epstein-files 1.0.14__py3-none-any.whl → 1.0.16__py3-none-any.whl

epstein_files/util/constants.py

@@ -198,7 +198,7 @@ EMAILERS = [
     'Vladimir Yudashkin',
 ]
 
-EMAILER_REGEXES = deepcopy(EMAILER_ID_REGEXES)
+EMAILER_REGEXES = deepcopy(EMAILER_ID_REGEXES)  # Keep a copy without the simple EMAILERS regexes
 
 # Add simple matching regexes for EMAILERS entries to EMAILER_REGEXES
 for emailer in EMAILERS:
@@ -208,73 +208,6 @@ for emailer in EMAILERS:
     EMAILER_REGEXES[emailer] = re.compile(emailer, re.IGNORECASE)
 
 
-##########################
-# OtherFile config stuff #
-##########################
-
-# strings
-FBI = 'FBI'
-MEME = 'meme of'
-PRESS_RELEASE = 'press release'
-RESUME_OF = 'professional resumé'
-SCREENSHOT = 'screenshot of'
-TRANSLATION = 'translation of'
-TWEET = 'tweet'
-
-# Legal cases
-BRUNEL_V_EPSTEIN = f"{JEAN_LUC_BRUNEL} v. {JEFFREY_EPSTEIN} and Tyler McDonald d/b/a YI.org"
-EDWARDS_V_DERSHOWITZ = f"{BRAD_EDWARDS} & {PAUL_G_CASSELL} v. {ALAN_DERSHOWITZ}"
-EPSTEIN_V_ROTHSTEIN_EDWARDS = f"Epstein v. Scott Rothstein, {BRAD_EDWARDS}, and L.M."
-GIUFFRE_V_DERSHOWITZ = f"{VIRGINIA_GIUFFRE} v. {ALAN_DERSHOWITZ}"
-GIUFFRE_V_EPSTEIN = f"{VIRGINIA_GIUFFRE} v. {JEFFREY_EPSTEIN}"
-GIUFFRE_V_MAXWELL = f"{VIRGINIA_GIUFFRE} v. {GHISLAINE_MAXWELL}"
-JANE_DOE_V_EPSTEIN_TRUMP = f"Jane Doe v. Donald Trump and {JEFFREY_EPSTEIN}"
-JANE_DOE_V_USA = 'Jane Doe #1 and Jane Doe #2 v. United States'
-NEW_YORK_V_EPSTEIN = f"New York v. {JEFFREY_EPSTEIN}"
-
-# Descriptions of non-email, non-text message files
-ARTICLE_DRAFT = 'draft of an article about'
-BOFA_MERRILL = f'{BOFA} / Merrill Lynch Report'
-BOFA_WEALTH_MGMT = f'{BOFA} Wealth Management'
-BROCKMAN_INC = 'Brockman, Inc.'
-CVRA = "Crime Victims' Rights Act [CVRA]"
-DAVID_BLAINE_VISA_LETTER = f"letter of recommendation for visa for a model"
-DERSH_GIUFFRE_TWEET = f"{TWEET} about {VIRGINIA_GIUFFRE}"
-DEUTSCHE_BANK_TAX_TOPICS = f'{DEUTSCHE_BANK} Wealth Management Tax Topics'
-DIANA_DEGETTE_CAMPAIGN = "Colorado legislator Diana DeGette's campaign"
-EPSTEIN_FOUNDATION = 'Jeffrey Epstein VI Foundation'
-FBI_REPORT = f"report on Epstein investigation (redacted)"
-FBI_SEIZED_PROPERTY = f"seized property inventory (redacted)"
-FEMALE_HEALTH_COMPANY = 'Female Health Company (FHX)'
-FIRE_AND_FURY = f"Fire And Fury"
-HARVARD_POETRY = f'{HARVARD} poetry stuff from {LISA_NEW}'
-HBS_APPLICATION = f"{HARVARD} Business School application letter"
-JASTA = 'JASTA'
-JASTA_SAUDI_LAWSUIT = f"{JASTA} lawsuit against Saudi Arabia by 9/11 victims"
-JP_MORGAN_EYE_ON_THE_MARKET = f"Eye On The Market"
-LAWRENCE_KRAUSS_ASU_ORIGINS = f"{LAWRENCE_KRAUSS}'s ASU Origins Project"
-LEXIS_NEXIS_CVRA_SEARCH = f"{LEXIS_NEXIS} search for case law around the {CVRA}"
-KEN_STARR_LETTER = f"letter to judge overseeing Epstein's criminal prosecution, mentions Alex Acosta"
-MICHAEL_WOLFF_EPSTEIN_ARTICLE_DRAFT = f"draft of an unpublished article about Epstein by {MICHAEL_WOLFF} written ca. 2014/2015"
-NIGHT_FLIGHT_BOOK = f'"Night Flight" (draft)'
-NOBEL_CHARITABLE_TRUST = 'Nobel Charitable Trust'
-OBAMA_JOKE = 'joke about Obama'
-PALM_BEACH_CODE_ENFORCEMENT = f'{PALM_BEACH} Code Enforcement'
-PALM_BEACH_PROPERTY_INFO = f"{PALM_BEACH} property info"
-PALM_BEACH_TSV = f"TSV of {PALM_BEACH} property"
-PALM_BEACH_WATER_COMMITTEE = f'{PALM_BEACH} Water Committee'
-PATTERSON_BOOK_SCANS = f'pages of "Filthy Rich: The Shocking True Story of {JEFFREY_EPSTEIN}"'
-REAL_DEAL_ARTICLE = 'article by Keith Larsen'
-SHIMON_POST_ARTICLE = f'selection of articles about the mideast'
-SINGLE_PAGE = 'single page of'
-STRANGE_BEDFELLOWS = "'Strange Bedfellows' list of invitees f. Johnny Depp, Woody Allen, Obama, and more"
-SWEDISH_LIFE_SCIENCES_SUMMIT = f"{BARBRO_C_EHNBOM}'s Swedish American Life Science Summit (SALSS)"
-TRUMP_DISCLOSURES = f"Donald Trump financial disclosures from U.S. Office of Government Ethics"
-UBS_CIO_REPORT = 'CIO Monthly Extended report'
-UN_GENERAL_ASSEMBLY = '67th U.N. General Assembly'
-WOMEN_EMPOWERMENT = f"Women Empowerment (WE) conference"
-ZUBAIR_AND_ANYA = f"{ZUBAIR_KHAN} and Anya Rasulova"
-
 # Atribution reasons
 BOLOTOVA_REASON = 'Same signature style as 029020 ("--" followed by "Sincerely Renata Bolotova")'
 KATHY_REASON = 'from "Kathy" about dems, sent from iPad'
@@ -369,6 +302,8 @@ TEXTS_CONFIG = CONFIRMED_TEXTS_CONFIG + UNCONFIRMED_TEXTS_CONFIG
 ################################################ EMAILS ################################################
 ########################################################################################################
 
+MICHAEL_WOLFF_EPSTEIN_ARTICLE_DRAFT = f"draft of an unpublished article about Epstein by {MICHAEL_WOLFF} written ca. 2014/2015"
+
 # Some emails have a lot of uninteresting CCs
 IRAN_DEAL_RECIPIENTS = ['Allen West', 'Rafael Bardaji', 'Philip Kafka', 'Herb Goodman', 'Grant Seeger', 'Lisa Albert', 'Janet Kafka', 'James Ramsey', 'ACT for America', 'John Zouzelka', 'Joel Dunn', 'Nate McClain', 'Bennet Greenwald', 'Taal Safdie', 'Uri Fouzailov', 'Neil Anderson', 'Nate White', 'Rita Hortenstine', 'Henry Hortenstine', 'Gary Gross', 'Forrest Miller', 'Bennett Schmidt', 'Val Sherman', 'Marcie Brown', 'Michael Horowitz', 'Marshall Funk']
 FLIGHT_IN_2012_PEOPLE = ['Francis Derby', 'Januiz Banasiak', 'Louella Rabuyo', 'Richard Barnnet']
@@ -601,6 +536,7 @@ EMAILS_CONFIG = [
     EmailCfg(id='033274', recipients=[JEFFREY_EPSTEIN]),  # this is a note sent to self
     EmailCfg(id='032780', recipients=[JEFFREY_EPSTEIN]),  # Bad OCR (nofix)
     EmailCfg(id='029324', recipients=[JEFFREY_EPSTEIN, "Jojo Fontanilla", "Lyn Fontanilla"]),  # Bad OCR (nofix)
+    EmailCfg(id='013482', recipients=[JEFFREY_EPSTEIN], is_fwded_article=True),  # other recipients redacted. "The view from the US: Stem cell therapy steps up a gear with firs"
     EmailCfg(id='033456', recipients=["Joel"], attribution_reason='Reply'),
     EmailCfg(id='033458', recipients=["Joel"], attribution_reason='Reply'),
     EmailCfg(id='033460', recipients=["Joel"], attribution_reason='Reply'),
@@ -691,7 +627,6 @@ EMAILS_CONFIG = [
     EmailCfg(id='021740', is_fwded_article=True),  # Miami Herald article about Epstein prosecutor
     EmailCfg(id='023126', is_fwded_article=True),  # Miami Herald on Alex Acosta
     EmailCfg(id='029625', is_fwded_article=True),  # Conchita Sarnoff Daily Beast Articles - Epstein Sex Trafficking Investigation and Settlement
-    EmailCfg(id='013482', is_fwded_article=True),  # The view from the US: Stem cell therapy steps up a gear with firs
     EmailCfg(id='029505', is_fwded_article=True),  # Foreign Policy Middle Eastern Monarchs Look at the Trump
     EmailCfg(id='029859', is_fwded_article=True),  # Palm Beach Post: Epstein paid three women $5.5 million to end lawsuits
     EmailCfg(id='031988', is_fwded_article=True),  # NYT review of Inside Job
@@ -866,6 +801,65 @@ EMAILS_CONFIG = [
 ####################################### OTHER FILES ############################################
 ################################################################################################
 
+# strings
+FBI = 'FBI'
+MEME = 'meme of'
+PRESS_RELEASE = 'press release'
+RESUME_OF = 'professional resumé'
+SCREENSHOT = 'screenshot of'
+TRANSLATION = 'translation of'
+TWEET = 'tweet'
+
+# Legal cases
+BRUNEL_V_EPSTEIN = f"{JEAN_LUC_BRUNEL} v. {JEFFREY_EPSTEIN} and Tyler McDonald d/b/a YI.org"
+EDWARDS_V_DERSHOWITZ = f"{BRAD_EDWARDS} & {PAUL_G_CASSELL} v. {ALAN_DERSHOWITZ}"
+EPSTEIN_V_ROTHSTEIN_EDWARDS = f"Epstein v. Scott Rothstein, {BRAD_EDWARDS}, and L.M."
+GIUFFRE_V_DERSHOWITZ = f"{VIRGINIA_GIUFFRE} v. {ALAN_DERSHOWITZ}"
+GIUFFRE_V_EPSTEIN = f"{VIRGINIA_GIUFFRE} v. {JEFFREY_EPSTEIN}"
+GIUFFRE_V_MAXWELL = f"{VIRGINIA_GIUFFRE} v. {GHISLAINE_MAXWELL}"
+JANE_DOE_V_EPSTEIN_TRUMP = f"Jane Doe v. Donald Trump and {JEFFREY_EPSTEIN}"
+JANE_DOE_V_USA = 'Jane Doe #1 and Jane Doe #2 v. United States'
+NEW_YORK_V_EPSTEIN = f"New York v. {JEFFREY_EPSTEIN}"
+
+# Descriptions of non-email, non-text message files
+ARTICLE_DRAFT = 'draft of an article about'
+BOFA_WEALTH_MGMT = f'{BOFA} Wealth Management'
+BROCKMAN_INC = 'Brockman, Inc.'
+CVRA = "Crime Victims' Rights Act [CVRA]"
+DAVID_BLAINE_VISA_LETTER = f"letter of recommendation for visa for a model"
+DERSH_GIUFFRE_TWEET = f"{TWEET} about {VIRGINIA_GIUFFRE}"
+DEUTSCHE_BANK_TAX_TOPICS = f'{DEUTSCHE_BANK} Wealth Management Tax Topics'
+DIANA_DEGETTE_CAMPAIGN = "Colorado legislator Diana DeGette's campaign"
+FBI_REPORT = f"report on Epstein investigation (redacted)"
+FBI_SEIZED_PROPERTY = f"seized property inventory (redacted)"
+FEMALE_HEALTH_COMPANY = 'Female Health Company (FHX)'
+FIRE_AND_FURY = f"Fire And Fury"
+HARVARD_POETRY = f'{HARVARD} poetry stuff from {LISA_NEW}'
+HBS_APPLICATION = f"{HARVARD} Business School application letter"
+JASTA = 'JASTA'
+JASTA_SAUDI_LAWSUIT = f"{JASTA} lawsuit against Saudi Arabia by 9/11 victims"
+JP_MORGAN_EYE_ON_THE_MARKET = f"Eye On The Market"
+LAWRENCE_KRAUSS_ASU_ORIGINS = f"{LAWRENCE_KRAUSS}'s ASU Origins Project"
+KEN_STARR_LETTER = f"letter to judge overseeing Epstein's criminal prosecution, mentions Alex Acosta"
+LEXIS_NEXIS_CVRA_SEARCH = f"{LEXIS_NEXIS} search for case law around the {CVRA}"
+NOBEL_CHARITABLE_TRUST = 'Nobel Charitable Trust'
+OBAMA_JOKE = 'joke about Obama'
+PALM_BEACH_CODE_ENFORCEMENT = f'{PALM_BEACH} Code Enforcement'
+PALM_BEACH_PROPERTY_INFO = f"{PALM_BEACH} property info"
+PALM_BEACH_TSV = f"TSV of {PALM_BEACH} property"
+PALM_BEACH_WATER_COMMITTEE = f'{PALM_BEACH} Water Committee'
+PATTERSON_BOOK_SCANS = f'pages of "Filthy Rich: The Shocking True Story of {JEFFREY_EPSTEIN}"'
+REAL_DEAL_ARTICLE = 'article by Keith Larsen'
+SHIMON_POST_ARTICLE = f'selection of articles about the mideast'
+STRANGE_BEDFELLOWS = "'Strange Bedfellows' list of invitees f. Johnny Depp, Woody Allen, Obama, and more"
+SWEDISH_LIFE_SCIENCES_SUMMIT = f"{BARBRO_C_EHNBOM}'s Swedish American Life Science Summit (SALSS)"
+TRUMP_DISCLOSURES = f"Donald Trump financial disclosures from U.S. Office of Government Ethics"
+UBS_CIO_REPORT = 'CIO Monthly Extended report'
+UN_GENERAL_ASSEMBLY = '67th U.N. General Assembly'
+WOMEN_EMPOWERMENT = f"Women Empowerment (WE) conference"
+ZUBAIR_AND_ANYA = f"{ZUBAIR_KHAN} and Anya Rasulova"
+
+
 OTHER_FILES_BOOKS = [
     DocCfg(id='017088', author=ALAN_DERSHOWITZ, description=f'"Taking the Stand: My Life in the Law" (draft)'),
     DocCfg(id='013501', author='Arnold J. Mandell', description=f'The Nearness Of Grace: A Personal Science Of Spiritual Transformation', date='2005-01-01'),
@@ -873,7 +867,7 @@ OTHER_FILES_BOOKS = [
     DocCfg(id='018438', author='Clarisse Thorn', description=f'The S&M Feminist'),
     DocCfg(id='019477', author=EDWARD_JAY_EPSTEIN, description=f'How America Lost Its Secrets: Edward Snowden, the Man, and the Theft'),
     DocCfg(id='020153', author=EDWARD_JAY_EPSTEIN, description=f'The Snowden Affair: A Spy Story In Six Parts'),
-    DocCfg(id='011472', author=EHUD_BARAK, description=NIGHT_FLIGHT_BOOK, date='2006-07-12', duplicate_ids=['027849']),  # date from _extract_timestamp()
+    DocCfg(id='011472', author=EHUD_BARAK, description=f'"Night Flight" (draft)', date='2006-07-12', duplicate_ids=['027849']),  # date from _extract_timestamp()
     DocCfg(id='010912', author=GORDON_GETTY, description=f'"Free Growth and Other Surprises" (draft)', date='2018-10-18'),
     DocCfg(id='010477', author=JAMES_PATTERSON, description=PATTERSON_BOOK_SCANS, date='2016-10-10'),
     DocCfg(id='010486', author=JAMES_PATTERSON, description=PATTERSON_BOOK_SCANS, date='2016-10-10'),
@@ -1002,7 +996,7 @@ OTHER_FILES_ARTICLES = [
     DocCfg(id='024997', author=SHIMON_POST, description=SHIMON_POST_ARTICLE, date='2011-09-08'),
     DocCfg(id='031941', author=SHIMON_POST, description=SHIMON_POST_ARTICLE, date='2011-11-17'),
     DocCfg(id='030829', author=f'South Florida Sun Sentinel', description=f'article about {BRAD_EDWARDS} and {JEFFREY_EPSTEIN}'),
-    DocCfg(id='021092', author='Tatler', description=f'{SINGLE_PAGE} of article about {GHISLAINE_MAXWELL} shredding documents', date='2019-08-15'),
+    DocCfg(id='021092', author='Tatler', description=f'single page of article about {GHISLAINE_MAXWELL} shredding documents', date='2019-08-15'),
     DocCfg(id='030333', author=f'The Independent', description=f'article about Prince Andrew, Epstein, and Epstein\'s butler who stole his address book'),
     DocCfg(id='010754', author=f'U.S. News', description=f"article about Yitzhak Rabin"),
     DocCfg(id='014498', author=VI_DAILY_NEWS, description='article', date='2016-12-13'),
@@ -1011,7 +1005,7 @@ OTHER_FILES_ARTICLES = [
     DocCfg(id='023046', author=VI_DAILY_NEWS, description='article', date='2019-02-27'),
     DocCfg(id='031170', author=VI_DAILY_NEWS, description='article', date='2019-03-06'),
     DocCfg(id='016506', author=VI_DAILY_NEWS, description='article', date='2019-02-28'),
-    DocCfg(id='016507', author=VI_DAILY_NEWS, description=f"'Perversion of Justice' by {JULIE_K_BROWN}", date='2018-12-19'),
+    DocCfg(id='016507', author=VI_DAILY_NEWS, description=f'"Perversion of Justice" by {JULIE_K_BROWN}', date='2018-12-19'),
     DocCfg(id='019212', author=WAPO, description=f'and Times Tribune articles about Bannon, Trump, and healthcare execs'),
     DocCfg(id='033379', author=WAPO, description=f'"How Washington Pivoted From Finger-Wagging to Appeasement" (about Viktor Orban)', date='2018-05-25'),
     DocCfg(
@@ -1026,12 +1020,12 @@ OTHER_FILES_ARTICLES = [
     DocCfg(id='026648', description=f'article about {JASTA} lawsuit against Saudi Arabia by 9/11 victims (Russian propaganda?)', date='2017-05-13'),
     DocCfg(id='032159', description=f"article about microfinance and cell phones in Zimbabwe, Strive Masiyiwa (Econet Wireless)"),
     DocCfg(id='030825', description=f'{ARTICLE_DRAFT} Syria'),
-    DocCfg(id='027051', description=f"German language article about the 2013 Lifeball / AIDS Gala", date='2013-01-01'),
+    DocCfg(id='027051', description=f"German article about the 2013 Lifeball / AIDS Gala", date='2013-01-01', attached_to_email_id='027049'),
     DocCfg(id='033480', description=f"John Bolton press clipping", date='2018-04-06', duplicate_ids=['033481']),
     DocCfg(id='013403', description=f"{LEXIS_NEXIS} result from The Evening Standard about Bernie Madoff", date='2009-12-24'),
     DocCfg(id='021093', description=f"page of unknown article about Epstein and Maxwell"),
-    DocCfg(id='031191', description=f"{SINGLE_PAGE} unknown article about Epstein and Trump's relationship in 1997"),
-    DocCfg(id='026520', description=f'Spanish language article about {SULTAN_BIN_SULAYEM}', date='2013-09-27'),
+    DocCfg(id='031191', description=f"single page of unknown article about Epstein and Trump's relationship in 1997"),
+    DocCfg(id='026520', description=f'Spanish article about {SULTAN_BIN_SULAYEM}', date='2013-09-27'),
     DocCfg(
         id='031736',
         description=f"{TRANSLATION} Arabic article by Abdulnaser Salamah 'Trump; Prince of Believers (Caliph)!'",
@@ -1186,16 +1180,16 @@ OTHER_FILES_LEGAL = [
 ]
 
 OTHER_FILES_CONFERENCES = [
-    DocCfg(id='014315', author=BOFA_MERRILL, description=f'2016 Future of Financials Conference, attached to 014312'),
+    DocCfg(id='014315', author=BOFA_MERRILL, description=f'2016 Future of Financials Conference', attached_to_email_id='014312'),
     DocCfg(id='026825', author=DEUTSCHE_BANK, description=f"Asset & Wealth Management featured speaker bios"),  # Really "Deutsche Asset" which may not be Deutsche Bank?
     DocCfg(id='023123', author=LAWRENCE_KRAUSS_ASU_ORIGINS, description=f"{STRANGE_BEDFELLOWS} (old draft)"),
     DocCfg(id='023120', author=LAWRENCE_KRAUSS_ASU_ORIGINS, description=STRANGE_BEDFELLOWS, duplicate_ids=['023121'], dupe_type='earlier'),
-    DocCfg(id='031359', author=NOBEL_CHARITABLE_TRUST, description=f"Earth Environment Convention about ESG investing"),
+    DocCfg(id='031359', author=NOBEL_CHARITABLE_TRUST, description=f'"Earth Environment Convention" about ESG investing'),
     DocCfg(id='031354', author=NOBEL_CHARITABLE_TRUST, description=f'"Thinking About the Environment and Technology" report 2011'),
     DocCfg(id='019300', author=SVETLANA_POZHIDAEVA, description=f'{WOMEN_EMPOWERMENT} f. {KATHRYN_RUEMMLER}', date='2019-04-05'),
     DocCfg(id='022267', author=SVETLANA_POZHIDAEVA, description=f'{WOMEN_EMPOWERMENT} founder essay about growing the seminar business'),
     DocCfg(id='022407', author=SVETLANA_POZHIDAEVA, description=f'{WOMEN_EMPOWERMENT} seminar pitch deck'),
-    DocCfg(id='017524', author=SWEDISH_LIFE_SCIENCES_SUMMIT, description=f"2012 program emailed to epstein BY {BARBRO_C_EHNBOM} in 031226", date='2012-08-18'),
+    DocCfg(id='017524', author=SWEDISH_LIFE_SCIENCES_SUMMIT, description=f"2012 program", date='2012-08-18', attached_to_email_id='031226'),
     DocCfg(id='026747', author=SWEDISH_LIFE_SCIENCES_SUMMIT, description=f"2017 program", date='2017-08-23'),
     DocCfg(id='014951', author='TED Talks', description=f"2017 program", date='2017-04-20'),
     DocCfg(id='024179', author=UN_GENERAL_ASSEMBLY, description=f'president and first lady schedule', date='2012-09-21'),
@@ -1411,7 +1405,7 @@ OTHER_FILES_POLITICS = [
         description=f"'Breaking Down Democracy: Goals, Strategies, and Methods of Modern Authoritarians'",
         date='2017-06-02',
     ),
-    DocCfg(id='026856', author='Kevin Rudd', description=f"speech 'Xi Jinping, China And The Global Order'", date='2018-06-26'),
+    DocCfg(id='026856', author='Kevin Rudd', description=f'speech "Xi Jinping, China And The Global Order"', date='2018-06-26'),
     DocCfg(id='026827', author='Scowcroft Group', description=f'report on ISIS', date='2015-11-14'),
     DocCfg(id='024294', author=STACEY_PLASKETT, description=f"campaign flier", date='2016-10-01'),
     DocCfg(
@@ -1424,7 +1418,7 @@ OTHER_FILES_POLITICS = [
     DocCfg(
         id='025849',
         author='US Office of Government Information Services',
-        description=f"Building a Bridge Between FOIA Requesters & Agencies",
+        description=f'"Building a Bridge Between FOIA Requesters & Agencies"',
     ),
     DocCfg(id='031670', description=f"letter from General Mike Flynn's lawyers to senators Mark Warner & Richard Burr about subpoena"),
     DocCfg(
@@ -1502,6 +1496,13 @@ OTHER_FILES_RESUMES = [
 OTHER_FILES_ARTS = [
     DocCfg(id='018703', author=ANDRES_SERRANO, description=f"artist statement about Trump objects"),
     DocCfg(id='023438', author=BROCKMAN_INC, description=f"announcement of auction of 'Noise' by Daniel Kahneman, Olivier Sibony, and Cass Sunstein"),
+    DocCfg(
+        id='025147',
+        author=BROCKMAN_INC,
+        description=f'hot list Frankfurt Book Fair (includes article about Silk Road/Ross Ulbricht)',
+        date='2016-10-23',
+        is_interesting=True,
+    ),
     DocCfg(id='030769', author='Independent Filmmaker Project (IFP)', description=f"2017 Gotham Awards invitation"),
     DocCfg(
         id='025205',
@@ -1526,12 +1527,6 @@ OTHER_FILES_MISC = [
         author=f'linkspirit (French?) and {LAWRENCE_KRAUSS}',
         is_interesting=True,  # we don't know who linkspirit is yet
     ),
-    DocCfg(
-        id='025147',
-        author=BROCKMAN_INC,
-        description=f'hot list Frankfurt Book Fair (includes article about Silk Road/Ross Ulbricht)',
-        date='2016-10-23',
-    ),
     DocCfg(id='022494', author='DOJ', description=f'Foreign Corrupt Practices Act (FCPA) Resource Guide'),
     DocCfg(id='023096', author=EPSTEIN_FOUNDATION, description=f'blog post', date='2012-11-15'),
     DocCfg(id='029326', author=EPSTEIN_FOUNDATION, description=f'{PRESS_RELEASE}', date='2013-02-15'),
@@ -1539,7 +1534,7 @@ OTHER_FILES_MISC = [
     DocCfg(id='027071', author=FEMALE_HEALTH_COMPANY, description=f"brochure requesting donations for female condoms in Uganda"),
     DocCfg(id='027074', author=FEMALE_HEALTH_COMPANY, description=f"pitch deck (USAID was a customer)"),
     DocCfg(id='032735', author=GORDON_GETTY, description=f"on Trump", date='2018-03-20'),  # Dated based on concurrent emails from Getty
-    DocCfg(id='025540', author=JEFFREY_EPSTEIN, description=f"rough draft of Epstein's side of the story?"),
+    DocCfg(id='025540', author=JEFFREY_EPSTEIN, description=f"rough draft of his side of the story"),
     DocCfg(id='026634', author='Michael Carrier', description=f"comments about an Apollo linked hedge fund 'DE Fund VIII'"),
     DocCfg(id='031425', author=SCOTT_J_LINK, description=f'completely redacted email from', is_interesting=False),
     DocCfg(id='020447', author='Working Group on Chinese Influence Activities in the U.S.', description=f'Promoting Constructive Vigilance'),

epstein_files/util/doc_cfg.py

@@ -36,7 +36,7 @@ FIELD_SORT_KEY = {
 }
 
 FINANCIAL_REPORTS_AUTHORS = [
-    BOFA,
+    BOFA_MERRILL,
     DEUTSCHE_BANK,
     ELECTRON_CAPITAL_PARTNERS,
     GOLDMAN_INVESTMENT_MGMT,
@@ -73,6 +73,7 @@ class DocCfg:
         is_synthetic (bool): True if this config was generated by the duplicate_cfgs() method
     """
     id: str
+    attached_to_email_id: str | None = None
     author: str | None = None
     category: str | None = None
     date: str | None = None
@@ -102,10 +103,10 @@ class DocCfg:
             return f"{msg} {self.description}" if self.description else msg
         elif self.author and self.description:
             if self.category in [ACADEMIA, BOOK]:
-                title = self.description if '"' in self.description else f"'{self.description}'"
+                title = self.description if '"' in self.description else f'"{self.description}"'
                 return f"{title} by {self.author}"
             elif self.category == FINANCE and self.author in FINANCIAL_REPORTS_AUTHORS:
-                return f"{self.author} report: '{self.description}'"
+                return f'{self.author} report: "{self.description}"'
             elif self.category == LEGAL and 'v.' in self.author:
                 return f"{self.author}: {self.description}"
         elif self.category and self.author is None and self.description is None:

epstein_files/util/env.py

@@ -8,10 +8,8 @@ from rich_argparse_plus import RichHelpFormatterPlus
 
 from epstein_files.util.logging import env_log_level, logger
 
-COUNT_WORDS_SCRIPT = 'epstein_word_count'
 DEFAULT_WIDTH = 145
-HTML_SCRIPTS = ['epstein_generate', COUNT_WORDS_SCRIPT]
-EPSTEIN_DOCS_DIR_ENV_VAR_NAME = 'EPSTEIN_DOCS_DIR'
+HTML_SCRIPTS = ['epstein_generate', 'epstein_word_count']
 
 
 RichHelpFormatterPlus.choose_theme('morning_glory')
@@ -50,6 +48,7 @@ args = parser.parse_args()
 
 
 # Verify Epstein docs can be found
+EPSTEIN_DOCS_DIR_ENV_VAR_NAME = 'EPSTEIN_DOCS_DIR'
 DOCS_DIR_ENV = environ.get(EPSTEIN_DOCS_DIR_ENV_VAR_NAME)
 DOCS_DIR = Path(DOCS_DIR_ENV or '').resolve()
 
@@ -65,13 +64,13 @@ is_env_var_set = lambda s: len(environ.get(s) or '') > 0
 is_html_script = current_script in HTML_SCRIPTS
 
 args.debug = args.deep_debug or args.debug or is_env_var_set('DEBUG')
+args.names = [None if n == 'None' else n for n in (args.names or [])]
 args.output_emails = args.output_emails or args.all_emails
 args.output_other = args.output_other or args.all_other_files or args.uninteresting
 args.overwrite_pickle = args.overwrite_pickle or (is_env_var_set('OVERWRITE_PICKLE') and not is_env_var_set('PICKLED'))
 args.width = args.width if is_html_script else None
-is_output_selected = any([arg.startswith('output_') and value for arg, value in vars(args).items()])
-is_output_selected = is_output_selected or args.json_metadata or args.colors_only
-specified_names: list[str | None] = [None if n == 'None' else n for n in (args.names or [])]
+is_any_output_selected = any([arg.startswith('output_') and value for arg, value in vars(args).items()])
+is_any_output_selected = is_any_output_selected or args.json_metadata or args.colors_only
 
 # Log level args
 if args.deep_debug:
@@ -86,9 +85,12 @@ elif not env_log_level:
 logger.info(f'Log level set to {logger.level}...')
 
 # Massage args that depend on other args to the appropriate state
-if current_script == 'epstein_generate' and not (is_output_selected or args.make_clean):
+if current_script == 'epstein_generate' and not (is_any_output_selected or args.make_clean):
     logger.warning(f"No output section chosen; outputting default selection of texts, selected emails, and other files...")
     args.output_texts = args.output_emails = args.output_other = True
 
 if args.debug:
-    logger.warning(f"Invocation args:\ncurrent_script={current_script}\nis_html_script={is_html_script},\nis_output_selected={is_output_selected}\nspecified_names={specified_names},\nargs={args}")
+    logger.warning(f"Invocation args:\ncurrent_script={current_script}\nis_html_script={is_html_script},\nis_output_selected={is_any_output_selected},\nargs={args}")
+
+if args.names:
+    logger.warning(f"Output restricted to {args.names}")

epstein_files/util/file_helper.py

@@ -12,7 +12,6 @@ KB = 1024
 MB = KB * KB
 
 file_size = lambda file_path: Path(file_path).stat().st_size
-file_size_str = lambda file_path: file_size_to_str(file_size(file_path))
 
 # Coerce methods handle both string and int arguments.
 coerce_file_name = lambda filename_or_id: coerce_file_stem(filename_or_id) + '.txt'
@@ -46,8 +45,12 @@ def extract_file_id(filename_or_id: int | str | Path) -> str:
     return file_match.group(1)
 
 
-def file_size_to_str(size: int) -> str:
-    digits = 2
+def file_size_str(file_path, digits: int | None = None):
+    return file_size_to_str(file_size(file_path), digits)
+
+
+def file_size_to_str(size: int, digits: int | None = None) -> str:
+    _digits = 2
 
     if size > MB:
         size_num = float(size) / MB
@@ -55,10 +58,11 @@ def file_size_to_str(size: int) -> str:
     elif size > KB:
         size_num = float(size) / KB
         size_str = 'kb'
-        digits = 1
+        _digits = 1
     else:
         return f"{size} b"
 
+    digits = _digits if digits is None else digits
     return f"{size_num:,.{digits}f} {size_str}"