Writing Python in a Nutshell, 4th Edition

Anna Martelli Ravenscroft and Alex Martelli

 A short history of the book
2003: 1st edition. (up to) v2.2; martellibot; 600 pages
2006: 2nd edition. (up to) v2.5; Alex (with Anna as tech reviewer); 738 pages
(eep, what growth!)
…long time lapse as Alex was not yet using Python 3…
2017: 3rd edition. v2.7 & 3.5; Alex & Anna, with Steve Holden as co-author; 772
pages (way too big!)
2023: 4th edition. V3.7 to 3.11; Alex, Anna, Steve, and Paul McGuire; 735 pages
(a little better!)
 The process of writing/editing/finalizing
● Location (authors): 2 x Sunnyvale, CA; 1 x Austin, TX; 1 near London, UK
● Tools: Google Spreadsheets, weekly Google Meet meetings, Google Docs…
(and more meetings, spreadsheets, docs…)
→ …meeting all chapter-by-chapter deadlines (to the editor’s surprise!-)
● Writing stylistic issues:
○ Oxford commas and active voice (avoid passives!)
○ Typography (italics? bold? which font should commas inside code be?)
○ Indicating version numbers, e.g.: 3.9+
● Reviewers and copyeditor (Rachel Head rocked!) and indexer, oh my!
 Some hard decisions
● Not covering
○ asyncio & friends (too important: deserves whole books of its own!)
○ key 3rd-party frameworks (django, flask, panda, PyTorch,
TensorFlow/Keras, …, any GUI, …)
● Two online-only chapters:
https://github.com/pynutshell/pynut4/tree/main/chapters
○ Packaging (chapter 24)
○ Extending and Embedding (chapter 25)
 Some not-so-hard decisions
● Github repo for all sample code from the book:
https://github.com/pynutshell/pynut4/tree/main
● Search PEP abstracts by keywords:
https://ptmcg.pythonanywhere.com/pep_search
● Search for language changes by version:
https://ptmcg.pythonanywhere.com/python_nutshell_app_a_search
A few of our favorite things
 Anna’s favorite new thing
The zoneinfo Module 3.9+
● Concrete implementation of timezones for use with tzinfo
● Uses system timezone data, or tzdata as a fallback
○ On Windows, you may need to pip install tzdata
○ Once tzdata installed, you don’t need to import tzdata
● Get a list of timezone names: zoneinfo.available_timezones()
● zoneinfo has one class ZoneInfo

>>> from zoneinfo import ZoneInfo

 Updating a Timezone
>>> d
datetime.datetime(2023,5,4,18,30,
tzinfo=zoneinfo.ZoneInfo(key='America/New_York'))
>>> d=d.replace(tzinfo=ZoneInfo("America/Los_Angeles"))
>>> d
datetime.datetime(2023,5,4,18,30,
tzinfo=zoneinfo.ZoneInfo(key='America/Los_Angeles'))
>>> dk=d.astimezone(tz=ZoneInfo("Asia/Katmandu"))
>>> dk
datetime.datetime(2023,5,5,7,15,
tzinfo=zoneinfo.ZoneInfo(key='Asia/Katmandu'))
# Katmandu offset is +5 hours 45 minutes from UTC (+12:45 ahead of PST)
 replace() vs astimezone()
replace
● returns new datetime with updated timezone without changing other
attributes
● pass keyword arg tzinfo= to change timezone
● can also be used to replace other attributes, e.g.:
>>> d=d.replace(hour=19, minute=0)
astimezone
● returns new datetime with updated time, date (as needed), & timezone
● won’t work with tzinfo=, requires keyword arg tz=
 Alex’s favorite (and… not-so-much) things

● Since 3.0…
○ text strings default to Unicode (good!)
○ identifiers may contain unicode (hmmm…)
■ They get normalized (NFKC, not NFC as rec'd in UAX#31!)
■ see PEP 672 (https://peps.python.org/pep-0672)
■ and devote your time (I estimate: several months!) to study
https://www.unicode.org/versions/Unicode15.0.0/UnicodeStandar
d-15.0.pdf and https://www.unicode.org/reports/ (all of them!)
Middle Dot “·” is a Valid Identifier Character

Not valid as a leading character

May cause difficulties due to visual similarity between · and .
Python in a Nutshell, 4th edition, Chapter 3, page 36
Unicode Identifiers Can Often “normalize to ASCII”

ᴾ𝘆𝙩𝚑𝓸𝔫 𝐹º𝑛ｔ 𝘔ⅸᵉ𝐫

http://ptmcg.pythonanywhere.com/font_mixer
18 Unicode Characters Normalize To ‘A’...
A 65 LATIN CAPITAL LETTER A
ᴬ 7468 MODIFIER LETTER CAPITAL A
Ⓐ 9398 CIRCLED LATIN CAPITAL LETTER A
Ａ 65313 FULLWIDTH LATIN CAPITAL LETTER A
𝐀 119808 MATHEMATICAL BOLD CAPITAL A
𝐴 119860 MATHEMATICAL ITALIC CAPITAL A
𝑨 119912 MATHEMATICAL BOLD ITALIC CAPITAL A
𝒜 119964 MATHEMATICAL SCRIPT CAPITAL A
𝓐 120016 MATHEMATICAL BOLD SCRIPT CAPITAL A
𝔄 120068 MATHEMATICAL FRAKTUR CAPITAL A
𝔸 120120 MATHEMATICAL DOUBLE-STRUCK CAPITAL A
𝕬 120172 MATHEMATICAL BOLD FRAKTUR CAPITAL A
𝖠 120224 MATHEMATICAL SANS-SERIF CAPITAL A
𝗔 120276 MATHEMATICAL SANS-SERIF BOLD CAPITAL A
𝘈 120328 MATHEMATICAL SANS-SERIF ITALIC CAPITAL A
𝘼 120380 MATHEMATICAL SANS-SERIF BOLD ITALIC CAPITAL A
𝙰 120432 MATHEMATICAL MONOSPACE CAPITAL A
🄰 127280 SQUARED LATIN CAPITAL LETTER A
But NOT Α (Greek capital letter α)!
 Some tidbits from Paul McGuire
Watch out for str.lstrip() and str.rstrip();
they might remove more than you want!
>>> print("https://shell.com".lstrip("https://"))
ell.com

removeprefix() 3.9+ does what you probably wanted:

>>> print("https://shell.com".removeprefix("https://"))
shell.com

There’s also a new removesuffix() 3.9+

COUNT HIGHER WITH INTS THAN WITH FLOATS

Floats lose integer

precision at 2**53

Ints can count up to

2**(2**63), or
roughly 10**(2.8e18)
https://www.build-python-from-source.com/
Found Choose
● this helpfulPython
website version
while Choosefor
● searching a few other options
common system libraries
● Generates text for a bash script (good for Linux
used while buiding Python
or WSL)
from source
 Python’s sort() Cited in Supreme Court Case
● Python’s sort() algorithm (known as
TimSort, named for its author Tim
Peters) was part of the source code
cited in the case of Oracle America v.
Google, 2012
● The case was generally decided for
Google, except for the infringement on
TimSort – the judgment against
Google for that was $0
A portion of our royalties for Python in a Nutshell
are donated to the PSF
 Discounts and book signings and Q&A

Discount code for 40% off, good thru May 31, 2023:
https://play.google.com/redeem?code=61P98Z8G5R20U
Upcoming book signings (tentative):
● North Bay Python https://2023.northbaypython.org/
● PyBay https://www.sfpythonmeetup.com/

Q&A