epstein-files 1.2.0__py3-none-any.whl → 1.2.1__py3-none-any.whl

epstein_files/util/highlighted_group.py

@@ -10,7 +10,7 @@ from epstein_files.util.constant.names import *
 from epstein_files.util.constant.strings import *
 from epstein_files.util.constant.urls import ARCHIVE_LINK_COLOR
 from epstein_files.util.constants import (EMAILER_ID_REGEXES, EPSTEIN_V_ROTHSTEIN_EDWARDS,
-     OSBORNE_LLP, REPLY_REGEX, SENT_FROM_REGEX, VIRGIN_ISLANDS)
+     OSBORNE_LLP, REPLY_REGEX, SENT_FROM_REGEX)
 from epstein_files.util.doc_cfg import *
 from epstein_files.util.data import without_falsey
 from epstein_files.util.env import args
@@ -20,17 +20,18 @@ CIVIL_ATTORNEY = 'civil attorney'
 CRIMINAL_DEFENSE_ATTORNEY = 'criminal defense attorney'
 CRIMINAL_DEFENSE_2008 = f"{CRIMINAL_DEFENSE_ATTORNEY} on 2008 case"
 EPSTEIN_LAWYER = 'lawyer'
-EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY = f"{CIVIL_ATTORNEY} in {EPSTEIN_V_ROTHSTEIN_EDWARDS}"
+EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY = f"{CIVIL_ATTORNEY} {EPSTEIN_V_ROTHSTEIN_EDWARDS}"
 ESTATE_EXECUTOR = 'estate executor'
 EPSTEIN_ESTATE_EXECUTOR = f"Epstein {ESTATE_EXECUTOR}"
+MC2_MODEL_MANAGEMENT = f"{JEAN_LUC_BRUNEL}'s MC2 Model Management"
 MIDEAST = 'mideast'
 QUESTION_MARKS_TXT = Text(QUESTION_MARKS, style='grey50')
 REGEX_STYLE_PREFIX = 'regex'
-SIMPLE_NAME_REGEX = re.compile(r"^[-\w ]+$", re.IGNORECASE)
+SIMPLE_NAME_REGEX = re.compile(r"^[-\w, ]+$", re.IGNORECASE)
+TECH_BRO = 'tech bro'
 
 CATEGORY_STYLE_MAPPING = {
     ARTICLE: JOURNALIST,
-    ARTS: ENTERTAINER,
     BOOK: JOURNALIST,
     LEGAL: EPSTEIN_LAWYER,
     POLITICS: LOBBYIST,
@@ -152,7 +153,7 @@ class HighlightedNames(HighlightedText):
             name_patterns = [remove_question_marks(name).replace(' ', r"\s+")]
 
         if ' ' in name:
-            for partial_name in [extract_first_name(name), extract_last_name(name)]:
+            for partial_name in [extract_first_name(name), extract_last_name(name), reversed_name(name)]:
                 if partial_name.lower() not in NAMES_TO_NOT_HIGHLIGHT and SIMPLE_NAME_REGEX.match(partial_name):
                     name_patterns.append(partial_name.replace(' ', r"\s+"))
 
@@ -219,17 +220,19 @@ HIGHLIGHTED_NAMES = [
         label=ACADEMIA,
         style='light_goldenrod2',
         emailers={
-            DAVID_HAIG: 'evolutionary geneticist?',
-            JOSCHA_BACH: 'cognitive science / AI research',
             'Daniel Kahneman': 'Nobel economic sciences laureate and cognitivie psychologist (?)',
-            'Ed Boyden': f'Associate Professor, {MIT_MEDIA_LAB} neurobiology',
+            DAVID_HAIG: 'evolutionary geneticist?',
+            'David Grosof': 'MIT Sloan School of Management',
+            'Ed Boyden': f'{MIT_MEDIA_LAB} neurobiology professor',
             'Harry Fisch': "men's health expert at New York-Presbyterian / Weill Cornell (?)",
-            LAWRENCE_KRAUSS: 'theoretical physicist',
+            JOSCHA_BACH: 'cognitive science / AI research',
+            LAWRENCE_KRAUSS: 'theoretical physicist with #MeToo problems',
             LINDA_STONE: f'ex-Microsoft, {MIT_MEDIA_LAB}',
             MARK_TRAMO: 'professor of neurology at UCLA',
             'Nancy Dahl': f'wife of {LAWRENCE_KRAUSS}',
             NEAL_KASSELL: 'professor of neurosurgery at University of Virginia',
             NOAM_CHOMSKY: f"professor of linguistics at MIT",
+            'Norman Finkelstein': 'scholar, well known critic of Israel',
             PETER_ATTIA: 'longevity medicine',
             ROBERT_TRIVERS: 'evolutionary biology',
             ROGER_SCHANK: 'Teachers College, Columbia University',
@@ -241,7 +244,6 @@ HIGHLIGHTED_NAMES = [
             r"Carl\s*Sagan",
             r"Columbia",
             r"Dan(iel|ny) Kahneman",
-            r"David Grosof",
             r"J(ames|im)\s*Watson",
             r"(Lord\s*)?Martin\s*Rees",
             r"Massachusetts\s*Institute\s*of\s*Technology",
@@ -290,10 +292,48 @@ HIGHLIGHTED_NAMES = [
             r"Zimbabwe(an)?",
         ],
     ),
+    HighlightedNames(
+        label=ARTS,
+        style='light_steel_blue3',
+        emailers={
+            ANDRES_SERRANO: "'Piss Christ' artist",
+            'Barry Josephson': 'American film producer, editor FamilySecurityMatters.org',
+            BILL_SIEGEL: 'documentary film producer and director',
+            DAVID_BLAINE: 'famous magician',
+            'Richard Merkin': 'painter, illustrator and arts educator',
+            STEVEN_PFEIFFER: 'Associate Director at Independent Filmmaker Project (IFP)',
+            'Steven Gaydos': 'American screenwriter and journalist',
+        },
+        patterns=[
+            r"(Art )?Spiegelman",
+            r"Artspace",
+            r"Bobby slayton",
+            r"bono\s*mick",
+            r"Errol(\s*Morris)?",
+            r"Etienne Binant",
+            r"(Frank\s)?Gehry",
+            r"Jagger",
+            r"(Jeffrey\s*)?Katzenberg",
+            r"(Johnny\s*)?Depp",
+            r"Kid Rock",
+            r"(Larry\s*)?Gagosian",
+            r"Lena\s*Dunham",
+            r"Madonna",
+            r"Mark\s*Burnett",
+            r"Phaidon",
+            r"Ramsey Elkholy",
+            r"Regan arts",
+            r"shirley maclaine",
+            r"Woody( Allen)?",
+            r"Zach Braff",
+        ],
+    ),
     HighlightedNames(
         label=BILL_GATES,
         style='turquoise4',
+        category=TECH_BRO,
         emailers={
+            BILL_GATES: 'ex-Microsoft, Gates Foundation, bgC3',
             BORIS_NIKOLIC: f'biotech VC partner of {BILL_GATES}, {EPSTEIN_ESTATE_EXECUTOR}',
         },
         patterns=[
@@ -344,6 +384,7 @@ HIGHLIGHTED_NAMES = [
         emailers={
             ALIREZA_ITTIHADIEH: 'CEO Freestream Aircraft Limited',
             BARBRO_C_EHNBOM: 'Swedish pharmaceuticals, SALSS',
+            BARRY_J_COHEN: None,
             'David Mitchell': 'Mitchell Holdings New York real estate developer',
             FRED_HADDAD: "co-founder of Heck's in West Virginia",
             GERALD_BARTON: "Maryland property developer Landmark Land Company",
@@ -482,7 +523,7 @@ HIGHLIGHTED_NAMES = [
             MERWIN_DELA_CRUZ: None,  # HOUSE_OVERSIGHT_032652 Groff says "Jojo and Merwin both requested off Nov. 25 and 26"
             NADIA_MARCINKO: "Epstein's pilot",
             'Sean J. Lancaster': 'airplane reseller',
-            ZUBAIR_KHAN: 'cybersecurity firm Tranchulas CEO, InsightsPod founder, Islamabad/Dubai',
+            ZUBAIR_KHAN: 'cybersecurity firm Tranchulas CEO, InsightsPod founder, Islamabad / Dubai',
         },
         patterns=[
             r"Adriana\s*Ross",
@@ -490,42 +531,6 @@ HIGHLIGHTED_NAMES = [
             r"(Sarah\s*)?Kellen", r"Vickers",  # Married name is Metiers
         ],
     ),
-    HighlightedNames(
-        label=ENTERTAINER,
-        style='light_steel_blue3',
-        emailers={
-            ANDRES_SERRANO: "'Piss Christ' artist",
-            'Barry Josephson': 'American film producer, editor FamilySecurityMatters.org',
-            BILL_SIEGEL: 'documentary film producer and director',
-            DAVID_BLAINE: 'famous magician',
-            'Richard Merkin': 'painter, illustrator and arts educator',
-            STEVEN_PFEIFFER: 'Associate Director at Independent Filmmaker Project (IFP)',
-        },
-        patterns=[
-            r"(Art )?Spiegelman",
-            r"Artspace",
-            r"Bobby slayton",
-            r"bono\s*mick",
-            r"Errol(\s*Morris)?",
-            r"Etienne Binant",
-            r"(Frank\s)?Gehry",
-            r"Jagger",
-            r"(Jeffrey\s*)?Katzenberg",
-            r"(Johnny\s*)?Depp",
-            r"Kid Rock",
-            r"(Larry\s*)?Gagosian",
-            r"Lena\s*Dunham",
-            r"Madonna",
-            r"Mark\s*Burnett",
-            r"Phaidon",
-            r"Ramsey Elkholy",
-            r"Regan arts",
-            r"shirley maclaine",
-            r"Steven Gaydos?",
-            r"Woody( Allen)?",
-            r"Zach Braff",
-        ],
-    ),
     HighlightedNames(
         label='Epstein',
         style='blue1',
@@ -565,7 +570,7 @@ HIGHLIGHTED_NAMES = [
             'Robert Gold': 'helped Epstein track down money belonging to Spanish families',
             'Roy Black': CRIMINAL_DEFENSE_2008,
             SCOTT_J_LINK: CRIMINAL_DEFENSE_ATTORNEY,
-            TONJA_HADDAD_COLEMAN: f'{EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY}, relation of Fred Haddad?',
+            TONJA_HADDAD_COLEMAN: f'{EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY}',  # relation of Fred Haddad?
         },
         patterns=[
             r"(Barry (E. )?)?Krischer",
@@ -593,8 +598,10 @@ HIGHLIGHTED_NAMES = [
         style='light_sky_blue3',
         emailers={
             ANDRZEJ_DUDA: 'former president of Poland',
+            'Caroline Lang': 'daughter of Jack Lang',
             "Edward Rod Larsen": f"son of {TERJE_ROD_LARSEN}",
             'Fabrice Aidan': f'diplomat who worked with {TERJE_ROD_LARSEN}',
+            'Jack Lang': 'former French Minister of National Education',
             MIROSLAV_LAJCAK: 'Russia-friendly Slovakian politician, friend of Steve Bannon',
             PETER_MANDELSON: 'UK politics',
             TERJE_ROD_LARSEN: 'Norwegian diplomat',
@@ -611,7 +618,6 @@ HIGHLIGHTED_NAMES = [
             r"Brit(ain|ish)",
             r"Brussels",
             r"Cannes",
-            r"(Caroline|Jack)?\s*Lang(, Caroline)?",
             r"Cypr(iot|us)",
             r"Davos",
             r"ECB",
@@ -662,12 +668,13 @@ HIGHLIGHTED_NAMES = [
         emailers={
             AMANDA_ENS: 'Citigroup',
             BRAD_WECHSLER: f"head of {LEON_BLACK}'s personal investment vehicle according to FT",
+            CECILIA_STEEN: None,
             DANIEL_SABBA: 'UBS Investment Bank',
             DAVID_FISZEL: 'CIO Honeycomb Asset Management',
             JES_STALEY: 'former CEO of Barclays',
             JIDE_ZEITLIN: 'former partner at Goldman Sachs, allegations of sexual misconduct',
             'Laurie Cameron': 'currency trading',
-            LEON_BLACK: 'Apollo CEO',
+            LEON_BLACK: 'Apollo CEO who paid Epstein tens of millions for tax advice',
             MARC_LEON: 'Luxury Properties Sari Morrocco',
             MELANIE_SPINELLA: 'representative of Leon Black',
             MORTIMER_ZUCKERMAN: 'business partner of Epstein, newspaper publisher',
@@ -683,6 +690,7 @@ HIGHLIGHTED_NAMES = [
             r"Apollo",
             r"Ari\s*Glass",
             r"Bank(\s*of\s*Scotland)",
+            r"Bear\s*Stearns",
             r"(Bernie\s*)?Madoff",
             r"Black(rock|stone)",
             r"B\s*of\s*A",
@@ -826,19 +834,21 @@ HIGHLIGHTED_NAMES = [
         style='bright_yellow',
         emailers={
             'Alain Forget': 'author of "How To Get Out Of This World ALIVE"',
+            'Alex Yablon': 'New York Magazine fact checker (?)',
             EDWARD_JAY_EPSTEIN: 'no relation, wrote books about spies',
             HENRY_HOLT: f"{MICHAEL_WOLFF}'s book publisher",
             JAMES_HILL: 'ABC News',
-            JENNIFER_JACQUET: 'Future Science',
+            JENNIFER_JACQUET: 'Future Science magazine',
             JOHN_BROCKMAN: 'literary agent and author specializing in scientific literature',
-            LANDON_THOMAS: 'New York Times',
+            LANDON_THOMAS: 'New York Times financial reporter',
             MICHAEL_WOLFF: 'Author of "Fire and Fury: Inside the Trump White House"',
             PAUL_KRASSNER: '60s counterculture guy',
+            'Peter Aldhous': 'Buzzfeed science reporter',
+            "Susan Edelman": 'New York Post reporter',
             'Tim Zagat': 'Zagat restaurant guide CEO',
         },
         patterns=[
             r"ABC(\s*News)?",
-            r"Alex\s*Yablon",
             r"Arianna(\s*Huffington)?",
             r"(Arthur\s*)?Kretchmer",
             r'Associated\s*Press',
@@ -872,6 +882,7 @@ HIGHLIGHTED_NAMES = [
             r"Huff(ington)?(\s*Po(st)?)?",
             r"Ingram, David",
             r"(James\s*)?(Hill|Patterson)",
+            r"Jesse Kornbluth",
             r"Jonathan\s*Karl",
             r"Julie\s*(K.?\s*)?Brown", r'jbrown@miamiherald.com',
             r"(Katie\s*)?Couric",
@@ -895,7 +906,6 @@ HIGHLIGHTED_NAMES = [
             r"(Sean\s*)?Hannity",
             r"Sulzberger",
             r"SunSentinel",
-            r"Susan Edelman",
             r"(The\s*)?Financial\s*Times",
             r"The\s*Guardian",
             r"TheHill",
@@ -956,6 +966,7 @@ HIGHLIGHTED_NAMES = [
         emailers={
             ANN_MARIE_VILLAFANA: 'Southern District of Florida (SDFL) U.S. Attorney',
             DANNY_FROST: 'Director of Communications at Manhattan D.A.',
+            'Police Code Enforcement': f"{PALM_BEACH} buildings code enforcement",
         },
         patterns=[
             r"AG",
@@ -984,7 +995,6 @@ HIGHLIGHTED_NAMES = [
             r"(General\s*)?P(a|e)traeus",
             r"IRS",
             r"(James\s*)?Comey",
-            r"Jeff(rey)?\s*Sessions",
             r"(Jennifer\s*Shasky\s*)?Calvery",
             r"((Judge|Mark)\s*)?(Carney|Filip)",
             r"(Judge\s*)?(Kenneth\s*)?(A\.?\s*)?Marra",
@@ -999,14 +1009,14 @@ HIGHLIGHTED_NAMES = [
             r"(Michael\s*)?Reiter",
             r"OGE",
             r"Office\s*of\s*Government\s*Ethics",
-            r"Police Code Enforcement",
+            r"police",
             r"(Preet\s*)?Bharara",
             r"SCOTUS",
             r"SD(FL|NY)",
             r"SEC",
             r"Secret\s*Service",
             r"Securities\s*and\s*Exchange\s*Commission",
-            r"Southern\s*District\s*of\s*(Florida|New\s*York)",
+            r"Southern\s*District(\s*of\s*(Florida|New\s*York))?",
             r"State\s*Dep(artmen)?t",
             r"Strzok",
             r"Supreme\s*Court",
@@ -1136,16 +1146,18 @@ HIGHLIGHTED_NAMES = [
         label='modeling',
         style='pale_violet_red1',
         emailers={
-            'Abi Schwinck': 'MC2 Model Management (?)',
+            'Abi Schwinck': f'{MC2_MODEL_MANAGEMENT} {QUESTION_MARKS}',
             DANIEL_SIAD: None,
             FAITH_KATES: 'Next Models co-founder',
             'Gianni Serazzi': 'fashion consultant?',
             HEATHER_MANN: 'South African former model, ex-girlfriend of Prince Andrew (?)',
-            JEAN_LUC_BRUNEL: 'MC2 Model Management founder, died by suicide in French jail',
-            JEFF_FULLER: 'president of MC2 Model Management USA',
+            JEAN_LUC_BRUNEL: f'MC2 Model Management founder, died by suicide in French jail',
+            JEFF_FULLER: f'president of {MC2_MODEL_MANAGEMENT} USA',
+            'lorraine@mc2mm.com': f'{MC2_MODEL_MANAGEMENT}',
+            'pink@mc2mm.com': f'{MC2_MODEL_MANAGEMENT}',
             MANUELA_MARTINEZ: 'Mega Partners (Brazilian agency)',
             MARIANA_IDZKOWSKA: None,
-            'Michael Sanka': 'MC2 Model Management (?)',
+            'Michael Sanka': f'{MC2_MODEL_MANAGEMENT} {QUESTION_MARKS}',
             'Vladimir Yudashkin': 'director of the 1 Mother Agency',
         },
         patterns=[
@@ -1202,6 +1214,7 @@ HIGHLIGHTED_NAMES = [
             r"George\s*(H\.?\s*)?(W\.?\s*)?Bush",
             r"(George\s*)?Nader",
             r"GOP",
+            r"Jeff(rey)?\s*Sessions",
             r"(John\s*(R.?\s*)?)Bolton",
             r"Kissinger",
             r"Kobach",
@@ -1245,7 +1258,7 @@ HIGHLIGHTED_NAMES = [
         emailers={
             'Dasha Zhukova': 'art collector, daughter of Alexander Zhukov',
             MASHA_DROKOVA: 'silicon valley VC, former Putin Youth member',
-            RENATA_BOLOTOVA: 'former aspiring model, now fund manager at New York State Insurance Fund',
+            RENATA_BOLOTOVA: 'former model, fund manager at New York State Insurance Fund',
             SVETLANA_POZHIDAEVA: "Epstein's Russian assistant who was recommended for a visa by Sergei Belyakov (FSB) and David Blaine",
         },
         patterns=[
@@ -1302,7 +1315,7 @@ HIGHLIGHTED_NAMES = [
         ],
     ),
     HighlightedNames(
-        label='tech bro',
+        label=TECH_BRO,
         style='bright_cyan',
         emailers={
             'Auren Hoffman': 'CEO of SafeGraph (firm that gathers location data from mobile devices) and LiveRamp',
@@ -1369,6 +1382,33 @@ HIGHLIGHTED_NAMES = [
             r"(The\s*)?Art\s*of\s*the\s*Deal",
         ],
     ),
+    HighlightedNames(
+        label='USVI',
+        style='sea_green1',
+        emailers={
+            CECILE_DE_JONGH: 'Virgin Islands first lady 2007-2015',
+            KENNETH_E_MAPP: 'Virgin Islands Governor',
+            STACEY_PLASKETT: 'Virgin Islands non-voting member of Congress',
+        },
+        patterns=[
+            r"Antigua",
+            r"Bahamas",
+            r"BVI",
+            r"Caribb?ean",
+            r"Dominican\s*Republic",
+            r"(Great|Little)\s*St.?\s*James",
+            r"Haiti(an)?",
+            r"(John\s*)deJongh(\s*Jr\.?)",
+            r"(Kenneth E\. )?Mapp",
+            r"PBI",
+            r"Puerto\s*Ric(an|o)",
+            r"S(ain)?t.?\s*Thomas",
+            r"USVI",
+            r"(?<!Epstein )VI",
+            r"(The\s*)?Virgin\s*Islands(\s*Daily\s*News)?",  # Hard to make this work right
+            r"(West\s*)?Palm\s*Beach(?!\s*(Daily|Post))",
+        ],
+    ),
     HighlightedNames(
         label='victim',
         style='orchid1',
@@ -1399,35 +1439,6 @@ HIGHLIGHTED_NAMES = [
             r"(Scott\s*)?Rothstein",
         ],
     ),
-    HighlightedNames(
-        label='USVI',
-        style='sea_green1',
-        emailers={
-            CECILE_DE_JONGH: 'Virgin Islands first lady 2007-2015',
-            KENNETH_E_MAPP: 'Virgin Islands Governor',
-            STACEY_PLASKETT: 'Virgin Islands non-voting member of Congress',
-        },
-        patterns=[
-            r"Antigua",
-            r"Bahamas",
-            r"BVI",
-            r"Caribb?ean",
-            r"Dominican\s*Republic",
-            r"(Great|Little)\s*St.?\s*James",
-            r"Haiti(an)?",
-            r"(John\s*)deJongh(\s*Jr\.?)",
-            r"(Kenneth E\. )?Mapp",
-            r"PBI",
-            r"Puerto\s*Ric(an|o)",
-            r"S(ain)?t.?\s*Thomas",
-            r"USVI",
-            r"(?<!Epstein )VI",
-            r"(The\s*)?Virgin\s*Islands(\s*Daily\s*News)?",  # Hard to make this work right
-            r"(West\s*)?Palm\s*Beach(?!\s*(Daily|Post))",
-        ],
-    ),
-
-    # Individuals
     HighlightedNames(
         label=STEVE_BANNON,
         style='color(58)',
@@ -1441,13 +1452,20 @@ HIGHLIGHTED_NAMES = [
             r"Biosphere",
         ],
     ),
+
+    # Individuals
     HighlightedNames(
+        emailers={STEVEN_HOFFENBERG: "Epstein's Towers Financial ponzi scheme partner, prison 18 years"},
         style='dark_olive_green3',
         category=FINANCE,
-        emailers={STEVEN_HOFFENBERG: "Epstein's Towers Financial ponzi scheme partner, prison 18 years"},
         patterns=[r"(steven?\s*)?hoffenberg?w?"],
     ),
-    HighlightedNames(emailers={GHISLAINE_MAXWELL: None}, patterns=[r"gmax(1@ellmax.com)?", r"(The )?TerraMar Project"], style='deep_pink3', category='Epstein'),
+    HighlightedNames(
+        emailers={GHISLAINE_MAXWELL: "Epstein's girlfriend, daughter of the spy Robert Maxwell"},
+        category='Epstein',
+        patterns=[r"gmax(1@ellmax.com)?", r"(The )?TerraMar Project"],
+        style='deep_pink3',
+    ),
     HighlightedNames(emailers={JABOR_Y: '"an influential man in Qatar"'}, category=MIDEAST, style='spring_green1'),
     HighlightedNames(emailers={KATHRYN_RUEMMLER: 'former Obama legal counsel'}, style='magenta2', category=FRIEND),
     HighlightedNames(emailers={MELANIE_WALKER: f"doctor, friend of {BILL_GATES}"}, style='pale_violet_red1', category=FRIEND),

epstein_files/util/logging.py

@@ -33,7 +33,7 @@ LOG_LEVEL_ENV_VAR = 'LOG_LEVEL'
 # Augment the standard log highlighter with 'epstein_filename' matcher
 class LogHighlighter(ReprHighlighter):
     highlights = ReprHighlighter.highlights + [
-        *[fr"(?P<{doc_type}>{doc_type}(Cfg)?)" for doc_type in DOC_TYPE_STYLES.keys()],
+        *[fr"(?P<{doc_type}>{doc_type}(Cfg|s)?)" for doc_type in DOC_TYPE_STYLES.keys()],
         "(?P<epstein_filename>" + FILE_NAME_REGEX.pattern + ')',
     ]
 

epstein_files/util/output.py

@@ -18,7 +18,7 @@ from epstein_files.util.constant.strings import AUTHOR, TIMESTAMP_STYLE
 from epstein_files.util.data import dict_sets_to_lists, uniquify
 from epstein_files.util.env import args
 from epstein_files.util.file_helper import log_file_write
-from epstein_files.util.logging import logger
+from epstein_files.util.logging import logger, exit_with_error
 from epstein_files.util.rich import *
 
 DEVICE_SIGNATURE_SUBTITLE = f"Email [italic]Sent from \\[DEVICE][/italic] Signature Breakdown"
@@ -69,6 +69,11 @@ INTERESTING_EMAIL_IDS = [
     #'023208',  # Extremely long Leon Black email chain
 ]
 
+INTERESTING_TEXT_IDS = [
+    '027275',  # "Crypto- Kerry- Qatar -sessions"
+    '027165',  # melaniee walker crypto health
+]
+
 
 def print_email_timeline(epstein_files: EpsteinFiles) -> None:
     """Print a table of all emails in chronological order."""
@@ -83,11 +88,17 @@ def print_email_timeline(epstein_files: EpsteinFiles) -> None:
         console.print(email)
 
 
-def print_emailers_info_png(epstein_files: EpsteinFiles) -> None:
+def print_emailers_info(epstein_files: EpsteinFiles) -> None:
+    """Print tbe summary table of everyone in the files to an image."""
     print_color_key()
     console.line()
     all_emailers = sorted(epstein_files.emailers(), key=lambda person: person.sort_key())
-    console.print(Person.emailer_info_table(all_emailers))
+    console.print(Person.emailer_info_table(all_emailers, show_epstein_total=True))
+
+    if not args.build:
+        logger.warning(f"Not writing .png file because --build is not set")
+        return
+
     svg_path = f"{EMAILERS_TABLE_PNG_PATH}.svg"
     console.save_svg(svg_path, theme=HTML_TERMINAL_THEME, title="Epstein Emailers")
     log_file_write(svg_path)
@@ -106,7 +117,10 @@ def print_emails_section(epstein_files: EpsteinFiles) -> list[Email]:
     people_to_print: list[Person]
 
     if args.names:
-        people_to_print = epstein_files.person_objs(args.names)
+        try:
+            people_to_print = epstein_files.person_objs(args.names)
+        except Exception as e:
+            exit_with_error(str(e))
     else:
         if args.all_emails:
             people_to_print = all_emailers
@@ -147,10 +161,10 @@ def print_emails_section(epstein_files: EpsteinFiles) -> list[Email]:
     if args.all_emails:
         _verify_all_emails_were_printed(epstein_files, printed_emails)
 
-    _print_email_device_info(epstein_files)
+    _print_email_device_signature_info(epstein_files)
     fwded_articles = [e for e in printed_emails if e.config and e.is_fwded_article()]
     log_msg = f"Rewrote {len(Email.rewritten_header_ids)} of {len(printed_emails)} email headers"
-    logger.warning(f"{log_msg}, {len(fwded_articles)} of the emails were forwarded articles.")
+    logger.warning(f"  -> {log_msg}, {len(fwded_articles)} of the Emails printed were forwarded articles.")
     return printed_emails
 
 
@@ -193,22 +207,30 @@ def print_json_stats(epstein_files: EpsteinFiles) -> None:
     print_json("count_by_month", count_by_month(epstein_files.all_documents()))
 
 
-def print_other_files_section(files: list[OtherFile], epstein_files: EpsteinFiles) -> None:
+def print_other_files_section(epstein_files: EpsteinFiles) -> list[OtherFile]:
     """Returns the OtherFile objects that were interesting enough to print."""
+    if args.uninteresting:
+        files = [f for f in epstein_files.other_files if not f.is_interesting()]
+    else:
+        files = [f for f in epstein_files.other_files if args.all_other_files or f.is_interesting()]
+
     title_pfx = '' if args.all_other_files else 'Selected '
-    category_table = OtherFile.count_by_category_table(files, title_pfx=title_pfx)
+    category_table = OtherFile.summary_table(files, title_pfx=title_pfx)
     other_files_preview_table = OtherFile.files_preview_table(files, title_pfx=title_pfx)
     print_section_header(f"{FIRST_FEW_LINES} of {len(files)} {title_pfx}Files That Are Neither Emails Nor Text Messages")
     print_other_page_link(epstein_files)
     print_centered(Padding(category_table, (2, 0)))
     console.print(other_files_preview_table)
+    return files
 
 
-def print_text_messages_section(imessage_logs: list[MessengerLog]) -> None:
+def print_text_messages_section(epstein_files: EpsteinFiles) -> list[MessengerLog]:
     """Print summary table and stats for text messages."""
+    imessage_logs = [log for log in epstein_files.imessage_logs if not args.names or log.author in args.names]
+
     if not imessage_logs:
-        logger.warning(f"No MessengerLog objects to output...")
-        return
+        logger.warning(f"No MessengerLogs found for {args.names}")
+        return imessage_logs
 
     print_section_header('All of His Text Messages')
     print_centered("(conversations are sorted chronologically based on timestamp of first message in the log file)", style='dim')
@@ -222,6 +244,8 @@ def print_text_messages_section(imessage_logs: list[MessengerLog]) -> None:
         console.print(Padding(log_file))
         console.line(2)
 
+    return imessage_logs
+
 
 def write_urls() -> None:
     """Write _URL style constant variables to URLS_ENV file so bash scripts can load as env vars."""
@@ -245,7 +269,7 @@ def write_urls() -> None:
     logger.warning(f"Wrote {len(url_vars)} URL variables to '{URLS_ENV}'\n")
 
 
-def _print_email_device_info(epstein_files: EpsteinFiles) -> None:
+def _print_email_device_signature_info(epstein_files: EpsteinFiles) -> None:
     print_subtitle_panel(DEVICE_SIGNATURE_SUBTITLE)
     console.print(_signature_table(epstein_files.email_device_signatures_to_authors(), (DEVICE_SIGNATURE, AUTHOR), ', '))
     console.print(_signature_table(epstein_files.email_authors_to_device_signatures(), (AUTHOR, DEVICE_SIGNATURE)))

epstein_files/util/rich.py

@@ -70,7 +70,7 @@ CONSOLE_ARGS = {
     'color_system': '256',
     'highlighter': EpsteinHighlighter(),
     'record': args.build,
-    'safe_box': False,
+    'safe_box': True,
     'theme': Theme(THEME_STYLES),
     'width': args.width,
 }
@@ -83,21 +83,21 @@ console = Console(**CONSOLE_ARGS)
 highlighter = CONSOLE_ARGS['highlighter']
 
 
-def add_cols_to_table(table: Table, col_names: list[str | dict]) -> None:
+def add_cols_to_table(table: Table, cols: list[str | dict], justify: str = 'center') -> None:
     """Left most col will be left justified, rest are center justified."""
-    for i, col in enumerate(col_names):
-        justify='left' if i == 0 else 'center'
+    for i, col in enumerate(cols):
+        col_justify = 'left' if i == 0 else justify
 
         if isinstance(col, dict):
             col_name = col['name']
-            kwargs = deepcopy(col)
-            kwargs['justify'] = kwargs.get('justify', justify)
-            del kwargs['name']
+            col_kwargs = deepcopy(col)
+            col_kwargs['justify'] = col_kwargs.get('justify', col_justify)
+            del col_kwargs['name']
         else:
             col_name = col
-            kwargs = {'justify': justify}
+            col_kwargs = {'justify': col_justify}
 
-        table.add_column(col_name, **kwargs)
+        table.add_column(col_name, **col_kwargs)
 
 
 def build_highlighter(pattern: str) -> EpsteinHighlighter:
@@ -187,7 +187,7 @@ def print_title_page_header() -> None:
         Text.from_markup(link_markup(other_site_url(), other_site_msg, f"{OTHER_SITE_LINK_STYLE} bold")),
         link_text_obj(WORD_COUNT_URL, 'most frequently used words in the emails and texts', AUX_SITE_LINK_STYLE),
         link_text_obj(JSON_METADATA_URL, 'author attribution explanations', AUX_SITE_LINK_STYLE),
-        link_text_obj(WORD_COUNT_URL, "epstein's json files", AUX_SITE_LINK_STYLE),
+        link_text_obj(JSON_FILES_URL, "epstein's json files", AUX_SITE_LINK_STYLE),
     ]
 
     for link in links:
@@ -199,7 +199,8 @@ def print_title_page_tables(epstein_files: 'EpsteinFiles') -> None:
     _print_external_links()
     console.line()
     _print_abbreviations_table()
-    epstein_files.print_files_summary()
+    print_centered(epstein_files.overview_table())
+    console.line()
     print_color_key()
     print_centered(f"if you think there's an attribution error or can deanonymize an {UNKNOWN} contact {CRYPTADAMUS_TWITTER}", 'grey46')
     print_centered('note this site is based on the OCR text provided by Congress which is not always the greatest', 'grey23')
@@ -252,7 +253,7 @@ def print_subtitle_panel(msg: str, style: str = 'black on white') -> None:
 def print_section_header(msg: str, style: str = SECTION_HEADER_STYLE, is_centered: bool = False) -> None:
     panel = Panel(Text(msg, justify='center'), expand=True, padding=(1, 1), style=style)
     panel = Align.center(panel) if is_centered else panel
-    console.print(Padding(panel, (3, 0, 1, 0)))
+    console.print(Padding(panel, (3, 5, 1, 5)))
 
 
 def print_starred_header(msg: str, num_stars: int = 7, num_spaces: int = 2, style: str = TITLE_STYLE) -> None:
@@ -311,8 +312,7 @@ def _print_external_links() -> None:
     raw_docs_link = join_texts([link_text_obj(RAW_OVERSIGHT_DOCS_GOOGLE_DRIVE_URL, 'raw files', style=f"{ARCHIVE_LINK_COLOR} dim")], encloser='()')
     print_centered(join_texts([presser_link, raw_docs_link]))
     print_centered(link_markup(JMAIL_URL, JMAIL) + " (read His Emails via Gmail interface)")
-    print_centered(link_markup(COFFEEZILLA_ARCHIVE_URL, 'Archive Of Epstein Materials') + " (Coffeezilla)")
-    print_centered(link_markup(COURIER_NEWSROOM_ARCHIVE_URL, 'Searchable Archive') + " (Courier Newsroom)")
+    print_centered(link_markup(EPSTEIN_DOCS_URL) + " (searchable archive)")
     print_centered(link_markup(EPSTEINIFY_URL) + " (raw document images)")
     print_centered(link_markup(EPSTEIN_WEB_URL) + " (character summaries)")
     print_centered(link_markup(EPSTEIN_MEDIA_URL) + " (raw document images)")

epstein_files/util/timer.py

@@ -1,5 +1,6 @@
 import time
 from dataclasses import dataclass, field
+from typing import Type
 
 from epstein_files.util.logging import logger
 
@@ -10,6 +11,19 @@ class Timer:
     checkpoint_at: float = field(default_factory=lambda: time.perf_counter())
     decimals: int = 2
 
+    def log_section_complete(self, label: str, all_docs: list, printed_docs: list) -> None:
+        num_skipped = len(all_docs) - len(printed_docs)
+        prefix = suffix = ''
+
+        if num_skipped == 0:
+            prefix = 'all '
+        elif num_skipped < 0:
+            suffix = f"(at least {num_skipped} {label}s printed more than once)"
+        else:
+            suffix = f"(skipped {num_skipped})"
+
+        self.print_at_checkpoint(f"Printed {prefix}{len(printed_docs)} {label}s {suffix}".strip())
+
     def print_at_checkpoint(self, msg: str) -> None:
         logger.warning(f"{msg} in {self.seconds_since_checkpoint_str()}...")
         self.checkpoint_at = time.perf_counter()