/project/pypiwin32/
-- ``pyth``: https://pyth.readthedocs.io/
-- ``python-docx`` (``import docx``): https://python-docx.readthedocs.io/
+- ``pypiwin32``: https://pypi.org/project/pypiwin32/ (Windows only)
 - ``weasyprint``: https://weasyprint.org/
 - ``xhtml2pdf``: https://xhtml2pdf.readthedocs.io/
+
+To *build* the library distribution (most people won't need this!), a few other
+development libraries are required:
+
+- ``sphinx``: https://www.sphinx-doc.org/
+- ``sphinx_rtd_theme``: https://github.com/readthedocs/sphinx_rtd_theme
+- ``twine``: https://pypi.org/project/twine/

setup.py

@@ -49,14 +49,17 @@
     # - SEE ALSO external_dependencies.rst
     "alembic",
     "appdirs>=1.4.0",
+    "arrow>=0.15",
     "beautifulsoup4",  # "import bs4" or "from bs4 import ..."
+    "chardet>=5.0.0",
     "colorlog",
     "isodate>=0.5.4",
     "numba",  # just-in-time compilation
     "numpy>=1.20.0,<2.0",  # 1.20.0 required for numpy.typing
     "openpyxl",
     "pandas",
     "pendulum>=2.1.1",
+    "pdfminer.six>=20191010",  # "import pdfminer"
     "prettytable",
     "psutil",
     "pygments",
@@ -68,6 +71,7 @@
     "semantic-version",
     "SQLAlchemy>=1.4,<3.0",
     "sqlparse",
+    "xlrd>=2.0.0",
 ]
 
 NOTES_RE_OTHER_REQUIREMENTS = """
@@ -79,45 +83,8 @@
 
 
 # -----------------------------------------------------------------------------
-# The following are NOT HANDLED GRACEFULLY; their absence will cause a runtime
-# ImportError, but we don't make them requirements as they need a compiler to
-# install (and one might want to use the rest of the library without them).
+# FOR OTHER REQUIREMENTS, AND OPTIONAL EXTRAS: SEE external_dependencies.rst
 # -----------------------------------------------------------------------------
-# - SEE ALSO external_dependencies.rst
-
-# arrow
-# bcrypt
-# colander
-# deform
-# Django>=4.2
-# dogpile.cache
-# pyramid
-# webob  # installed by pyramid
-
-
-# -----------------------------------------------------------------------------
-# The following are OPTIONAL; their absence will be handled gracefully, so
-# they are not requirements, but we note them here:
-# -----------------------------------------------------------------------------
-# - SEE ALSO external_dependencies.rst
-
-# mmh3
-# pdfkit
-# pdfminer
-# pypiwin32
-# pyth
-# python-docx   # "import docx"
-# weasyprint
-# xhtml2pdf
-
-
-# -----------------------------------------------------------------------------
-# FOR LIBRARY DEVELOPMENT
-# -----------------------------------------------------------------------------
-
-# sphinx
-# sphinx_rtd_theme
-# twine
 
 """