Match current state of name parsing branch

python · AA-Turner · Apr 24, 2020 · Apr 25, 2020 · Apr 25, 2020 · Apr 25, 2020
commit 116f3a7761849d34ddcd9c60dee00c2907d79e23
diff --git a/AUTHORS.csv b/AUTHORS.csv
@@ -1,249 +1,12 @@
-"Full Name", "Surna
8000
me First", "Name Reference"
-"Aahz", "Aahz", "Aahz"
-"James C. Ahlstrom", "Ahlstrom, James C.", "Ahlstrom"
-"Jim Althoff", "Althoff, Jim", "Althoff"
-"Kevin Altis", "Altis, Kevin", "Altis"
-"Chris Angelico", "Angelico, Chris", "Angelico"
-"Philipp Angerer", "Angerer, Philipp", "Angerer"
-"David Ascher", "Ascher, David", "Ascher"
-"Peter Astrand", "Astrand, Peter", "Astrand"
-"Carl Banks", "Banks, Carl", "Banks"
-"Christopher Barker", "Barker, Christopher", "Barker"
-"Paul Barrett", "Barrett, Paul", "Barrett"
-"Facundo Batista", "Batista, Facundo", "Batista"
-"Anthony Baxter", "Baxter, Anthony", "Baxter"
-"Stefan Behnel", "Behnel, Stefan", "Behnel"
-"Thomas Bellman", "Bellman, Thomas", "Bellman"
-"Alexander Belopolsky", "Belopolsky, Alexander", "Belopolsky"
-"Eli Bendersky", "Bendersky, Eli", "Bendersky"
-"Cory Benfield", "Benfield, Cory", "Benfield"
-"Steven Bethard", "Bethard, Steven", "Bethard"
-"Stéphane Bidoul", "Bidoul, Stéphane", "Bidoul"
-"Stefano Borini", "Borini, Stefano", "Borini"
-"Georg Brandl", "Brandl, Georg", "Brandl"
-"Erik M. Bray", "Bray, Erik M.", "Bray"
-"Gerald Britton", "Britton, Gerald", "Britton"
-"Oleg Broytman", "Broytman, Oleg", "Broytman"
-"Benoit Bryon", "Bryon, Benoit", "Bryon"
-"Brandt Bucher", "Bucher, Brandt", "Bucher"
-"Brett Cannon", "Cannon, Brett", "Cannon"
-"Justin Cappos", "Cappos, Justin", "Cappos"
-"Josiah Carlson", "Carlson, Josiah", "Carlson"
-"W Isaac Carroll", "Carroll, W Isaac", "Carroll"
-"Matt Chisholm", "Chisholm, Matt", "Chisholm"
-"Nick Coghlan", "Coghlan, Nick", "Coghlan"
-"Dave Cole", "Cole, Dave", "Cole"
-"Robert Collins", "Collins, Robert", "Collins"
-"Paul Colomiets", "Colomiets, Paul", "Colomiets"
-"Mario Corchero", "Corchero, Mario", "Corchero"
-"Christopher A. Craig", "Craig, Christopher A.", "Craig"
-"Laura Creighton", "Creighton, Laura", "Creighton"
-"Steven D'Aprano", "D'Aprano, Steven", "D'Aprano"
-"Kushal Das", "Das, Kushal", "Das"
-"Ned Deily", "Deily, Ned", "Deily"
-"Tim Delaney", "Delaney, Tim", "Delaney"
-"Lois Anne DeLong", "DeLong, Lois Anne", "DeLong"
-"Jeroen Demeyer", "Demeyer, Jeroen", "Demeyer"
-"Vladimir Diaz", "Diaz, Vladimir", "Diaz"
-"Jack Diederich", "Diederich, Jack", "Diederich"
-"Steve Dower", "Dower, Steve", "Dower"
-"Walter Dörwald", "Dörwald, Walter", "Dörwald"
-"Fred L. Drake, Jr.", "Drake, Fred L., Jr.", "Drake"
-"Michael P. Dubner", "Dubner, Michael P.", "Dubner"
-"Paul F. Dubois", "Dubois, Paul F.", "Dubois"
-"Ernest W. Durbin III", "Durbin, Ernest W., III", "Durbin"
-"P.J. Eby", "Eby, P.J.", "Eby"
 "Phillip J. Eby", "Eby, Phillip J.", "Eby"
-"Tal Einat", "Einat, Tal", "Einat"
-"Micah Elliott", "Elliott, Micah", "Elliott"
-"Jeff Epler", "Epler, Jeff", "Epler"
-"David Eppstein", "Eppstein, David", "Eppstein"
-"Clark C. Evans", "Evans, Clark C.", "Evans"
-"Gregory Ewing", "Ewing, Gregory", "Ewing"
-"Greg Ewing", "Ewing, Greg", "Ewing"
-"Martijn Faassen", "Faassen, Martijn", "Faassen"
-"Ben Finney", "Finney, Ben", "Finney"
-"Michael Foord", "Foord, Michael", "Foord"
-"Ethan Furman", "Furman, Ethan", "Furman"
-"Pablo Galindo", "Galindo, Pablo", "Galindo"
-"Paul Ganssle", "Ganssle, Paul", "Ganssle"
-"Alex Gaynor", "Gaynor, Alex", "Gaynor"
-"Pradyun Gedam", "Gedam, Pradyun", "Gedam"
-"Damien George", "George, Damien", "George"
-"Frédéric B. Giacometti", "Giacometti, Frédéric B.", "Giacometti"
-"Scott Gilbert", "Gilbert, Scott", "Gilbert"
-"Ryan Gonzalez", "Gonzalez, Ryan", "Gonzalez"
-"David Goodger", "Goodger, David", "Goodger"
-"Grant Griffin", "Griffin, Grant", "Griffin"
-"Mark E. Haase", "Haase, Mark E.", "Haase"
-"Mark Hammond", "Hammond, Mark", "Hammond"
-"Peter Harris", "Harris, Peter", "Harris"
-"Larry Hastings", "Hastings, Larry", "Hastings"
-"Christian Heimes", "Heimes, Christian", "Heimes"
-"Thomas Heller", "Heller, Thomas", "Heller"
-"Doug Hellmann", "Hellmann, Doug", "Hellmann"
-"Magnus Lie Hetland", "Hetland, Magnus Lie", "Hetland"
-"Raymond Hettinger", "Hettinger, Raymond", "Hettinger"
-"Neil Hodgson", "Hodgson, Neil", "Hodgson"
-"Daniel Holth", "Holth, Daniel", "Holth"
-"Philip House", "House, Philip", "House"
-"Laurens Van Houtven", "Van Houtven, Laurens", "Houtven"
-"Ben Hoyt", "Hoyt, Ben", "Hoyt"
-"Miro Hrončok", "Hrončok, Miro", "Hrončok"
-"Michael Hudson", "Hudson, Michael", "Hudson"
-"Jeremy Hylton", "Hylton, Jeremy", "Hylton"
-"Inada Naoki", "Inada, Naoki", "Inada"
-"Dustin Ingram", "Ingram, Dustin", "Ingram"
-"Atsuo Ishimoto", "Ishimoto, Atsuo", "Ishimoto"
-"Jack Jansen", "Jansen, Jack", "Jansen"
-"Chris Jerdonek", "Jerdonek, Chris", "Jerdonek"
-"Joseph Jevnik", "Jevnik, Joseph", "Jevnik"
-"Jim J. Jewett", "Jewett, Jim J.", "Jewett"
-"Jim Jewett", "Jewett, Jim", "Jewett"
-"Ewa Jodlowska", "Jodlowska, Ewa", "Jodlowska"
-"Richard Jones", "Jones, Richard", "Jones"
-"Konstantin Kashin", "Kashin, Konstantin", "Kashin"
-"Reid Kleckner", "Kleckner, Reid", "Kleckner"
-"Thomas Kluyver", "Kluyver, Thomas", "Kluyver"
-"Stepan Koltsov", "Koltsov, Stepan", "Koltsov"
-"Stefan Krah", "Krah, Stefan", "Krah"
-"Sebastian Kreft", "Kreft, Sebastian", "Kreft"
-"Holger Krekel", "Krekel, Holger", "Krekel"
-"A.M. Kuchling", "Kuchling, A.M.", "Kuchling"
-"Trishank Karthik Kuppusamy", "Kuppusamy, Trishank Karthik", "Kuppusamy"
-"Robert Kuska", "Kuska, Robert", "Kuska"
-"Joshua Landau", "Landau, Joshua", "Landau"
-"Łukasz Langa", "Langa, Łukasz", "Langa"
-"Michael Lee", "Lee, Michael", "Lee"
-"Jukka Lehtosalo", "Lehtosalo, Jukka", "Lehtosalo"
-"Marc-André Lemburg", "Lemburg, Marc-André", "Lemburg"
-"Ivan Levkivskyi", "Levkivskyi, Ivan", "Levkivskyi"
-"Gregory Lielens", "Lielens, Gregory", "Lielens"
-"Björn Lindqvist", "Lindqvist, Björn", "Lindqvist"
-"Joshua Lock", "Lock, Joshua", "Lock"
-"Tony Lownds", "Lownds, Tony", "Lownds"
-"Martin von Löwis", "von Löwis, Martin", "von Löwis"
-"Martin v. Löwis", "\v. Löwis, Martin", "\v. Löwis"
-"Mariatta", "Mariatta", "Mariatta"
-"Alex Martelli", "Martelli, Alex", "Martelli"
-"Joseph Martinot-Lagarde", "Martinot-Lagarde, Joseph", "Martinot-Lagarde"
-"Lino Mastrodomenico", "Mastrodomenico, Lino", "Mastrodomenico"
-"Patrick Maupin", "Maupin, Patrick", "Maupin"
-"Andrew McClelland", "McClelland, Andrew", "McClelland"
-"Charles R. McCreary", "McCreary, Charles R.", "McCreary"
-"Chris McDonough", "McDonough, Chris", "McDonough"
-"Robert T. McGibbon", "McGibbon, Robert T.", "McGibbon"
-"Gordon McMillan", "McMillan, Gordon", "McMillan"
-"Andrew McNamara", "McNamara, Andrew", "McNamara"
-"Ezio Melotti", "Melotti, Ezio", "Melotti"
-"Mark Mendoza", "Mendoza, Mark", "Mendoza"
-"Markus Meskanen", "Meskanen, Markus", "Meskanen"
-"Mike Meyer", "Meyer, Mike", "Meyer"
-"Carl Meyer", "Meyer, Carl", "Meyer"
-"Trent Mick", "Mick, Trent", "Mick"
-"Mike G. Miller", "Miller, Mike G.", "Miller"
-"Skip Montanaro", "Montanaro, Skip", "Montanaro"
-"Peter Moody", "Moody, Peter", "Moody"
-"Marina Moore", "Moore, Marina", "Moore"
-"Paul Moore", "Moore, Paul", "Moore"
-"R David Murray", "Murray, R David", "Murray"
-"Charles-François Natali", "Natali, Charles-François", "Natali"
-"Lysandros Nikolaou", "Nikolaou, Lysandros", "Nikolaou"
-"Jesse Noller", "Noller, Jesse", "Noller"
-"Ben North", "North, Ben", "North"
-"Neal Norwitz", "Norwitz, Neal", "Norwitz"
-"Dirkjan Ochtman", "Ochtman, Dirkjan", "Ochtman"
-"Travis Oliphant", "Oliphant, Travis", "Oliphant"
-"Jason Orendorff", "Orendorff, Jason", "Orendorff"
-"Tomáš Orsava", "Orsava, Tomáš", "Orsava"
-"Richard Oudkerk", "Oudkerk, Richard", "Oudkerk"
-"Ronald Oussoren", "Oussoren, Ronald", "Oussoren"
-"Julien Palard", "Palard, Julien", "Palard"
-"Samuele Pedroni", "Pedroni, Samuele", "Pedroni"
-"Berker Peksag", "Peksag, Berker", "Peksag"
-"Michel Pelletier", "Pelletier, Michel", "Pelletier"
-"Tim Peters", "Peters, Tim", "Peters"
-"Benjamin Peterson", "Peterson, Benjamin", "Peterson"
-"Jason Petrone", "Petrone, Jason", "Petrone"
-"Antoine Pitrou", "Pitrou, Antoine", "Pitrou"
-"Marcel Plch", "Plch, Marcel", "Plch"
-"James Polley", "Polley, James", "Polley"
-"Philippe PRADOS", "Prados, Philippe", "Prados"
-"Elvis Pranskevichus", "Pranskevichus, Elvis", "Pranskevichus"
-"Paul Prescod", "Prescod, Paul", "Prescod"
-"(James) Eric Pruitt", "Pruitt, (James) Eric", "Pruitt"
-"Lukas Puehringer", "Puehringer, Lukas", "Puehringer"
-"Brian Quinlan", "Quinlan, Brian", "Quinlan"
-"Terry Reedy", "Reedy, Terry", "Reedy"
-"Lennart Regebro", "Regebro, Lennart", "Regebro"
-"Sean Reifschneider", "Reifschneider, Sean", "Reifschneider"
-"Christian R. Reis", "Reis, Christian R.", "Reis"
-"Jonathan Riehl", "Riehl, Jonathan", "Riehl"
-"Lisa Roach", "Roach, Lisa", "Roach"
-"Andre Roberge", "Roberge, Andre", "Roberge"
-"Armin Ronacher", "Ronacher, Armin", "Ronacher"
-"Guido van Rossum", "van Rossum, Guido (GvR)", "GvR"
-"Just van Rossum", "van Rossum, Just (JvR)", "JvR"
-"Todd Rovito", "Rovito, Todd", "Rovito"
-"Lie Ryan", "Ryan, Lie", "Ryan"
-"Vinay Sajip", "Sajip, Vinay", "Sajip"
-"Pablo Galindo Salgado", "Salgado, Pablo Galindo", "Salgado"
-"Neil Schemenauer", "Schemenauer, Neil", "Schemenauer"
-"Peter Schneider-Kamp", "Schneider-Kamp, Peter", "Schneider-Kamp"
-"Ed Schofield", "Schofield, Ed", "Schofield"
-"Yury Selivanov", "Selivanov, Yury", "Selivanov"
-"Jiwon Seo", "Seo, Jiwon", "Seo"
-"Mark Shannon", "Shannon, Mark", "Shannon"
-"Cameron Simpson", "Simpson, Cameron", "Simpson"
-"Greg Slodkowicz", "Slodkowicz, Greg", "Slodkowicz"
-"Nathaniel J. Smith", "Smith, Nathaniel J.", "Smith"
-"Gregory P. Smith", "Smith, Gregory P.", "Smith"
-"Kevin D. Smith", "Smith, Kevin D.", "Smith"
-"Ethan Smith", "Smith, Ethan", "Smith"
-"Nathaniel Smith", "Smith, Nathaniel", "Smith"
-"Eric V. Smith", "Smith, Eric V.", "Smith"
-"Eric Snow", "Snow, Eric", "Snow"
-"Calvin Spealman", "Spealman, Calvin", "Spealman"
-"Kerrick Staley", "Staley, Kerrick", "Staley"
-"Greg Stein", "Stein, Greg", "Stein"
-"Victor Stinner", "Stinner, Victor", "Stinner"
-"Serhiy Storchaka", "Storchaka, Serhiy", "Storchaka"
-"Donald Stufft", "Stufft, Donald", "Stufft"
-"Daniel Stutzbach", "Stutzbach, Daniel", "Stutzbach"
-"Michael J. Sullivan", "Sullivan, Michael J.", "Sullivan"
-"Roman Suzi", "Suzi, Roman", "Suzi"
-"Dennis Sweeney", "Sweeney, Dennis", "Sweeney"
-"Talin", "Talin", "Talin"
-"Steven Taschuk", "Taschuk, Steven", "Taschuk"
-"Batuhan Taskaya", "Taskaya, Batuhan", "Taskaya"
-"Martin Teichmann", "Teichmann, Martin", "Teichmann"
-"The Python core team and community", "The Python core team and community", "The Python core team and community"
-"Geoffrey Thomas", "Thomas, Geoffrey", "Thomas"
-"Oren Tirosh", "Tirosh, Oren", "Tirosh"
-"Stephen J. Turnbull", "Turnbull, Stephen J.", "Turnbull"
-"Daniel Urban", "Urban, Daniel", "Urban"
-"Eric N. Vander Weele", "Vander Weele, Eric N.", "Vander Weele"
-"Till Varoquaux", "Varoquaux, Till", "Varoquaux"
-"Alexandre Vassalotti", "Vassalotti, Alexandre", "Vassalotti"
-"Mike Verdone", "Verdone, Mike", "Verdone"
-"Dino Viehland", "Viehland, Dino", "Viehland"
-"Petr Viktorin", "Viktorin, Petr", "Viktorin"
-"Zachary Ware", "Ware, Zachary", "Ware"
-"Gregory R. Warnes", "Warnes, Gregory R.", "Warnes"
-"Barry Warsaw", "Warsaw, Barry", "Warsaw"
-"Terence Way", "Way, Terence", "Way"
-"Cliff Wells", "Wells, Cliff", "Wells"
-"Jervis Whitley", "Whitley, Jervis", "Whitley"
-"Mark  Williams", "Williams, Mark", "Williams"
-"Carol Willing", "Willing, Carol", "Willing"
-"Greg Wilson", "Wilson, Greg", "Wilson"
-"Collin Winter", "Winter, Collin", "Winter"
-"Thomas Wouters", "Wouters, Thomas", "Wouters"
-"Masayuki Yamamoto", "Yamamoto, Masayuki", "Yamamoto"
-"Jeffrey Yasskin", "Yasskin, Jeffrey", "Yasskin"
-"Ka-Ping Yee", "Yee, Ka-Ping", "Yee"
-"Moshe Zadka", "Zadka, Moshe", "Zadka"
-"Koos Zevenhoven", "Zevenhoven, Koos", "Zevenhoven"
-"Huaiyu Zhu", "Zhu, Huaiyu", "Zhu"
-"Shannon Zhu", "Zhu, Shannon", "Zhu"
-"Tarek Ziadé", "Ziadé, Tarek", "Ziadé"
+Full Name, Surname First, Name Reference
+Ernest W. Durbin III, "Durbin, Ernest W., III", Durbin
+Inada Naoki, "Inada, Naoki", Inada
+Guido van Rossum, "van Rossum, Guido (GvR)", GvR
+Just van Rossum, "van Rossum, Just (JvR)", JvR
+The Python core team and community, The Python core team and community, python-dev
+P.J. Eby, "Eby, Phillip J.", Eby
+Greg Ewing,  "Ewing, Gregory", Ewing
+Jim Jewett, "Jewett, Jim J.", Jewett
+Nathaniel Smith, "Smith, Nathaniel J.", Smith
+Martin v. Löwis, "von Löwis, Martin", von Löwis
+
diff --git a/pep_extensions/pep_zero_generator/pep_0_parser.py b/pep_extensions/pep_zero_generator/pep_0_parser.py
@@ -1,4 +1,3 @@
-# -*- coding: utf-8 -*-
 """Code for handling object representation of a PEP."""
 import re
 import textwrap
@@ -50,16 +49,38 @@ class Author:
             The author's email address.
     """
 
-    def __init__(self, author_and_email_tuple: Tuple[str, str], authors_lookup: dict):
+    def __init__(self, author_and_email_tuple: Tuple[str, str], authors_exceptions: dict):
         """Parse the name and email address of an author."""
+        self.first = self.last = ''
+
         name, email = author_and_email_tuple
         self.first_last: str = name.strip()
         self.email: str = email.lower()
 
-        name_dict = authors_lookup[self.first_last]
+        name_dict = authors_exceptions.get(self.first_last)
+        if name_dict:
+            self.last_first = name_dict["Surname First"]
+            self.nick = self.last = name_dict["Name Reference"]
+        else:
+            self.set_name_parts()
+
+    def set_name_parts(self):
+        name_dict = self._parse_name(self.first_last)
+        suffix = name_dict.get("suffix")
+        if "name" in name_dict:
+            self.last_first = name_dict["name"]
+            self.nick = name_dict["name"]
+        else:
+            self.first = name_dict["forename"].rstrip()
+            self.last = name_dict["surname"]
+            if self.last[1] == ".":
+                # Add an escape to avoid docutils turning `v.` into `22.`.
+                self.last = "\\" + self.last
+            self.last_first = ", ".join([self.last, self.first])
+            self.nick = self.last
 
-        self.last_first: str = name_dict["Surname First"]
-        self.nick: str = name_dict["Name Reference"]
+        if suffix:
+            self.last_first += f", {suffix}"
 
     def __hash__(self):
         return hash(self.first_last)
@@ -82,6 +103,61 @@ def sort_by(self) -> str:
             base = last.lower()
         return unicodedata.normalize("NFKD", base)
 
+    @staticmethod
+    def _parse_name(full_name):
+        """Decompose a full name into parts.
+
+        If a mononym (e.g, 'Aahz') then return the full name. If there are
+        suffixes in the name (e.g. ', Jr.' or 'III'), then find and extract
+        them. If there is a middle initial followed by a full stop, then
+        combine the following words into a surname (e.g. N. Vander Weele). If
+        there is a leading, lowercase portion to the last name (e.g. 'van' or
+        'von') then include it in the surname.
+
+        """
+        possible_suffixes = ["Jr", "Jr.", "II", "III"]
+
+        suffix_partition = full_name.partition(",")
+        pre_suffix = suffix_partition[0].strip()
+        suffix = suffix_partition[2].strip()
+
+        name_parts = pre_suffix.split(" ")
+        num_parts = len(name_parts)
+        name = {"suffix": suffix}
+
+        if num_parts == 0:
+            raise ValueError("Name is empty!")
+        elif num_parts == 1:
+            name.update(name=name_parts[0])
+        elif num_parts == 2:
+            name.update(forename=name_parts[0], surname=name_parts[1])
+        elif num_parts > 2:
+            # handles III etc.
+            if name_parts[-1] in possible_suffixes:
+                new_suffix = " ".join([*name_parts[-1:], suffix]).strip()
+                name_parts.pop(-1)
+                name.update(suffix=new_suffix)
+
+            # handles von, van, v. etc.
+            if name_parts[-2].islower():
+                forename = " ".join(name_parts[:-2])
+                surname = " ".join(name_parts[-2:])
+                name.update(forename=forename, surname=surname)
+
+            # handles double surnames after a middle initial (e.g. N. Vander Weele)
+            elif any(s.endswith(".") for s in name_parts):
+                split_position = [i for i, x in enumerate(name_parts) if x.endswith(".")][-1] + 1
+                forename = " ".join(name_parts[:split_position])
+                surname = " ".join(name_parts[split_position:])
+                name.update(forename=forename, surname=surname)
+
+            else:
+                forename = " ".join(name_parts[:-1])
+                surname = " ".join(name_parts[-1:])
+                name.update(forename=forename, surname=surname)
+
+        return name
+
 
 class PEP:
     """Representation of PEPs.

diff --git a/pep_extensions/pep_zero_generator/pep_0_writer.py b/pep_extensions/pep_zero_generator/pep_0_writer.py
@@ -2,6 +2,7 @@
 import datetime
 import unicodedata
 from functools import partial
+from itertools import groupby
 from operator import attrgetter
 from typing import List
 
@@ -172,9 +173,9 @@ def verify_email_addresses(peps):
 
     @staticmethod
     def sort_authors(authors_dict):
-        authors_list = list(authors_dict.keys())
-        authors_list.sort(key=attrgetter("sort_by"))
-        return authors_list
+        authors_list = sorted(authors_dict.keys(), key=attrgetter("sort_by"))
+        unique_authors = [next(a) for k, a in groupby(authors_list, key=attrgetter("last_first"))]
+        return unique_authors
 
     @staticmethod
     def normalized_last_first(name):