epstein-files 1.1.3__py3-none-any.whl → 1.2.0__py3-none-any.whl

epstein_files/util/highlighted_group.py

@@ -1,8 +1,8 @@
 import json
 import re
-from abc import ABC
 from dataclasses import dataclass, field
 
+from rich.console import Console
 from rich.highlighter import RegexHighlighter
 from rich.text import Text
 
@@ -12,17 +12,19 @@ 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)
 from epstein_files.util.doc_cfg import *
-from epstein_files.util.data import extract_last_name, without_falsey
+from epstein_files.util.data import without_falsey
+from epstein_files.util.env import args
 from epstein_files.util.logging import logger
 
 CIVIL_ATTORNEY = 'civil attorney'
 CRIMINAL_DEFENSE_ATTORNEY = 'criminal defense attorney'
 CRIMINAL_DEFENSE_2008 = f"{CRIMINAL_DEFENSE_ATTORNEY} on 2008 case"
-EPSTEIN_LAWYER = 'Epstein lawyer'
-EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY = f"{CIVIL_ATTORNEY} working on {EPSTEIN_V_ROTHSTEIN_EDWARDS}"
+EPSTEIN_LAWYER = 'lawyer'
+EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY = f"{CIVIL_ATTORNEY} in {EPSTEIN_V_ROTHSTEIN_EDWARDS}"
 ESTATE_EXECUTOR = 'estate executor'
 EPSTEIN_ESTATE_EXECUTOR = f"Epstein {ESTATE_EXECUTOR}"
-QUESTION_MARKS_TXT = Text(QUESTION_MARKS, style='dim')
+MIDEAST = 'mideast'
+QUESTION_MARKS_TXT = Text(QUESTION_MARKS, style='grey50')
 REGEX_STYLE_PREFIX = 'regex'
 SIMPLE_NAME_REGEX = re.compile(r"^[-\w ]+$", re.IGNORECASE)
 
@@ -38,20 +40,20 @@ CATEGORY_STYLE_MAPPING = {
 
 CATEGORY_STYLES = {
     JSON: 'dark_red',
-    JUNK: 'grey19',
     'letter': 'medium_orchid1'
 }
 
+debug_console = Console(color_system='256')
+
 
 @dataclass(kw_only=True)
 class BaseHighlight:
     """
-    Regex and style information.
+    Regex and style information for things we want to highlight.
 
     Attributes:
-        label (str): RegexHighlighter match group name, defaults to 1st 'emailers' key if only 1 emailer provided
+        label (str): RegexHighlighter match group name
         pattern (str): regex pattern identifying strings matching this group
-        regex (re.Pattern): matches self.pattern
         style (str): Rich style to apply to text matching this group
         theme_style_name (str): The style name that must be a part of the rich.Console's theme
     """
@@ -79,9 +81,6 @@ class HighlightedText(BaseHighlight):
     Attributes:
         label (str): RegexHighlighter match group name, defaults to 1st 'emailers' key if only 1 emailer provided
         patterns (list[str]): regex patterns identifying strings matching this group
-        regex (re.Pattern): matches self.pattern
-        style (str): Rich style to apply to text matching this group
-        theme_style_name (str): The style name that must be a part of the rich.Console's theme
     """
     patterns: list[str] = field(default_factory=list)
     _pattern: str = field(init=False)
@@ -112,6 +111,7 @@ class HighlightedNames(HighlightedText):
     """
     category: str = ''
     emailers: dict[str, str | None] = field(default_factory=dict)
+    should_match_first_last_name: bool = True
 
     def __post_init__(self):
         if not (self.emailers or self.patterns):
@@ -126,40 +126,42 @@ class HighlightedNames(HighlightedText):
         self._pattern = '|'.join([self._emailer_pattern(e) for e in self.emailers] + self.patterns)
         self.regex = re.compile(fr"\b({self._match_group_var}({self._pattern})s?)\b", re.IGNORECASE)
 
-    def get_info(self, name: str) -> str | None:
-        """Label and additional info for 'name' if 'name' is in self.emailers."""
-        info_pieces = [
-            self.category or ('' if len(self.emailers) == 1 else self.label.replace('_', ' ')),
-            self.emailers.get(name),
-        ]
+    def category_str(self) -> str:
+        if self.category:
+            return self.category
+        elif len(self.emailers) == 1 and self.label == [k for k in self.emailers.keys()][0]:
+            return ''
+        else:
+            return self.label.replace('_', ' ')
 
+    def info_for(self, name: str, include_category: bool = False) -> str | None:
+        """Label and additional info for 'name' if 'name' is in self.emailers."""
+        info_pieces = [self.category_str()] if include_category else []
+        info_pieces.append(self.emailers.get(name) or '')
         info_pieces = without_falsey(info_pieces)
         return ', '.join(info_pieces) if info_pieces else None
 
     def _emailer_pattern(self, name: str) -> str:
         """Pattern matching 'name'. Extends value in EMAILER_ID_REGEXES with first/last name if it exists."""
-        name = remove_question_marks(name)
-        last_name = extract_last_name(name)
-        first_name = name.removesuffix(f" {last_name}")
+        if not self.should_match_first_last_name:
+            return name
 
         if name in EMAILER_ID_REGEXES:
-            pattern = EMAILER_ID_REGEXES[name].pattern
+            name_patterns = [EMAILER_ID_REGEXES[name].pattern]
+        else:
+            name_patterns = [remove_question_marks(name).replace(' ', r"\s+")]
 
-            # Include regex for first and last names
-            for partial_name in [first_name, last_name]:
-                if SIMPLE_NAME_REGEX.match(partial_name) and partial_name.lower() not in NAMES_TO_NOT_HIGHLIGHT:
-                    pattern += fr"|{partial_name}"
+        if ' ' in name:
+            for partial_name in [extract_first_name(name), extract_last_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+"))
 
-            return pattern
-        elif ' ' not in name:
-            return name
+        pattern = '|'.join(name_patterns)
 
-        name_patterns = [
-            n.replace(' ', r"\s+") for n in [name, first_name, last_name]
-            if n.lower() not in NAMES_TO_NOT_HIGHLIGHT
-        ]
+        if args.deep_debug:
+            debug_console.print(Text('').append(f"{name:25s}", style=self.style).append(f" '{pattern}'", style='dim'))
 
-        return '|'.join(name_patterns)
+        return pattern
 
     def __str__(self) -> str:
         return super().__str__()
@@ -211,7 +213,7 @@ HIGHLIGHTED_NAMES = [
     ManualHighlight(
         label='email_subject',
         style='light_yellow3',
-        pattern=r"^(> )?Subject: (?P<email_subject>.*)",
+        pattern=r"^(> )?(Classification|Flag|Subject): (?P<email_subject>.*)",
     ),
     HighlightedNames(
         label=ACADEMIA,
@@ -225,31 +227,35 @@ HIGHLIGHTED_NAMES = [
             LAWRENCE_KRAUSS: 'theoretical physicist',
             LINDA_STONE: f'ex-Microsoft, {MIT_MEDIA_LAB}',
             MARK_TRAMO: 'professor of neurology at UCLA',
-            'Nancy Dahl': 'wife of Lawrence Krauss',
+            'Nancy Dahl': f'wife of {LAWRENCE_KRAUSS}',
             NEAL_KASSELL: 'professor of neurosurgery at University of Virginia',
+            NOAM_CHOMSKY: f"professor of linguistics at MIT",
             PETER_ATTIA: 'longevity medicine',
             ROBERT_TRIVERS: 'evolutionary biology',
             ROGER_SCHANK: 'Teachers College, Columbia University',
+            'Valeria Chomsky': f"wife of {NOAM_CHOMSKY}",
         },
         patterns=[
-            r"Alain Forget",
+            r"Bard\s+((Early )?College|High School|Schools)",
             r"Brotherton",
             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",
             r"Media\s*Lab",
-            r"Minsky",
+            r"(Marvin\s*)?Minsky",
             r"MIT(\s*Media\s*Lab)?",
-            r"((Noam|Valeria)\s*)?Chomsky",
             r"Norman\s*Finkelstein",
+            r"Oxford(?! Analytica)",
             r"Praluent",
+            r"Princeton(\s*University)?",
             r"Regeneron",
             r"(Richard\s*)?Dawkins",
             r"Sanofi",
-            r"Stanford",
+            r"Stanford(\s*University)?(\s*Hospital)?",
             r"(Stephen\s*)?Hawking",
             r"(Steven?\s*)?Pinker",
             r"UCLA",
@@ -284,33 +290,51 @@ HIGHLIGHTED_NAMES = [
             r"Zimbabwe(an)?",
         ],
     ),
+    HighlightedNames(
+        label=BILL_GATES,
+        style='turquoise4',
+        emailers={
+            BORIS_NIKOLIC: f'biotech VC partner of {BILL_GATES}, {EPSTEIN_ESTATE_EXECUTOR}',
+        },
+        patterns=[
+            r"BG",
+            r"b?g?C3",
+            r"(Bill\s*((and|or|&)\s*Melinda\s*)?)?Gates(\s*Foundation)?",
+            r"Melinda(\s*Gates)?",
+            r"Microsoft",
+            r"MSFT",
+        ],
+    ),
     HighlightedNames(
         label='bitcoin',
         style='orange1 bold',
         emailers={
-            'Jeffrey Wernick': 'former COO of Parler, involved in numerous crypto companies like Bitforex',
+            JEFFREY_WERNICK: 'former COO of Parler, involved in numerous crypto companies like Bitforex',
             JEREMY_RUBIN: 'developer/researcher',
             JOI_ITO: f"former head of {MIT_MEDIA_LAB} and MIT Digital Currency Initiative",
             ANTHONY_SCARAMUCCI: 'Skybridge Capital, FTX investor',
         },
         patterns=[
             r"Balaji",
-            r"bitcoin",
+            r"bitcoin(\s*Foundation)?",
             r"block ?chain(\s*capital)?",
             r"Brock(\s*Pierce)?",
             r"coins?",
             r"cr[iy]?pto(currenc(y|ies))?",
+            r"Digital\s*Currenc(ies|y)(\s*Initiative)?",
             r"e-currency",
-            r"(Gavin )?Andressen",
+            r"(Gavin )?Andress?en",
             r"(Howard\s+)?Lutnic?k",
             r"Libra",
             r"Madars",
+            r"Mi(chael|ke)\s*Novogratz",
             r"(Patrick\s*)?Murck",
             r"(Ross\s*)?Ulbricht",
             r"Silk\s*Road",
             r"SpanCash",
             r"Tether",
             r"virtual\s*currenc(ies|y)",
+            r"Wladimir( van der Laan)?",
             r"(zero\s+knowledge\s+|zk)pro(of|tocols?)",
         ],
     ),
@@ -320,10 +344,10 @@ HIGHLIGHTED_NAMES = [
         emailers={
             ALIREZA_ITTIHADIEH: 'CEO Freestream Aircraft Limited',
             BARBRO_C_EHNBOM: 'Swedish pharmaceuticals, SALSS',
-            'David Mitchell': 'president of New York real estate development firm Mitchell Holdings',
+            '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",
-            GORDON_GETTY: 'heir of oil tycoon J. Paul Getty',
+            GORDON_GETTY: 'heir to oil tycoon J. Paul Getty',
             NICHOLAS_RIBIS: 'Hilton CEO, former president of Trump Organization',
             'Philip Kafka': 'president of Prince Concepts (and son of Terry Kafka?)',
             ROBERT_LAWRENCE_KUHN: 'investment banker, China expert',
@@ -380,7 +404,7 @@ HIGHLIGHTED_NAMES = [
             r"Ministry\sof\sState\sSecurity",
             r"Mongolian?",
             r"MSS",
-            r"North\s*Korea",
+            r"North\s*Korean?",
             r"Peking",
             r"PRC",
             r"Pyongyang",
@@ -390,10 +414,10 @@ HIGHLIGHTED_NAMES = [
         ],
     ),
     HighlightedNames(
-        label=DEEPAK_CHOPRA,
+        label='deepak',
         style='dark_sea_green4',
         emailers={
-            CAROLYN_RANGEL: 'assistant',
+            CAROLYN_RANGEL: f"Deepak Chopra's assistant {QUESTION_MARKS}",
             DEEPAK_CHOPRA: 'woo woo',
         },
     ),
@@ -405,6 +429,7 @@ HIGHLIGHTED_NAMES = [
         },
         patterns=[
             r"(Al\s*)?Franken",
+            r"Al\s*Gore",
             r"(Barac?k )?Obama",
             r"((Bill|Hillart?y)\s*)?Clinton",
             r"((Chuck|Charles)\s*)?S(ch|hc)umer",
@@ -416,6 +441,7 @@ HIGHLIGHTED_NAMES = [
             r"Eric Holder",
             r"George\s*Mitchell",
             r"(George\s*)?Soros",
+            r"Hakeem\s*Jeffries",
             r"Hill?ary",
             r"HRC",
             r"(Jo(e|seph)\s*)?Biden",
@@ -426,35 +452,43 @@ HIGHLIGHTED_NAMES = [
             r"(Nancy )?Pelosi",
             r"Ron\s*Dellums",
             r"Schumer",
-            r"(Tim\s*)?Geithner",
+            r"(Tim(othy)?\s*)?Geithner",
+            r"Tom McMillen",
             r"Vernon\s*Jordan",
         ],
     ),
     HighlightedNames(
-        label='Dubin family',
+        label='Dubins',
         style='medium_orchid1',
         emailers={
             GLENN_DUBIN: "Highbridge Capital Management, married to Epstein's ex-gf Eva",
             EVA: "possibly Epstein's ex-girlfriend (?)",
+            'Eva Dubin': f"Epstein's ex-girlfriend now married to {GLENN_DUBIN}",
         },
         patterns=[r"((Celina|Eva( Anderss?on)?|Glenn) )?Dubin"],
     ),
     HighlightedNames(
         label='employee',
-        style='deep_sky_blue4',
+        style='medium_purple4',
         emailers={
             'Alfredo Rodriguez': "Epstein's butler, stole the journal",
             ERIC_ROTH: 'jet decorator',
             GWENDOLYN_BECK: 'Epstein fund manager in the 90s',
+            JANUSZ_BANASIAK: "Epstein's house manager",
             JEAN_HUGUEN: 'interior design at Alberto Pinto Cabinet',
-            LAWRANCE_VISOSKI: 'pilot',
-            LESLEY_GROFF: 'assistant',
+            LAWRANCE_VISOSKI: "Epstein's pilot",
+            LESLEY_GROFF: f"Epstein's assistant",
             'Linda Pinto': 'interior design at Alberto Pinto Cabinet',
             MERWIN_DELA_CRUZ: None,  # HOUSE_OVERSIGHT_032652 Groff says "Jojo and Merwin both requested off Nov. 25 and 26"
-            NADIA_MARCINKO: 'pilot',
+            NADIA_MARCINKO: "Epstein's pilot",
             'Sean J. Lancaster': 'airplane reseller',
+            ZUBAIR_KHAN: 'cybersecurity firm Tranchulas CEO, InsightsPod founder, Islamabad/Dubai',
         },
-        patterns=[r"Merwin"],
+        patterns=[
+            r"Adriana\s*Ross",
+            r"Merwin",
+            r"(Sarah\s*)?Kellen", r"Vickers",  # Married name is Metiers
+        ],
     ),
     HighlightedNames(
         label=ENTERTAINER,
@@ -464,7 +498,6 @@ HIGHLIGHTED_NAMES = [
             'Barry Josephson': 'American film producer, editor FamilySecurityMatters.org',
             BILL_SIEGEL: 'documentary film producer and director',
             DAVID_BLAINE: 'famous magician',
-            HENRY_HOLT: f"{MICHAEL_WOLFF}'s book publisher",
             'Richard Merkin': 'painter, illustrator and arts educator',
             STEVEN_PFEIFFER: 'Associate Director at Independent Filmmaker Project (IFP)',
         },
@@ -493,14 +526,28 @@ HIGHLIGHTED_NAMES = [
             r"Zach Braff",
         ],
     ),
+    HighlightedNames(
+        label='Epstein',
+        style='blue1',
+        emailers={
+            JEFFREY_EPSTEIN: None,
+            MARK_EPSTEIN: 'brother of Jeffrey',
+        },
+        patterns=[
+            r"JEGE",
+            r"LSJ",
+        ],
+    ),
     HighlightedNames(
         label=EPSTEIN_LAWYER,
         style='purple',
         emailers={
             'Alan S Halperin': 'partner at Paul, Weiss',
+            ALAN_DERSHOWITZ: 'Harvard Law School professor',
             ARDA_BESKARDES: 'NYC immigration attorney allegedly involved in sex-trafficking operations',
             BENNET_MOSKOWITZ: f'represented the {EPSTEIN_ESTATE_EXECUTOR}s',
             BRAD_KARP: 'head of the law firm Paul Weiss',
+            'Connie Zaguirre': f"office of {ROBERT_D_CRITTON_JR}",
             DAVID_SCHOEN: f"{CRIMINAL_DEFENSE_ATTORNEY} after 2019 arrest",
             DEBBIE_FEIN: EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY,
             'Erika Kellerhals': 'attorney in St. Thomas',
@@ -509,24 +556,26 @@ HIGHLIGHTED_NAMES = [
             JACKIE_PERCZEK: CRIMINAL_DEFENSE_2008,
             JAY_LEFKOWITZ: f"Kirkland & Ellis partner, {CRIMINAL_DEFENSE_2008}",
             JESSICA_CADWELL: f'paralegal to {ROBERT_D_CRITTON_JR}',  # house_oversight_030464
+            KEN_STARR: 'head of the Monica Lewinsky investigation against Bill Clinton',
             LILLY_SANCHEZ: CRIMINAL_DEFENSE_ATTORNEY,
             MARTIN_WEINBERG: CRIMINAL_DEFENSE_ATTORNEY,
             MICHAEL_MILLER: 'Steptoe LLP partner',
             REID_WEINGARTEN: 'Steptoe LLP partner',
             ROBERT_D_CRITTON_JR: CRIMINAL_DEFENSE_ATTORNEY,
-            'Robert Gold': 'helped Epstein track down millions belonging to wealthy Spanish families',
+            '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}, maybe daughter of Fred Haddad?',
+            TONJA_HADDAD_COLEMAN: f'{EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY}, relation of Fred Haddad?',
         },
         patterns=[
             r"(Barry (E. )?)?Krischer",
+            r"dersh",
             r"Kate Kelly",
             r"Kirkland\s*&\s*Ellis",
             r"(Leon\s*)?Jaworski",
             r"Michael J. Pike",
             r"Paul,?\s*Weiss",
-            r"Steptoe(\s*LLP)?",
+            r"Steptoe(\s*& Johnson)?(\s*LLP)?",
             r"Wein(berg|garten)",
         ],
     ),
@@ -544,10 +593,12 @@ HIGHLIGHTED_NAMES = [
         style='light_sky_blue3',
         emailers={
             ANDRZEJ_DUDA: 'former president of Poland',
+            "Edward Rod Larsen": f"son of {TERJE_ROD_LARSEN}",
+            'Fabrice Aidan': f'diplomat who worked with {TERJE_ROD_LARSEN}',
             MIROSLAV_LAJCAK: 'Russia-friendly Slovakian politician, friend of Steve Bannon',
             PETER_MANDELSON: 'UK politics',
             TERJE_ROD_LARSEN: 'Norwegian diplomat',
-            THORBJORN_JAGLAND: 'former prime minister of Norway and head of the Nobel Peace Prize Committee',
+            THORBJORN_JAGLAND: 'former prime minister of Norway, Nobel Peace Prize Committee',
         },
         patterns=[
             r"(Angela )?Merk(el|le)",
@@ -572,6 +623,7 @@ HIGHLIGHTED_NAMES = [
             r"Germany?",
             r"Gillard",
             r"Gree(ce|k)",
+            r"Ibiza",
             r"Ital(ian|y)",
             r"Jacques",
             r"Kiev",
@@ -590,32 +642,18 @@ HIGHLIGHTED_NAMES = [
             r"Polish",
             r"pope",
             r"(Sebastian )?Kurz",
-            r"(Vi(c|k)tor\s+)?Orbah?n",
-            r"Edward Rod Larsen",
+            r"Stockholm",
             r"Strasbourg",
             r"Strauss[- ]?Kahn",
             r"Swed(en|ish)(?![-\s]+American Life Scienc)",
-            r"Switzerland",
+            r"Swi(ss|tzerland)",
             r"(Tony\s)?Blair",
             r"U\.K\.",
             r"Ukrain(e|ian)",
+            r"(Vi(c|k)tor\s+)?Orbah?n",
             r"Vienna",
             r"Zug",
-        ],
-    ),
-    HighlightedNames(
-        label='famous lawyer',
-        style='medium_purple3',
-        emailers={
-            ALAN_DERSHOWITZ: 'Harvard Law School professor and all around (in)famous American lawyer',
-            'Fabrice Aidan': f'diplomat who worked with {TERJE_ROD_LARSEN}',
-            KEN_STARR: 'head of the Monica Lewinsky investigation against Bill Clinton',
-        },
-        patterns=[
-            r"(David\s*)?Bo[il]es",
-            r"dersh",
-            r"(Gloria\s*)?Allred",
-            r"(Mi(chael|ke)\s*)?Avenatti",
+            r"Zurich",
         ],
     ),
     HighlightedNames(
@@ -650,6 +688,7 @@ HIGHLIGHTED_NAMES = [
             r"B\s*of\s*A",
             r"Boothbay(\sFund\sManagement)?",
             r"Chase\s*Bank",
+            r"Conrad B",
             r"Credit\s*Suisse",
             r"DB",
             r"Deutsche?\s*(Asset|Bank)",
@@ -661,13 +700,14 @@ HIGHLIGHTED_NAMES = [
             r"HSBC",
             r"Invesco",
             r"(Janet\s*)?Yellen",
-            r"(Jerome\s*)?Powell(?!M\. Cabot)",
+            r"(Jerome\s*)?Powell(?! M\. Cabot)",
             r"(Jimmy\s*)?Cayne",
             r"JPMC?",
             r"j\.?p\.?\s*morgan(\.?com|\s*Chase)?",
             r"Madoff",
             r"Merrill(\s*Lynch)?",
-            r"(Michael\s*)?(Cembalest|Milken)",
+            r"(Michael\s*)?Cembalest",
+            r"(Mi(chael|ke)\s*)?Milken(\s*Conference)?",
             r"Mizrahi\s*Bank",
             r"MLPF&S",
             r"Morgan Stanley",
@@ -681,11 +721,11 @@ HIGHLIGHTED_NAMES = [
         ],
     ),
     HighlightedNames(
-        label='friend',
+        label=FRIEND,
         style='tan',
         emailers={
             DANGENE_AND_JENNIE_ENTERPRISE: 'founders of the members-only CORE club',
-            DAVID_STERN: f'emailed Epstein from Moscow, appears to know chairman of {DEUTSCHE_BANK}',
+            DAVID_STERN: f'emailed Epstein from Moscow, knows chairman of {DEUTSCHE_BANK} (?)',
             JONATHAN_FARKAS: "heir to the Alexander's department store fortune",
             'linkspirit': 'Skype username of someone Epstein communicated with',
             'Peter Thomas Roth': 'student of Epstein at Dalton, skincare company founder',
@@ -694,6 +734,7 @@ HIGHLIGHTED_NAMES = [
         },
         patterns=[
             r"Andrew Farkas",
+            r"Jonanthan and Kimberly Farkus",
             r"Thomas\s*(J\.?\s*)?Barrack(\s*Jr)?",
         ],
     ),
@@ -708,7 +749,7 @@ HIGHLIGHTED_NAMES = [
             LISA_NEW: f'professor of poetry, wife of {LARRY_SUMMERS}, AKA "Elisa New"',
             'Lisa Randall': 'theoretical physicist',
             MARTIN_NOWAK: 'professor of mathematics and biology',
-            MOSHE_HOFFMAN: 'lecturer and research scholar in behavioral and evolutionary economics',
+            MOSHE_HOFFMAN: 'behavioral and evolutionary economics',
         },
         patterns=[
             r"Cambridge",
@@ -726,7 +767,6 @@ HIGHLIGHTED_NAMES = [
         emailers={
             ANIL_AMBANI: 'chairman of Reliance Group',
             VINIT_SAHNI: None,
-            ZUBAIR_KHAN: 'cybersecurity firm Tranchulas CEO, InsightsPod founder, Islamabad/Dubai',
         },
         patterns=[
             r"Abraaj",
@@ -747,7 +787,7 @@ HIGHLIGHTED_NAMES = [
         label='Israel',
         style='dodger_blue2',
         emailers={
-            EHUD_BARAK: 'former primer minister',
+            EHUD_BARAK: 'former prime minister of Israel, Epstein business partner',
             'Mitchell Bard': 'director of the American-Israeli Cooperative Enterprise (AICE)',
             'Nili Priell Barak': 'wife of Ehud Barak',
         },
@@ -785,7 +825,9 @@ HIGHLIGHTED_NAMES = [
         label=JOURNALIST,
         style='bright_yellow',
         emailers={
-            EDWARD_JAY_EPSTEIN: 'no relation, wrote about the kinds of crimes Epstein was involved in',
+            'Alain Forget': 'author of "How To Get Out Of This World ALIVE"',
+            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',
             JOHN_BROCKMAN: 'literary agent and author specializing in scientific literature',
@@ -800,24 +842,30 @@ HIGHLIGHTED_NAMES = [
             r"Arianna(\s*Huffington)?",
             r"(Arthur\s*)?Kretchmer",
             r'Associated\s*Press',
+            r"Axios",
             r"BBC",
-            r"Bloomberg",
             r"Breitbart",
+            r"BuzzFeed",
+            r"CBS(\s*(4|Corp|News))?"
             r"Charlie\s*Rose",
             r"China\s*Daily",
             r"CNBC",
             r"CNN(politics?)?",
-            r"Con[cs]hita",
-            r"Sarnoff",
-            r"(?<!Virgin[-\s]Islands[-\s])Daily\s*(Beast|Mail|News|Telegraph)",
+            r"Con[cs]hita", r"Sarnoff",
+            r"Daily Business Review",
+            r"(?<!Virgin[-\s]Islands[-\s])(The\s*)?Daily\s*(Beast|Mail|News|Telegraph)",
             r"(David\s*)?(Pecker|Pegg)",
             r"David\s*Brooks",
             r"Ed\s*Krassenstein",
             r"(Emily\s*)?Michot",
             r"Ezra\s*Klein",
+            r"Forbes",
+            r"Fortune\s*Magazine",
+            r"Fox\s*News(\.com)?",
             r"FrontPage Magazine",
             r"FT",
             r"(George\s*)?Stephanopoulus",
+            r"Ger(ald|ry)\s*Baker",
             r"Globe\s*and\s*Mail",
             r"Good\s*Morning\s*America",
             r"Graydon(\s*Carter)?",
@@ -829,20 +877,29 @@ HIGHLIGHTED_NAMES = [
             r"(Katie\s*)?Couric",
             r"Keith\s*Larsen",
             r"L\.?A\.?\s*Times",
+            r"Law(360|\.com|fare)",
+            r"(Les\s*)?Moonves",
+            r"MarketWatch",
             r"Miami\s*Herald",
+            r"(Mi(chael|ke)\s*)?Bloomberg",
             r"(Michele\s*)?Dargan",
+            r"Morning News USA",
             r"(National\s*)?Enquirer",
+            r"News(max|week)",
             r"NYer",
             r"Palm\s*Beach\s*(Daily\s*News|Post)",
             r"PERVERSION\s*OF\s*JUSTICE",
             r"Politico",
             r"Pro\s*Publica",
+            r"Reuters",
             r"(Sean\s*)?Hannity",
             r"Sulzberger",
             r"SunSentinel",
             r"Susan Edelman",
             r"(The\s*)?Financial\s*Times",
             r"The\s*Guardian",
+            r"TheHill",
+            r"(The\s*)?Mail\s*On\s*Sunday",
             r"(The\s*)?N(ew\s*)?Y(ork\s*)?(P(ost)?|T(imes)?)",
             r"(The\s*)?New\s*Yorker",
             r"(The\s*)?Wall\s*Street\s*Journal",
@@ -857,6 +914,18 @@ HIGHLIGHTED_NAMES = [
             r"[-\w.]+@(bbc|independent|mailonline|mirror|thetimes)\.co\.uk",
         ],
     ),
+    HighlightedNames(
+        label=JUNK,
+        style='gray46',
+        emailers={
+            'asmallworld@travel.asmallworld.net': None,
+            "digest-noreply@quora.com": None,
+            'editorialstaff@flipboard.com': None,
+            'How To Academy': None,
+            'Jokeland': None,
+        },
+        should_match_first_last_name=False,
+    ),
     HighlightedNames(
         label='Latin America',
         style='yellow',
@@ -866,11 +935,12 @@ HIGHLIGHTED_NAMES = [
             r"Bra[sz]il(ian)?",
             r"Caracas",
             r"Castro",
+            r"Chile",
             r"Colombian?",
             r"Cuban?",
             r"El\s*Salvador",
             r"((Enrique )?Pena )?Nieto",
-            r"Lat(in)?\s*Am(erica)?",
+            r"Lat(in)?\s*Am(erican?)?",
             r"Lula",
             r"Mexic(an|o)",
             r"(Nicolas\s+)?Maduro",
@@ -881,7 +951,7 @@ HIGHLIGHTED_NAMES = [
         ],
     ),
     HighlightedNames(
-        label='law enforcement',
+        label='government',
         style='color(24) bold',
         emailers={
             ANN_MARIE_VILLAFANA: 'Southern District of Florida (SDFL) U.S. Attorney',
@@ -911,6 +981,7 @@ HIGHLIGHTED_NAMES = [
             r"FINRA",
             r"FOIA",
             r"FTC",
+            r"(General\s*)?P(a|e)traeus",
             r"IRS",
             r"(James\s*)?Comey",
             r"Jeff(rey)?\s*Sessions",
@@ -950,6 +1021,7 @@ HIGHLIGHTED_NAMES = [
         label=LOBBYIST,
         style='light_coral',
         emailers={
+            BOB_CROWE: 'partner at Nelson Mullins',
             'Joshua Cooper Ramo': 'co-CEO of Henry Kissinger Associates',
             KATHERINE_KEATING: 'Daughter of former Australian PM',
             MOHAMED_WAHEED_HASSAN: 'former president of the Maldives',
@@ -959,14 +1031,13 @@ HIGHLIGHTED_NAMES = [
             'Stanley Rosenberg': 'former President of the Massachusetts Senate',
         },
         patterns=[
-            r"[BR]ob Crowe",
             r"CSIS",
             r"(Kevin\s*)?Rudd",
             r"Stanley Rosenberg",
         ],
     ),
     HighlightedNames(
-        label='mideast',
+        label=MIDEAST,
         style='dark_sea_green4',
         emailers={
             ANAS_ALRASHEED: 'former information minister of Kuwait (???)',
@@ -990,6 +1061,7 @@ HIGHLIGHTED_NAMES = [
             r"Basiji?",
             r"Beirut",
             r"Benghazi",
+            r"Brunei",
             r"Cairo",
             r"Chagoury",
             r"Damascus",
@@ -1049,7 +1121,7 @@ HIGHLIGHTED_NAMES = [
             r"sheikh",
             r"shia",
             r"(Sultan\s*)?Yacoub",
-            r"Sultan",
+            r"Sultan(?! Ahmed)",
             r"Syrian?",
             r"(Tarek\s*)?El\s*Sayed",
             r"Tehran",
@@ -1078,21 +1150,23 @@ HIGHLIGHTED_NAMES = [
         },
         patterns=[
             r"\w+@mc2mm.com",
-            r"model(ed|ing)",
+            r"MC2",
             r"(Nicole\s*)?Junkerman",  # Also a venture fund manager now
+            r"Tigrane",
         ],
     ),
     HighlightedNames(
         label=PUBLICIST,
         style='orange_red1',
         emailers={
-            AL_SECKEL: 'husband of Isabel Maxwell, Mindshift conference organizer who fell off a cliff',
-            'Barnaby Marsh': 'co-founder of Saint Partners, a philanthropy services company',
-            CHRISTINA_GALBRAITH: f"{REPUTATION_MGMT}, worked with Tyler Shears",
-            IAN_OSBORNE: f'{OSBORNE_LLP} reputation repairer hired by Epstein in 2011',
+            AL_SECKEL: "Isabel Maxwell's husband, Mindshift conference, fell off a cliff",
+            'Barnaby Marsh': 'co-founder of philanthropy services company Saint Partners',
+            CHRISTINA_GALBRAITH: f"{EPSTEIN_FOUNDATION} Media/PR, worked with {TYLER_SHEARS}",  # Title in 031441
+            IAN_OSBORNE: f'{OSBORNE_LLP} reputation repairer hired in 2011',
+            MATTHEW_HILTZIK: 'crisis PR at Hiltzik Strategies',
             MICHAEL_SITRICK: 'crisis PR',
             'Owen Blicksilver': 'OBPR, Inc.',
-            PEGGY_SIEGAL: 'socialite',
+            PEGGY_SIEGAL: 'socialite, movie promoter',
             'R. Couri Hay': None,
             ROSS_GOW: 'Acuity Reputation Management',
             TYLER_SHEARS: f"{REPUTATION_MGMT}, worked on with {CHRISTINA_GALBRAITH}",
@@ -1100,6 +1174,7 @@ HIGHLIGHTED_NAMES = [
         patterns=[
             r"(Matt(hew)? )?Hiltzi[gk]",
             REPUTATION_MGMT,
+            r"Reputation.com",
         ],
     ),
     HighlightedNames(
@@ -1123,7 +1198,7 @@ HIGHLIGHTED_NAMES = [
             r"(?<!Merwin Dela )Cruz",
             r"Devin\s*Nunes",
             r"(Don\s*)?McGa[hn]n",
-            r"Fox\s*News",
+            r"Gary\s*Cohn",
             r"George\s*(H\.?\s*)?(W\.?\s*)?Bush",
             r"(George\s*)?Nader",
             r"GOP",
@@ -1131,7 +1206,7 @@ HIGHLIGHTED_NAMES = [
             r"Kissinger",
             r"Kobach",
             r"Kolfage",
-            r"Kudlow",
+            r"(Larry\s*)?Kudlow",
             r"Lewandowski",
             r"(Marco\s)?Rubio",
             r"(Mark\s*)Meadows",
@@ -1160,7 +1235,7 @@ HIGHLIGHTED_NAMES = [
         style='indian_red',
         emailers={
             ARIANE_DE_ROTHSCHILD: 'heiress',
-            JOHNNY_EL_HACHEM: f'works with {ARIANE_DE_ROTHSCHILD}',
+            JOHNNY_EL_HACHEM: f'Edmond de Rothschild Private Equity',
         },
         patterns=['AdR'],
     ),
@@ -1169,7 +1244,7 @@ HIGHLIGHTED_NAMES = [
         style='red bold',
         emailers={
             'Dasha Zhukova': 'art collector, daughter of Alexander Zhukov',
-            MASHA_DROKOVA: 'silicon valley VC, former Putin Youth',
+            MASHA_DROKOVA: 'silicon valley VC, former Putin Youth member',
             RENATA_BOLOTOVA: 'former aspiring model, now 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",
         },
@@ -1186,7 +1261,7 @@ HIGHLIGHTED_NAMES = [
             r"KGB",
             r"Kislyak",
             r"Kremlin",
-            r"Kuznetsova",
+            r"(Anastasia\s*)?Kuznetsova",
             r"Lavrov",
             r"Lukoil",
             r"Moscow",
@@ -1208,7 +1283,6 @@ HIGHLIGHTED_NAMES = [
             r"(Vitaly\s*)?Churkin",
         ],
     ),
-
     HighlightedNames(
         label='Southeast Asia',
         style='light_salmon3 bold',
@@ -1218,9 +1292,10 @@ HIGHLIGHTED_NAMES = [
             r"Cambodian?",
             r"Laos",
             r"Malaysian?",
+            r"Maldives",
             r"Myan?mar",
             r"Philippines",
-            r"South\s*Korea",
+            r"South\s*Korean?",
             r"Tai(pei|wan)",
             r"Thai(land)?",
             r"Vietnam(ese)?",
@@ -1245,7 +1320,10 @@ HIGHLIGHTED_NAMES = [
             r"Eric\s*Schmidt",
             r"Greylock(\s*Partners)?",
             r"(?<!(ustin|Moshe)\s)Hoffmand?",
+            r"(Jeff\s*)?Bezos",
+            r"Larry Page",
             r"LinkedIn",
+            r"(Marc\s*)?Andreess?en",
             r"(Mark\s*)?Zuckerberg",
             r"Masa(yoshi)?(\sSon)?",
             r"Najeev",
@@ -1273,15 +1351,15 @@ HIGHLIGHTED_NAMES = [
             r"\bDJ?T\b",
             r"Donald J. Tramp",
             r"(Donald\s+(J\.\s+)?)?Trump(ism|\s*(Org(anization)?|Properties)(\s*LLC)?)?",
-            r"Don(ald| *Jr)(?! Rubin)",
+            r"Don(ald| *Jr)(?! (B|Rubin))",
             r"Ivank?a",
-            r"Jared",
-            r"Kushner",
+            r"Jared", r"(?<!Tony )Kushner",
             r"(Madeleine\s*)?Westerhout",
             r"Mar[-\s]*a[-\s]*Lago",
             r"(Marla\s*)?Maples",
             r"(Matt(hew)? )?Calamari",
             r"\bMatt C\b",
+            r"Michael\s*Cohen",
             r"Melania",
             r"(Michael (J.? )?)?Boccio",
             r"Paul Rampell",
@@ -1295,6 +1373,8 @@ HIGHLIGHTED_NAMES = [
         label='victim',
         style='orchid1',
         patterns=[
+            r"(David\s*)?Bo[il]es",
+            r"(Gloria\s*)?Allred",
             r"(Jane|Tiffany)\s*Doe",
             r"Katie\s*Johnson",
             r"(Virginia\s+((L\.?|Roberts)\s+)?)?Giuffre",
@@ -1313,18 +1393,19 @@ HIGHLIGHTED_NAMES = [
         patterns=[
             r"(Alan(\s*P.)?|MINTZ)\s*FRAADE",
             r"(J\.?\s*)?(Stan(ley)?\s*)?Pottinger",
+            r"(Mi(chael|ke)\s*)?Avenatti",
             r"Paul\s*(G.\s*)?Cassell",
             r"Rothstein\s*Rosenfeldt\s*Adler",
             r"(Scott\s*)?Rothstein",
         ],
     ),
     HighlightedNames(
-        label=VIRGIN_ISLANDS,
+        label='USVI',
         style='sea_green1',
         emailers={
-            CECILE_DE_JONGH: 'first lady 2007-2015',
-            KENNETH_E_MAPP: 'Governor',
-            STACEY_PLASKETT: 'non-voting member of Congress',
+            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",
@@ -1345,48 +1426,35 @@ HIGHLIGHTED_NAMES = [
             r"(West\s*)?Palm\s*Beach(?!\s*(Daily|Post))",
         ],
     ),
-    HighlightedNames(
-        label=BILL_GATES,
-        style='turquoise4',
-        emailers={
-            BORIS_NIKOLIC: f'biotech VC partner of {BILL_GATES}, {EPSTEIN_ESTATE_EXECUTOR}',
-        },
-        patterns=[
-            r"BG",
-            r"b?g?C3",
-            r"(Bill\s*((and|or)\s*Melinda\s*)?)?Gates",
-            r"Melinda(\s*Gates)?",
-            r"Microsoft",
-            r"MSFT",
-        ],
-    ),
+
+    # Individuals
     HighlightedNames(
         label=STEVE_BANNON,
         style='color(58)',
-        category=POLITICS,
+        category='Republican',
         emailers={
-            STEVE_BANNON: "Trump campaign manager, memecoin grifter, convicted criminal",
+            SEAN_BANNON: f"{STEVE_BANNON}'s brother",
+            STEVE_BANNON: "Trump campaign manager, memecoin grifter",
         },
         patterns=[
             r"(American\s*)?Dharma",
-            r"((Steve|Sean)\s*)?Bannon?",
+            r"Biosphere",
         ],
     ),
     HighlightedNames(
         style='dark_olive_green3',
         category=FINANCE,
-        emailers={STEVEN_HOFFENBERG: "Epstein's ponzi scheme partner at Towers Financial, prison for 18 years"},
+        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"TerraMar"], style='deep_pink3'),
-    HighlightedNames(emailers={JABOR_Y: '"an influential man in Qatar"'}, category='mideast', style='spring_green1'),
-    HighlightedNames(emailers={JEFFREY_EPSTEIN: None}, patterns=[r"JEGE", r"LSJ", r"Mark (L. )?Epstein"], style='blue1'),
-    HighlightedNames(emailers={KATHRYN_RUEMMLER: 'former Obama legal counsel'}, style='magenta2'),
-    HighlightedNames(emailers={MELANIE_WALKER: 'doctor'}, style='pale_violet_red1'),
-    HighlightedNames(emailers={PAULA: "Epstein's ex-girlfriend who is now in the opera world"}, label='paula', style='pink1'),
-    HighlightedNames(emailers={PRINCE_ANDREW: 'British royal family'}, style='dodger_blue1'),
-    HighlightedNames(emailers={SOON_YI_PREVIN: 'wife of Woody Allen'}, style='hot_pink'),
-    HighlightedNames(emailers={SULTAN_BIN_SULAYEM: 'CEO of DP World, chairman of ports in Dubai'}, style='green1'),
+    HighlightedNames(emailers={GHISLAINE_MAXWELL: None}, patterns=[r"gmax(1@ellmax.com)?", r"(The )?TerraMar Project"], style='deep_pink3', category='Epstein'),
+    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),
+    HighlightedNames(emailers={PAULA: "Epstein's ex-girlfriend who is now in the opera world"}, label='paula', style='pink1', category=FRIEND),
+    HighlightedNames(emailers={PRINCE_ANDREW: 'British royal family'}, style='dodger_blue1', category='Europe'),
+    HighlightedNames(emailers={SOON_YI_PREVIN: 'wife of Woody Allen'}, style='hot_pink', category=FRIEND),
+    HighlightedNames(emailers={SULTAN_BIN_SULAYEM: 'CEO of DP World, chairman of ports in Dubai'}, style='green1', category=MIDEAST),
 
     # HighlightedText not HighlightedNames bc of word boundary issue
     HighlightedText(
@@ -1409,7 +1477,7 @@ HIGHLIGHTED_TEXTS = [
     HighlightedText(
         label='header_field',
         style='plum4',
-        patterns=[r'^(> )?(Date|From|Sent|To|C[cC]|Importance|Subject|Bee|B[cC]{2}|Attachments):'],
+        patterns=[r'^(> )?(Date|From|Sent|To|C[cC]|Importance|Reply-To|Subject|Bee|B[cC]{2}|Attachments|Flag|Classification):'],
     ),
     HighlightedText(
         label='http_links',
@@ -1456,6 +1524,7 @@ HIGHLIGHTED_TEXTS = [
 ]
 
 ALL_HIGHLIGHTS = HIGHLIGHTED_NAMES + HIGHLIGHTED_TEXTS
+JUNK_EMAILERS = [name for name in [hn for hn in HIGHLIGHTED_NAMES if hn.label == JUNK][0].emailers.keys()]
 
 
 class EpsteinHighlighter(RegexHighlighter):
@@ -1464,23 +1533,6 @@ class EpsteinHighlighter(RegexHighlighter):
     highlights = [highlight_group.regex for highlight_group in ALL_HIGHLIGHTS]
 
 
-def get_category_txt_for_name(name: str | None) -> Text | None:
-    highlight_group = _get_highlight_group_for_name(name)
-
-    if highlight_group and isinstance(highlight_group, HighlightedNames):
-        category = highlight_group.category or highlight_group.label
-
-        if category != name:
-            return styled_category(category)
-
-
-def get_info_for_name(name: str | None) -> str | None:
-    highlight_group = _get_highlight_group_for_name(name)
-
-    if highlight_group and isinstance(highlight_group, HighlightedNames) and name:
-        return highlight_group.get_info(name)
-
-
 def get_style_for_category(category: str) -> str | None:
     if category in CATEGORY_STYLES:
         return CATEGORY_STYLES[category]
@@ -1494,20 +1546,27 @@ def get_style_for_category(category: str) -> str | None:
             return highlight_group.style
 
 
-def get_style_for_name(name: str | None, default_style: str = DEFAULT, allow_bold: bool = True) -> str:
-    highlight_group = _get_highlight_group_for_name(name or UNKNOWN)
+def get_style_for_name(name: str | None, default_style: str = DEFAULT_NAME_STYLE, allow_bold: bool = True) -> str:
+    highlight_group = get_highlight_group_for_name(name or UNKNOWN)
     style = highlight_group.style if highlight_group else default_style
-    return style if allow_bold else style.replace('bold', '').strip()
+    style = style if allow_bold else style.replace('bold', '').strip()
+    logger.debug(f"get_style_for_name('{name}', '{default_style}', '{allow_bold}') yielded '{style}'")
+    return style
 
 
 def styled_category(category: str | None) -> Text:
     if not category:
         return QUESTION_MARKS_TXT
 
-    return Text(category, get_style_for_category(category) or 'wheat4')
+    category_str = 'resumé' if category == 'resume' else category.lower()
+    return Text(category_str, get_style_for_category(category) or 'wheat4')
+
 
+def styled_name(name: str | None, default_style: str = DEFAULT_NAME_STYLE) -> Text:
+    return Text(name or UNKNOWN, style=get_style_for_name(name, default_style=default_style))
 
-def _get_highlight_group_for_name(name: str | None) -> HighlightedNames | None:
+
+def get_highlight_group_for_name(name: str | None) -> HighlightedNames | None:
     if name is None:
         return None
 
@@ -1520,6 +1579,9 @@ def _print_highlighted_names_repr() -> None:
     for hn in HIGHLIGHTED_NAMES:
         if isinstance(hn, HighlightedNames):
             print(indented(repr(hn)) + ',')
+            print(f"pattern: '{hn.regex.pattern}'")
 
     import sys
     sys.exit()
+
+#_print_highlighted_names_repr()