epstein-files 1.2.5__py3-none-any.whl → 1.5.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (37) hide show
  1. epstein_files/__init__.py +55 -23
  2. epstein_files/documents/communication.py +9 -5
  3. epstein_files/documents/document.py +231 -135
  4. epstein_files/documents/doj_file.py +242 -0
  5. epstein_files/documents/doj_files/full_text.py +166 -0
  6. epstein_files/documents/email.py +289 -232
  7. epstein_files/documents/emails/email_header.py +35 -16
  8. epstein_files/documents/emails/emailers.py +223 -0
  9. epstein_files/documents/imessage/text_message.py +2 -3
  10. epstein_files/documents/json_file.py +18 -14
  11. epstein_files/documents/messenger_log.py +23 -39
  12. epstein_files/documents/other_file.py +54 -48
  13. epstein_files/epstein_files.py +65 -29
  14. epstein_files/person.py +151 -94
  15. epstein_files/util/constant/names.py +37 -10
  16. epstein_files/util/constant/output_files.py +2 -0
  17. epstein_files/util/constant/strings.py +14 -7
  18. epstein_files/util/constant/urls.py +17 -0
  19. epstein_files/util/constants.py +556 -391
  20. epstein_files/util/data.py +2 -0
  21. epstein_files/util/doc_cfg.py +44 -33
  22. epstein_files/util/env.py +34 -19
  23. epstein_files/util/file_helper.py +30 -6
  24. epstein_files/util/helpers/debugging_helper.py +13 -0
  25. epstein_files/util/helpers/env_helpers.py +21 -0
  26. epstein_files/util/highlighted_group.py +121 -37
  27. epstein_files/util/layout/left_bar_panel.py +26 -0
  28. epstein_files/util/logging.py +28 -13
  29. epstein_files/util/output.py +49 -40
  30. epstein_files/util/rich.py +30 -3
  31. epstein_files/util/word_count.py +7 -7
  32. {epstein_files-1.2.5.dist-info → epstein_files-1.5.0.dist-info}/METADATA +16 -3
  33. epstein_files-1.5.0.dist-info/RECORD +40 -0
  34. {epstein_files-1.2.5.dist-info → epstein_files-1.5.0.dist-info}/entry_points.txt +1 -1
  35. epstein_files-1.2.5.dist-info/RECORD +0 -34
  36. {epstein_files-1.2.5.dist-info → epstein_files-1.5.0.dist-info}/LICENSE +0 -0
  37. {epstein_files-1.2.5.dist-info → epstein_files-1.5.0.dist-info}/WHEEL +0 -0
@@ -7,11 +7,11 @@ from rich.console import Console
7
7
  from rich.highlighter import RegexHighlighter
8
8
  from rich.text import Text
9
9
 
10
+ from epstein_files.documents.emails.emailers import EMAILER_ID_REGEXES
10
11
  from epstein_files.util.constant.names import *
11
12
  from epstein_files.util.constant.strings import *
12
13
  from epstein_files.util.constant.urls import ARCHIVE_LINK_COLOR
13
- from epstein_files.util.constants import (EMAILER_ID_REGEXES, EPSTEIN_V_ROTHSTEIN_EDWARDS,
14
- OSBORNE_LLP, REPLY_REGEX, SENT_FROM_REGEX)
14
+ from epstein_files.util.constants import EPSTEIN_V_ROTHSTEIN_EDWARDS, OSBORNE_LLP, REPLY_REGEX, SENT_FROM_REGEX
15
15
  from epstein_files.util.data import sort_dict, without_falsey
16
16
  from epstein_files.util.doc_cfg import *
17
17
  from epstein_files.util.env import args
@@ -31,6 +31,8 @@ REGEX_STYLE_PREFIX = 'regex'
31
31
  SIMPLE_NAME_REGEX = re.compile(r"^[-\w, ]+$", re.IGNORECASE)
32
32
  TECH_BRO = 'tech bro'
33
33
 
34
+ VICTIM_COLOR = 'orchid1'
35
+
34
36
  CATEGORY_STYLE_MAPPING = {
35
37
  ARTICLE: JOURNALIST,
36
38
  BOOK: JOURNALIST,
@@ -160,7 +162,7 @@ class HighlightedNames(HighlightedText):
160
162
 
161
163
  pattern = '|'.join(name_patterns)
162
164
 
163
- if args.deep_debug:
165
+ if args.deep_debug and args.colors_only:
164
166
  debug_console.print(Text('').append(f"{name:25s}", style=self.style).append(f" '{pattern}'", style='dim'))
165
167
 
166
168
  return pattern
@@ -215,7 +217,7 @@ HIGHLIGHTED_NAMES = [
215
217
  ManualHighlight(
216
218
  label='email_subject',
217
219
  style='light_yellow3',
218
- pattern=r"^(> )?(Classification|Flag|Subject): (?P<email_subject>.*)",
220
+ pattern=r"^(> )?(Classification|Flag|Subject|Sujet ?): (?P<email_subject>.*)",
219
221
  ),
220
222
  HighlightedNames(
221
223
  label=ACADEMIA,
@@ -245,12 +247,13 @@ HIGHLIGHTED_NAMES = [
245
247
  r"Bard\s+((Early )?College|High School|Schools)",
246
248
  r"Brotherton",
247
249
  r"Carl\s*Sagan",
248
- r"Columbia",
250
+ r"Columbia(\s*(Business\s*School|University))?",
249
251
  r"Dan(iel|ny) Kahneman",
250
252
  r"(Francis\s*)?Crick",
251
253
  r"J(ames|im)\s*Watson",
252
254
  r"(Lord\s*)?Martin\s*Rees",
253
255
  r"Massachusetts\s*Institute\s*of\s*Technology",
256
+ r"Mayo\s*Clinic",
254
257
  r"Media\s*Lab",
255
258
  r"(Marvin\s*)?Minsky",
256
259
  r"MIT(\s*Media\s*Lab)?",
@@ -260,10 +263,11 @@ HIGHLIGHTED_NAMES = [
260
263
  r"Princeton(\s*University)?",
261
264
  r"Regeneron",
262
265
  r"(Richard\s*)?Dawkins",
266
+ r"Rockefeller\s*University",
263
267
  r"(Sandy\s*)?Pentland", # Media Lab
264
268
  r"Sanofi",
265
269
  r"Stanford(\s*University)?(\s*Hospital)?",
266
- r"(Stephen\s*)?Hawking",
270
+ r"(Ste(ph|v)en\s*)?Hawking",
267
271
  r"(Steven?\s*)?Pinker",
268
272
  r"Texas\s*A&M",
269
273
  r"Tulane",
@@ -319,6 +323,7 @@ HIGHLIGHTED_NAMES = [
319
323
  patterns=[
320
324
  r"(Art )?Spiegelman",
321
325
  r"Artspace",
326
+ r"Ayn\s*Rand",
322
327
  r"Bobby slayton",
323
328
  r"bono\s*mick",
324
329
  r"Errol(\s*Morris)?",
@@ -367,6 +372,8 @@ HIGHLIGHTED_NAMES = [
367
372
  label='bitcoin',
368
373
  style='orange1 bold',
369
374
  emailers={
375
+ AMIR_TAAKI: f"bitcoin bro, partner of {BROCK_PIERCE} (?)",
376
+ BROCK_PIERCE: 'crypto bro, Bannon business partner, Tether co-founder, arrested in house full of machine guns & "8,000 fotografias de pornopgraphia infantil"',
370
377
  JEFFREY_WERNICK: 'former COO of Parler, involved in numerous crypto companies like Bitforex',
371
378
  JEREMY_RUBIN: 'developer/researcher',
372
379
  JOI_ITO: f"former head of {MIT_MEDIA_LAB} and MIT Digital Currency Initiative",
@@ -374,11 +381,12 @@ HIGHLIGHTED_NAMES = [
374
381
  },
375
382
  patterns=[
376
383
  r"Balaji",
384
+ r"Bart\s*Stephens", # co-founder, Blockchain Capital
385
+ r"Bioptix", # Now RIOT Blockchain
377
386
  r"bitcoin(\s*Foundation)?",
378
- r"block ?chain(\s*capital)?",
379
387
  r"Brian Forde",
380
388
  r"Brock(\s*Pierce)?",
381
- r"coins?",
389
+ r"coins?(\s*Center)?",
382
390
  r"Cory\s*Fields", # bitcoin dev
383
391
  r"cr[iy]?pto(currenc(y|ies))?",
384
392
  r"Digital\s*Currenc(ies|y)(\s*Initiative)?",
@@ -390,12 +398,16 @@ HIGHLIGHTED_NAMES = [
390
398
  r"Madars",
391
399
  r"Mi(chael|ke)\s*Novogratz",
392
400
  r"(Patrick\s*)?Murck",
393
- r"Ron Rivest",
401
+ r"(Hester\s*)?Peirce",
402
+ r"(?-i:RIOT)", # (?-i:) makes it case sensitive
403
+ r"Ron\s*Rivest",
404
+ r"block ?chain(\s*capital)?",
394
405
  r"(Ross\s*)?Ulbricht",
395
406
  r"Silk\s*Road",
396
407
  r"SpanCash",
397
408
  r"Tether",
398
409
  r"virtual\s*currenc(ies|y)",
410
+ r"Wire\s*ca\n?rd",
399
411
  r"Wladimir( van der Laan)?", # bitcoin dev
400
412
  r"(zero\s+knowledge\s+|zk)pro(of|tocols?)",
401
413
  ],
@@ -413,9 +425,11 @@ HIGHLIGHTED_NAMES = [
413
425
  'Philip Kafka': 'president of Prince Concepts (and son of Terry Kafka?)',
414
426
  ROBERT_LAWRENCE_KUHN: 'investment banker, China expert',
415
427
  TERRY_KAFKA: 'CEO of Impact Outdoor (highway billboards)',
416
- TOM_PRITZKER: 'brother of J.B. Pritzker',
428
+ TOM_PRITZKER: 'chairman of The Pritzker Organization and Hyatt Hotels',
417
429
  },
418
430
  patterns=[
431
+ r"Arthur Klein",
432
+ r"(Barry\s*)?Honig",
419
433
  r"((Bill|David)\s*)?Koch(\s*(Bro(s|thers)|Industries))?",
420
434
  r"Gruterite",
421
435
  r"((John|Patricia)\s*)?Kluge",
@@ -423,6 +437,7 @@ HIGHLIGHTED_NAMES = [
423
437
  r"(Mi(chael|ke)\s*)?Ovitz",
424
438
  r"(Steve\s+)?Wynn",
425
439
  r"(Les(lie)?\s+)?Wexner",
440
+ r"Michael\s*Klein",
426
441
  r"New Leaf Ventures",
427
442
  r"Park Partners",
428
443
  r"SALSS",
@@ -500,6 +515,7 @@ HIGHLIGHTED_NAMES = [
500
515
  r"Dem(ocrat(ic)?)?",
501
516
  r"(Diana\s*)?DeGette",
502
517
  r"DNC",
518
+ r"(Ed(ward)?\s*)?Mezvinsky",
503
519
  r"Elena\s*Kagan",
504
520
  r"(Eliott?\s*)?Spitzer(, Eliot)?",
505
521
  r"Eric Holder",
@@ -550,7 +566,6 @@ HIGHLIGHTED_NAMES = [
550
566
  MERWIN_DELA_CRUZ: None, # HOUSE_OVERSIGHT_032652 Groff says "Jojo and Merwin both requested off Nov. 25 and 26"
551
567
  NADIA_MARCINKO: "Epstein's pilot",
552
568
  'Sean J. Lancaster': 'airplane reseller',
553
- ZUBAIR_KHAN: 'Tranchulas cybersecurity, InsightsPod founder, Islamabad / Dubai',
554
569
  },
555
570
  patterns=[
556
571
  r"Adriana\s*Ross",
@@ -566,8 +581,9 @@ HIGHLIGHTED_NAMES = [
566
581
  MARK_EPSTEIN: 'brother of Jeffrey',
567
582
  },
568
583
  patterns=[
569
- r"JEGE(\s*Inc)",
584
+ r"JEGE(\s*Inc)?",
570
585
  r"LSJ",
586
+ r"Zorro(\s*Ranch)?",
571
587
  ],
572
588
  ),
573
589
  HighlightedNames(
@@ -579,6 +595,7 @@ HIGHLIGHTED_NAMES = [
579
595
  ARDA_BESKARDES: 'NYC immigration attorney allegedly involved in sex-trafficking operations',
580
596
  BENNET_MOSKOWITZ: f'represented the {EPSTEIN_ESTATE_EXECUTOR}s',
581
597
  BRAD_KARP: 'head of the law firm Paul Weiss',
598
+ CHRISTIAN_EVERDELL: f"{GHISLAINE_MAXWELL}'s lawyer ca. 2021, Cohen & Gresser",
582
599
  'Connie Zaguirre': f"office of {ROBERT_D_CRITTON_JR}",
583
600
  DAVID_SCHOEN: f"{CRIMINAL_DEFENSE_ATTORNEY} after 2019 arrest",
584
601
  DEBBIE_FEIN: EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY,
@@ -599,6 +616,7 @@ HIGHLIGHTED_NAMES = [
599
616
  'Robert Gold': 'helped Epstein track down money belonging to Spanish families',
600
617
  'Roy Black': CRIMINAL_DEFENSE_2008,
601
618
  SCOTT_J_LINK: CRIMINAL_DEFENSE_ATTORNEY,
619
+ STACEY_RICHMAN: f"New York {CRIMINAL_DEFENSE_ATTORNEY}",
602
620
  TONJA_HADDAD_COLEMAN: f'{EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY}', # relation of Fred Haddad?
603
621
  },
604
622
  patterns=[
@@ -610,6 +628,7 @@ HIGHLIGHTED_NAMES = [
610
628
  r"Michael J. Pike",
611
629
  r"Paul,?\s*Weiss",
612
630
  r"Steptoe(\s*& Johnson)?(\s*LLP)?",
631
+ r"Sull(ivan)?\s*(&|and)?\s*Crom(well)?",
613
632
  r"Wein(berg|garten)",
614
633
  ],
615
634
  ),
@@ -638,8 +657,10 @@ HIGHLIGHTED_NAMES = [
638
657
  },
639
658
  patterns=[
640
659
  r"AfD",
660
+ r"Alfa(\s*Bank)",
641
661
  r"(Angela )?Merk(el|le)",
642
662
  r"Austria",
663
+ r"Belgi(an|um)",
643
664
  r"(Benjamin\s*)?Harnwell",
644
665
  r"Berlin",
645
666
  r"Borge",
@@ -649,6 +670,7 @@ HIGHLIGHTED_NAMES = [
649
670
  r"Brussels",
650
671
  r"Cannes",
651
672
  r"Cypr(iot|us)",
673
+ r"David\s*Cameron",
652
674
  r"Davos",
653
675
  r"ECB",
654
676
  r"England",
@@ -663,6 +685,8 @@ HIGHLIGHTED_NAMES = [
663
685
  r"Ital(ian|y)",
664
686
  r"Jacques",
665
687
  r"Kiev",
688
+ r"Latvian?",
689
+ r"Lithuanian?",
666
690
  r"Le\s*Pen",
667
691
  r"London",
668
692
  r"Macron",
@@ -672,11 +696,13 @@ HIGHLIGHTED_NAMES = [
672
696
  r"NATO",
673
697
  r"(Nicholas\s*)?Sarkozy",
674
698
  r"Nigel(\s*Farage)?",
699
+ r"(Northern\s*)?Ireland",
675
700
  r"Norw(ay|egian)",
676
701
  r"Oslo",
677
702
  r"Paris",
678
703
  r"Polish",
679
704
  r"pope",
705
+ r"Portugal",
680
706
  r"Scotland",
681
707
  r"(Sebastian )?Kurz",
682
708
  r"Stockholm",
@@ -685,6 +711,7 @@ HIGHLIGHTED_NAMES = [
685
711
  r"Swed(en|ish)(?![-\s]+American Life Scienc)",
686
712
  r"Swi(ss|tzerland)",
687
713
  r"(Tony\s)?Blair",
714
+ r"United\s*Kingdom",
688
715
  r"U\.K\.",
689
716
  r"Ukrain(e|ian)",
690
717
  r"Venice",
@@ -748,6 +775,7 @@ HIGHLIGHTED_NAMES = [
748
775
  r"(Janet\s*)?Yellen",
749
776
  r"(Jerome\s*)?Powell(?! M\. Cabot)",
750
777
  r"(Jimmy\s*)?Cayne",
778
+ r"Joon\s*Yun",
751
779
  r"JPMC?",
752
780
  r"j\.?p\.?\s*morgan(\.?com|\s*Chase)?",
753
781
  r"Madoff",
@@ -760,8 +788,12 @@ HIGHLIGHTED_NAMES = [
760
788
  r"(Peter L. )?Scher",
761
789
  r"(Ray\s*)?Dalio",
762
790
  r"(Richard\s*)?LeFrak",
763
- r"(Ste(phen|ve)\s*)?Schwart?z?man",
791
+ r"Rockefeller(?! University)(\s*Foundation)?",
792
+ r"SBNY",
764
793
  r"Serageldin",
794
+ r"Signature\s*Bank",
795
+ r"(Ste(phen|ve)\s*)?Schwart?z?man",
796
+ r"Susquehanna",
765
797
  r"UBS",
766
798
  r"us.gio@jpmorgan.com",
767
799
  r"Wall\s*Street(?!\s*Jour)",
@@ -789,7 +821,9 @@ HIGHLIGHTED_NAMES = [
789
821
  label='government',
790
822
  style='color(24) bold',
791
823
  emailers={
824
+ ALISON_J_NATHAN: "judge in New York's Southern District",
792
825
  ANN_MARIE_VILLAFANA: 'Southern District of Florida (SDFL) U.S. Attorney',
826
+ CHRISTOPHER_DILORIO: 'self described whistleblower',
793
827
  DANNY_FROST: 'Director of Communications at Manhattan D.A.',
794
828
  'Police Code Enforcement': f"{PALM_BEACH} buildings code enforcement",
795
829
  },
@@ -799,7 +833,9 @@ HIGHLIGHTED_NAMES = [
799
833
  r'Alice\s*Fisher|Fisher, Alice',
800
834
  r"AML",
801
835
  r"(Andrew\s*)?(McCabe|Natsios)",
836
+ r"(Assistant\s+)?State\s*Attorney",
802
837
  r"Attorney General",
838
+ r'Barbara\s*Burns', # AUSA
803
839
  r"((Bob|Robert)\s*)?Mueller",
804
840
  r"(Byung\s)?Pak",
805
841
  r"Case 1:19-cv-03377(-LAP)?",
@@ -812,17 +848,19 @@ HIGHLIGHTED_NAMES = [
812
848
  r"Dep(artmen)?t\.?\s*of\s*(the\s*)?(Justice|Treasury)",
813
849
  r"DHS",
814
850
  r"DOJ",
851
+ r"EDGAR (Filing|Search)", # SEC database is EDGAR
815
852
  r"FBI",
816
853
  r"FCPA",
817
854
  r"FDIC",
818
855
  r"FDLE",
819
856
  r"Federal\s*Bureau\s*of\s*Investigation",
820
857
  r"FinCEN",
821
- r"FINRA",
858
+ r"(www\.)?FINRA(\.org)?",
822
859
  r"FOIA",
823
860
  r"FTC",
824
861
  r"(General\s*)?P(a|e)traeus",
825
862
  r"Geoff\s*Ling",
863
+ r"Homeland\s*Security",
826
864
  r"IRS",
827
865
  r"(James\s*)?Comey",
828
866
  r"(Jennifer\s*Shasky\s*)?Calvery",
@@ -840,11 +878,12 @@ HIGHLIGHTED_NAMES = [
840
878
  r"(Michael\s*)?Reiter",
841
879
  r"OGE",
842
880
  r"Office\s*of\s*Government\s*Ethics",
881
+ r"PBPD",
843
882
  r"police",
844
883
  r"(Preet\s*)?Bharara",
845
884
  r"SCOTUS",
846
885
  r"SD(FL|NY)",
847
- r"SEC",
886
+ r"SEC(\.gov)?",
848
887
  r"Secret\s*Service",
849
888
  r"Securities\s*and\s*Exchange\s*Commission",
850
889
  r"Southern\s*District(\s*of\s*(Florida|New\s*York))?",
@@ -919,6 +958,7 @@ HIGHLIGHTED_NAMES = [
919
958
  r"(eh|(Ehud|Nili Priell)\s*)?barak",
920
959
  r"EB",
921
960
  r"Ehud\s*Barack",
961
+ r"Hapoalim",
922
962
  r"Israeli?",
923
963
  r"Jerusalem",
924
964
  r"J\s*Street",
@@ -952,7 +992,7 @@ HIGHLIGHTED_NAMES = [
952
992
  'Alain Forget': 'author of "How To Get Out Of This World ALIVE"',
953
993
  'Alex Yablon': 'New York Magazine fact checker (?)',
954
994
  EDWARD_JAY_EPSTEIN: 'no relation, wrote books about spies',
955
- HENRY_HOLT: f"{MICHAEL_WOLFF}'s book publisher",
995
+ HENRY_HOLT: f"{MICHAEL_WOLFF}'s book publisher (company not a person)",
956
996
  JAMES_HILL: 'ABC News',
957
997
  JENNIFER_JACQUET: 'Future Science magazine',
958
998
  JOHN_BROCKMAN: 'literary agent and author specializing in scientific literature',
@@ -972,14 +1012,14 @@ HIGHLIGHTED_NAMES = [
972
1012
  r'Associated\s*Press',
973
1013
  r"Axios",
974
1014
  r"BBC",
975
- r"(Bob|Robert)\s*Costa",
1015
+ r"(Bob|Robert)\s*(Costa|Woodward)",
976
1016
  r"Breitbart",
977
1017
  r"BuzzFeed(\s*News)?",
978
1018
  r"C-?Span",
979
1019
  r"CBS(\s*(4|Corp|News))?",
980
1020
  r"Charlie\s*Rose",
981
1021
  r"China\s*Daily",
982
- r"CNBC",
1022
+ r"(C|MS)?NBC(\s*News)?",
983
1023
  r"CNN(politics?)?",
984
1024
  r"Con[cs]hita", r"Sarnoff",
985
1025
  r"Daily Business Review",
@@ -1000,6 +1040,7 @@ HIGHLIGHTED_NAMES = [
1000
1040
  r"Globe\s*and\s*Mail",
1001
1041
  r"Good\s*Morning\s*America",
1002
1042
  r"Graydon(\s*Carter)?",
1043
+ r"Hollywood\s*Reporter",
1003
1044
  r"Huff(ington)?(\s*Po(st)?)?",
1004
1045
  r"Ingram, David",
1005
1046
  r"James\s*Hill",
@@ -1007,6 +1048,7 @@ HIGHLIGHTED_NAMES = [
1007
1048
  r"Jesse Kornbluth",
1008
1049
  r"John\s*Connolly",
1009
1050
  r"Jonathan\s*Karl",
1051
+ r"Journal of Criminal Law and Criminology",
1010
1052
  r"Julie\s*(K.?\s*)?Brown", r'jbrown@miamiherald.com',
1011
1053
  r"(Katie\s*)?Couric",
1012
1054
  r"Keith\s*Larsen",
@@ -1025,7 +1067,6 @@ HIGHLIGHTED_NAMES = [
1025
1067
  r"PERVERSION\s*OF\s*JUSTICE",
1026
1068
  r"Politico",
1027
1069
  r"Pro\s*Publica",
1028
- r"Reuters",
1029
1070
  r"(Sean\s*)?Hannity",
1030
1071
  r"Sharon Churcher", # Daily Mail
1031
1072
  r"Sulzberger",
@@ -1038,7 +1079,9 @@ HIGHLIGHTED_NAMES = [
1038
1079
  r"(The\s*)?New\s*Yorker",
1039
1080
  r"(The\s*)?Wall\s*Street\s*Journal",
1040
1081
  r"(The\s*)?Wa(shington\s*)?Po(st)?",
1082
+ r"(Thomson\s*)?Reuters",
1041
1083
  r"(Uma\s*)?Sanghvi",
1084
+ r"USA\s*Today",
1042
1085
  r"Vanity\s*Fair",
1043
1086
  r"Viceland",
1044
1087
  r"Vick[iy]\s*Ward",
@@ -1072,6 +1115,7 @@ HIGHLIGHTED_NAMES = [
1072
1115
  r"Chile",
1073
1116
  r"Colombian?",
1074
1117
  r"Cuban?",
1118
+ r"el chapo",
1075
1119
  r"El\s*Salvador",
1076
1120
  r"((Enrique )?Pena )?Nieto",
1077
1121
  r"Lat(in)?\s*Am(erican?)?",
@@ -1113,12 +1157,17 @@ HIGHLIGHTED_NAMES = [
1113
1157
  r"Arizona(?! State University)",
1114
1158
  r"Aspen",
1115
1159
  r"Berkeley",
1160
+ r"Boston",
1116
1161
  r"Brooklyn",
1117
1162
  r"California",
1118
1163
  r"Canada",
1119
1164
  r"Cape Cod",
1165
+ r"Charlottesville",
1166
+ r"Colorado",
1120
1167
  r"Connecticut",
1121
1168
  r"Florida",
1169
+ r"Jersey\s*City",
1170
+ r"Los Angeles",
1122
1171
  r"Loudoun\s*County?",
1123
1172
  r"Martha's\s*Vineyard",
1124
1173
  r"Miami(?!\s?Herald)",
@@ -1128,9 +1177,12 @@ HIGHLIGHTED_NAMES = [
1128
1177
  r"NY(C|\s*State)",
1129
1178
  r"Orange\s*County",
1130
1179
  r"Oregon",
1180
+ r"Palo Alto",
1181
+ r"Pennsylvania",
1131
1182
  r"Phoenix",
1132
1183
  r"Portland",
1133
- r"Santa\s*Fe",
1184
+ r"San Francisco",
1185
+ r"Sant[ae]\s*Fe",
1134
1186
  r"Telluride",
1135
1187
  r"Teterboro",
1136
1188
  r"Texas(?! A&M)",
@@ -1146,7 +1198,7 @@ HIGHLIGHTED_NAMES = [
1146
1198
  style='dark_sea_green4',
1147
1199
  emailers={
1148
1200
  ANAS_ALRASHEED: 'former information minister of Kuwait (???)',
1149
- AZIZA_ALAHMADI: 'Abu Dhabi Department of Culture & Tourism',
1201
+ AZIZA_ALAHMADI: 'Abu Dhabi Department of Culture & Tourism, assistant of Al Sabbagh',
1150
1202
  RAAFAT_ALSABBAGH: 'Saudi royal advisor',
1151
1203
  SHAHER_ABDULHAK_BESHER: 'Yemeni billionaire',
1152
1204
  },
@@ -1157,6 +1209,7 @@ HIGHLIGHTED_NAMES = [
1157
1209
  r"Afghanistan",
1158
1210
  r"Al[-\s]?Qa[ei]da",
1159
1211
  r"Ahmadinejad",
1212
+ r"(Rakhat )?Aliyev",
1160
1213
  r"Arab",
1161
1214
  r"Aramco",
1162
1215
  r"Armenia",
@@ -1185,6 +1238,7 @@ HIGHLIGHTED_NAMES = [
1185
1238
  r"Hamas",
1186
1239
  r"Hezbollah",
1187
1240
  r"HBJ",
1241
+ r"Hourani",
1188
1242
  r"Houthi",
1189
1243
  r"Imran\s+Khan",
1190
1244
  r"Iran(ian)?([-\s]Contra)?",
@@ -1207,10 +1261,11 @@ HIGHLIGHTED_NAMES = [
1207
1261
  r"MB(N|S|Z)",
1208
1262
  r"Mid(dle)?\s*East(ern)?",
1209
1263
  r"Mohammed\s+bin\s+Salman",
1210
- r"Morocco",
1264
+ r"Morocc(an|o)",
1211
1265
  r"Mubarak",
1212
1266
  r"Muslim(\s*Brotherhood)?",
1213
1267
  r"Nayaf",
1268
+ r"Nazarbayev",
1214
1269
  r"Pakistani?",
1215
1270
  r"Omar",
1216
1271
  r"(Osama\s*)?Bin\s*Laden",
@@ -1230,10 +1285,10 @@ HIGHLIGHTED_NAMES = [
1230
1285
  r"Syrian?",
1231
1286
  r"(Tarek\s*)?El\s*Sayed",
1232
1287
  r"Tehran",
1288
+ r"Timur\s*Kulibayev",
1233
1289
  r"Tripoli",
1234
1290
  r"Tunisian?",
1235
- r"Turk(ey|ish)",
1236
- r"Turks(?! & Caicos)",
1291
+ r"Turk(ey|ish)?(?!s & Caicos)",
1237
1292
  r"UAE",
1238
1293
  r"((Iraq|Iran|Kuwait|Qatar|Yemen)i?)",
1239
1294
  ],
@@ -1281,6 +1336,7 @@ HIGHLIGHTED_NAMES = [
1281
1336
  },
1282
1337
  patterns=[
1283
1338
  r"(Matt(hew)? )?Hiltzi[gk]",
1339
+ r"Philip\s*Barden",
1284
1340
  r"PR\s*Newswire",
1285
1341
  REPUTATION_MGMT,
1286
1342
  r"Reputation.com",
@@ -1308,6 +1364,7 @@ HIGHLIGHTED_NAMES = [
1308
1364
  r"Broidy",
1309
1365
  r"(Chris\s)?Christie",
1310
1366
  r"(?<!Merwin Dela )Cruz",
1367
+ r"Darrell\s*Issa",
1311
1368
  r"Devin\s*Nunes",
1312
1369
  r"(Don\s*)?McGa[hn]n",
1313
1370
  r"Erik Prince",
@@ -1315,6 +1372,8 @@ HIGHLIGHTED_NAMES = [
1315
1372
  r"George\s*(H\.?\s*)?(W\.?\s*)?Bush",
1316
1373
  r"(George\s*)?Nader",
1317
1374
  r"GOP",
1375
+ r"((Chair|Jay|Joseph)\s*)?Clayton", # SEC chair, now SDNY
1376
+ r"((Bill|William)\s*)?Hinman"
1318
1377
  r"Jeff(rey)?\s*Sessions",
1319
1378
  r"(John\s*(R.?\s*)?)?Bolton",
1320
1379
  r"Kasich",
@@ -1333,7 +1392,7 @@ HIGHLIGHTED_NAMES = [
1333
1392
  r"(Michael\s)?Hayden",
1334
1393
  r"((General|Mike)\s*)?(Flynn|Pence)",
1335
1394
  r"(Mitt\s*)?Romney",
1336
- r"Mnuchin",
1395
+ r"(Steven?\s*)?Mnuchin",
1337
1396
  r"(Newt\s*)Gingrich",
1338
1397
  r"Nikki",
1339
1398
  r"Haley",
@@ -1346,7 +1405,9 @@ HIGHLIGHTED_NAMES = [
1346
1405
  r"(Rex\s*)?Till?erson",
1347
1406
  r"(?<!Cynthia )(Richard\s*)?Nixon",
1348
1407
  r"RNC",
1408
+ r"(Roy|Stephen)\s*Moore",
1349
1409
  r"Tea\s*Party",
1410
+ r"Wilbur\s*Ross",
1350
1411
  ],
1351
1412
  ),
1352
1413
  HighlightedNames(
@@ -1364,7 +1425,7 @@ HIGHLIGHTED_NAMES = [
1364
1425
  emailers={
1365
1426
  'Dasha Zhukova': 'art collector, daughter of Alexander Zhukov',
1366
1427
  MASHA_DROKOVA: 'silicon valley VC, former Putin Youth member',
1367
- RENATA_BOLOTOVA: 'former model, fund manager at New York State Insurance Fund',
1428
+ RENATA_BOLOTOVA: 'former model, fund manager at New York State Insurance Fund, Рената Болотова',
1368
1429
  SVETLANA_POZHIDAEVA: "Epstein's Russian assistant who was recommended for a visa by Sergei Belyakov (FSB) and David Blaine",
1369
1430
  },
1370
1431
  patterns=[
@@ -1383,20 +1444,21 @@ HIGHLIGHTED_NAMES = [
1383
1444
  r"(Anastasia\s*)?Kuznetsova",
1384
1445
  r"Lavrov",
1385
1446
  r"Lukoil",
1447
+ r'(Semion\s*)?Mogilevich',
1386
1448
  r"Moscow",
1387
1449
  r"(Natalia\s*)?Veselnitskaya",
1388
1450
  r"(Oleg\s*)?Deripaska",
1389
1451
  r"Oleksandr Vilkul",
1390
1452
  r"Onexim", # Prokhorov investment vehicle
1391
1453
  r"Prokhorov",
1454
+ r"Rakishev",
1392
1455
  r"Rosneft",
1393
1456
  r"RT",
1394
1457
  r"St.?\s*?Petersburg",
1395
1458
  r'Svet',
1396
- r"Russian?",
1459
+ r"Russ?ian?",
1397
1460
  r"Sberbank",
1398
1461
  r"Soviet(\s*Union)?",
1399
- r"Timur\s*Kulibayev",
1400
1462
  r"USSR",
1401
1463
  r"Vlad(imir)?(?! Yudash)",
1402
1464
  r"(Vladimir\s*)?Putin",
@@ -1435,6 +1497,7 @@ HIGHLIGHTED_NAMES = [
1435
1497
  REID_HOFFMAN: 'PayPal mafia member, founder of LinkedIn',
1436
1498
  STEVEN_SINOFSKY: 'ex-Microsoft, loves bitcoin',
1437
1499
  VINCENZO_IOZZO: 'CEO of the identity-security company SlashID',
1500
+ ZUBAIR_KHAN: 'Tranchulas cybersecurity, InsightsPod founder, Islamabad / Dubai',
1438
1501
  },
1439
1502
  patterns=[
1440
1503
  r"AG?I",
@@ -1443,6 +1506,7 @@ HIGHLIGHTED_NAMES = [
1443
1506
  r"Danny\s*Hillis",
1444
1507
  r"deep learning",
1445
1508
  r"Drew\s*Houston",
1509
+ r"Eliezer\s*Yudkowsky",
1446
1510
  r"Eric\s*Schmidt",
1447
1511
  r"Greylock(\s*Partners)?",
1448
1512
  r"(?<!(ustin|Moshe)\s)Hoffmand?",
@@ -1462,6 +1526,7 @@ HIGHLIGHTED_NAMES = [
1462
1526
  r"Softbank",
1463
1527
  r"SpaceX",
1464
1528
  r"Tim\s*Ferriss?",
1529
+ r"Vision\s*Fund",
1465
1530
  r"WikiLeak(ed|s)",
1466
1531
  ],
1467
1532
  ),
@@ -1476,6 +1541,7 @@ HIGHLIGHTED_NAMES = [
1476
1541
  r"@?realDonaldTrump",
1477
1542
  r"(Alan\s*)?Weiss?elberg",
1478
1543
  r"Alex\s*Jones",
1544
+ r"(Brad(ley)?\s*)Parscale",
1479
1545
  r"\bDJ?T\b",
1480
1546
  r"Donald J. Tramp",
1481
1547
  r"(Donald\s+(J\.\s+)?)?Trump(ism|\s*(Org(anization)?|Properties)(\s*LLC)?)?",
@@ -1518,22 +1584,24 @@ HIGHLIGHTED_NAMES = [
1518
1584
  r"(Kenneth E\. )?Mapp",
1519
1585
  r"PBI",
1520
1586
  r"Puerto\s*Ric(an|o)",
1587
+ r"San\s*Juan",
1521
1588
  r"S(ain)?t.?\s*Thomas",
1522
1589
  r"USVI",
1523
1590
  r"(?<!stein |vis-a-)VI(?!s-a-)",
1524
- r"(The\s*)?Virgin\s*Is(al|la)nds(\s*Daily\s*News)?", # Hard to make this work right
1591
+ r"(The\s*)?Virgin\s*Is(al|la)nd?s(\s*Daily\s*News)?", # Hard to make this work right
1525
1592
  r"(West\s*)?Palm\s*Beach(\s*County)?(?!\s*(Daily|Post))",
1526
1593
  ],
1527
1594
  ),
1528
1595
  HighlightedNames(
1529
1596
  label='victim',
1530
- style='orchid1',
1597
+ style=VICTIM_COLOR,
1531
1598
  patterns=[
1532
- r"#metoo",
1599
+ r"child\s*pornography",
1533
1600
  r"(David\s*)?Bo[il]es(,?\s*Schiller( & Flexner)?)?",
1534
1601
  r"(Gloria\s*)?Allred",
1535
1602
  r"(Jane|Tiffany)\s*Doe",
1536
1603
  r"Katie\s*Johnson",
1604
+ r"Minor\s*Victim",
1537
1605
  r"pedophile",
1538
1606
  r"Stephanie\s*Clifford",
1539
1607
  r"Stormy\s*Daniels",
@@ -1546,7 +1614,9 @@ HIGHLIGHTED_NAMES = [
1546
1614
  style='medium_orchid1',
1547
1615
  emailers={
1548
1616
  BRAD_EDWARDS: ROTHSTEIN_ROSENFELDT_ADLER,
1617
+ 'Douglas Wigdor': f'lawsuit against {LEON_BLACK}, Wigdor LLP',
1549
1618
  'Grant J. Smith': ROTHSTEIN_ROSENFELDT_ADLER,
1619
+ 'Jeanne M. Christensen': f'lawsuit against {LEON_BLACK}, Wigdor LLP',
1550
1620
  JACK_SCAROLA: 'Searcy Denney Scarola Barnhart & Shipley',
1551
1621
  KEN_JENNE: ROTHSTEIN_ROSENFELDT_ADLER,
1552
1622
  },
@@ -1557,6 +1627,7 @@ HIGHLIGHTED_NAMES = [
1557
1627
  r"Paul\s*(G.\s*)?Cassell",
1558
1628
  r"Rothstein\s*Rosenfeldt\s*Adler",
1559
1629
  r"(Scott\s*)?Rothstein",
1630
+ r"Wigdor(Law)?",
1560
1631
  ],
1561
1632
  ),
1562
1633
  HighlightedNames(
@@ -1583,7 +1654,11 @@ HIGHLIGHTED_NAMES = [
1583
1654
  HighlightedNames(
1584
1655
  emailers={GHISLAINE_MAXWELL: "Epstein's girlfriend, daughter of the spy Robert Maxwell"},
1585
1656
  category='Epstein',
1586
- patterns=[r"gmax(1@ellmax.com)?", r"(The )?TerraMar Project"],
1657
+ patterns=[
1658
+ r"gmax(1@ellmax.com)?",
1659
+ r"(The )?TerraMar Project",
1660
+ r"(Scott\s*)?Borgenson",
1661
+ ],
1587
1662
  style='deep_pink3',
1588
1663
  ),
1589
1664
  HighlightedNames(emailers={JABOR_Y: '"an influential man in Qatar"'}, category=MIDEAST, style='spring_green1'),
@@ -1595,6 +1670,11 @@ HIGHLIGHTED_NAMES = [
1595
1670
  HighlightedNames(emailers={SULTAN_BIN_SULAYEM: 'chairman of ports in Dubai, CEO of DP World'}, style='green1', category=MIDEAST),
1596
1671
 
1597
1672
  # HighlightedText not HighlightedNames bc of word boundary issue
1673
+ HighlightedText(
1674
+ label='metoo',
1675
+ style=VICTIM_COLOR,
1676
+ patterns=[r"#metoo"]
1677
+ ),
1598
1678
  HighlightedText(
1599
1679
  label='phone_number',
1600
1680
  style='bright_green',
@@ -1615,7 +1695,7 @@ HIGHLIGHTED_TEXTS = [
1615
1695
  HighlightedText(
1616
1696
  label='header_field',
1617
1697
  style='plum4',
1618
- patterns=[r'^>? ?(Date|From|Sent|To|C[cC]|Importance|Reply[- ]?To|Subject|Bee|B[cC]{2}|Attachments|Flag|Classification|((A|De(stinataire)?|Envoye|Expe(cl|d)iteur|Objet|Q) ?)):'],
1698
+ patterns=[r'^[>• ]{,4}(Date ?|From|Sent|To|C[cC]|Importance|Reply[- ]?To|Subject|Bee|B[cC]{2}|Attachments|Flag|Classification|[Il]nline-[Il]mages|((A|Debut du message transfer[&e]|De(stinataire)?|Envoye|Expe(cl|d)iteur|Objet|Q|Sujet) ?)):|^on behalf of'],
1619
1699
  ),
1620
1700
  HighlightedText(
1621
1701
  label='http_links',
@@ -1625,16 +1705,20 @@ HIGHLIGHTED_TEXTS = [
1625
1705
  HighlightedText(
1626
1706
  label='quoted_reply_line',
1627
1707
  style='dim',
1628
- patterns=[REPLY_REGEX.pattern],
1708
+ patterns=[
1709
+ REPLY_REGEX.pattern, r"^(> )?wrote:$",
1710
+ r"CONFIDENTIAL FOR ATTORNEY'S EYES ONLY(\nDO NOT COPY)?",
1711
+ r"PRIVILEGED ?- ?ATTORNEY WORK.*(\nCONFIDENTIAL - SUBJECT TO.*)?",
1712
+ ],
1629
1713
  ),
1630
1714
  HighlightedText(
1631
1715
  label='redacted',
1632
1716
  style='grey58',
1633
- patterns=[fr"{REDACTED}|Privileged - Redacted"],
1717
+ patterns=[fr"{REDACTED}|<?Privileged - Redacted>?"],
1634
1718
  ),
1635
1719
  HighlightedText(
1636
1720
  label='sent_from',
1637
- style='gray42 italic',
1721
+ style='light_cyan3 italic dim',
1638
1722
  patterns=[SENT_FROM_REGEX.pattern],
1639
1723
  ),
1640
1724
  HighlightedText(
@@ -1652,7 +1736,7 @@ HIGHLIGHTED_TEXTS = [
1652
1736
  ManualHighlight(
1653
1737
  label='email_attachments',
1654
1738
  style='gray30 italic',
1655
- pattern=r"^(> )?Attachments: (?P<email_attachments>.*)",
1739
+ pattern=r"^(> )?(Attachments|[Il]nline-[Il]mages): (?P<email_attachments>.*)",
1656
1740
  ),
1657
1741
  ManualHighlight(
1658
1742
  label='email_timestamp',