epstein-files 1.2.1__py3-none-any.whl → 1.2.5__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.
- epstein_files/__init__.py +31 -6
- epstein_files/documents/document.py +5 -1
- epstein_files/documents/email.py +192 -203
- epstein_files/epstein_files.py +2 -3
- epstein_files/person.py +29 -9
- epstein_files/util/constant/names.py +9 -7
- epstein_files/util/constant/output_files.py +8 -5
- epstein_files/util/constant/strings.py +2 -1
- epstein_files/util/constant/urls.py +13 -2
- epstein_files/util/constants.py +35 -11
- epstein_files/util/data.py +1 -0
- epstein_files/util/env.py +9 -6
- epstein_files/util/highlighted_group.py +263 -117
- epstein_files/util/output.py +18 -9
- epstein_files/util/rich.py +7 -2
- epstein_files/util/word_count.py +1 -1
- {epstein_files-1.2.1.dist-info → epstein_files-1.2.5.dist-info}/METADATA +1 -1
- epstein_files-1.2.5.dist-info/RECORD +34 -0
- epstein_files-1.2.1.dist-info/RECORD +0 -34
- {epstein_files-1.2.1.dist-info → epstein_files-1.2.5.dist-info}/LICENSE +0 -0
- {epstein_files-1.2.1.dist-info → epstein_files-1.2.5.dist-info}/WHEEL +0 -0
- {epstein_files-1.2.1.dist-info → epstein_files-1.2.5.dist-info}/entry_points.txt +0 -0
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
import json
|
|
2
2
|
import re
|
|
3
|
+
from collections import defaultdict
|
|
3
4
|
from dataclasses import dataclass, field
|
|
4
5
|
|
|
5
6
|
from rich.console import Console
|
|
@@ -11,8 +12,8 @@ from epstein_files.util.constant.strings import *
|
|
|
11
12
|
from epstein_files.util.constant.urls import ARCHIVE_LINK_COLOR
|
|
12
13
|
from epstein_files.util.constants import (EMAILER_ID_REGEXES, EPSTEIN_V_ROTHSTEIN_EDWARDS,
|
|
13
14
|
OSBORNE_LLP, REPLY_REGEX, SENT_FROM_REGEX)
|
|
15
|
+
from epstein_files.util.data import sort_dict, without_falsey
|
|
14
16
|
from epstein_files.util.doc_cfg import *
|
|
15
|
-
from epstein_files.util.data import without_falsey
|
|
16
17
|
from epstein_files.util.env import args
|
|
17
18
|
from epstein_files.util.logging import logger
|
|
18
19
|
|
|
@@ -153,7 +154,7 @@ class HighlightedNames(HighlightedText):
|
|
|
153
154
|
name_patterns = [remove_question_marks(name).replace(' ', r"\s+")]
|
|
154
155
|
|
|
155
156
|
if ' ' in name:
|
|
156
|
-
for partial_name in [
|
|
157
|
+
for partial_name in [reversed_name(name), extract_first_name(name), extract_last_name(name)]: # Order matters
|
|
157
158
|
if partial_name.lower() not in NAMES_TO_NOT_HIGHLIGHT and SIMPLE_NAME_REGEX.match(partial_name):
|
|
158
159
|
name_patterns.append(partial_name.replace(' ', r"\s+"))
|
|
159
160
|
|
|
@@ -239,11 +240,14 @@ HIGHLIGHTED_NAMES = [
|
|
|
239
240
|
'Valeria Chomsky': f"wife of {NOAM_CHOMSKY}",
|
|
240
241
|
},
|
|
241
242
|
patterns=[
|
|
243
|
+
r"Andy\s*Lippman", # Media Lab
|
|
244
|
+
r"Arizona\s*State\s*University",
|
|
242
245
|
r"Bard\s+((Early )?College|High School|Schools)",
|
|
243
246
|
r"Brotherton",
|
|
244
247
|
r"Carl\s*Sagan",
|
|
245
248
|
r"Columbia",
|
|
246
249
|
r"Dan(iel|ny) Kahneman",
|
|
250
|
+
r"(Francis\s*)?Crick",
|
|
247
251
|
r"J(ames|im)\s*Watson",
|
|
248
252
|
r"(Lord\s*)?Martin\s*Rees",
|
|
249
253
|
r"Massachusetts\s*Institute\s*of\s*Technology",
|
|
@@ -256,10 +260,13 @@ HIGHLIGHTED_NAMES = [
|
|
|
256
260
|
r"Princeton(\s*University)?",
|
|
257
261
|
r"Regeneron",
|
|
258
262
|
r"(Richard\s*)?Dawkins",
|
|
263
|
+
r"(Sandy\s*)?Pentland", # Media Lab
|
|
259
264
|
r"Sanofi",
|
|
260
265
|
r"Stanford(\s*University)?(\s*Hospital)?",
|
|
261
266
|
r"(Stephen\s*)?Hawking",
|
|
262
267
|
r"(Steven?\s*)?Pinker",
|
|
268
|
+
r"Texas\s*A&M",
|
|
269
|
+
r"Tulane",
|
|
263
270
|
r"UCLA",
|
|
264
271
|
],
|
|
265
272
|
),
|
|
@@ -276,19 +283,23 @@ HIGHLIGHTED_NAMES = [
|
|
|
276
283
|
patterns=[
|
|
277
284
|
r"Buhari",
|
|
278
285
|
r"Econet(\s*Wireless)",
|
|
286
|
+
r"Ethiopian?",
|
|
279
287
|
r"Ghana(ian)?",
|
|
280
288
|
r"Glencore",
|
|
281
289
|
r"Goodluck Jonathan",
|
|
282
290
|
r"Johannesburg",
|
|
283
|
-
r"
|
|
291
|
+
r"Kenyan?",
|
|
284
292
|
r"Nigerian?",
|
|
285
293
|
r"Okey Enelamah",
|
|
294
|
+
r"(Paul\s*)?Kagame",
|
|
295
|
+
r"Rwandan?",
|
|
286
296
|
r"Senegal(ese)?",
|
|
287
297
|
r"Serengeti",
|
|
288
298
|
r"(South\s*)?African?",
|
|
289
299
|
r"(Strive\s*)?Masiyiwa",
|
|
290
|
-
r"
|
|
300
|
+
r"Tanzanian?",
|
|
291
301
|
r"Ugandan?",
|
|
302
|
+
r"(Yoweri\s*)?Museveni",
|
|
292
303
|
r"Zimbabwe(an)?",
|
|
293
304
|
],
|
|
294
305
|
),
|
|
@@ -300,6 +311,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
300
311
|
'Barry Josephson': 'American film producer, editor FamilySecurityMatters.org',
|
|
301
312
|
BILL_SIEGEL: 'documentary film producer and director',
|
|
302
313
|
DAVID_BLAINE: 'famous magician',
|
|
314
|
+
'David Brenner': 'American comedian and actor',
|
|
303
315
|
'Richard Merkin': 'painter, illustrator and arts educator',
|
|
304
316
|
STEVEN_PFEIFFER: 'Associate Director at Independent Filmmaker Project (IFP)',
|
|
305
317
|
'Steven Gaydos': 'American screenwriter and journalist',
|
|
@@ -312,6 +324,9 @@ HIGHLIGHTED_NAMES = [
|
|
|
312
324
|
r"Errol(\s*Morris)?",
|
|
313
325
|
r"Etienne Binant",
|
|
314
326
|
r"(Frank\s)?Gehry",
|
|
327
|
+
r"Harvey\s*Weinstein", r"wientstein", r"Weinstein\s*Co(s?|mpany)",
|
|
328
|
+
r"IFP",
|
|
329
|
+
r"Independent\s*Filmmaker\s*Project",
|
|
315
330
|
r"Jagger",
|
|
316
331
|
r"(Jeffrey\s*)?Katzenberg",
|
|
317
332
|
r"(Johnny\s*)?Depp",
|
|
@@ -320,6 +335,8 @@ HIGHLIGHTED_NAMES = [
|
|
|
320
335
|
r"Lena\s*Dunham",
|
|
321
336
|
r"Madonna",
|
|
322
337
|
r"Mark\s*Burnett",
|
|
338
|
+
r"New York Film Festival",
|
|
339
|
+
r"Peter Getzels",
|
|
323
340
|
r"Phaidon",
|
|
324
341
|
r"Ramsey Elkholy",
|
|
325
342
|
r"Regan arts",
|
|
@@ -340,6 +357,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
340
357
|
r"BG",
|
|
341
358
|
r"b?g?C3",
|
|
342
359
|
r"(Bill\s*((and|or|&)\s*Melinda\s*)?)?Gates(\s*Foundation)?",
|
|
360
|
+
r"Kofi\s*Rashid",
|
|
343
361
|
r"Melinda(\s*Gates)?",
|
|
344
362
|
r"Microsoft",
|
|
345
363
|
r"MSFT",
|
|
@@ -358,23 +376,27 @@ HIGHLIGHTED_NAMES = [
|
|
|
358
376
|
r"Balaji",
|
|
359
377
|
r"bitcoin(\s*Foundation)?",
|
|
360
378
|
r"block ?chain(\s*capital)?",
|
|
379
|
+
r"Brian Forde",
|
|
361
380
|
r"Brock(\s*Pierce)?",
|
|
362
381
|
r"coins?",
|
|
382
|
+
r"Cory\s*Fields", # bitcoin dev
|
|
363
383
|
r"cr[iy]?pto(currenc(y|ies))?",
|
|
364
384
|
r"Digital\s*Currenc(ies|y)(\s*Initiative)?",
|
|
365
385
|
r"e-currency",
|
|
366
|
-
r"(Gavin )?Andress?en",
|
|
386
|
+
r"(Gavin )?Andress?en", # bitcoin dev
|
|
367
387
|
r"(Howard\s+)?Lutnic?k",
|
|
388
|
+
r"(Jim\s*)Pallotta", # Media lab advisory board
|
|
368
389
|
r"Libra",
|
|
369
390
|
r"Madars",
|
|
370
391
|
r"Mi(chael|ke)\s*Novogratz",
|
|
371
392
|
r"(Patrick\s*)?Murck",
|
|
393
|
+
r"Ron Rivest",
|
|
372
394
|
r"(Ross\s*)?Ulbricht",
|
|
373
395
|
r"Silk\s*Road",
|
|
374
396
|
r"SpanCash",
|
|
375
397
|
r"Tether",
|
|
376
398
|
r"virtual\s*currenc(ies|y)",
|
|
377
|
-
r"Wladimir( van der Laan)?",
|
|
399
|
+
r"Wladimir( van der Laan)?", # bitcoin dev
|
|
378
400
|
r"(zero\s+knowledge\s+|zk)pro(of|tocols?)",
|
|
379
401
|
],
|
|
380
402
|
),
|
|
@@ -385,11 +407,9 @@ HIGHLIGHTED_NAMES = [
|
|
|
385
407
|
ALIREZA_ITTIHADIEH: 'CEO Freestream Aircraft Limited',
|
|
386
408
|
BARBRO_C_EHNBOM: 'Swedish pharmaceuticals, SALSS',
|
|
387
409
|
BARRY_J_COHEN: None,
|
|
388
|
-
'David Mitchell': 'Mitchell Holdings New York real estate developer',
|
|
389
|
-
FRED_HADDAD: "co-founder of Heck's in West Virginia",
|
|
410
|
+
'David Mitchell': 'Mitchell Holdings, New York real estate developer',
|
|
390
411
|
GERALD_BARTON: "Maryland property developer Landmark Land Company",
|
|
391
412
|
GORDON_GETTY: 'heir to oil tycoon J. Paul Getty',
|
|
392
|
-
NICHOLAS_RIBIS: 'Hilton CEO, former president of Trump Organization',
|
|
393
413
|
'Philip Kafka': 'president of Prince Concepts (and son of Terry Kafka?)',
|
|
394
414
|
ROBERT_LAWRENCE_KUHN: 'investment banker, China expert',
|
|
395
415
|
TERRY_KAFKA: 'CEO of Impact Outdoor (highway billboards)',
|
|
@@ -398,7 +418,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
398
418
|
patterns=[
|
|
399
419
|
r"((Bill|David)\s*)?Koch(\s*(Bro(s|thers)|Industries))?",
|
|
400
420
|
r"Gruterite",
|
|
401
|
-
r"(John\s*)?Kluge",
|
|
421
|
+
r"((John|Patricia)\s*)?Kluge",
|
|
402
422
|
r"Marc Rich",
|
|
403
423
|
r"(Mi(chael|ke)\s*)?Ovitz",
|
|
404
424
|
r"(Steve\s+)?Wynn",
|
|
@@ -407,6 +427,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
407
427
|
r"Park Partners",
|
|
408
428
|
r"SALSS",
|
|
409
429
|
r"Swedish[-\s]*American\s*Life\s*Science\s*Summit",
|
|
430
|
+
r"Trilateral Commission",
|
|
410
431
|
r"Valhi",
|
|
411
432
|
r"(Yves\s*)?Bouvier",
|
|
412
433
|
],
|
|
@@ -418,6 +439,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
418
439
|
r"CBD",
|
|
419
440
|
r"cannabis",
|
|
420
441
|
r"marijuana",
|
|
442
|
+
r"psychedelic",
|
|
421
443
|
r"THC",
|
|
422
444
|
r"WEED(guide|maps)?[^s]?",
|
|
423
445
|
],
|
|
@@ -474,6 +496,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
474
496
|
r"(Barac?k )?Obama",
|
|
475
497
|
r"((Bill|Hillart?y)\s*)?Clinton",
|
|
476
498
|
r"((Chuck|Charles)\s*)?S(ch|hc)umer",
|
|
499
|
+
r"Debbie\s*Wasserman\s*Schultz",
|
|
477
500
|
r"Dem(ocrat(ic)?)?",
|
|
478
501
|
r"(Diana\s*)?DeGette",
|
|
479
502
|
r"DNC",
|
|
@@ -491,6 +514,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
491
514
|
r"(Matteo\s*)?Salvini",
|
|
492
515
|
r"Maxine\s*Waters",
|
|
493
516
|
r"(Nancy )?Pelosi",
|
|
517
|
+
r"Open Society( Global Board)?",
|
|
494
518
|
r"Ron\s*Dellums",
|
|
495
519
|
r"Schumer",
|
|
496
520
|
r"(Tim(othy)?\s*)?Geithner",
|
|
@@ -506,16 +530,19 @@ HIGHLIGHTED_NAMES = [
|
|
|
506
530
|
EVA: "possibly Epstein's ex-girlfriend (?)",
|
|
507
531
|
'Eva Dubin': f"Epstein's ex-girlfriend now married to {GLENN_DUBIN}",
|
|
508
532
|
},
|
|
509
|
-
patterns=[r"((Celina|Eva( Anderss?on)?|Glenn) )?Dubin"],
|
|
533
|
+
patterns=[r"((Celina|Eva( Anderss?on)?|Glenn?) )?Dubin"],
|
|
510
534
|
),
|
|
511
535
|
HighlightedNames(
|
|
512
536
|
label='employee',
|
|
513
537
|
style='medium_purple4',
|
|
514
538
|
emailers={
|
|
515
539
|
'Alfredo Rodriguez': "Epstein's butler, stole the journal",
|
|
516
|
-
|
|
540
|
+
'Bernard Kruger': "Epstein's doctor",
|
|
541
|
+
EDUARDO_ROBLES: f'home builder at Creative Kingdom Dubai',
|
|
542
|
+
ERIC_ROTH: 'jet decorator at International Jet',
|
|
517
543
|
GWENDOLYN_BECK: 'Epstein fund manager in the 90s',
|
|
518
544
|
JANUSZ_BANASIAK: "Epstein's house manager",
|
|
545
|
+
"John Allessi": "Epstein's houseman",
|
|
519
546
|
JEAN_HUGUEN: 'interior design at Alberto Pinto Cabinet',
|
|
520
547
|
LAWRANCE_VISOSKI: "Epstein's pilot",
|
|
521
548
|
LESLEY_GROFF: f"Epstein's assistant",
|
|
@@ -523,7 +550,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
523
550
|
MERWIN_DELA_CRUZ: None, # HOUSE_OVERSIGHT_032652 Groff says "Jojo and Merwin both requested off Nov. 25 and 26"
|
|
524
551
|
NADIA_MARCINKO: "Epstein's pilot",
|
|
525
552
|
'Sean J. Lancaster': 'airplane reseller',
|
|
526
|
-
ZUBAIR_KHAN: '
|
|
553
|
+
ZUBAIR_KHAN: 'Tranchulas cybersecurity, InsightsPod founder, Islamabad / Dubai',
|
|
527
554
|
},
|
|
528
555
|
patterns=[
|
|
529
556
|
r"Adriana\s*Ross",
|
|
@@ -539,7 +566,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
539
566
|
MARK_EPSTEIN: 'brother of Jeffrey',
|
|
540
567
|
},
|
|
541
568
|
patterns=[
|
|
542
|
-
r"JEGE",
|
|
569
|
+
r"JEGE(\s*Inc)",
|
|
543
570
|
r"LSJ",
|
|
544
571
|
],
|
|
545
572
|
),
|
|
@@ -556,7 +583,9 @@ HIGHLIGHTED_NAMES = [
|
|
|
556
583
|
DAVID_SCHOEN: f"{CRIMINAL_DEFENSE_ATTORNEY} after 2019 arrest",
|
|
557
584
|
DEBBIE_FEIN: EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY,
|
|
558
585
|
'Erika Kellerhals': 'attorney in St. Thomas',
|
|
586
|
+
FRED_HADDAD: "co-founder of Heck's in West Virginia",
|
|
559
587
|
GERALD_LEFCOURT: f'friend of {ALAN_DERSHOWITZ}',
|
|
588
|
+
'Howard Rubenstein': f"Epstein's former spokesman",
|
|
560
589
|
JACK_GOLDBERGER: CRIMINAL_DEFENSE_2008,
|
|
561
590
|
JACKIE_PERCZEK: CRIMINAL_DEFENSE_2008,
|
|
562
591
|
JAY_LEFKOWITZ: f"Kirkland & Ellis partner, {CRIMINAL_DEFENSE_2008}",
|
|
@@ -599,7 +628,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
599
628
|
emailers={
|
|
600
629
|
ANDRZEJ_DUDA: 'former president of Poland',
|
|
601
630
|
'Caroline Lang': 'daughter of Jack Lang',
|
|
602
|
-
|
|
631
|
+
EDWARD_ROD_LARSEN: f"son of {TERJE_ROD_LARSEN}",
|
|
603
632
|
'Fabrice Aidan': f'diplomat who worked with {TERJE_ROD_LARSEN}',
|
|
604
633
|
'Jack Lang': 'former French Minister of National Education',
|
|
605
634
|
MIROSLAV_LAJCAK: 'Russia-friendly Slovakian politician, friend of Steve Bannon',
|
|
@@ -608,6 +637,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
608
637
|
THORBJORN_JAGLAND: 'former prime minister of Norway, Nobel Peace Prize Committee',
|
|
609
638
|
},
|
|
610
639
|
patterns=[
|
|
640
|
+
r"AfD",
|
|
611
641
|
r"(Angela )?Merk(el|le)",
|
|
612
642
|
r"Austria",
|
|
613
643
|
r"(Benjamin\s*)?Harnwell",
|
|
@@ -647,6 +677,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
647
677
|
r"Paris",
|
|
648
678
|
r"Polish",
|
|
649
679
|
r"pope",
|
|
680
|
+
r"Scotland",
|
|
650
681
|
r"(Sebastian )?Kurz",
|
|
651
682
|
r"Stockholm",
|
|
652
683
|
r"Strasbourg",
|
|
@@ -656,6 +687,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
656
687
|
r"(Tony\s)?Blair",
|
|
657
688
|
r"U\.K\.",
|
|
658
689
|
r"Ukrain(e|ian)",
|
|
690
|
+
r"Venice",
|
|
659
691
|
r"(Vi(c|k)tor\s+)?Orbah?n",
|
|
660
692
|
r"Vienna",
|
|
661
693
|
r"Zug",
|
|
@@ -684,8 +716,11 @@ HIGHLIGHTED_NAMES = [
|
|
|
684
716
|
'Skip Rimer': 'Milken Institute (Michael Milken)',
|
|
685
717
|
'Steven Elkman': DEUTSCHE_BANK,
|
|
686
718
|
'Vahe Stepanian': 'Cetera Financial Group',
|
|
719
|
+
VINIT_SAHNI: f"analyst at {DEUTSCHE_BANK} and {GOLDMAN_SACHS}",
|
|
687
720
|
},
|
|
688
721
|
patterns=[
|
|
722
|
+
r"Ace\s*Greenberg",
|
|
723
|
+
r"AIG",
|
|
689
724
|
r"((anti.?)?money\s+)?launder(s?|ers?|ing)?(\s+money)?",
|
|
690
725
|
r"Apollo",
|
|
691
726
|
r"Ari\s*Glass",
|
|
@@ -705,8 +740,11 @@ HIGHLIGHTED_NAMES = [
|
|
|
705
740
|
r"FRBNY",
|
|
706
741
|
r"Goldman(\s*Sachs)",
|
|
707
742
|
r"GRAT",
|
|
743
|
+
r"Gratitude (America|& Enhanced)", # Leon Black and/or Epstein charity?
|
|
744
|
+
r"Hank\s*Greenburg",
|
|
708
745
|
r"HSBC",
|
|
709
746
|
r"Invesco",
|
|
747
|
+
r"Jamie\s*D(imon)?",
|
|
710
748
|
r"(Janet\s*)?Yellen",
|
|
711
749
|
r"(Jerome\s*)?Powell(?! M\. Cabot)",
|
|
712
750
|
r"(Jimmy\s*)?Cayne",
|
|
@@ -715,17 +753,18 @@ HIGHLIGHTED_NAMES = [
|
|
|
715
753
|
r"Madoff",
|
|
716
754
|
r"Merrill(\s*Lynch)?",
|
|
717
755
|
r"(Michael\s*)?Cembalest",
|
|
718
|
-
r"(Mi(chael|ke)\s*)?Milken(\s*Conference)?",
|
|
756
|
+
r"(Mi(chael|ke)\s*)?Milken(\s*Conference|Institute)?",
|
|
719
757
|
r"Mizrahi\s*Bank",
|
|
720
758
|
r"MLPF&S",
|
|
721
759
|
r"Morgan Stanley",
|
|
722
760
|
r"(Peter L. )?Scher",
|
|
723
761
|
r"(Ray\s*)?Dalio",
|
|
724
762
|
r"(Richard\s*)?LeFrak",
|
|
725
|
-
r"
|
|
763
|
+
r"(Ste(phen|ve)\s*)?Schwart?z?man",
|
|
726
764
|
r"Serageldin",
|
|
727
765
|
r"UBS",
|
|
728
766
|
r"us.gio@jpmorgan.com",
|
|
767
|
+
r"Wall\s*Street(?!\s*Jour)",
|
|
729
768
|
],
|
|
730
769
|
),
|
|
731
770
|
HighlightedNames(
|
|
@@ -746,6 +785,82 @@ HIGHLIGHTED_NAMES = [
|
|
|
746
785
|
r"Thomas\s*(J\.?\s*)?Barrack(\s*Jr)?",
|
|
747
786
|
],
|
|
748
787
|
),
|
|
788
|
+
HighlightedNames(
|
|
789
|
+
label='government',
|
|
790
|
+
style='color(24) bold',
|
|
791
|
+
emailers={
|
|
792
|
+
ANN_MARIE_VILLAFANA: 'Southern District of Florida (SDFL) U.S. Attorney',
|
|
793
|
+
DANNY_FROST: 'Director of Communications at Manhattan D.A.',
|
|
794
|
+
'Police Code Enforcement': f"{PALM_BEACH} buildings code enforcement",
|
|
795
|
+
},
|
|
796
|
+
patterns=[
|
|
797
|
+
r"AG",
|
|
798
|
+
r"(Alicia\s*)?Valle",
|
|
799
|
+
r'Alice\s*Fisher|Fisher, Alice',
|
|
800
|
+
r"AML",
|
|
801
|
+
r"(Andrew\s*)?(McCabe|Natsios)",
|
|
802
|
+
r"Attorney General",
|
|
803
|
+
r"((Bob|Robert)\s*)?Mueller",
|
|
804
|
+
r"(Byung\s)?Pak",
|
|
805
|
+
r"Case 1:19-cv-03377(-LAP)?",
|
|
806
|
+
r"(CENT|NORTH|SOUTH)COM",
|
|
807
|
+
r"CFTC?",
|
|
808
|
+
r"CIA",
|
|
809
|
+
r"CIS",
|
|
810
|
+
r"CVRA",
|
|
811
|
+
r"DARPA",
|
|
812
|
+
r"Dep(artmen)?t\.?\s*of\s*(the\s*)?(Justice|Treasury)",
|
|
813
|
+
r"DHS",
|
|
814
|
+
r"DOJ",
|
|
815
|
+
r"FBI",
|
|
816
|
+
r"FCPA",
|
|
817
|
+
r"FDIC",
|
|
818
|
+
r"FDLE",
|
|
819
|
+
r"Federal\s*Bureau\s*of\s*Investigation",
|
|
820
|
+
r"FinCEN",
|
|
821
|
+
r"FINRA",
|
|
822
|
+
r"FOIA",
|
|
823
|
+
r"FTC",
|
|
824
|
+
r"(General\s*)?P(a|e)traeus",
|
|
825
|
+
r"Geoff\s*Ling",
|
|
826
|
+
r"IRS",
|
|
827
|
+
r"(James\s*)?Comey",
|
|
828
|
+
r"(Jennifer\s*Shasky\s*)?Calvery",
|
|
829
|
+
r"((Judge|Mark)\s*)?(Carney|Filip)",
|
|
830
|
+
r"(Judge\s*)?(Kenneth\s*)?(A\.?\s*)?Marra",
|
|
831
|
+
r"(Justice|Treasury)\s*Dep(t|artment)",
|
|
832
|
+
r"(Kirk )?Blouin",
|
|
833
|
+
r"KYC",
|
|
834
|
+
r"(Lann?a\s*)?Belohlavek",
|
|
835
|
+
r"NIH",
|
|
836
|
+
r"NPA",
|
|
837
|
+
r"NS(A|C)",
|
|
838
|
+
r"OCC",
|
|
839
|
+
r"OFAC",
|
|
840
|
+
r"(Michael\s*)?Reiter",
|
|
841
|
+
r"OGE",
|
|
842
|
+
r"Office\s*of\s*Government\s*Ethics",
|
|
843
|
+
r"police",
|
|
844
|
+
r"(Preet\s*)?Bharara",
|
|
845
|
+
r"SCOTUS",
|
|
846
|
+
r"SD(FL|NY)",
|
|
847
|
+
r"SEC",
|
|
848
|
+
r"Secret\s*Service",
|
|
849
|
+
r"Securities\s*and\s*Exchange\s*Commission",
|
|
850
|
+
r"Southern\s*District(\s*of\s*(Florida|New\s*York))?",
|
|
851
|
+
r"State\s*Dep(artmen)?t",
|
|
852
|
+
r"Strzok",
|
|
853
|
+
r"Supreme\s*Court",
|
|
854
|
+
r"Treasury\s*(Dep(artmen)?t|Secretary)",
|
|
855
|
+
r"TSA",
|
|
856
|
+
r"U\.?S\.? attorney",
|
|
857
|
+
r"USAID",
|
|
858
|
+
r"US\s*(AF|Army|Air\s*Force)",
|
|
859
|
+
r"Walter\s*Reed(\s*Army\s*Institute\s*of\s*Research)?",
|
|
860
|
+
r"(William\s*J\.?\s*)?Zloch",
|
|
861
|
+
r"WRAIR",
|
|
862
|
+
],
|
|
863
|
+
),
|
|
749
864
|
HighlightedNames(
|
|
750
865
|
label=HARVARD,
|
|
751
866
|
style='light_goldenrod3',
|
|
@@ -773,8 +888,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
773
888
|
label='India',
|
|
774
889
|
style='bright_green',
|
|
775
890
|
emailers={
|
|
776
|
-
ANIL_AMBANI: 'chairman of Reliance Group',
|
|
777
|
-
VINIT_SAHNI: None,
|
|
891
|
+
ANIL_AMBANI: 'billionaire chairman of Reliance Group',
|
|
778
892
|
},
|
|
779
893
|
patterns=[
|
|
780
894
|
r"Abraaj",
|
|
@@ -797,16 +911,18 @@ HIGHLIGHTED_NAMES = [
|
|
|
797
911
|
emailers={
|
|
798
912
|
EHUD_BARAK: 'former prime minister of Israel, Epstein business partner',
|
|
799
913
|
'Mitchell Bard': 'director of the American-Israeli Cooperative Enterprise (AICE)',
|
|
800
|
-
|
|
914
|
+
NILI_PRIELL_BARAK: 'wife of Ehud Barak',
|
|
801
915
|
},
|
|
802
916
|
patterns=[
|
|
803
917
|
r"AIPAC",
|
|
804
918
|
r"Bibi",
|
|
805
919
|
r"(eh|(Ehud|Nili Priell)\s*)?barak",
|
|
920
|
+
r"EB",
|
|
806
921
|
r"Ehud\s*Barack",
|
|
807
922
|
r"Israeli?",
|
|
808
923
|
r"Jerusalem",
|
|
809
924
|
r"J\s*Street",
|
|
925
|
+
r"Menachem\s*Begin",
|
|
810
926
|
r"Mossad",
|
|
811
927
|
r"Netanyahu",
|
|
812
928
|
r"(Sheldon\s*)?Adelson",
|
|
@@ -849,14 +965,18 @@ HIGHLIGHTED_NAMES = [
|
|
|
849
965
|
},
|
|
850
966
|
patterns=[
|
|
851
967
|
r"ABC(\s*News)?",
|
|
968
|
+
r"Alexandra Wolfe|Wolfe, Alexandra",
|
|
969
|
+
r"AlterNet",
|
|
852
970
|
r"Arianna(\s*Huffington)?",
|
|
853
971
|
r"(Arthur\s*)?Kretchmer",
|
|
854
972
|
r'Associated\s*Press',
|
|
855
973
|
r"Axios",
|
|
856
974
|
r"BBC",
|
|
975
|
+
r"(Bob|Robert)\s*Costa",
|
|
857
976
|
r"Breitbart",
|
|
858
|
-
r"BuzzFeed",
|
|
859
|
-
r"
|
|
977
|
+
r"BuzzFeed(\s*News)?",
|
|
978
|
+
r"C-?Span",
|
|
979
|
+
r"CBS(\s*(4|Corp|News))?",
|
|
860
980
|
r"Charlie\s*Rose",
|
|
861
981
|
r"China\s*Daily",
|
|
862
982
|
r"CNBC",
|
|
@@ -869,6 +989,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
869
989
|
r"Ed\s*Krassenstein",
|
|
870
990
|
r"(Emily\s*)?Michot",
|
|
871
991
|
r"Ezra\s*Klein",
|
|
992
|
+
r"Fire\s*and\s*Fury",
|
|
872
993
|
r"Forbes",
|
|
873
994
|
r"Fortune\s*Magazine",
|
|
874
995
|
r"Fox\s*News(\.com)?",
|
|
@@ -881,8 +1002,10 @@ HIGHLIGHTED_NAMES = [
|
|
|
881
1002
|
r"Graydon(\s*Carter)?",
|
|
882
1003
|
r"Huff(ington)?(\s*Po(st)?)?",
|
|
883
1004
|
r"Ingram, David",
|
|
884
|
-
r"
|
|
1005
|
+
r"James\s*Hill",
|
|
1006
|
+
r"(James\s*)?Patterson",
|
|
885
1007
|
r"Jesse Kornbluth",
|
|
1008
|
+
r"John\s*Connolly",
|
|
886
1009
|
r"Jonathan\s*Karl",
|
|
887
1010
|
r"Julie\s*(K.?\s*)?Brown", r'jbrown@miamiherald.com',
|
|
888
1011
|
r"(Katie\s*)?Couric",
|
|
@@ -892,7 +1015,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
892
1015
|
r"(Les\s*)?Moonves",
|
|
893
1016
|
r"MarketWatch",
|
|
894
1017
|
r"Miami\s*Herald",
|
|
895
|
-
r"(Mi(chael|ke)\s*)?
|
|
1018
|
+
r"(Mi(chael|ke)\s*)?Bloomber[gq](\s*News)?",
|
|
896
1019
|
r"(Michele\s*)?Dargan",
|
|
897
1020
|
r"Morning News USA",
|
|
898
1021
|
r"(National\s*)?Enquirer",
|
|
@@ -904,13 +1027,14 @@ HIGHLIGHTED_NAMES = [
|
|
|
904
1027
|
r"Pro\s*Publica",
|
|
905
1028
|
r"Reuters",
|
|
906
1029
|
r"(Sean\s*)?Hannity",
|
|
1030
|
+
r"Sharon Churcher", # Daily Mail
|
|
907
1031
|
r"Sulzberger",
|
|
908
1032
|
r"SunSentinel",
|
|
909
1033
|
r"(The\s*)?Financial\s*Times",
|
|
910
1034
|
r"The\s*Guardian",
|
|
911
1035
|
r"TheHill",
|
|
912
1036
|
r"(The\s*)?Mail\s*On\s*Sunday",
|
|
913
|
-
r"(The\s*)?N(ew\s*)?Y(ork
|
|
1037
|
+
r"(The\s*)?N(ew\s*)?Y(ork)?\s*(Magazine|Observer|P(ost)?|T(imes)?)",
|
|
914
1038
|
r"(The\s*)?New\s*Yorker",
|
|
915
1039
|
r"(The\s*)?Wall\s*Street\s*Journal",
|
|
916
1040
|
r"(The\s*)?Wa(shington\s*)?Po(st)?",
|
|
@@ -952,7 +1076,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
952
1076
|
r"((Enrique )?Pena )?Nieto",
|
|
953
1077
|
r"Lat(in)?\s*Am(erican?)?",
|
|
954
1078
|
r"Lula",
|
|
955
|
-
r"Mexic(an|o)",
|
|
1079
|
+
r"(?<!New )Mexic(an|o)",
|
|
956
1080
|
r"(Nicolas\s+)?Maduro",
|
|
957
1081
|
r"Panama( Papers)?",
|
|
958
1082
|
r"Peru(vian)?",
|
|
@@ -960,90 +1084,61 @@ HIGHLIGHTED_NAMES = [
|
|
|
960
1084
|
r"Zambrano",
|
|
961
1085
|
],
|
|
962
1086
|
),
|
|
963
|
-
HighlightedNames(
|
|
964
|
-
label='government',
|
|
965
|
-
style='color(24) bold',
|
|
966
|
-
emailers={
|
|
967
|
-
ANN_MARIE_VILLAFANA: 'Southern District of Florida (SDFL) U.S. Attorney',
|
|
968
|
-
DANNY_FROST: 'Director of Communications at Manhattan D.A.',
|
|
969
|
-
'Police Code Enforcement': f"{PALM_BEACH} buildings code enforcement",
|
|
970
|
-
},
|
|
971
|
-
patterns=[
|
|
972
|
-
r"AG",
|
|
973
|
-
r"(Alicia\s*)?Valle",
|
|
974
|
-
r"AML",
|
|
975
|
-
r"(Andrew\s*)?McCabe",
|
|
976
|
-
r"Attorney General",
|
|
977
|
-
r"((Bob|Robert)\s*)?Mueller",
|
|
978
|
-
r"(Byung\s)?Pak",
|
|
979
|
-
r"Case 1:19-cv-03377(-LAP)?",
|
|
980
|
-
r"CFTC?",
|
|
981
|
-
r"CIA",
|
|
982
|
-
r"CIS",
|
|
983
|
-
r"CVRA",
|
|
984
|
-
r"Dep(artmen)?t\.?\s*of\s*(the\s*)?(Justice|Treasury)",
|
|
985
|
-
r"DHS",
|
|
986
|
-
r"DOJ",
|
|
987
|
-
r"FBI",
|
|
988
|
-
r"FCPA",
|
|
989
|
-
r"FDIC",
|
|
990
|
-
r"Federal\s*Bureau\s*of\s*Investigation",
|
|
991
|
-
r"FinCEN",
|
|
992
|
-
r"FINRA",
|
|
993
|
-
r"FOIA",
|
|
994
|
-
r"FTC",
|
|
995
|
-
r"(General\s*)?P(a|e)traeus",
|
|
996
|
-
r"IRS",
|
|
997
|
-
r"(James\s*)?Comey",
|
|
998
|
-
r"(Jennifer\s*Shasky\s*)?Calvery",
|
|
999
|
-
r"((Judge|Mark)\s*)?(Carney|Filip)",
|
|
1000
|
-
r"(Judge\s*)?(Kenneth\s*)?(A\.?\s*)?Marra",
|
|
1001
|
-
r"(Justice|Treasury)\s*Dep(t|artment)",
|
|
1002
|
-
r"(Kirk )?Blouin",
|
|
1003
|
-
r"KYC",
|
|
1004
|
-
r"(Lann?a\s*)?Belohlavek",
|
|
1005
|
-
r"NIH",
|
|
1006
|
-
r"NS(A|C)",
|
|
1007
|
-
r"OCC",
|
|
1008
|
-
r"OFAC",
|
|
1009
|
-
r"(Michael\s*)?Reiter",
|
|
1010
|
-
r"OGE",
|
|
1011
|
-
r"Office\s*of\s*Government\s*Ethics",
|
|
1012
|
-
r"police",
|
|
1013
|
-
r"(Preet\s*)?Bharara",
|
|
1014
|
-
r"SCOTUS",
|
|
1015
|
-
r"SD(FL|NY)",
|
|
1016
|
-
r"SEC",
|
|
1017
|
-
r"Secret\s*Service",
|
|
1018
|
-
r"Securities\s*and\s*Exchange\s*Commission",
|
|
1019
|
-
r"Southern\s*District(\s*of\s*(Florida|New\s*York))?",
|
|
1020
|
-
r"State\s*Dep(artmen)?t",
|
|
1021
|
-
r"Strzok",
|
|
1022
|
-
r"Supreme\s*Court",
|
|
1023
|
-
r"Treasury\s*(Dep(artmen)?t|Secretary)",
|
|
1024
|
-
r"TSA",
|
|
1025
|
-
r"U\.?S\.? attorney",
|
|
1026
|
-
r"USAID",
|
|
1027
|
-
r"(William\s*J\.?\s*)?Zloch",
|
|
1028
|
-
],
|
|
1029
|
-
),
|
|
1030
1087
|
HighlightedNames(
|
|
1031
1088
|
label=LOBBYIST,
|
|
1032
1089
|
style='light_coral',
|
|
1033
1090
|
emailers={
|
|
1034
1091
|
BOB_CROWE: 'partner at Nelson Mullins',
|
|
1035
1092
|
'Joshua Cooper Ramo': 'co-CEO of Henry Kissinger Associates',
|
|
1036
|
-
KATHERINE_KEATING: '
|
|
1093
|
+
KATHERINE_KEATING: 'daughter of former Australian prime minister',
|
|
1037
1094
|
MOHAMED_WAHEED_HASSAN: 'former president of the Maldives',
|
|
1038
1095
|
OLIVIER_COLOM: 'France',
|
|
1039
|
-
'Paul Keating': 'former
|
|
1096
|
+
'Paul Keating': 'former prime minister of Australia',
|
|
1040
1097
|
PUREVSUREN_LUNDEG: 'Mongolian ambassador to the UN',
|
|
1041
1098
|
'Stanley Rosenberg': 'former President of the Massachusetts Senate',
|
|
1042
1099
|
},
|
|
1043
1100
|
patterns=[
|
|
1044
1101
|
r"CSIS",
|
|
1102
|
+
r"elisabeth\s*feliho",
|
|
1045
1103
|
r"(Kevin\s*)?Rudd",
|
|
1046
1104
|
r"Stanley Rosenberg",
|
|
1105
|
+
r"Vinoda\s*Basnayake",
|
|
1106
|
+
],
|
|
1107
|
+
),
|
|
1108
|
+
HighlightedNames(
|
|
1109
|
+
label='locations',
|
|
1110
|
+
style='cornsilk1',
|
|
1111
|
+
patterns=[
|
|
1112
|
+
r"Alabama",
|
|
1113
|
+
r"Arizona(?! State University)",
|
|
1114
|
+
r"Aspen",
|
|
1115
|
+
r"Berkeley",
|
|
1116
|
+
r"Brooklyn",
|
|
1117
|
+
r"California",
|
|
1118
|
+
r"Canada",
|
|
1119
|
+
r"Cape Cod",
|
|
1120
|
+
r"Connecticut",
|
|
1121
|
+
r"Florida",
|
|
1122
|
+
r"Loudoun\s*County?",
|
|
1123
|
+
r"Martha's\s*Vineyard",
|
|
1124
|
+
r"Miami(?!\s?Herald)",
|
|
1125
|
+
r"Nantucket",
|
|
1126
|
+
r"New\s*(Jersey|Mexico)",
|
|
1127
|
+
r"(North|South)\s*Carolina",
|
|
1128
|
+
r"NY(C|\s*State)",
|
|
1129
|
+
r"Orange\s*County",
|
|
1130
|
+
r"Oregon",
|
|
1131
|
+
r"Phoenix",
|
|
1132
|
+
r"Portland",
|
|
1133
|
+
r"Santa\s*Fe",
|
|
1134
|
+
r"Telluride",
|
|
1135
|
+
r"Teterboro",
|
|
1136
|
+
r"Texas(?! A&M)",
|
|
1137
|
+
r"Toronto",
|
|
1138
|
+
r"Tu(sc|cs)on",
|
|
1139
|
+
r"Vermont",
|
|
1140
|
+
r"Washington(\s*D\.?C)?(?!\s*Post)",
|
|
1141
|
+
r"Westchester",
|
|
1047
1142
|
],
|
|
1048
1143
|
),
|
|
1049
1144
|
HighlightedNames(
|
|
@@ -1114,7 +1209,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1114
1209
|
r"Mohammed\s+bin\s+Salman",
|
|
1115
1210
|
r"Morocco",
|
|
1116
1211
|
r"Mubarak",
|
|
1117
|
-
r"Muslim",
|
|
1212
|
+
r"Muslim(\s*Brotherhood)?",
|
|
1118
1213
|
r"Nayaf",
|
|
1119
1214
|
r"Pakistani?",
|
|
1120
1215
|
r"Omar",
|
|
@@ -1138,6 +1233,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1138
1233
|
r"Tripoli",
|
|
1139
1234
|
r"Tunisian?",
|
|
1140
1235
|
r"Turk(ey|ish)",
|
|
1236
|
+
r"Turks(?! & Caicos)",
|
|
1141
1237
|
r"UAE",
|
|
1142
1238
|
r"((Iraq|Iran|Kuwait|Qatar|Yemen)i?)",
|
|
1143
1239
|
],
|
|
@@ -1185,8 +1281,10 @@ HIGHLIGHTED_NAMES = [
|
|
|
1185
1281
|
},
|
|
1186
1282
|
patterns=[
|
|
1187
1283
|
r"(Matt(hew)? )?Hiltzi[gk]",
|
|
1284
|
+
r"PR\s*Newswire",
|
|
1188
1285
|
REPUTATION_MGMT,
|
|
1189
1286
|
r"Reputation.com",
|
|
1287
|
+
r"(Robert L\. )?Dilenschneider",
|
|
1190
1288
|
],
|
|
1191
1289
|
),
|
|
1192
1290
|
HighlightedNames(
|
|
@@ -1201,8 +1299,10 @@ HIGHLIGHTED_NAMES = [
|
|
|
1201
1299
|
r"Alberto\sGonzale[sz]",
|
|
1202
1300
|
r"(Alex\s*)?Acosta",
|
|
1203
1301
|
r"(Ben\s*)?Sasse",
|
|
1302
|
+
r"Betsy Devos",
|
|
1204
1303
|
r"((Bill|William)\s*)?Barr",
|
|
1205
1304
|
r"Bill\s*Shine",
|
|
1305
|
+
r"Blackwater",
|
|
1206
1306
|
r"(Bob\s*)?Corker",
|
|
1207
1307
|
r"(Brett\s*)?Kavanaugh",
|
|
1208
1308
|
r"Broidy",
|
|
@@ -1210,12 +1310,16 @@ HIGHLIGHTED_NAMES = [
|
|
|
1210
1310
|
r"(?<!Merwin Dela )Cruz",
|
|
1211
1311
|
r"Devin\s*Nunes",
|
|
1212
1312
|
r"(Don\s*)?McGa[hn]n",
|
|
1313
|
+
r"Erik Prince",
|
|
1213
1314
|
r"Gary\s*Cohn",
|
|
1214
1315
|
r"George\s*(H\.?\s*)?(W\.?\s*)?Bush",
|
|
1215
1316
|
r"(George\s*)?Nader",
|
|
1216
1317
|
r"GOP",
|
|
1217
1318
|
r"Jeff(rey)?\s*Sessions",
|
|
1218
|
-
r"(John\s*(R.?\s*)?)Bolton",
|
|
1319
|
+
r"(John\s*(R.?\s*)?)?Bolton",
|
|
1320
|
+
r"Kasich",
|
|
1321
|
+
r"Keith\s*Schiller",
|
|
1322
|
+
r"Kelly(\s*Anne?)?\s*Conway|Kellyanne",
|
|
1219
1323
|
r"Kissinger",
|
|
1220
1324
|
r"Kobach",
|
|
1221
1325
|
r"Kolfage",
|
|
@@ -1230,6 +1334,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1230
1334
|
r"((General|Mike)\s*)?(Flynn|Pence)",
|
|
1231
1335
|
r"(Mitt\s*)?Romney",
|
|
1232
1336
|
r"Mnuchin",
|
|
1337
|
+
r"(Newt\s*)Gingrich",
|
|
1233
1338
|
r"Nikki",
|
|
1234
1339
|
r"Haley",
|
|
1235
1340
|
r"(Paul\s*)?(Manafort|Volcker)",
|
|
@@ -1238,8 +1343,9 @@ HIGHLIGHTED_NAMES = [
|
|
|
1238
1343
|
r"Reagan",
|
|
1239
1344
|
r"Reince", r"Priebus",
|
|
1240
1345
|
r"Republican",
|
|
1241
|
-
r"(Rex\s*)?
|
|
1346
|
+
r"(Rex\s*)?Till?erson",
|
|
1242
1347
|
r"(?<!Cynthia )(Richard\s*)?Nixon",
|
|
1348
|
+
r"RNC",
|
|
1243
1349
|
r"Tea\s*Party",
|
|
1244
1350
|
],
|
|
1245
1351
|
),
|
|
@@ -1281,6 +1387,8 @@ HIGHLIGHTED_NAMES = [
|
|
|
1281
1387
|
r"(Natalia\s*)?Veselnitskaya",
|
|
1282
1388
|
r"(Oleg\s*)?Deripaska",
|
|
1283
1389
|
r"Oleksandr Vilkul",
|
|
1390
|
+
r"Onexim", # Prokhorov investment vehicle
|
|
1391
|
+
r"Prokhorov",
|
|
1284
1392
|
r"Rosneft",
|
|
1285
1393
|
r"RT",
|
|
1286
1394
|
r"St.?\s*?Petersburg",
|
|
@@ -1288,8 +1396,9 @@ HIGHLIGHTED_NAMES = [
|
|
|
1288
1396
|
r"Russian?",
|
|
1289
1397
|
r"Sberbank",
|
|
1290
1398
|
r"Soviet(\s*Union)?",
|
|
1399
|
+
r"Timur\s*Kulibayev",
|
|
1291
1400
|
r"USSR",
|
|
1292
|
-
r"
|
|
1401
|
+
r"Vlad(imir)?(?! Yudash)",
|
|
1293
1402
|
r"(Vladimir\s*)?Putin",
|
|
1294
1403
|
r"Women\s*Empowerment",
|
|
1295
1404
|
r"Xitrans",
|
|
@@ -1300,6 +1409,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1300
1409
|
label='Southeast Asia',
|
|
1301
1410
|
style='light_salmon3 bold',
|
|
1302
1411
|
patterns=[
|
|
1412
|
+
r"Australian?(?! Ave)",
|
|
1303
1413
|
r"Bangkok",
|
|
1304
1414
|
r"Burm(a|ese)",
|
|
1305
1415
|
r"Cambodian?",
|
|
@@ -1307,6 +1417,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1307
1417
|
r"Malaysian?",
|
|
1308
1418
|
r"Maldives",
|
|
1309
1419
|
r"Myan?mar",
|
|
1420
|
+
r"New\s*Zealand",
|
|
1310
1421
|
r"Philippines",
|
|
1311
1422
|
r"South\s*Korean?",
|
|
1312
1423
|
r"Tai(pei|wan)",
|
|
@@ -1327,8 +1438,10 @@ HIGHLIGHTED_NAMES = [
|
|
|
1327
1438
|
},
|
|
1328
1439
|
patterns=[
|
|
1329
1440
|
r"AG?I",
|
|
1441
|
+
r"Artificial\s*(General\s*)?Intelligence",
|
|
1330
1442
|
r"Chamath", r"Palihapitiya",
|
|
1331
1443
|
r"Danny\s*Hillis",
|
|
1444
|
+
r"deep learning",
|
|
1332
1445
|
r"Drew\s*Houston",
|
|
1333
1446
|
r"Eric\s*Schmidt",
|
|
1334
1447
|
r"Greylock(\s*Partners)?",
|
|
@@ -1357,10 +1470,12 @@ HIGHLIGHTED_NAMES = [
|
|
|
1357
1470
|
style='red3 bold',
|
|
1358
1471
|
emailers={
|
|
1359
1472
|
'Bruce Moskowitz': "'Trump's health guy' according to Epstein",
|
|
1473
|
+
NICHOLAS_RIBIS: 'Hilton CEO, former president of Trump Organization',
|
|
1360
1474
|
},
|
|
1361
1475
|
patterns=[
|
|
1362
1476
|
r"@?realDonaldTrump",
|
|
1363
1477
|
r"(Alan\s*)?Weiss?elberg",
|
|
1478
|
+
r"Alex\s*Jones",
|
|
1364
1479
|
r"\bDJ?T\b",
|
|
1365
1480
|
r"Donald J. Tramp",
|
|
1366
1481
|
r"(Donald\s+(J\.\s+)?)?Trump(ism|\s*(Org(anization)?|Properties)(\s*LLC)?)?",
|
|
@@ -1372,7 +1487,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1372
1487
|
r"(Marla\s*)?Maples",
|
|
1373
1488
|
r"(Matt(hew)? )?Calamari",
|
|
1374
1489
|
r"\bMatt C\b",
|
|
1375
|
-
r"Michael\s*Cohen",
|
|
1490
|
+
r"Michael\s*(D\.?\s*)?Cohen",
|
|
1376
1491
|
r"Melania",
|
|
1377
1492
|
r"(Michael (J.? )?)?Boccio",
|
|
1378
1493
|
r"Paul Rampell",
|
|
@@ -1398,25 +1513,30 @@ HIGHLIGHTED_NAMES = [
|
|
|
1398
1513
|
r"Dominican\s*Republic",
|
|
1399
1514
|
r"(Great|Little)\s*St.?\s*James",
|
|
1400
1515
|
r"Haiti(an)?",
|
|
1516
|
+
r"Jamaican?",
|
|
1401
1517
|
r"(John\s*)deJongh(\s*Jr\.?)",
|
|
1402
1518
|
r"(Kenneth E\. )?Mapp",
|
|
1403
1519
|
r"PBI",
|
|
1404
1520
|
r"Puerto\s*Ric(an|o)",
|
|
1405
1521
|
r"S(ain)?t.?\s*Thomas",
|
|
1406
1522
|
r"USVI",
|
|
1407
|
-
r"(?<!
|
|
1408
|
-
r"(The\s*)?Virgin\s*
|
|
1409
|
-
r"(West\s*)?Palm\s*Beach(?!\s*(Daily|Post))",
|
|
1523
|
+
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
|
|
1525
|
+
r"(West\s*)?Palm\s*Beach(\s*County)?(?!\s*(Daily|Post))",
|
|
1410
1526
|
],
|
|
1411
1527
|
),
|
|
1412
1528
|
HighlightedNames(
|
|
1413
1529
|
label='victim',
|
|
1414
1530
|
style='orchid1',
|
|
1415
1531
|
patterns=[
|
|
1416
|
-
r"
|
|
1532
|
+
r"#metoo",
|
|
1533
|
+
r"(David\s*)?Bo[il]es(,?\s*Schiller( & Flexner)?)?",
|
|
1417
1534
|
r"(Gloria\s*)?Allred",
|
|
1418
1535
|
r"(Jane|Tiffany)\s*Doe",
|
|
1419
1536
|
r"Katie\s*Johnson",
|
|
1537
|
+
r"pedophile",
|
|
1538
|
+
r"Stephanie\s*Clifford",
|
|
1539
|
+
r"Stormy\s*Daniels",
|
|
1420
1540
|
r"(Virginia\s+((L\.?|Roberts)\s+)?)?Giuffre",
|
|
1421
1541
|
r"Virginia\s+Roberts",
|
|
1422
1542
|
],
|
|
@@ -1455,7 +1575,7 @@ HIGHLIGHTED_NAMES = [
|
|
|
1455
1575
|
|
|
1456
1576
|
# Individuals
|
|
1457
1577
|
HighlightedNames(
|
|
1458
|
-
emailers={STEVEN_HOFFENBERG: "Epstein's Towers Financial ponzi
|
|
1578
|
+
emailers={STEVEN_HOFFENBERG: "Epstein's Towers Financial ponzi partner, prison for 18 years"},
|
|
1459
1579
|
style='dark_olive_green3',
|
|
1460
1580
|
category=FINANCE,
|
|
1461
1581
|
patterns=[r"(steven?\s*)?hoffenberg?w?"],
|
|
@@ -1471,8 +1591,8 @@ HIGHLIGHTED_NAMES = [
|
|
|
1471
1591
|
HighlightedNames(emailers={MELANIE_WALKER: f"doctor, friend of {BILL_GATES}"}, style='pale_violet_red1', category=FRIEND),
|
|
1472
1592
|
HighlightedNames(emailers={PAULA: "Epstein's ex-girlfriend who is now in the opera world"}, label='paula', style='pink1', category=FRIEND),
|
|
1473
1593
|
HighlightedNames(emailers={PRINCE_ANDREW: 'British royal family'}, style='dodger_blue1', category='Europe'),
|
|
1474
|
-
HighlightedNames(emailers={SOON_YI_PREVIN: 'wife of Woody Allen'}, style='hot_pink', category=
|
|
1475
|
-
HighlightedNames(emailers={SULTAN_BIN_SULAYEM: '
|
|
1594
|
+
HighlightedNames(emailers={SOON_YI_PREVIN: 'wife of Woody Allen'}, style='hot_pink', category=ARTS),
|
|
1595
|
+
HighlightedNames(emailers={SULTAN_BIN_SULAYEM: 'chairman of ports in Dubai, CEO of DP World'}, style='green1', category=MIDEAST),
|
|
1476
1596
|
|
|
1477
1597
|
# HighlightedText not HighlightedNames bc of word boundary issue
|
|
1478
1598
|
HighlightedText(
|
|
@@ -1495,7 +1615,7 @@ HIGHLIGHTED_TEXTS = [
|
|
|
1495
1615
|
HighlightedText(
|
|
1496
1616
|
label='header_field',
|
|
1497
1617
|
style='plum4',
|
|
1498
|
-
patterns=[r'
|
|
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) ?)):'],
|
|
1499
1619
|
),
|
|
1500
1620
|
HighlightedText(
|
|
1501
1621
|
label='http_links',
|
|
@@ -1549,6 +1669,41 @@ class EpsteinHighlighter(RegexHighlighter):
|
|
|
1549
1669
|
"""Finds and colors interesting keywords based on the above config."""
|
|
1550
1670
|
base_style = f"{REGEX_STYLE_PREFIX}."
|
|
1551
1671
|
highlights = [highlight_group.regex for highlight_group in ALL_HIGHLIGHTS]
|
|
1672
|
+
highlight_counts = defaultdict(int)
|
|
1673
|
+
|
|
1674
|
+
def highlight(self, text: Text) -> None:
|
|
1675
|
+
"""overrides https://rich.readthedocs.io/en/latest/_modules/rich/highlighter.html#RegexHighlighter"""
|
|
1676
|
+
highlight_regex = text.highlight_regex
|
|
1677
|
+
|
|
1678
|
+
for re_highlight in self.highlights:
|
|
1679
|
+
highlight_regex(re_highlight, style_prefix=self.base_style)
|
|
1680
|
+
|
|
1681
|
+
if args.debug and isinstance(re_highlight, re.Pattern):
|
|
1682
|
+
for match in re_highlight.finditer(text.plain):
|
|
1683
|
+
type(self).highlight_counts[(match.group(1) or 'None').replace('\n', ' ')] += 1
|
|
1684
|
+
|
|
1685
|
+
def print_highlight_counts(self, console: Console) -> None:
|
|
1686
|
+
"""Print counts of how many times strings were highlighted."""
|
|
1687
|
+
highlight_counts = deepcopy(self.highlight_counts)
|
|
1688
|
+
weak_date_regex = re.compile(r"^(\d\d?/|20|http|On ).*")
|
|
1689
|
+
|
|
1690
|
+
for highlighted, count in sort_dict(highlight_counts):
|
|
1691
|
+
if highlighted is None or weak_date_regex.match(highlighted):
|
|
1692
|
+
continue
|
|
1693
|
+
|
|
1694
|
+
try:
|
|
1695
|
+
console.print(f"{highlighted:25s} highlighted {count} times")
|
|
1696
|
+
except Exception as e:
|
|
1697
|
+
logger.error(f"Failed to print highlight count {count} for {highlighted}")
|
|
1698
|
+
|
|
1699
|
+
|
|
1700
|
+
def get_highlight_group_for_name(name: str | None) -> HighlightedNames | None:
|
|
1701
|
+
if name is None:
|
|
1702
|
+
return None
|
|
1703
|
+
|
|
1704
|
+
for highlight_group in HIGHLIGHTED_NAMES:
|
|
1705
|
+
if highlight_group.regex.search(name):
|
|
1706
|
+
return highlight_group
|
|
1552
1707
|
|
|
1553
1708
|
|
|
1554
1709
|
def get_style_for_category(category: str) -> str | None:
|
|
@@ -1584,15 +1739,6 @@ def styled_name(name: str | None, default_style: str = DEFAULT_NAME_STYLE) -> Te
|
|
|
1584
1739
|
return Text(name or UNKNOWN, style=get_style_for_name(name, default_style=default_style))
|
|
1585
1740
|
|
|
1586
1741
|
|
|
1587
|
-
def get_highlight_group_for_name(name: str | None) -> HighlightedNames | None:
|
|
1588
|
-
if name is None:
|
|
1589
|
-
return None
|
|
1590
|
-
|
|
1591
|
-
for highlight_group in HIGHLIGHTED_NAMES:
|
|
1592
|
-
if highlight_group.regex.search(name):
|
|
1593
|
-
return highlight_group
|
|
1594
|
-
|
|
1595
|
-
|
|
1596
1742
|
def _print_highlighted_names_repr() -> None:
|
|
1597
1743
|
for hn in HIGHLIGHTED_NAMES:
|
|
1598
1744
|
if isinstance(hn, HighlightedNames):
|