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
@@ -4,12 +4,17 @@ from typing import cast
4
4
 
5
5
  from dateutil.parser import parse
6
6
 
7
+ from epstein_files.documents.doj_files.full_text import EFTA00009622_TEXT
7
8
  from epstein_files.util.constant.names import *
8
9
  from epstein_files.util.constant.strings import *
9
10
  from epstein_files.util.doc_cfg import DocCfg, EmailCfg, TextCfg
11
+ from epstein_files.util.env import args
10
12
  from epstein_files.util.logging import logger
11
13
 
12
14
  FALLBACK_TIMESTAMP = parse("1/1/2051 12:01:01 AM")
15
+ MAX_CHARS_TO_PRINT = 4000
16
+ NO_TRUNCATE = -1
17
+ TRUNCATED_CHARS = int(MAX_CHARS_TO_PRINT / 3)
13
18
 
14
19
  HEADER_ABBREVIATIONS = {
15
20
  "AD": "Abu Dhabi",
@@ -19,6 +24,7 @@ HEADER_ABBREVIATIONS = {
19
24
  'bgC3': 'Bill Gates Ventures (renamed in 2018)',
20
25
  "Brock": 'Brock Pierce (crypto bro with a very sordid past)',
21
26
  "DB": "Deutsche Bank (maybe??)",
27
+ "GCC": "Gulf Cooperation Council",
22
28
  "GRAT": "Grantor Retained Annuity Trust (tax shelter)",
23
29
  'HBJ': "Sheikh Hamad bin Jassim (former Qatari prime minister)",
24
30
  'Jabor': '"an influential man in Qatar"',
@@ -51,169 +57,6 @@ HEADER_ABBREVIATIONS = {
51
57
  # Emailers Config Stuff #
52
58
  #########################
53
59
 
54
- # Emailers
55
- EMAILER_ID_REGEXES: dict[str, re.Pattern] = {
56
- ALAN_DERSHOWITZ: re.compile(r'(alan.{1,7})?dershowi(lz?|t?z)|AlanDersh', re.IGNORECASE),
57
- ALIREZA_ITTIHADIEH: re.compile(r'Alireza.[Il]ttihadieh', re.IGNORECASE),
58
- AMANDA_ENS: re.compile(r'ens, amanda?|Amanda.Ens', re.IGNORECASE),
59
- ANAS_ALRASHEED: re.compile(r'anas\s*al\s*rashee[cd]', re.IGNORECASE),
60
- ANIL_AMBANI: re.compile(r'Anil.Ambani', re.IGNORECASE),
61
- ANN_MARIE_VILLAFANA: re.compile(r'Villafana, Ann Marie|(A(\.|nn) Marie )?Villafa(c|n|ri)a', re.IGNORECASE),
62
- ANTHONY_SCARAMUCCI: re.compile(r"mooch|(Anthony ('The Mooch' )?)?Scaramucci", re.IGNORECASE),
63
- ARIANE_DE_ROTHSCHILD: re.compile(r'AdeR|((Ariane|Edmond) (de )?)?Roths(ch|hc)?ild|Ariane', re.IGNORECASE),
64
- BARBRO_C_EHNBOM: re.compile(r'behnbom@aol.com|(Barbro\s.*)?Ehnbom', re.IGNORECASE),
65
- BARRY_J_COHEN: re.compile(r'barry\s*((j.?|james)\s*)?cohen?', re.IGNORECASE),
66
- BENNET_MOSKOWITZ: re.compile(r'Moskowitz.*Bennet|Bennet.*Moskowitz', re.IGNORECASE),
67
- BOB_CROWE: re.compile(r"[BR]ob Crowe", re.IGNORECASE),
68
- BORIS_NIKOLIC: re.compile(r'(boris )?nikolic?', re.IGNORECASE),
69
- BRAD_EDWARDS: re.compile(r'Brad(ley)?(\s*J(.?|ames))?\s*Edwards', re.IGNORECASE),
70
- BRAD_KARP: re.compile(r'Brad (S.? )?Karp|Karp, Brad', re.IGNORECASE),
71
- DANGENE_AND_JENNIE_ENTERPRISE: re.compile(r'Dangene and Jennie Enterprise?', re.IGNORECASE),
72
- DANNY_FROST: re.compile(r'Frost, Danny|frostd@dany.nyc.gov|Danny\s*Frost', re.IGNORECASE),
73
- DARREN_INDYKE: re.compile(r'darren$|Darren\s*(K\.?\s*)?[il]n[dq]_?yke?|dkiesq', re.IGNORECASE),
74
- DAVID_FISZEL: re.compile(r'David\s*Fis?zel', re.IGNORECASE),
75
- DAVID_HAIG: re.compile(fr'{DAVID_HAIG}|Haig, David', re.IGNORECASE),
76
- DAVID_STERN: re.compile(r'David Stern?', re.IGNORECASE),
77
- EDUARDO_ROBLES: re.compile(r'Ed(uardo)?\s*Robles', re.IGNORECASE),
78
- EDWARD_JAY_EPSTEIN: re.compile(r'Edward (Jay )?Epstein', re.IGNORECASE),
79
- EHUD_BARAK: re.compile(r'(ehud|e?h)\s*barak|\behud', re.IGNORECASE),
80
- FAITH_KATES: re.compile(r'faith kates?', re.IGNORECASE),
81
- GERALD_BARTON: re.compile(r'Gerald.*Barton', re.IGNORECASE),
82
- GERALD_LEFCOURT: re.compile(r'Gerald\s*(B\.?\s*)?Lefcourt', re.IGNORECASE),
83
- GHISLAINE_MAXWELL: re.compile(r'g ?max(well)?|Ghislaine|Maxwell', re.IGNORECASE),
84
- HEATHER_MANN: re.compile(r'Heather Mann?', re.IGNORECASE),
85
- INTELLIGENCE_SQUARED: re.compile(r'intelligence\s*squared', re.IGNORECASE),
86
- JACKIE_PERCZEK: re.compile(r'jackie percze[kl]?', re.IGNORECASE),
87
- JABOR_Y: re.compile(r'[ji]abor\s*y?', re.IGNORECASE),
88
- JAMES_HILL: re.compile(r"hill, james e.|james.e.hill@abc.com", re.IGNORECASE),
89
- JANUSZ_BANASIAK: re.compile(r"Janu[is]z Banasiak", re.IGNORECASE),
90
- JEAN_LUC_BRUNEL: re.compile(r'Jean[- ]Luc Brunel?|JeanLuc', re.IGNORECASE),
91
- JEFF_FULLER: re.compile(r"jeff@mc2mm.com|Jeff Fuller", re.IGNORECASE),
92
- JEFFREY_EPSTEIN: re.compile(r'[djl]\s?ee[vy]acation[©@]?g?(mail.com)?|Epstine|\bJEE?\b|Jeffrey E((sp|ps)tein?)?( VI Foundation)?|jeeproject@yahoo.com|J Jep|Jeffery Edwards|(?<!(Mark L.|ard Jay) )Epstein', re.IGNORECASE),
93
- JESSICA_CADWELL: re.compile(r'Jessica Cadwell?', re.IGNORECASE),
94
- JOHNNY_EL_HACHEM: re.compile(r'el hachem johnny|johnny el hachem', re.IGNORECASE),
95
- JOI_ITO: re.compile(r'ji@media.mit.?edu|(joichi|joi)( Ito)?', re.IGNORECASE),
96
- JONATHAN_FARKAS: re.compile(r'Jonathan Fark(a|u)(s|il)', re.IGNORECASE),
97
- KATHRYN_RUEMMLER: re.compile(r'Kathr?yn? Ruemmler?', re.IGNORECASE),
98
- KEN_STARR: re.compile(r'starr, ken|Ken(neth\s*(W.\s*)?)?\s+starr?|starr', re.IGNORECASE),
99
- LANDON_THOMAS: re.compile(r'lando[nr] thomas( jr)?|thomas jr.?, lando[nr]', re.IGNORECASE),
100
- LARRY_SUMMERS: re.compile(r'(La(wrence|rry).{1,5})?Summers?|^LH$|LHS|[Il]hsofficel?', re.IGNORECASE),
101
- LAWRANCE_VISOSKI: re.compile(r'La(rry|wrance) Visoski?|Lvjet', re.IGNORECASE),
102
- LAWRENCE_KRAUSS: re.compile(r'Lawrence Kraus[es]?|[jl]awkrauss|kruase', re.IGNORECASE),
103
- LEON_BLACK: re.compile(r'Leon\s*Black?|(?<!Marc )Leon(?! (Botstein|Jaworski|Wieseltier))', re.IGNORECASE),
104
- LILLY_SANCHEZ: re.compile(r'Lilly.*Sanchez', re.IGNORECASE),
105
- LISA_NEW: re.compile(r'E?Lisa New?\b', re.IGNORECASE),
106
- MANUELA_MARTINEZ: re.compile(fr'Manuela (- Mega Partners|Martinez)', re.IGNORECASE),
107
- MARIANA_IDZKOWSKA: re.compile(r'Mariana [Il]d[źi]kowska?', re.IGNORECASE),
108
- MARK_EPSTEIN: re.compile(r'Mark (L\. )?(Epstein|Lloyd)', re.IGNORECASE),
109
- MARC_LEON: re.compile(r'Marc[.\s]+(Kensington|Leon)|Kensington2', re.IGNORECASE),
110
- MARTIN_NOWAK: re.compile(r'(Martin.*?)?No[vw]ak|Nowak, Martin', re.IGNORECASE),
111
- MARTIN_WEINBERG: re.compile(r'martin.*?weinberg', re.IGNORECASE),
112
- "Matthew Schafer": re.compile(r"matthew\.?schafer?", re.IGNORECASE),
113
- MELANIE_SPINELLA: re.compile(r'M?elanie Spine[Il]{2}a', re.IGNORECASE),
114
- MICHAEL_BUCHHOLTZ: re.compile(r'Michael.*Buchholtz', re.IGNORECASE),
115
- MICHAEL_MILLER: re.compile(r'Micha(el)? Miller|Miller, Micha(el)?', re.IGNORECASE),
116
- MICHAEL_SITRICK: re.compile(r'(Mi(chael|ke).{0,5})?[CS]itrick', re.IGNORECASE),
117
- MICHAEL_WOLFF: re.compile(r'Michael\s*Wol(f[ef]e?|i)|Wolff', re.IGNORECASE),
118
- MIROSLAV_LAJCAK: re.compile(r"Miro(slav)?(\s+Laj[cč][aá]k)?"),
119
- MOHAMED_WAHEED_HASSAN: re.compile(r'Mohamed Waheed(\s+Hassan)?', re.IGNORECASE),
120
- NADIA_MARCINKO: re.compile(r"Na[dď]i?a\s+Marcinko(v[aá])?", re.IGNORECASE),
121
- NEAL_KASSELL: re.compile(r'Neal\s*Kassell?', re.IGNORECASE),
122
- NICHOLAS_RIBIS: re.compile(r'Nic(holas|k)[\s._]Ribi?s?|Ribbis', re.IGNORECASE),
123
- OLIVIER_COLOM: re.compile(fr'Colom, Olivier|{OLIVIER_COLOM}', re.IGNORECASE),
124
- PAUL_BARRETT: re.compile(r'Paul Barre(d|tt)', re.IGNORECASE),
125
- PAUL_KRASSNER: re.compile(r'Pa\s?ul Krassner', re.IGNORECASE),
126
- PAUL_MORRIS: re.compile(r'morris, paul|Paul Morris', re.IGNORECASE),
127
- PAULA: re.compile(r'^Paula( Heil Fisher)?$', re.IGNORECASE),
128
- PEGGY_SIEGAL: re.compile(r'Peggy Siegal?', re.IGNORECASE),
129
- PETER_ATTIA: re.compile(r'Peter Attia?', re.IGNORECASE),
130
- PETER_MANDELSON: re.compile(r"((Lord|Peter) )?Mandelson", re.IGNORECASE),
131
- 'pink@mc2mm.com': re.compile(r"^Pink$|pink@mc2mm\.com", re.IGNORECASE),
132
- PRINCE_ANDREW: re.compile(r'Prince Andrew|The Duke', re.IGNORECASE),
133
- REID_WEINGARTEN: re.compile(r'Weingarten, Rei[cdi]|Rei[cdi] Weingarten', re.IGNORECASE),
134
- RICHARD_KAHN: re.compile(r'rich(ard)? kahn?', re.IGNORECASE),
135
- ROBERT_D_CRITTON_JR: re.compile(r'Robert D.? Critton,? Jr.?', re.IGNORECASE),
136
- ROBERT_LAWRENCE_KUHN: re.compile(r'Robert\s*(Lawrence)?\s*Kuhn', re.IGNORECASE),
137
- ROBERT_TRIVERS: re.compile(r'tri[vy]ersr@gmail|Robert\s*Trivers?', re.IGNORECASE),
138
- ROSS_GOW: re.compile(fr"Ross(acuity)? Gow|(ross@)?acuity\s*reputation(\.com)?", re.IGNORECASE),
139
- SAMUEL_LEFF: re.compile(r"Sam(uel)?(/Walli)? Leff", re.IGNORECASE),
140
- SCOTT_J_LINK: re.compile(r'scott j. link?', re.IGNORECASE),
141
- SEAN_BANNON: re.compile(r'sean bannon?', re.IGNORECASE),
142
- SHAHER_ABDULHAK_BESHER: re.compile(r'\bShaher( Abdulhak Besher)?\b', re.IGNORECASE),
143
- SOON_YI_PREVIN: re.compile(r'Soon[- ]Yi Previn?', re.IGNORECASE),
144
- STEPHEN_HANSON: re.compile(r'ste(phen|ve) hanson?|Shanson900', re.IGNORECASE),
145
- STEVE_BANNON: re.compile(r'steve banno[nr]?', re.IGNORECASE),
146
- STEVEN_SINOFSKY: re.compile(r'Steven Sinofsky?', re.IGNORECASE),
147
- SULTAN_BIN_SULAYEM: re.compile(r'Sultan (Ahmed )?bin Sulaye?m?', re.IGNORECASE),
148
- TERJE_ROD_LARSEN: re.compile(r"Terje(( (R[øo]e?d[- ])?)?Lars[eo]n)?", re.IGNORECASE),
149
- TERRY_KAFKA: re.compile(r'Terry Kafka?', re.IGNORECASE),
150
- THANU_BOONYAWATANA: re.compile(r"Thanu (BOONYAWATANA|Cnx)", re.IGNORECASE),
151
- THORBJORN_JAGLAND: re.compile(r'(Thor.{3,8})?Jag[il]and?', re.IGNORECASE),
152
- TONJA_HADDAD_COLEMAN: re.compile(r"To(nj|rl)a Haddad Coleman|haddadfm@aol.com", re.IGNORECASE),
153
- VINCENZO_IOZZO: re.compile(r"Vincenzo [IL]ozzo", re.IGNORECASE),
154
- }
155
-
156
- # If found as substring consider them the author
157
- EMAILERS = [
158
- 'Anne Boyles',
159
- AL_SECKEL,
160
- AZIZA_ALAHMADI,
161
- BILL_GATES,
162
- BILL_SIEGEL,
163
- BRAD_WECHSLER,
164
- CHRISTINA_GALBRAITH,
165
- DANIEL_SABBA,
166
- 'Danny Goldberg',
167
- DAVID_SCHOEN,
168
- DEBBIE_FEIN,
169
- DEEPAK_CHOPRA,
170
- GLENN_DUBIN,
171
- GORDON_GETTY,
172
- 'Kevin Bright',
173
- 'Jack Lang',
174
- JACK_SCAROLA,
175
- JAY_LEFKOWITZ,
176
- JES_STALEY,
177
- JOHN_PAGE,
178
- 'Jokeland',
179
- JOSCHA_BACH,
180
- 'Kathleen Ruderman',
181
- KENNETH_E_MAPP,
182
- 'Larry Cohen',
183
- LESLEY_GROFF,
184
- 'lorraine@mc2mm.com',
185
- LINDA_STONE,
186
- 'Lyn Fontanilla',
187
- MARK_TRAMO,
188
- MELANIE_WALKER,
189
- MERWIN_DELA_CRUZ,
190
- 'Michael Simmons', # Not the only "To:"
191
- 'middle.east.update@hotmail.com',
192
- 'Nancy Cain',
193
- 'Nancy Dahl',
194
- 'Nancy Portland',
195
- 'Oliver Goodenough',
196
- 'Peter Aldhous',
197
- 'Peter Green',
198
- ROGER_SCHANK,
199
- 'Roy Black',
200
- STEVEN_PFEIFFER,
201
- 'Steven Victor MD',
202
- 'Susan Edelman',
203
- TOM_BARRACK,
204
- 'Vladimir Yudashkin',
205
- ]
206
-
207
- EMAILER_REGEXES = deepcopy(EMAILER_ID_REGEXES) # Keep a copy without the simple EMAILERS regexes
208
-
209
- # Add simple matching regexes for EMAILERS entries to EMAILER_REGEXES
210
- for emailer in EMAILERS:
211
- if emailer in EMAILER_REGEXES:
212
- raise RuntimeError(f"Can't overwrite emailer regex for '{emailer}'")
213
-
214
- EMAILER_REGEXES[emailer] = re.compile(emailer, re.IGNORECASE)
215
-
216
-
217
60
  # Atribution reasons
218
61
  BOLOTOVA_REASON = 'Same signature style as 029020 ("--" followed by "Sincerely Renata Bolotova")'
219
62
  KATHY_REASON = 'from "Kathy" about dems, sent from iPad'
@@ -308,15 +151,47 @@ TEXTS_CONFIG = CONFIRMED_TEXTS_CONFIG + UNCONFIRMED_TEXTS_CONFIG
308
151
  ################################################ EMAILS ################################################
309
152
  ########################################################################################################
310
153
 
154
+ MICHAEL_WOLFF_EPSTEIN_ARTICLE_DRAFT = f"draft of an unpublished article about Epstein by {MICHAEL_WOLFF} written ca. 2014/2015"
311
155
  # Some emails have a lot of uninteresting CCs
312
156
  FLIGHT_IN_2012_PEOPLE: list[Name] = ['Francis Derby', JANUSZ_BANASIAK, 'Louella Rabuyo', 'Richard Barnnet']
313
157
  IRAN_DEAL_RECIPIENTS: list[Name] = ['Allen West', 'Rafael Bardaji', 'Philip Kafka', 'Herb Goodman', 'Grant Seeger', 'Lisa Albert', 'Janet Kafka', 'James Ramsey', 'ACT for America', 'John Zouzelka', 'Joel Dunn', 'Nate McClain', 'Bennet Greenwald', 'Taal Safdie', 'Uri Fouzailov', 'Neil Anderson', 'Nate White', 'Rita Hortenstine', 'Henry Hortenstine', 'Gary Gross', 'Forrest Miller', 'Bennett Schmidt', 'Val Sherman', 'Marcie Brown', 'Michael Horowitz', 'Marshall Funk']
314
- MICHAEL_WOLFF_EPSTEIN_ARTICLE_DRAFT = f"draft of an unpublished article about Epstein by {MICHAEL_WOLFF} written ca. 2014/2015"
158
+
159
+ TRIVERS_CCS: list[Name] = [
160
+ "Alan Rogers",
161
+ "Anna Dreber",
162
+ "Anula Jayasuriya",
163
+ "Bill Prezant",
164
+ "Bobby McCormick",
165
+ "Clive Crook",
166
+ "Dane Stangler",
167
+ "Ron Bailey",
168
+ "Ditsa Pines",
169
+ "David Darst",
170
+ "Gerry Ohrstrom",
171
+ "Paul Romer",
172
+ "John Mallen",
173
+ "Jim Halligan",
174
+ "Lee Silver",
175
+ "Monika Gruter Cheney",
176
+ "Marguerite Atkins",
177
+ "Matt Ridley",
178
+ "Mike Cagney",
179
+ "Evan Smith",
180
+ "Roger Edelen",
181
+ "Oliver Goodenough",
182
+ "Paul Zak",
183
+ "Peter J Richerson",
184
+ "Clair Brown",
185
+ "Terry Anderson",
186
+ "Tim Kane",
187
+ "Rob Hanson",
188
+ "president@usfca.edu",
189
+ ]
190
+
315
191
 
316
192
  EMAILS_CONFIG = [
317
193
  # 026294 and 026296 might also be Ittihadieh based on timing
318
194
  EmailCfg(id='032436', author=ALIREZA_ITTIHADIEH, attribution_reason='Signature'),
319
- # 032542 026078 026080 026083 026086 026090 might also be Anas based on discussion of Dubai and Kuwait
320
195
  EmailCfg(id='032543', author=ANAS_ALRASHEED, attribution_reason='Later reply 033000 has quote'),
321
196
  EmailCfg(id='026167', author=ANAS_ALRASHEED, attribution_reason='name visible in 033022 reply'),
322
197
  EmailCfg(id='032571', author=ANAS_ALRASHEED, attribution_reason='name visible in 033022 reply'),
@@ -355,6 +230,12 @@ EMAILS_CONFIG = [
355
230
  EmailCfg(id='032676', author=ANAS_ALRASHEED, attribution_reason='name visible in 029113 reply'),
356
231
  EmailCfg(id='026237', author=ANAS_ALRASHEED, attribution_reason='name visible in 029113 reply'),
357
232
  EmailCfg(id='032682', author=ANAS_ALRASHEED, attribution_reason='name visible in 029113 reply'),
233
+ EmailCfg(id='032542', author=ANAS_ALRASHEED, attribution_reason='discussion of Kuwait and timing', is_attribution_uncertain=True),
234
+ EmailCfg(id='026078', author=ANAS_ALRASHEED, attribution_reason='discussion of Kuwait and timing', is_attribution_uncertain=True),
235
+ EmailCfg(id='026080', author=ANAS_ALRASHEED, attribution_reason='discussion of Kuwait and timing', is_attribution_uncertain=True),
236
+ EmailCfg(id='026083', author=ANAS_ALRASHEED, attribution_reason='discussion of Kuwait and timing', is_attribution_uncertain=True),
237
+ EmailCfg(id='026086', author=ANAS_ALRASHEED, attribution_reason='discussion of Kuwait and timing', is_attribution_uncertain=True),
238
+ EmailCfg(id='026090', author=ANAS_ALRASHEED, attribution_reason='discussion of Kuwait and timing', is_attribution_uncertain=True),
358
239
  EmailCfg(id='026064', author=ARIANE_DE_ROTHSCHILD, attribution_reason='signature'),
359
240
  EmailCfg(id='026069', author=ARIANE_DE_ROTHSCHILD, attribution_reason='signature'),
360
241
  EmailCfg(id='030741', author=ARIANE_DE_ROTHSCHILD, attribution_reason='signature'),
@@ -371,33 +252,40 @@ EMAILS_CONFIG = [
371
252
  fwded_text_after='Transcript: Phone call between President',
372
253
  ),
373
254
  EmailCfg(id='033328', author=AZIZA_ALAHMADI, attribution_reason='"Regards, Aziza" at bottom'),
374
- EmailCfg(id='026659', author=BARBRO_C_EHNBOM, attribution_reason='Reply'),
375
- EmailCfg(id='031215', author=BARBRO_C_EHNBOM, duplicate_ids=['026745'], dupe_type='redacted'), # the same except for 'your Anna!'. author must be specified because email address is redacted in 026745 so it needs the config
376
- EmailCfg(id='026764', author=BARRY_J_COHEN), # Bad OCR (nofix)
377
- EmailCfg(id='031206', author=BENNET_MOSKOWITZ, duplicate_ids=['031227']),
378
- EmailCfg(id='031442', author=CHRISTINA_GALBRAITH, duplicate_ids=['031996']),
255
+ EmailCfg(
256
+ id='031215',
257
+ author=BARBRO_C_EHNBOM,
258
+ duplicate_ids=['026745'],
259
+ dupe_type='redacted',
260
+ comment="the same except for 'your Anna!'. author must be specified because email address is redacted in 026745 so it needs the config",
261
+ ),
262
+ EmailCfg(id='031206', duplicate_ids=['031227']),
263
+ EmailCfg(id='031591', duplicate_ids=['031442', '031996']),
379
264
  EmailCfg(
380
265
  id='019446',
381
266
  author=CHRISTINA_GALBRAITH,
382
267
  attribution_reason='shows from "Christina media/PR" which fits',
383
268
  is_attribution_uncertain=True,
384
269
  ),
385
- EmailCfg(id='026625', author=DARREN_INDYKE, actual_text='Hysterical.'),
386
270
  EmailCfg(
387
271
  id='026624',
388
272
  author=DARREN_INDYKE,
389
- recipients=[JEFFREY_EPSTEIN],
390
- timestamp=parse('2016-10-01 16:40:00'),
273
+ date='2016-10-01 16:40:00',
391
274
  duplicate_ids=['031708'],
275
+ recipients=[JEFFREY_EPSTEIN],
276
+ subject='Donald Trump Rape Lawsuit Refiled With New Witness I Law News',
392
277
  ),
393
278
  EmailCfg(
394
279
  id='031278',
395
280
  actual_text='',
396
281
  author=DARREN_INDYKE,
397
- description=f"heavily redacted email, quoted replies are from {STEVEN_HOFFENBERG} about James Patterson's book",
398
- recipients=['Charles Michael'],
399
- timestamp=parse('2016-08-17 11:26:00'),
282
+ recipients=["Charles Michael"],
400
283
  attribution_reason='Quoted replies are in 019109',
284
+ date='2016-08-17 11:26:00',
285
+ description="heavily redacted email, quoted replies are from Steven Hoffenberg about James Patterson's book",
286
+ subject='FW: Privileged and Confidential - Fwd: JAMES PATTERSON NEW BOOK TELLING FEDS COVER UP OF BILLIONAIRE JEFF EPSTEIN CHILD RAPES RELEASE DATE OCT 10 2016 STEVEN HOFFENBERG IS ON THE BOOK WRITING TEAM !!!!',
287
+ truncate_to=2500,
288
+ comment='Hoffenberg',
401
289
  ),
402
290
  EmailCfg(id='026290', author=DAVID_SCHOEN, attribution_reason='Signature'),
403
291
  EmailCfg(id='031339', author=DAVID_SCHOEN, attribution_reason='Signature'),
@@ -405,43 +293,43 @@ EMAILS_CONFIG = [
405
293
  EmailCfg(id='031560', author=DAVID_SCHOEN, attribution_reason='Signature'),
406
294
  EmailCfg(id='026287', author=DAVID_SCHOEN, attribution_reason='Signature'),
407
295
  EmailCfg(id='033419', author=DAVID_SCHOEN, attribution_reason='Signature'),
408
- EmailCfg(id='031460', author=EDWARD_JAY_EPSTEIN, attribution_reason=f"quoted reply has edwardjayepstein.com", is_fwded_article=True),
296
+ EmailCfg(id='031460', author=EDWARD_JAY_EPSTEIN, attribution_reason='quoted reply has edwardjayepstein.com', is_fwded_article=True),
409
297
  EmailCfg(
410
298
  id='030475',
411
299
  author=FAITH_KATES,
412
- attribution_reason=f'{NEXT_MANAGEMENT} legal signature',
300
+ attribution_reason='Next Management LLC legal signature',
413
301
  duplicate_ids=['030575'],
414
- dupe_type='redacted'
302
+ dupe_type='redacted',
415
303
  ),
416
- EmailCfg(id='026547', author=GERALD_BARTON, recipients=[JEFFREY_EPSTEIN]), # Bad OCR # TODO: email header is really jacked up
304
+ EmailCfg(id='026547', author=GERALD_BARTON, recipients=[JEFFREY_EPSTEIN]),
417
305
  EmailCfg(id='029969', author=GWENDOLYN_BECK, attribution_reason='signature "Longevity & Successful Aging"'),
418
306
  EmailCfg(id='029968', author=GWENDOLYN_BECK, attribution_reason='signature "beckresearchlabs.com"', duplicate_ids=['031120']),
419
- EmailCfg(id='029970', author=GWENDOLYN_BECK, attribution_reason='signed "Longevity & Successful Agin"'),
420
- EmailCfg(id='029960', author=GWENDOLYN_BECK, attribution_reason='signature "Beck Center for Longevity & Aging"'),
307
+ EmailCfg(id='029970', author=GWENDOLYN_BECK, attribution_reason='signed "Longevity & Successful Aging"', truncate_to=400),
308
+ EmailCfg(id='029960', author=GWENDOLYN_BECK, attribution_reason='signature "Beck Center for Longevity & Aging"', truncate_to=498),
421
309
  EmailCfg(id='029959', author=GWENDOLYN_BECK, attribution_reason='signature "Beck Center for Longevity & Aging"'),
422
- EmailCfg(id='033360', author=HENRY_HOLT, attribution_reason='in signature'), # Henry Holt is a company not a person
310
+ EmailCfg(id='033360', author=HENRY_HOLT, attribution_reason='in signature'),
423
311
  EmailCfg(id='033384', author=JACK_GOLDBERGER, attribution_reason='Might be Paul Prosperi?', is_attribution_uncertain=True),
424
312
  EmailCfg(id='026024', author=JEAN_HUGUEN, attribution_reason='Signature'),
425
313
  EmailCfg(id='021823', author=JEAN_LUC_BRUNEL, attribution_reason='Reply'),
426
- EmailCfg(id='022949', author=JEFFREY_EPSTEIN), # Bad OCR (nofix)
427
- EmailCfg(id='031624', author=JEFFREY_EPSTEIN), # Bad OCR (nofix)
428
- EmailCfg(id='031996', author=JEFFREY_EPSTEIN, recipients=[CHRISTINA_GALBRAITH], attribution_reason='bounced', duplicate_ids=['031442']),
429
- EmailCfg(id='018726', author=JEFFREY_EPSTEIN, timestamp=parse('2018-06-08 08:36:00')), # nofix
430
- EmailCfg(id='032283', author=JEFFREY_EPSTEIN, timestamp=parse('2016-09-14 08:04:00')), # nofix
431
- EmailCfg(id='026943', author=JEFFREY_EPSTEIN, timestamp=parse('2019-05-22 05:47:00')), # nofix
314
+ EmailCfg(id='031624', author=JEFFREY_EPSTEIN),
315
+ EmailCfg(id='018726', author=JEFFREY_EPSTEIN, date='2018-06-08 08:36:00'),
316
+ EmailCfg(id='032283', author=JEFFREY_EPSTEIN, date='2016-09-14 08:04:00'),
317
+ EmailCfg(id='026943', author=JEFFREY_EPSTEIN, date='2019-05-22 05:47:00'),
432
318
  EmailCfg(
433
319
  id='023208',
434
320
  author=JEFFREY_EPSTEIN,
435
- description=f"very long email chain about Leon Black's finances and things like Gratitude America",
436
- fwded_text_after='Date: Tue, Oct 27',
437
321
  recipients=[BRAD_WECHSLER, MELANIE_SPINELLA],
322
+ description="very long email chain about Leon Black's finances and things like Gratitude America",
438
323
  duplicate_ids=['023291'],
324
+ fwded_text_after='Date: Tue, Oct 27',
325
+ truncate_to=NO_TRUNCATE,
326
+ comment="Long discussion about leon black's finances",
439
327
  ),
440
328
  EmailCfg(
441
329
  id='032214',
442
330
  author=JEFFREY_EPSTEIN,
443
- actual_text='Agreed',
444
331
  recipients=[MIROSLAV_LAJCAK],
332
+ actual_text='Agreed',
445
333
  attribution_reason='Quoted reply has signature',
446
334
  ),
447
335
  EmailCfg(id='029582', author=JEFFREY_EPSTEIN, recipients=[RENATA_BOLOTOVA], attribution_reason=BOLOTOVA_REASON),
@@ -450,15 +338,24 @@ EMAILS_CONFIG = [
450
338
  EmailCfg(id='031826', author=JEFFREY_EPSTEIN, actual_text='I have'),
451
339
  EmailCfg(id='030768', author=JEFFREY_EPSTEIN, actual_text='ok'),
452
340
  EmailCfg(id='022938', author=JEFFREY_EPSTEIN, actual_text='what do you suggest?'),
453
- EmailCfg(id='031791', author=JESSICA_CADWELL, attribution_reason='signature'),
454
- EmailCfg(id='028851', author=JOI_ITO, recipients=[JEFFREY_EPSTEIN], timestamp=parse('2014-04-27 06:00:00')),
455
341
  EmailCfg(
456
- id='028849',
457
- attribution_reason='Conversation with Joi Ito',
458
- author=JOI_ITO,
459
- description=f"{JOI_ITO} reaching out to Epstein for an immediate phone call after news about illicit Russian money",
460
- recipients=[JEFFREY_EPSTEIN],
461
- timestamp=parse('2014-04-27 07:41:00'), # Filled in from 028847
342
+ id='031791',
343
+ author=JESSICA_CADWELL,
344
+ attribution_reason='signature',
345
+ truncate_to=NO_TRUNCATE,
346
+ comment='First email in Jessica Cadwell chain about service of legal documents',
347
+ ),
348
+ EmailCfg(
349
+ id='028850',
350
+ description='Joi Ito reaching out to Epstein for an immediate phone call after news about illicit Russian money',
351
+ duplicate_ids=['028851'],
352
+ dupe_type='quoted',
353
+ ),
354
+ EmailCfg(
355
+ id='028848',
356
+ description='Joi Ito reaching out to Epstein for an immediate phone call after news about illicit Russian money',
357
+ duplicate_ids=['028849'],
358
+ dupe_type='quoted',
462
359
  ),
463
360
  EmailCfg(id='028507', author=JONATHAN_FARKAS, attribution_reason='reply signed "best Jonathan"'),
464
361
  EmailCfg(id='033282', author=JONATHAN_FARKAS, attribution_reason='reply signed "thanks Jonathan"', duplicate_ids=['033484']),
@@ -470,9 +367,9 @@ EMAILS_CONFIG = [
470
367
  EmailCfg(id='032386', author=KATHRYN_RUEMMLER, attribution_reason=KATHY_REASON, is_attribution_uncertain=True),
471
368
  EmailCfg(id='032727', author=KATHRYN_RUEMMLER, attribution_reason=KATHY_REASON, is_attribution_uncertain=True),
472
369
  EmailCfg(id='030478', author=LANDON_THOMAS),
473
- EmailCfg(id='029013', author=LARRY_SUMMERS, recipients=[JEFFREY_EPSTEIN]), # Bad OCR (nofix)
370
+ EmailCfg(id='029013', author=LARRY_SUMMERS, recipients=[JEFFREY_EPSTEIN]),
474
371
  EmailCfg(id='029196', author=LAWRENCE_KRAUSS, recipients=[JEFFREY_EPSTEIN], actual_text='Talk in 40?'),
475
- EmailCfg(id='033593', author=LAWRANCE_VISOSKI, attribution_reason='Signature'),
372
+ EmailCfg(id='033593', author=LAWRANCE_VISOSKI, attribution_reason='Signature', truncate_to=NO_TRUNCATE, comment='visoski email about planes'),
476
373
  EmailCfg(id='033370', author=LAWRANCE_VISOSKI, attribution_reason=LARRY_REASON),
477
374
  EmailCfg(id='033495', author=LAWRANCE_VISOSKI, attribution_reason=LARRY_REASON),
478
375
  EmailCfg(id='033487', author=LAWRANCE_VISOSKI, recipients=[JEFFREY_EPSTEIN]),
@@ -490,35 +387,55 @@ EMAILS_CONFIG = [
490
387
  EmailCfg(id='026609', author='Mark Green', attribution_reason='Actually a fwd, Mark Green is in signature'),
491
388
  EmailCfg(id='030472', author=MARTIN_WEINBERG, attribution_reason='Maybe. in reply', is_attribution_uncertain=True),
492
389
  EmailCfg(id='032563', author=MASHA_DROKOVA, attribution_reason='replied to in 033014'),
493
- EmailCfg(id='032564', author=MASHA_DROKOVA, attribution_reason='follow up to 032563 about huffpo article with link'),
390
+ EmailCfg(
391
+ id='032564',
392
+ attribution_reason='follow up to 032563 about huffpo article with link',
393
+ author=MASHA_DROKOVA,
394
+ description='an archived version of the HuffPost link is here: https://archive.is/hJxT3 '
395
+ ),
494
396
  EmailCfg(id='031544', author=MASHA_DROKOVA, attribution_reason='follow up to 032563 about huffpo article with link'),
495
397
  EmailCfg(id='032605', author=MASHA_DROKOVA, attribution_reason="re: PR interview, 031544 says she'll be in NY at that time"),
496
398
  EmailCfg(id='032606', author=MASHA_DROKOVA, attribution_reason="re: PR interview, 031544 says she'll be in NY at that time"),
497
399
  EmailCfg(id='032607', author=MASHA_DROKOVA, attribution_reason="re: PR interview, 031544 says she'll be in NY at that time"),
498
400
  EmailCfg(id='032609', author=MASHA_DROKOVA, attribution_reason="re: PR interview, 031544 says she'll be in NY at that time"),
499
- EmailCfg(id='032604', author=MASHA_DROKOVA, attribution_reason="timing, subject (interviews/articles), and sequential ID", is_attribution_uncertain=True),
500
- EmailCfg(id='032581', author=MASHA_DROKOVA, attribution_reason="timing, subject (interviews/articles), and sequential ID", is_attribution_uncertain=True),
401
+ EmailCfg(
402
+ id='032604',
403
+ author=MASHA_DROKOVA,
404
+ attribution_reason='timing, subject (interviews/articles), and sequential ID',
405
+ is_attribution_uncertain=True,
406
+ ),
407
+ EmailCfg(
408
+ id='032581',
409
+ author=MASHA_DROKOVA,
410
+ attribution_reason='timing, subject (interviews/articles), and sequential ID',
411
+ is_attribution_uncertain=True,
412
+ ),
501
413
  EmailCfg(id='030235', author=MELANIE_WALKER, attribution_reason='In fwd'),
502
414
  EmailCfg(id='032343', author=MELANIE_WALKER, attribution_reason='Name seen in later reply 032346'),
503
415
  EmailCfg(id='032212', author=MIROSLAV_LAJCAK, attribution_reason='signature'),
504
- EmailCfg(id='021814', author=NADIA_MARCINKO, attribution_reason='reply'), #, actual_text="I'm a pilot...I prefer sex slave to copilot ;)"),
505
- EmailCfg(id='021808', author=NADIA_MARCINKO, attribution_reason='reply'),
506
- EmailCfg(id='022190', author=NADIA_MARCINKO, attribution_reason='reply'),
416
+ EmailCfg(id='021814', author=NADIA_MARCINKO, attribution_reason='reply'),
417
+ EmailCfg(id='021808', author=NADIA_MARCINKO, attribution_reason='reply', truncate_to=NO_TRUNCATE),
507
418
  EmailCfg(id='021818', author=NADIA_MARCINKO, attribution_reason='reply'),
508
- EmailCfg(id='022197', author=NADIA_MARCINKO, attribution_reason='reply'),
509
419
  EmailCfg(id='022214', author=NADIA_MARCINKO, attribution_reason='Reply header'),
510
420
  EmailCfg(id='021811', author=NADIA_MARCINKO, attribution_reason='Signature and email address in the message'),
511
421
  EmailCfg(id='028487', author=NORMAN_D_RAU, attribution_reason='Fwded from "to" address', duplicate_ids=['026612']),
512
422
  EmailCfg(
513
423
  id='024923',
514
- author=PAUL_KRASSNER,
515
- recipients=['George Krassner', 'Nick Kazan', 'Mrisman02', 'Rebecca Risman', 'Linda W. Grossman'],
516
- duplicate_ids=['031973']
424
+ recipients=["George Krassner", "Nick Kazan", "Mrisman02", "Rebecca Risman", "Linda W. Grossman"],
425
+ duplicate_ids=['031973'],
426
+ comment='krassner',
517
427
  ),
518
- EmailCfg(id='032457', author=PAUL_KRASSNER, recipients=[JEFFREY_EPSTEIN, 'Nancy Cain']), # Bad OCR (nofix)
428
+ EmailCfg(id='032457', author=PAUL_KRASSNER, recipients=[JEFFREY_EPSTEIN, "Nancy Cain"]),
519
429
  EmailCfg(id='029981', author=PAULA, attribution_reason='Name in reply + opera reference (Fisher now works in opera)'),
520
430
  EmailCfg(id='030482', author=PAULA, attribution_reason=PAULA_REASON),
521
431
  EmailCfg(id='033383', author=PAUL_PROSPERI, attribution_reason='Reply'),
432
+ EmailCfg(
433
+ id='012684',
434
+ author=PAUL_PROSPERI,
435
+ attribution_reason='one week after Epstein asked Prosperi for info on Kluge auction',
436
+ is_attribution_uncertain=True,
437
+ is_fwded_article=True,
438
+ ),
522
439
  EmailCfg(
523
440
  id='033561',
524
441
  author=PAUL_PROSPERI,
@@ -528,8 +445,16 @@ EMAILS_CONFIG = [
528
445
  EmailCfg(id='031694', author=PEGGY_SIEGAL, attribution_reason='quoted', is_attribution_uncertain=True),
529
446
  EmailCfg(id='032219', author=PEGGY_SIEGAL, attribution_reason='Signed "Peggy"'),
530
447
  EmailCfg(id='029020', author=RENATA_BOLOTOVA, attribution_reason='Signature'),
531
- EmailCfg(id='029605', author=RENATA_BOLOTOVA, attribution_reason=BOLOTOVA_REASON),
532
- EmailCfg(id='029606', author=RENATA_BOLOTOVA, attribution_reason=BOLOTOVA_REASON),
448
+ EmailCfg(
449
+ id='029605',
450
+ author=RENATA_BOLOTOVA,
451
+ attribution_reason=BOLOTOVA_REASON,
452
+ ),
453
+ EmailCfg(
454
+ id='029606',
455
+ author=RENATA_BOLOTOVA,
456
+ attribution_reason=BOLOTOVA_REASON,
457
+ ),
533
458
  EmailCfg(id='029604', author=RENATA_BOLOTOVA, attribution_reason='Continued in 239606 etc'),
534
459
  EmailCfg(
535
460
  id='033584',
@@ -538,66 +463,77 @@ EMAILS_CONFIG = [
538
463
  attribution_reason='Refs paper by Trivers',
539
464
  duplicate_ids=['033169'],
540
465
  ),
541
- EmailCfg(
542
- id='026320',
543
- author=SEAN_BANNON,
544
- attribution_reason="From protonmail, Bannon wrote 'just sent from my protonmail' in 027067",
545
- ),
546
- EmailCfg(id='029003', author=SOON_YI_PREVIN, attribution_reason='"Sent from Soon-Yi\'s iPhone"'),
547
- EmailCfg(id='029005', author=SOON_YI_PREVIN, attribution_reason='"Sent from Soon-Yi\'s iPhone"'),
548
- EmailCfg(id='029007', author=SOON_YI_PREVIN, attribution_reason='"Sent from Soon-Yi\'s iPhone"'),
549
- EmailCfg(id='029010', author=SOON_YI_PREVIN, attribution_reason='"Sent from Soon-Yi\'s iPhone"'),
550
- EmailCfg(id='032296', author=SOON_YI_PREVIN, attribution_reason='"Sent from Soon-Yi\'s iPhone"'),
551
- EmailCfg(id='033292', author=SOON_YI_PREVIN, attribution_reason='mentions "Woody\'s movie"', is_attribution_uncertain=True),
466
+ EmailCfg(id='026320', author=SEAN_BANNON, attribution_reason="From protonmail, Bannon wrote 'just sent from my protonmail' in 027067"),
467
+ EmailCfg(id='029003', author=SOON_YI_PREVIN, attribution_reason="\"Sent from Soon-Yi's iPhone\""),
468
+ EmailCfg(id='029005', author=SOON_YI_PREVIN, attribution_reason="\"Sent from Soon-Yi's iPhone\""),
469
+ EmailCfg(id='029007', author=SOON_YI_PREVIN, attribution_reason="\"Sent from Soon-Yi's iPhone\""),
470
+ EmailCfg(id='029010', author=SOON_YI_PREVIN, attribution_reason="\"Sent from Soon-Yi's iPhone\""),
471
+ EmailCfg(id='032296', author=SOON_YI_PREVIN, attribution_reason="\"Sent from Soon-Yi's iPhone\""),
472
+ EmailCfg(id='033292', author=SOON_YI_PREVIN, attribution_reason="mentions \"Woody's movie\"", is_attribution_uncertain=True),
552
473
  EmailCfg(
553
474
  id='019109',
554
475
  author=STEVEN_HOFFENBERG,
555
476
  recipients=["Players2"],
556
- timestamp=parse('2016-08-11 09:36:01'),
557
- attribution_reason=f"Actually a fwd by Charles Michael but {STEVEN_HOFFENBERG} email more interesting",
477
+ description='looks like a memo from Charles Michael containing copy/paste of email contents?',
478
+ date='2016-08-11 09:36:01',
479
+ subject='FW: Privileged and Confidential - Fwd: JAMES PATTERSON NEW BOOK TELLING FEDS COVER UP OF BILLIONAIRE JEFF EPSTEIN CHILD RAPES RELEASE DATE OCT 10 2016 STEVEN HOFFENBERG IS ON THE BOOK WRITING TEAM !!!!',
558
480
  ),
559
481
  EmailCfg(
560
482
  id='026620',
561
- attribution_reason='ends with "Respectfully, terry"',
562
483
  author=TERRY_KAFKA,
563
- fwded_text_after='From: Mike Cohen',
564
484
  recipients=[JEFFREY_EPSTEIN, MARK_EPSTEIN, MICHAEL_BUCHHOLTZ] + IRAN_DEAL_RECIPIENTS,
565
- subject='Fw: The Iran Nuclear Deal',
485
+ attribution_reason='ends with "Respectfully, terry"',
566
486
  duplicate_ids=['028482'],
487
+ fwded_text_after='From: Mike Cohen',
488
+ subject='Fw: The Iran Nuclear Deal',
567
489
  ),
568
490
  EmailCfg(id='029992', author=TERRY_KAFKA, attribution_reason='Quoted reply'),
569
491
  EmailCfg(id='029985', author=TERRY_KAFKA, attribution_reason='Quoted reply in 029992'),
570
492
  EmailCfg(id='020666', author=TERRY_KAFKA, attribution_reason="Ends with 'Terry'"),
571
- EmailCfg(id='026014', author=ZUBAIR_KHAN, recipients=[JEFFREY_EPSTEIN], timestamp=parse('2016-11-04 17:46:00')),
572
493
  EmailCfg(id='033021', recipients=[ANAS_ALRASHEED], attribution_reason='visible in 033022'),
573
494
  EmailCfg(id='027063', recipients=[ANTHONY_BARRETT]),
574
495
  EmailCfg(id='030764', recipients=[ARIANE_DE_ROTHSCHILD], attribution_reason='Reply'),
575
496
  EmailCfg(id='026431', recipients=[ARIANE_DE_ROTHSCHILD], attribution_reason='Reply'),
576
497
  EmailCfg(id='032876', recipients=[CECILIA_STEEN], attribution_reason='unredacted in 032267'),
577
498
  EmailCfg(id='026466', recipients=[DIANE_ZIMAN], attribution_reason='Quoted reply'),
578
- EmailCfg(id='031607', recipients=[EDWARD_JAY_EPSTEIN], attribution_reason=f"quoted reply has edwardjayepstein.com"),
499
+ EmailCfg(id='031607', recipients=[EDWARD_JAY_EPSTEIN], is_fwded_article=True, attribution_reason='quoted reply has edwardjayepstein.com'),
579
500
  EmailCfg(
580
501
  id='030525',
581
502
  recipients=[FAITH_KATES],
582
- attribution_reason=f'Reply in 030414 has {NEXT_MANAGEMENT} legal signature',
503
+ attribution_reason='Reply in 030414 has Next Management LLC legal signature',
583
504
  duplicate_ids=['030581'],
584
505
  ),
585
- EmailCfg(id='025329', recipients=['George Krassner', 'Nancy Cain', 'Tom', 'Marie Moneysmith', 'Steven Gaydos', 'Linda W. Grossman', 'Holly Krassner Dawson', 'Daniel Dawson', 'Danny Goldberg', 'Caryl Ratner', 'Kevin Bright', 'Michael Simmons', SAMUEL_LEFF, 'Bob Fass', 'Lynnie Tofte Fass', 'Barb Cowles', 'Lee Quarnstrom']),
586
- EmailCfg(id='033568', recipients=['George Krassner', 'Daniel Dawson', 'Danny Goldberg', 'Tom', 'Kevin Bright', 'Walli Leff', 'Michael Simmons', 'Lee Quarnstrom', 'Lanny Swerdlow', 'Larry Sloman', 'W&K', 'Harry Shearer', 'Jay Levin']),
506
+ EmailCfg(
507
+ id='025329',
508
+ recipients=["George Krassner", "Nancy Cain", "Tom", "Marie Moneysmith", "Steven Gaydos", "Linda W. Grossman", "Holly Krassner Dawson", "Daniel Dawson", "Danny Goldberg", "Caryl Ratner", "Kevin Bright", "Michael Simmons", SAMUEL_LEFF, "Bob Fass", "Lynnie Tofte Fass", "Barb Cowles", "Lee Quarnstrom"],
509
+ ),
510
+ EmailCfg(
511
+ id='033568',
512
+ recipients=["George Krassner", "Daniel Dawson", "Danny Goldberg", "Tom", "Kevin Bright", "Walli Leff", "Michael Simmons", "Lee Quarnstrom", "Lanny Swerdlow", "Larry Sloman", "W&K", "Harry Shearer", "Jay Levin"],
513
+ subject="Fwd: Daryl Cagle's Blog",
514
+ ),
587
515
  EmailCfg(id='026426', recipients=[JEAN_HUGUEN], attribution_reason='Reply'),
588
516
  EmailCfg(id='022202', recipients=[JEAN_LUC_BRUNEL], attribution_reason='Follow up / reply', duplicate_ids=['029975']),
589
- EmailCfg(id='022187', recipients=[JEFFREY_EPSTEIN]), # Bad OCR (nofix)
590
- EmailCfg(id='031489', recipients=[JEFFREY_EPSTEIN]), # Bad OCR (unfixable)
591
- EmailCfg(id='030347', recipients=[JEFFREY_EPSTEIN]), # Bad OCR (nofix)
592
- EmailCfg(id='030367', recipients=[JEFFREY_EPSTEIN]), # Bad OCR (nofix)
593
- EmailCfg(id='033274', recipients=[JEFFREY_EPSTEIN]), # this is a note sent to self
594
- EmailCfg(id='032780', recipients=[JEFFREY_EPSTEIN]), # Bad OCR (nofix)
595
- EmailCfg(id='029324', recipients=[JEFFREY_EPSTEIN, "Jojo Fontanilla", "Lyn Fontanilla"]), # Bad OCR (nofix)
596
- EmailCfg(id='013482', recipients=[JEFFREY_EPSTEIN], is_fwded_article=True), # other recipients redacted. "The view from the US: Stem cell therapy steps up a gear with firs"
597
- EmailCfg(id='029558', recipients=[JEFFREY_EPSTEIN, KATHERINE_KEATING], attribution_reason='BCC', fwded_text_after='Creativity is central'),
517
+ EmailCfg(id='030347', recipients=[JEFFREY_EPSTEIN]),
518
+ EmailCfg(id='033274', recipients=[JEFFREY_EPSTEIN]),
519
+ EmailCfg(id='032780', recipients=[JEFFREY_EPSTEIN]),
520
+ EmailCfg(id='033386', recipients=[JEFFREY_EPSTEIN, None], duplicate_ids=['033599']),
521
+ EmailCfg(id='029324', recipients=[JEFFREY_EPSTEIN, "Jojo Fontanilla", "Lyn Fontanilla"]),
522
+ EmailCfg(id='013482', recipients=[JEFFREY_EPSTEIN], is_fwded_article=True),
523
+ EmailCfg(
524
+ id='029558',
525
+ recipients=[JEFFREY_EPSTEIN, KATHERINE_KEATING],
526
+ attribution_reason='BCC',
527
+ fwded_text_after='Creativity is central',
528
+ ),
598
529
  EmailCfg(id='033456', recipients=["Joel"], attribution_reason='Reply'),
599
530
  EmailCfg(id='033458', recipients=["Joel"], attribution_reason='Reply'),
600
531
  EmailCfg(id='033460', recipients=["Joel"], attribution_reason='Reply'),
532
+ EmailCfg(
533
+ id='025589',
534
+ recipients=[GORDON_GETTY, JEFFREY_EPSTEIN] + TRIVERS_CCS,
535
+ attribution_reason='Reply',
536
+ ),
601
537
  EmailCfg(
602
538
  id='021090',
603
539
  recipients=[JONATHAN_FARKAS],
@@ -608,15 +544,15 @@ EMAILS_CONFIG = [
608
544
  id='033073',
609
545
  recipients=[KATHRYN_RUEMMLER],
610
546
  attribution_reason='to "Kathy" about dems, sent from iPad',
611
- is_attribution_uncertain=True, # It's actually Kathy R. as the recipient that's the uncertain part
547
+ is_attribution_uncertain=True,
612
548
  ),
613
549
  EmailCfg(
614
550
  id='032939',
615
551
  recipients=[KATHRYN_RUEMMLER],
616
552
  attribution_reason='to "Kathy" about dems, sent from iPad',
617
- is_attribution_uncertain=True, # It's actually Kathy R. as the recipient that's the uncertain part
553
+ is_attribution_uncertain=True,
618
554
  ),
619
- EmailCfg(id='030522', recipients=[LANDON_THOMAS], attribution_reason='reply header', is_fwded_article=True), # Vicky Ward article
555
+ EmailCfg(id='030522', recipients=[LANDON_THOMAS], attribution_reason='reply header', is_fwded_article=True),
620
556
  EmailCfg(id='031413', recipients=[LANDON_THOMAS], attribution_reason='reply header'),
621
557
  EmailCfg(id='033591', recipients=[LAWRANCE_VISOSKI], attribution_reason='Reply signature', duplicate_ids=['033591']),
622
558
  EmailCfg(id='027097', recipients=[LAWRANCE_VISOSKI], attribution_reason='Reply signature', duplicate_ids=['028787']),
@@ -624,113 +560,159 @@ EMAILS_CONFIG = [
624
560
  EmailCfg(id='022250', recipients=[LESLEY_GROFF], attribution_reason='Reply'),
625
561
  EmailCfg(id='030242', recipients=[MARIANA_IDZKOWSKA], duplicate_ids=['032048'], dupe_type='redacted'),
626
562
  EmailCfg(id='033027', recipients=[MASHA_DROKOVA], attribution_reason="re: PR interview, 031544 says she'll be in NY at that time"),
627
- EmailCfg(id='033025', recipients=[MASHA_DROKOVA], attribution_reason="timing, subject (interviews/articles), and sequential ID", is_attribution_uncertain=True),
563
+ EmailCfg(
564
+ id='033025',
565
+ recipients=[MASHA_DROKOVA],
566
+ attribution_reason='timing, subject (interviews/articles), and sequential ID',
567
+ is_attribution_uncertain=True,
568
+ ),
628
569
  EmailCfg(id='030368', recipients=[MELANIE_SPINELLA], attribution_reason='Actually a self fwd from jeffrey to jeffrey'),
629
570
  EmailCfg(id='030369', recipients=[MELANIE_SPINELLA], attribution_reason='Actually a self fwd from jeffrey to jeffrey'),
630
571
  EmailCfg(id='030371', recipients=[MELANIE_SPINELLA], attribution_reason='Actually a self fwd from jeffrey to jeffrey'),
631
572
  EmailCfg(id='022258', recipients=[NADIA_MARCINKO], attribution_reason='Reply header'),
632
573
  EmailCfg(id='022193', recipients=[NADIA_MARCINKO], attribution_reason='reply'),
574
+ EmailCfg(id='022197', recipients=[NADIA_MARCINKO], attribution_reason='reply'),
575
+ EmailCfg(id='022190', recipients=[NADIA_MARCINKO], attribution_reason='reply'),
633
576
  EmailCfg(id='030572', recipients=[PAULA], attribution_reason='quoted in 030482', is_attribution_uncertain=True),
634
- EmailCfg(id='030506', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True),
635
- EmailCfg(id='030507', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True),
636
- EmailCfg(id='030508', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True),
637
- EmailCfg(id='030509', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True),
577
+ EmailCfg(id='030506', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True,),
578
+ EmailCfg(id='030507', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True,),
579
+ EmailCfg(id='030508', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True,),
580
+ EmailCfg(id='030509', recipients=[PAULA], attribution_reason=PAULA_REASON, is_attribution_uncertain=True,),
638
581
  EmailCfg(id='030096', recipients=[PETER_MANDELSON], attribution_reason='reply header'),
639
582
  EmailCfg(id='032951', recipients=[RAAFAT_ALSABBAGH, None], attribution_reason='Redacted'),
640
583
  EmailCfg(id='029581', recipients=[RENATA_BOLOTOVA], attribution_reason=BOLOTOVA_REASON),
641
584
  EmailCfg(id='019334', recipients=[STEVE_BANNON], attribution_reason='quoted reply'),
642
585
  EmailCfg(id='021106', recipients=[STEVE_BANNON], attribution_reason='Reply'),
643
-
644
- # Misc configs
645
- EmailCfg(id='029344', actual_text='I thought of you when I read this article. Was this your idea? Alan'),
646
- EmailCfg(id='032358', actual_text=REDACTED), # Completely redacted
647
- EmailCfg(id='033050', actual_text='schwartman'),
648
- EmailCfg(id='031036', description=f"{BARBRO_C_EHNBOM} related donation and Swedish girls discussion"),
649
- EmailCfg(id='022219', description="discussion of attempts to clean up Epstein's Google search results"),
586
+ EmailCfg(id='029344', actual_text='I thought of you when I read this article. Was this your idea? Alan', is_fwded_article=True),
587
+ EmailCfg(id='032358', actual_text=REDACTED),
588
+ EmailCfg(
589
+ id='031036',
590
+ description='Barbro C. Ehnbom related donation and Swedish girls discussion',
591
+ is_interesting=True,
592
+ comment='Barbro Ehnbom talking about Swedish girl',
593
+ ),
594
+ EmailCfg(id='032946', description='discussion of obtaining a Moroccan visa for an unnamed woman', is_interesting=True),
595
+ EmailCfg(id='031320', description='discussion of routing Gratitude foundation money through Peggy Siegal', is_interesting=True),
596
+ EmailCfg(id='023627', description=MICHAEL_WOLFF_EPSTEIN_ARTICLE_DRAFT, is_fwded_article=True, is_interesting=True, truncate_to=16800),
597
+ EmailCfg(id='032671', description='connections to Fusion GPS who produced the Steele Dossier on Trump'),
598
+ EmailCfg(id='033052', description='connections to Fusion GPS who produced the Steele Dossier on Trump'),
599
+ EmailCfg(id='022219', description="discussion of attempts to clean up Epstein's Google search results", truncate_to=2404),
600
+ EmailCfg(id='031472', description="discussion of publishing an audio recording related to the coup in Turkey"),
601
+ EmailCfg(id='031333', description='Fort Knox conspiracy theory, looks like a Russian disinfo article', is_fwded_article=True),
602
+ EmailCfg(id='031335', description='Fort Knox conspiracy theory, looks like a Russian disinfo article', is_fwded_article=True),
650
603
  EmailCfg(id='030648', description="is the 'roger' Epstein is trying to meet Roger Stone?"),
651
604
  EmailCfg(id='030762', description="is the 'roger' Epstein is trying to meet Roger Stone?"),
652
605
  EmailCfg(id='030649', description="is the 'roger' Epstein is trying to meet Roger Stone?"),
653
606
  EmailCfg(id='026026', description="is the 'roger' Epstein is trying to meet Roger Stone?"),
654
607
  EmailCfg(id='026030', description="is the 'roger' Epstein is trying to meet Roger Stone?"),
655
608
  EmailCfg(id='026033', description="is the 'roger' Epstein is trying to meet Roger Stone?"),
656
- EmailCfg(id='031320', description=f"Epstein and {RICHARD_KAHN} appear to be discussing routing donatings through {PEGGY_SIEGAL}"),
609
+ EmailCfg(id='031011', description='jokes about Chicago corruption', duplicate_ids=['031090'], truncate_to=TRUNCATED_CHARS),
657
610
  EmailCfg(id='016693', description='signed "MM"'),
658
- EmailCfg(id='028524', is_fwded_article=True, description='Zach Braff op-ed on Woody Allen in NYT'),
659
- EmailCfg(id='031333', is_fwded_article=True, description='Fort Knox conspiracy theory, looks like a Russian disinfo article'), # Russia Says IMF Chief Jailed For Discovering All US Gold is Gone
660
- EmailCfg(id='031335', is_fwded_article=True, description='Fort Knox conspiracy theory, looks like a Russian disinfo article'), # DOMINQUE STRAUSS-KAHN ARRESTED, NOT BECAUSE HE RAPED A MAID, BUT BECAUSE HE HAD EVIDENCE US HAS NO GOLD IN FORT KNOX.
661
- EmailCfg(id='023627', is_fwded_article=True, description=MICHAEL_WOLFF_EPSTEIN_ARTICLE_DRAFT),
662
- EmailCfg(id='026298', is_fwded_article=True, duplicate_ids=['026499']), # Written by someone else?
663
- EmailCfg(id='029692', is_fwded_article=True, duplicate_ids=['029779']), # WaPo article
664
- EmailCfg(id='022344', is_fwded_article=True, duplicate_ids=['028529']), # Bill Gates is most admired from Nikolic
665
- EmailCfg(id='018197', is_fwded_article=True, duplicate_ids=['028648']), # Ray Takeyh article fwd
666
- EmailCfg(id='028728', is_fwded_article=True, duplicate_ids=['027102']), # WSJ forward to Larry Summers
667
- EmailCfg(id='028781', is_fwded_article=True, duplicate_ids=['013460']), # Atlantic on Jim Yong Kim, Obama's World Bank Pick
668
- EmailCfg(id='025041', is_fwded_article=True, duplicate_ids=['028675']), # Obama agenda
669
- EmailCfg(id='031136', is_fwded_article=True, duplicate_ids=['028791']), # 'Smart Money is Fleeing US Stocks'
670
- EmailCfg(id='031779', is_fwded_article=True, duplicate_ids=['026938']), # Sarah Silverman on AI
671
- EmailCfg(id='029849', is_fwded_article=True, duplicate_ids=['033482']), # Fareed Zakaria: Trump sells America short),
672
- EmailCfg(id='032023', is_fwded_article=True, duplicate_ids=['032012']), # American-Israeli Cooperative Enterprise Newsletter
673
- EmailCfg(id='021758', is_fwded_article=True, duplicate_ids=['030616']), # Radar Online article about Epstein's early prison release
674
- EmailCfg(id='033297', is_fwded_article=True, duplicate_ids=['033586']), # Sultan Sulayem fwding article about Trump and Russia
675
- EmailCfg(id='026829', is_fwded_article=True), # Taxes
676
- EmailCfg(id='020443', is_fwded_article=True), # WSJ Deplorables Bannon
677
- EmailCfg(id='030372', is_fwded_article=True), # Bannon China Iran
678
- EmailCfg(id='030983', is_fwded_article=True), # Power Line blog Alex Acosta and Jeffrey Epstein Plea Deal Analysis
679
- EmailCfg(id='031774', is_fwded_article=True), # Krassner fwd of Palmer Report article
680
- EmailCfg(id='033345', is_fwded_article=True), # Krassner fwd of Palmer Report article
681
- EmailCfg(id='029903', is_fwded_article=True), # Krassner fwd of Ann Coulter article about Epstein
682
- EmailCfg(id='030266', is_fwded_article=True), # Krassner fwd of article about Dershowitz
683
- EmailCfg(id='030868', is_fwded_article=True), # 'He doesn't like this sh*t': Trump reportedly hates his job and his staff after 1 month
684
- EmailCfg(id='026755', is_fwded_article=True), # HuffPo
685
- EmailCfg(id='016218', is_fwded_article=True), # AT&T confirms it paid Trump lawyer Cohen for insights on Trump
686
- EmailCfg(id='030528', is_fwded_article=True), # Vicky Ward article
687
- EmailCfg(id='030460', is_fwded_article=True), # Vicky Ward article
688
- EmailCfg(id='028508', is_fwded_article=True), # nanosatellites article
689
- EmailCfg(id='019845', is_fwded_article=True), # Pro Publica article on Preet Bharara
690
- EmailCfg(id='029021', is_fwded_article=True), # article about bannon sent by Alain Forget
691
- EmailCfg(id='031688', is_fwded_article=True), # Bill Siegel fwd of email about hamas
692
- EmailCfg(id='026551', is_fwded_article=True), # Sultan bin Sulayem "Ayatollah between the sheets"
693
- EmailCfg(id='031768', is_fwded_article=True), # Sultan bin Sulayem 'Horseface'
694
- EmailCfg(id='031569', is_fwded_article=True), # Article by Kathryn Alexeeff fwded to Peter Thiel
695
- EmailCfg(id='029689', is_fwded_article=True), # Tunisia article to Larry Summers
696
- EmailCfg(id='014525', is_fwded_article=True), # Really more of a mailing list from Paul Morris?
697
- EmailCfg(id='024384', is_fwded_article=True), # Interview with Bill Siegal re: Islam
698
- EmailCfg(id='030200', is_fwded_article=True), # Lawfare indicting a president
699
- EmailCfg(id='029509', is_fwded_article=True), # Deepak Chopra LSD, Quantum Healing
700
- EmailCfg(id='026778', is_fwded_article=True), # tax alert
701
- EmailCfg(id='023001', is_fwded_article=True), # Miami Herald article timeline of the sex abuse case
702
- EmailCfg(id='013405', is_fwded_article=True), # Articles about epstein case
703
- EmailCfg(id='021740', is_fwded_article=True), # Miami Herald article about Epstein prosecutor
704
- EmailCfg(id='023126', is_fwded_article=True), # Miami Herald on Alex Acosta
705
- EmailCfg(id='029625', is_fwded_article=True), # Conchita Sarnoff Daily Beast Articles - Epstein Sex Trafficking Investigation and Settlement
706
- EmailCfg(id='029505', is_fwded_article=True), # Foreign Policy Middle Eastern Monarchs Look at the Trump
707
- EmailCfg(id='029859', is_fwded_article=True), # Palm Beach Post: Epstein paid three women $5.5 million to end lawsuits
708
- EmailCfg(id='031988', is_fwded_article=True), # NYT review of Inside Job
709
- EmailCfg(id='029901', is_fwded_article=True), # THE EDGE question
710
- EmailCfg(id='031399', is_fwded_article=True), # Miami U.S. Attorney's Office recuses itself from Jeffrey Epstein case
711
- EmailCfg(id='031705', is_fwded_article=True), # Thomas Friedman why not in vegas?
712
- EmailCfg(id='016801', is_fwded_article=True), # Capital Market Outlook
713
- EmailCfg(id='023564', is_fwded_article=True), # BBG ;Leon Black's Tax-Overhaul Dilemma Could Alter Wall Street Model
714
- EmailCfg(id='025231', is_fwded_article=True), # Newsmax: Laffer, Laffer: Obama Must Use Reaganomics to Save Economy The only way President Barack Obama can solve
715
- EmailCfg(id='031472', is_fwded_article=True), # WSJ: Lawyers for Imam Wanted by Turkish authorities Fear for Their Client's Life
716
- EmailCfg(id='012684', is_fwded_article=True), # Trump in talks to buy socialite Kluge's Charlottesville vineyard
717
- EmailCfg(id='028536', is_fwded_article=True), # Palm Beach Post FBI Epstein files say he gave info. Does it explain sweetheart deal?
718
- EmailCfg(id='030326', is_fwded_article=True), # NYP Congressional candidate compares Melania Trump to prostitute
719
- EmailCfg(id='030519', is_fwded_article=True), # Daily Mail on Prince Andrew
720
- EmailCfg(id='030878', is_fwded_article=True), # Steve Bannon almost appeared in Michael Moore's 'Fahrenheit 11/9'
721
- EmailCfg(id='024300', is_fwded_article=True), # Bookstore owner calls police after customer confronted Steve Bannon
722
- EmailCfg(id='026924', is_fwded_article=True), # The Onion
723
- EmailCfg(id='033311', is_fwded_article=True), # 2016 election polls
724
- EmailCfg(id='026580', is_fwded_article=True), # NPR: Antigua: Land Of Sun, Sand, And Super Cheap
725
- EmailCfg(id='031340', is_fwded_article=True), # Article about Alex Jones threatening Robert Mueller
726
- EmailCfg(id='030209', is_fwded_article=True), # Atlantic Council Syria: Blackberry Diplomacy
727
- EmailCfg(id='026605', is_fwded_article=True), # Article about Ruemmler turning down attorney general job by NEDRA PICKLER
728
- EmailCfg(id='031990', is_fwded_article=True), # newsmax on ken starr
729
- EmailCfg(id='029433', is_fwded_article=True), # Estate Planning After the Enactment of the Tax Cuts and Jobs Act
730
- EmailCfg(id='032475', timestamp=parse('2017-02-15 13:31:25')),
731
- EmailCfg(id='030373', timestamp=parse('2018-10-03 01:49:27')),
732
-
733
- # Configure duplicates
611
+ EmailCfg(id='028524', description='Zach Braff op-ed on Woody Allen in NYT', is_fwded_article=True),
612
+ EmailCfg(id='026298', duplicate_ids=['026499'], is_fwded_article=True, comment='Written by someone else?'),
613
+ EmailCfg(id='029692', duplicate_ids=['029779'], is_fwded_article=True, comment='WaPo article'),
614
+ EmailCfg(id='022344', duplicate_ids=['028529'], is_fwded_article=True, comment='Bill Gates is most admired from Nikolic'),
615
+ EmailCfg(id='018197', duplicate_ids=['028648'], is_fwded_article=True, comment='Ray Takeyh article fwd'),
616
+ EmailCfg(id='028728', duplicate_ids=['027102'], is_fwded_article=True, comment='WSJ forward to Larry Summers'),
617
+ EmailCfg(id='028781', duplicate_ids=['013460'], is_fwded_article=True, comment="Atlantic on Jim Yong Kim, Obama's World Bank Pick"),
618
+ EmailCfg(id='025041', duplicate_ids=['028675'], is_fwded_article=True, comment='Obama agenda'),
619
+ EmailCfg(id='031136', duplicate_ids=['028791'], is_fwded_article=True, comment="'Smart Money is Fleeing US Stocks'"),
620
+ EmailCfg(id='031779', duplicate_ids=['026938'], is_fwded_article=True, comment='Sarah Silverman on AI'),
621
+ EmailCfg(id='029849', duplicate_ids=['033482'], is_fwded_article=True, comment='Fareed Zakaria: Trump sells America short),'),
622
+ EmailCfg(id='032023', duplicate_ids=['032012'], is_fwded_article=True, comment='American-Israeli Cooperative Enterprise Newsletter'),
623
+ EmailCfg(id='021758', duplicate_ids=['030616'], is_fwded_article=True, comment="Radar Online article about Epstein's early prison release"),
624
+ EmailCfg(id='033297', duplicate_ids=['033586'], is_fwded_article=True, comment='Sultan Sulayem article about Trump and Russia'),
625
+ EmailCfg(id='031428', duplicate_ids=['031388'], is_fwded_article=True),
626
+ EmailCfg(id='033528', duplicate_ids=['033517'], is_fwded_article=True),
627
+ EmailCfg(id='030238', duplicate_ids=['031130'], is_fwded_article=True),
628
+ EmailCfg(id='031112', duplicate_ids=['030876'], is_fwded_article=True),
629
+ EmailCfg(id='031423', duplicate_ids=['025361'], is_fwded_article=True),
630
+ EmailCfg(id='026755', is_fwded_article=True, truncate_to=TRUNCATED_CHARS, comment='Epstein self fwd'),
631
+ EmailCfg(id='026778', is_fwded_article=True, truncate_to=TRUNCATED_CHARS, comment='Kahn taxes'),
632
+ EmailCfg(id='032458', truncate_to=NO_TRUNCATE, description='discussion of acquiring pieces for Epstein\'s art collection'),
633
+ EmailCfg(id='032464', truncate_to=NO_TRUNCATE, description='redacted discussion about art advisor Etienne Binant'),
634
+ # TODO: just use a search string for truncation
635
+ EmailCfg(id='032964', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
636
+ EmailCfg(id='032968', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
637
+ EmailCfg(id='032467', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
638
+ EmailCfg(id='032480', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
639
+ EmailCfg(id='032486', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
640
+ EmailCfg(id='032491', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
641
+ EmailCfg(id='032975', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
642
+ EmailCfg(id='032496', truncate_to=700, description='redacted discussion about art advisor Etienne Binant'),
643
+ EmailCfg(id='025643', is_fwded_article=True, comment='Alice Fisher stated to be in the mix of potential candidates for new FBI director.'),
644
+ EmailCfg(id='014523', is_fwded_article=True, comment='finance research'),
645
+ EmailCfg(id='014857', is_fwded_article=True, comment='finance research'),
646
+ EmailCfg(id='029458', is_fwded_article=True, comment='finance research'),
647
+ EmailCfg(id='026893', is_fwded_article=True, comment='finance research'),
648
+ EmailCfg(id='033362', is_fwded_article=True, comment='finance research'),
649
+ EmailCfg(id='030865', is_fwded_article=True, comment='Deutsche bank research'),
650
+ EmailCfg(id='021231', is_fwded_article=True, comment='11 places with worse economy than US'),
651
+ EmailCfg(id='029905', is_fwded_article=True, comment='Ann Coulter'),
652
+ EmailCfg(id='026829', is_fwded_article=True, comment='Taxes'),
653
+ EmailCfg(id='020443', is_fwded_article=True, comment='WSJ Deplorables Bannon'),
654
+ EmailCfg(id='030372', is_fwded_article=True, comment='Bannon China Iran'),
655
+ EmailCfg(id='030983', is_fwded_article=True, comment='Power Line blog Alex Acosta and Jeffrey Epstein Plea Deal Analysis'),
656
+ EmailCfg(id='031774', is_fwded_article=True, comment='Krassner fwd of Palmer Report article'),
657
+ EmailCfg(id='033345', is_fwded_article=True, comment='Krassner fwd of Palmer Report article'),
658
+ EmailCfg(id='029903', is_fwded_article=True, comment='Krassner fwd of Ann Coulter article about Epstein'),
659
+ EmailCfg(id='030266', is_fwded_article=True, comment='Krassner fwd of article about Dershowitz'),
660
+ EmailCfg(id='030868', is_fwded_article=True, comment="'He doesn't like this sh*t': Trump reportedly hates his job and his staff after 1 month"),
661
+ EmailCfg(id='016218', is_fwded_article=True, comment='AT&T confirms it paid Trump lawyer Cohen for insights on Trump'),
662
+ EmailCfg(id='030528', is_fwded_article=True, comment='Vicky Ward article'),
663
+ EmailCfg(id='030460', is_fwded_article=True, comment='Vicky Ward article'),
664
+ EmailCfg(id='028508', is_fwded_article=True, comment='nanosatellites article'),
665
+ EmailCfg(id='019845', is_fwded_article=True, comment='Pro Publica article on Preet Bharara'),
666
+ EmailCfg(id='029021', is_fwded_article=True, comment='article about bannon sent by Alain Forget'),
667
+ EmailCfg(id='031688', is_fwded_article=True, comment='Bill Siegel fwd of email about hamas'),
668
+ EmailCfg(id='026551', is_fwded_article=True, comment='Sultan bin Sulayem "Ayatollah between the sheets"'),
669
+ EmailCfg(id='031768', is_fwded_article=True, comment="Sultan bin Sulayem 'Horseface'"),
670
+ EmailCfg(id='031569', is_fwded_article=True, comment='Article by Kathryn Alexeeff fwded to Peter Thiel'),
671
+ EmailCfg(id='029689', is_fwded_article=True, comment='Tunisia article to Larry Summers'),
672
+ EmailCfg(id='014525', is_fwded_article=True, comment='Really more of a mailing list from Paul Morris?'),
673
+ EmailCfg(id='024384', is_fwded_article=True, comment='Interview with Bill Siegal re: Islam'),
674
+ EmailCfg(id='030200', is_fwded_article=True, comment='Lawfare indicting a president'),
675
+ EmailCfg(id='029509', is_fwded_article=True, comment='Deepak Chopra LSD, Quantum Healing'),
676
+ EmailCfg(id='023001', is_fwded_article=True, comment='Miami Herald article timeline of the sex abuse case'),
677
+ EmailCfg(id='013405', is_fwded_article=True, comment='Articles about epstein case'),
678
+ EmailCfg(id='021740', is_fwded_article=True, comment='Miami Herald article about Epstein prosecutor'),
679
+ EmailCfg(id='023126', is_fwded_article=True, comment='Miami Herald on Alex Acosta'),
680
+ EmailCfg(id='029625', is_fwded_article=True, comment='Conchita Sarnoff Daily Beast Articles - Epstein Sex Trafficking Investigation and Settlement'),
681
+ EmailCfg(id='029505', is_fwded_article=True, comment='Foreign Policy Middle Eastern Monarchs Look at the Trump'),
682
+ EmailCfg(id='029859', is_fwded_article=True, comment='Palm Beach Post: Epstein paid three women $5.5 million to end lawsuits'),
683
+ EmailCfg(id='031988', is_fwded_article=True, comment='NYT review of Inside Job'),
684
+ EmailCfg(id='029901', is_fwded_article=True, comment='THE EDGE question'),
685
+ EmailCfg(id='031399', is_fwded_article=True, comment="Miami U.S. Attorney's Office recuses itself from Jeffrey Epstein case"),
686
+ EmailCfg(id='031705', is_fwded_article=True, comment='Thomas Friedman why not in vegas?'),
687
+ EmailCfg(id='016801', is_fwded_article=True, comment='Capital Market Outlook'),
688
+ EmailCfg(id='023564', is_fwded_article=True, comment="BBG ;Leon Black's Tax-Overhaul Dilemma Could Alter Wall Street Model"),
689
+ EmailCfg(id='025231', is_fwded_article=True, comment='Newsmax: Laffer, Laffer: Obama Must Use Reaganomics to Save Economy The only way President Barack Obama can solve'),
690
+ EmailCfg(id='028536', is_fwded_article=True, comment='Palm Beach Post FBI Epstein files say he gave info. Does it explain sweetheart deal?'),
691
+ EmailCfg(id='030326', is_fwded_article=True, comment='NYP Congressional candidate compares Melania Trump to prostitute'),
692
+ EmailCfg(id='030519', is_fwded_article=True, comment='Daily Mail on Prince Andrew'),
693
+ EmailCfg(id='030878', is_fwded_article=True, comment="Steve Bannon almost appeared in Michael Moore's 'Fahrenheit 11/9'"),
694
+ EmailCfg(id='024300', is_fwded_article=True, comment='Bookstore owner calls police after customer confronted Steve Bannon'),
695
+ EmailCfg(id='026924', is_fwded_article=True, comment='The Onion'),
696
+ EmailCfg(id='022624', is_fwded_article=True, comment='Disgusting: Clinton Snared In Pedophile Ring'),
697
+ EmailCfg(id='022673', is_fwded_article=True, comment='Epstein + Clinton'),
698
+ EmailCfg(id='033301', is_fwded_article=True, comment='description of Trump jet'),
699
+ EmailCfg(id='021729', is_fwded_article=True, comment='Acosta rebuke'),
700
+ EmailCfg(id='023635', is_fwded_article=True, comment='Landon Thomas finance related'),
701
+ EmailCfg(id='026637', is_fwded_article=True, comment='Landon Thomas finance related'),
702
+ EmailCfg(id='021764', is_fwded_article=True, comment='He was 50 and they were girls'),
703
+ EmailCfg(id='033311', is_fwded_article=True, truncate_to=TRUNCATED_CHARS, comment='Kahn taxes'),
704
+ EmailCfg(id='026580', is_fwded_article=True, comment='NPR: Antigua: Land Of Sun, Sand, And Super Cheap'),
705
+ EmailCfg(id='031340', is_fwded_article=True, comment='Article about Alex Jones threatening Robert Mueller'),
706
+ EmailCfg(id='030209', is_fwded_article=True, comment='Atlantic Council Syria: Blackberry Diplomacy'),
707
+ EmailCfg(id='026605', is_fwded_article=True, comment='Article about Ruemmler turning down attorney general job by NEDRA PICKLER'),
708
+ EmailCfg(id='031990', is_fwded_article=True, comment='newsmax on ken starr'),
709
+ EmailCfg(id='029433', is_fwded_article=True, truncate_to=TRUNCATED_CHARS, comment='Kahn taxes'),
710
+ EmailCfg(id='030927', is_fwded_article=True, comment="don't talk to the FBI"),
711
+ EmailCfg(id='033329', is_fwded_article=True, comment='Chinese economists'),
712
+ EmailCfg(id='031764', is_fwded_article=True, truncate_to=3500, comment='broidy malaysia'),
713
+ EmailCfg(id='032475', date='2017-02-15 13:31:25'),
714
+ EmailCfg(id='030373', date='2018-10-03 01:49:27'),
715
+ EmailCfg(id='032325', duplicate_ids=['026014'], dupe_type='quoted', is_interesting=True, comment='Zubair'),
734
716
  EmailCfg(id='026631', duplicate_ids=['026632'], dupe_type='quoted'),
735
717
  EmailCfg(id='028768', duplicate_ids=['026563'], dupe_type='redacted'),
736
718
  EmailCfg(id='027056', duplicate_ids=['028762'], dupe_type='redacted'),
@@ -749,19 +731,16 @@ EMAILS_CONFIG = [
749
731
  EmailCfg(id='033512', duplicate_ids=['033361']),
750
732
  EmailCfg(id='030299', duplicate_ids=['021794']),
751
733
  EmailCfg(id='033575', duplicate_ids=['012898']),
752
- EmailCfg(id='031428', is_fwded_article=True, duplicate_ids=['031388']),
753
734
  EmailCfg(id='031980', duplicate_ids=['019409']),
754
735
  EmailCfg(id='033486', duplicate_ids=['033156']),
755
736
  EmailCfg(id='025790', duplicate_ids=['031994']),
756
737
  EmailCfg(id='028497', duplicate_ids=['026228']),
757
- EmailCfg(id='033528', is_fwded_article=True, duplicate_ids=['033517']),
758
738
  EmailCfg(id='019412', duplicate_ids=['028621']),
759
739
  EmailCfg(id='027053', duplicate_ids=['028765']),
760
740
  EmailCfg(id='027049', duplicate_ids=['028773']),
761
741
  EmailCfg(id='033580', duplicate_ids=['033207']),
762
742
  EmailCfg(id='028506', duplicate_ids=['025547']),
763
- EmailCfg(id='028784', duplicate_ids=['026549']),
764
- EmailCfg(id='033386', duplicate_ids=['033599']),
743
+ EmailCfg(id='028784', duplicate_ids=['026549'], is_interesting=True, comment='seminars: Money / Power'),
765
744
  EmailCfg(id='023024', duplicate_ids=['030622']),
766
745
  EmailCfg(id='030618', duplicate_ids=['023026']),
767
746
  EmailCfg(id='028780', duplicate_ids=['026834']),
@@ -772,15 +751,14 @@ EMAILS_CONFIG = [
772
751
  EmailCfg(id='030587', duplicate_ids=['030514']),
773
752
  EmailCfg(id='031089', duplicate_ids=['018084']),
774
753
  EmailCfg(id='031088', duplicate_ids=['030885']),
775
- EmailCfg(id='030238', duplicate_ids=['031130']),
776
754
  EmailCfg(id='030859', duplicate_ids=['031067']),
777
755
  EmailCfg(id='030635', duplicate_ids=['031134']),
778
- EmailCfg(id='028494', duplicate_ids=['026234']),
756
+ EmailCfg(id='028494', duplicate_ids=['026234'], truncate_to=NO_TRUNCATE, comment='Email about being in palm beach w/trump people'),
779
757
  EmailCfg(id='030311', duplicate_ids=['021790']),
780
758
  EmailCfg(id='033508', duplicate_ids=['029880']),
781
759
  EmailCfg(id='030493', duplicate_ids=['030612']),
782
760
  EmailCfg(id='032051', duplicate_ids=['031771']),
783
- EmailCfg(id='031217', duplicate_ids=['021761']),
761
+ EmailCfg(id='031217', duplicate_ids=['021761'], truncate_to=1800, comment='1st email for dersh, has long article'),
784
762
  EmailCfg(id='031346', duplicate_ids=['031426']),
785
763
  EmailCfg(id='031345', duplicate_ids=['031427']),
786
764
  EmailCfg(id='031343', duplicate_ids=['031432']),
@@ -806,12 +784,10 @@ EMAILS_CONFIG = [
806
784
  EmailCfg(id='027032', duplicate_ids=['028531']),
807
785
  EmailCfg(id='026777', duplicate_ids=['028493']),
808
786
  EmailCfg(id='029837', duplicate_ids=['029255']),
809
- EmailCfg(id='031423', duplicate_ids=['025361']),
810
787
  EmailCfg(id='029299', duplicate_ids=['033594']),
811
788
  EmailCfg(id='030904', duplicate_ids=['031069']),
812
789
  EmailCfg(id='030006', duplicate_ids=['031165']),
813
790
  EmailCfg(id='025215', duplicate_ids=['031159']),
814
- EmailCfg(id='031011', duplicate_ids=['031090']),
815
791
  EmailCfg(id='032068', duplicate_ids=['018158']),
816
792
  EmailCfg(id='031213', duplicate_ids=['031221']),
817
793
  EmailCfg(id='016595', duplicate_ids=['016690']),
@@ -827,7 +803,6 @@ EMAILS_CONFIG = [
827
803
  EmailCfg(id='028620', duplicate_ids=['027094']),
828
804
  EmailCfg(id='032456', duplicate_ids=['033579']),
829
805
  EmailCfg(id='030315', duplicate_ids=['030255']),
830
- EmailCfg(id='031112', duplicate_ids=['030876']),
831
806
  EmailCfg(id='030614', duplicate_ids=['030491']),
832
807
  EmailCfg(id='033585', duplicate_ids=['032279']),
833
808
  EmailCfg(id='031220', duplicate_ids=['031189']),
@@ -835,22 +810,21 @@ EMAILS_CONFIG = [
835
810
  EmailCfg(id='033230', duplicate_ids=['033577']),
836
811
  EmailCfg(id='032125', duplicate_ids=['023971']),
837
812
  EmailCfg(id='031230', duplicate_ids=['031203']),
838
- EmailCfg(id='028752', duplicate_ids=['026569']),
813
+ EmailCfg(id='028752', duplicate_ids=['026569'], subject='Re: Program & Attendee list'),
839
814
  EmailCfg(id='031773', duplicate_ids=['032050']),
840
815
  EmailCfg(id='021400', duplicate_ids=['031983']),
841
- EmailCfg(id='026548', duplicate_ids=['033491']),
816
+ EmailCfg(id='026548', duplicate_ids=['033491', '033495'], comment='033495 is HTML garbage version'),
842
817
  EmailCfg(id='029752', duplicate_ids=['023550']),
843
818
  EmailCfg(id='030339', duplicate_ids=['030592']),
844
- EmailCfg(id='032250', duplicate_ids=['033589']),
845
-
846
- # Emails that need a little help determining how to separate the actual text from fwded text
819
+ EmailCfg(id='032250', duplicate_ids=['033589'], truncate_to=1000, comment='Wolff article'),
847
820
  EmailCfg(id='013415', fwded_text_after='Darren K. Indyke'),
848
821
  EmailCfg(id='024624', fwded_text_after='On Tue, May 14'),
849
- EmailCfg(id='025888', fwded_text_after='Jul 24, 2015'),
822
+ EmailCfg(id='025888', fwded_text_after='Jul 24, 2015', comment='sultan asking about cholesterol drug'),
823
+ EmailCfg(id='026362', fwded_text_after='Jul 24, 2015', comment='sultan asking about cholesterol drug'),
850
824
  EmailCfg(id='016413', fwded_text_after='In a former warehouse'),
851
825
  EmailCfg(id='025548', fwded_text_after='Edward Jay Epstein'),
852
826
  EmailCfg(id='032806', fwded_text_after='• Sep 13, 2018'),
853
- EmailCfg(id='024251', fwded_text_after='Debate Schedule'),
827
+ EmailCfg(id='024251', fwded_text_after='Debate Schedule', truncate_to=TRUNCATED_CHARS, comment='Kahn taxes'),
854
828
  EmailCfg(id='028943', fwded_text_after='-Lisa'),
855
829
  EmailCfg(id='029431', fwded_text_after='I am writing now'),
856
830
  EmailCfg(id='020437', fwded_text_after='Will Cohen Cooperate'),
@@ -859,7 +833,7 @@ EMAILS_CONFIG = [
859
833
  EmailCfg(id='030324', fwded_text_after='For Federal Programs'),
860
834
  EmailCfg(id='022766', fwded_text_after='--- On Wed, 4/22/15'),
861
835
  EmailCfg(id='025606', fwded_text_after='> On May 6,'),
862
- EmailCfg(id='022977', fwded_text_after='Top of Form'),
836
+ EmailCfg(id='022977', fwded_text_after='Top of Form', truncate_to=1800, comment='Krassner with huge attachments field'),
863
837
  EmailCfg(id='033420', fwded_text_after='Slowing economy could increase pressure on'),
864
838
  EmailCfg(id='019203', fwded_text_after='This end-of-the-year'),
865
839
  EmailCfg(id='022207', fwded_text_after='Web Images Videos Maps'),
@@ -871,9 +845,131 @@ EMAILS_CONFIG = [
871
845
  EmailCfg(id='031742', fwded_text_after="Trump's former campaign manager Paul Manafort"),
872
846
  EmailCfg(id='028925', fwded_text_after='> on Jan 4, 2015'),
873
847
  EmailCfg(id='029773', fwded_text_after='Omar Quadhafi', duplicate_ids=['012685']),
874
- EmailCfg(id='012197_4', fwded_text_after="Thanks -- Jay"),
848
+ EmailCfg(id='012197_4', fwded_text_after='Thanks -- Jay', is_fwded_article=False),
849
+ EmailCfg(id='026449', description="comments about Trump's 'former bridge girl and toy'", is_interesting=True),
850
+ EmailCfg(id='030727', description='Epstein requesting help for Russia and discussion of an unknown dead Chinese shareholder', is_interesting=True),
851
+ EmailCfg(id='033377', description="Epstein says he knows Russian oligarch Oleg Deripaska", is_interesting=True),
852
+ EmailCfg(id='029098', description=f"{MASHA_DROKOVA}'s skype contact is in Moscow", is_interesting=True),
853
+ EmailCfg(id='032229', description='Michael Wolff offers ideas for entrapping Trump', is_interesting=True),
854
+ EmailCfg(id='023454', description='Email invitation sent to tech CEOs and Epstein', is_interesting=True, truncate_to=1878),
855
+ EmailCfg(id='032842', description='discussion of "stealing" a casino in Atlantic City', is_interesting=True),
856
+ EmailCfg(id='030630', is_interesting=True, comment="'What happens with zubair's project?'"),
857
+ EmailCfg(id='033178', is_interesting=True, comment="'How is it going with Zubair?'"),
858
+ EmailCfg(id='022396', is_interesting=True, comment='Ukraine friend'),
859
+ EmailCfg(id='026505', is_interesting=True, comment='I know how dirty trump is'),
860
+ EmailCfg(id='029679', is_interesting=True, comment="Trump's driver was the bag man"),
861
+ EmailCfg(id='026258', is_interesting=True, comment='Bannon cripto coin issues'),
862
+ EmailCfg(id='026260', is_interesting=True, comment='Bannon cripto coin issues'),
863
+ EmailCfg(id='032359', is_interesting=True, comment='Jabor e-currency'),
864
+ EmailCfg(id='031451', is_interesting=True, comment='"would you like photso of donald and girls in bikinis in my kitchen."'),
865
+ EmailCfg(id='031596', is_interesting=True, comment='"would you like photso of donald and girls in bikinis in my kitchen."'),
866
+ EmailCfg(id='031601', is_interesting=True, comment='Old gf i gave to donald', truncate_to=2000),
867
+ EmailCfg(id='030725', is_interesting=True, comment='David Stern in Moscow'),
868
+ EmailCfg(id='030714', is_interesting=True, comment='Bannon, Russian Dugan shout out'),
869
+ EmailCfg(id='031659', is_interesting=True, comment='"i have met some very bad people „ none as bad as trump"'),
870
+ EmailCfg(id='030245', is_interesting=True, comment='Epstein rationalizes his behavior in an open letter to the world'),
871
+ EmailCfg(id='030781', is_interesting=True, comment='Bannon email about crypto coin issues'),
872
+ EmailCfg(id='032906', is_interesting=True, comment='David Blaine email'),
873
+ EmailCfg(id='026036', is_interesting=True, truncate_to=6000, comment='Gino Yu blockchain mention'),
874
+ EmailCfg(id='029609', is_interesting=True, comment='Joi Ito'),
875
+ EmailCfg(id='025233', is_interesting=True, comment='Reputation.com discussion'),
876
+ EmailCfg(id='017827', is_interesting=True, comment='Bannon / Peggy Siegal email about netflix doc on Epstein'),
877
+ EmailCfg(id='030222', is_interesting=True, comment='Ross Gow / Ghislaine correspondence'),
878
+ EmailCfg(
879
+ id='026028',
880
+ description=f"Epstein introduces {LARRY_SUMMERS} to possibly corrupt Senegalese official Karim Wade",
881
+ is_interesting=True,
882
+ ),
883
+ EmailCfg(id='029545', is_interesting=True, comment='Tyler Shears reputation'),
884
+ EmailCfg(id='025812', is_interesting=True, comment='Tyler Shears reputation'),
885
+ EmailCfg(
886
+ id='029914',
887
+ description='Epstein and Lord Mandelson discuss Russian investments',
888
+ is_interesting=True,
889
+ truncate_to=4500,
890
+ ),
891
+ EmailCfg(id='033453', is_interesting=True, description='possibly an email from one of the women who sued Trump'),
892
+ EmailCfg(id='029342', is_interesting=True, truncate_to=2000, comment='Hakeem Jeffries'),
893
+ EmailCfg(id='031326', is_interesting=True, comment='"dog that hasn\'t barked is trump"'),
894
+ EmailCfg(id='033171', is_interesting=True, comment='Zubair'),
895
+ EmailCfg(
896
+ id='032319',
897
+ dupe_type='quoted',
898
+ duplicate_ids=['032283'],
899
+ is_interesting=True,
900
+ comment='Zubair',
901
+ ),
902
+ EmailCfg(
903
+ id='031152',
904
+ description='discussion of notoriously corrupt Kazakh politician Rakhat Aliyev whom Epstein apparently knew',
905
+ is_interesting=True,
906
+ ),
907
+ EmailCfg(id='030745', description="planning a public statement for Ghislaine", truncate_to=NO_TRUNCATE),
908
+ EmailCfg(id='028589', truncate_to=NO_TRUNCATE, comment='Long thread with Reid Weingarten'),
909
+ EmailCfg(id='026059', truncate_to=2650, comment='Rothschild'),
910
+ EmailCfg(id='032643', truncate_to=NO_TRUNCATE, comment='Anas al Rasheed'),
911
+ EmailCfg(id='031619', truncate_to=652, comment='Reply to grab em by the pussy story'),
912
+ EmailCfg(id='021096', truncate_to=700, comment='Sinofsky article quote'),
913
+ EmailCfg(id='032865', truncate_to=445, comment='Barton reply'),
914
+ EmailCfg(id='027126', truncate_to=1000, comment='Summers'),
915
+ EmailCfg(id='030950', truncate_to=4500, comment='Ian Osborne'),
916
+ EmailCfg(id='029684', truncate_to=402, comment='Maldives reply'),
917
+ EmailCfg(id='018045', truncate_to=TRUNCATED_CHARS, comment='invite'),
918
+ EmailCfg(id='017574', truncate_to=4000, comment='Lisa Randall invite'),
919
+ EmailCfg(id='030589', truncate_to=1000, comment='Brett Jaffe Fwd'),
920
+ EmailCfg(id='025655', truncate_to=400, comment='reply to article'),
921
+ EmailCfg(id='026451', truncate_to=500, comment='reply to article'),
922
+ EmailCfg(id='023717', truncate_to=489, comment='reply to article'),
923
+ EmailCfg(id='022265', truncate_to=NO_TRUNCATE),
924
+ EmailCfg(id='026243', truncate_to=NO_TRUNCATE),
925
+ EmailCfg(id='029680', truncate_to=900, comment='Maldives'),
926
+ EmailCfg(id='029534', truncate_to=900, comment='Maldives'),
927
+ EmailCfg(id='028760', truncate_to=900, comment='Epstein in an argument with someone'),
928
+ EmailCfg(id='028757', truncate_to=900, comment='Epstein in an argument with someone'),
929
+ EmailCfg(id='027059', truncate_to=NO_TRUNCATE, comment='Jean Luc Brunel and Boris Nikolic'),
930
+ EmailCfg(id='027028', truncate_to=1000, comment='Tom Pritzer penny pritzker'),
931
+ EmailCfg(id='029910', truncate_to=NO_TRUNCATE, comment='Tom Pritzer Aspen'),
932
+ EmailCfg(id='025163', truncate_to=NO_TRUNCATE, comment='Tom Pritzer'),
933
+
934
+ # DOJ files
935
+ EmailCfg(id='EFTA00935996', recipients=[RENATA_BOLOTOVA], attribution_reason='"sneaky dog"'),
936
+ EmailCfg(id='EFTA02731737', date='2023-06-30T16:05:00'),
937
+ EmailCfg(id='EFTA02731689', author=UNKNOWN, recipients=[None], date='2023-06-09 20:14:00'),
938
+ EmailCfg(id='EFTA02731475', date='2023-05-31T20:53:00'),
939
+ EmailCfg(id='EFTA02731732', date='2024-03-06T12:21:00'),
940
+ EmailCfg(id='EFTA02731485', date='2023-06-12T13:53:00'),
941
+ EmailCfg(id='EFTA02731617', date='2021-04-28T15:05:41'),
942
+ EmailCfg(id='EFTA02730483', date='2023-07-11T08:25:00'), # TODO: actually reply timewtamp
943
+ EmailCfg(id='EFTA02730481', date='2023-07-07T11:01:00'), # TODO: actually reply timewtamp
944
+ EmailCfg(id='EFTA02731754', date='2024-03-06T23:24:00'), # TODO: actually reply timewtamp
945
+ EmailCfg(id='EFTA02731735', date='2024-03-04T05:04:00'), # TODO: actually reply timewtamp
946
+ EmailCfg(id='EFTA02731577', date='2024-10-16T00:00:00'), # TODO: actually reply timewtamp
947
+ EmailCfg(id='EFTA02730468', date='2019-07-11T08:25:00'), # TODO: This is just wrong
948
+ # Generated basd on OtheFile extract_timestamp()
949
+ EmailCfg(id='EFTA02731783', date='2022-01-21 17:28:00'),
950
+ EmailCfg(id='EFTA02731587', date='2022-01-21 17:28:00'),
951
+ EmailCfg(id='EFTA02731729', date='2021-08-17 00:00:00'),
952
+ EmailCfg(id='EFTA02731578', date='2021-05-28 10:00:00'),
953
+ EmailCfg(id='EFTA02730473', date='2013-04-24 16:32:00'),
954
+ EmailCfg(id='EFTA02731699', date='2021-05-27 10:19:00'),
955
+ EmailCfg(id='EFTA02731583', date='2022-01-21 17:28:00'),
956
+ EmailCfg(id='EFTA02731552', date='2021-05-26 16:12:00'),
957
+ EmailCfg(id='EFTA00039888', date='2019-05-14 16:49:00'),
958
+ EmailCfg(id='EFTA02731684', date='2021-05-11 15:27:00'),
959
+ EmailCfg(id='EFTA02731697', date='2021-06-07 17:33:00'),
960
+ EmailCfg(id='EFTA02731733', date='2021-05-17 17:29:00'),
961
+ EmailCfg(id='EFTA00040145', date='2021-11-09 17:24:30'),
962
+ # EmailCfg(id='EFTA02730468', date='2004-02-03 00:00:00'), # TODO: ???
963
+ EmailCfg(id='EFTA02731528', date='2021-05-06 09:39:15'),
964
+ EmailCfg(id='EFTA02730485', date='2021-12-03 00:00:00'),
965
+ EmailCfg(id='EFTA00039689', truncate_to=NO_TRUNCATE),
966
+ EmailCfg(id='EFTA00995559', author=RENATA_BOLOTOVA, attribution_reason='poorly redacted signature'),
875
967
  ]
876
968
 
969
+ if args.constantize:
970
+ for email_cfg in EMAILS_CONFIG:
971
+ print(f"{email_cfg},")
972
+
877
973
 
878
974
  ################################################################################################
879
975
  ####################################### OTHER FILES ############################################
@@ -910,7 +1006,7 @@ DEUTSCHE_BANK_TAX_TOPICS = f'{DEUTSCHE_BANK} Wealth Management Tax Topics'
910
1006
  DIANA_DEGETTE_CAMPAIGN = "Colorado legislator Diana DeGette's campaign"
911
1007
  FBI_REPORT = f"report on Epstein investigation (redacted)"
912
1008
  FBI_SEIZED_PROPERTY = f"seized property inventory (redacted)"
913
- FEMALE_HEALTH_COMPANY = 'Female Health Company (FHX)'
1009
+ FEMALE_HEALTH_COMPANY = 'Female Health Company (FHC)'
914
1010
  FIRE_AND_FURY = f"Fire And Fury"
915
1011
  HARVARD_POETRY = f'{HARVARD} poetry stuff from {LISA_NEW}'
916
1012
  HBS_APPLICATION = f"{HARVARD} Business School application letter"
@@ -940,7 +1036,7 @@ ZUBAIR_AND_ANYA = f"{ZUBAIR_KHAN} and Anya Rasulova"
940
1036
  OTHER_FILES_BOOKS = [
941
1037
  DocCfg(id='017088', author=ALAN_DERSHOWITZ, description=f'"Taking the Stand: My Life in the Law" (draft)'),
942
1038
  DocCfg(id='013501', author='Arnold J. Mandell', description=f'The Nearness Of Grace: A Personal Science Of Spiritual Transformation', date='2005-01-01'),
943
- DocCfg(id='012899', author='Ben Goertzel', description=f'Engineering General Intelligence: A Path to Advanced AGI Via Embodied Learning and Cognitive Synergy'),
1039
+ DocCfg(id='012899', author='Ben Goertzel', description=f'Engineering General Intelligence: A Path to Advanced AGI Via Embodied Learning and Cognitive Synergy', date='2013-09-19'),
944
1040
  DocCfg(id='018438', author='Clarisse Thorn', description=f'The S&M Feminist'),
945
1041
  DocCfg(id='019477', author=EDWARD_JAY_EPSTEIN, description=f'How America Lost Its Secrets: Edward Snowden, the Man, and the Theft'),
946
1042
  DocCfg(id='020153', author=EDWARD_JAY_EPSTEIN, description=f'The Snowden Affair: A Spy Story In Six Parts'),
@@ -1024,7 +1120,7 @@ OTHER_FILES_ARTICLES = [
1024
1120
  DocCfg(id='013435', author=PALM_BEACH_DAILY_NEWS, description=f"article about Epstein's address book", date='2011-03-11'),
1025
1121
  DocCfg(id='013440', author=PALM_BEACH_DAILY_NEWS, description=f"article about Epstein's gag order", date='2011-07-13'),
1026
1122
  DocCfg(id='029238', author=PALM_BEACH_DAILY_NEWS, description=f"article about Epstein's plea deal"),
1027
- DocCfg(id='021775', author=PALM_BEACH_POST, description="article about 'He Was 50. And They Were Girls'"),
1123
+ DocCfg(id='021775', author=PALM_BEACH_POST, description="article about 'He Was 50. And They Were Girls'", attached_to_email_id='021764'),
1028
1124
  DocCfg(id='022989', author=PALM_BEACH_POST, description="article about alleged rape of 13 year old by Trump"),
1029
1125
  DocCfg(id='022987', author=PALM_BEACH_POST, description="article about just a headline on Trump and Epstein"),
1030
1126
  DocCfg(id='015028', author=PALM_BEACH_POST, description="article about reopening Epstein's criminal case"),
@@ -1042,14 +1138,14 @@ OTHER_FILES_ARTICLES = [
1042
1138
  DocCfg(id='010715', author=PEGGY_SIEGAL, description=f"Oscar Diary April", date='2012-02-27'),
1043
1139
  DocCfg(id='019849', author=PEGGY_SIEGAL, description=f"Oscar Diary April", date='2017-02-27', duplicate_ids=['019864']),
1044
1140
  DocCfg(id='026851', author='Politifact', description=f"lying politicians chart", date='2016-07-26'),
1045
- DocCfg(id='033253', author=ROBERT_LAWRENCE_KUHN, description=f'{BBC} article about Rohingya in Myanmar'),
1141
+ DocCfg(id='033253', author=ROBERT_LAWRENCE_KUHN, description=f'{BBC} article about Rohingya in Myanmar', attached_to_email_id='033252'),
1046
1142
  DocCfg(id='026887', author=ROBERT_LAWRENCE_KUHN, description=f'{BBC} "New Tariffs - Trade War"'),
1047
1143
  DocCfg(id='026877', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "New Tariffs - Trade War"'),
1048
1144
  DocCfg(id='026868', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "Quest Means Business New China Tariffs — Trade War"', date='2018-09-18'),
1049
1145
  DocCfg(id='023707', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "Quest Means Business U.S. and China Agree to Pause Trade War"', date='2018-12-03'),
1050
- DocCfg(id='029176', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "U.S. China Tariffs - Trade War"'),
1051
- DocCfg(id='032638', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "Xi Jinping and the New Politburo Committee"'),
1052
- DocCfg(id='023666', author=ROBERT_LAWRENCE_KUHN, description=f"sizzle reel / television appearances"),
1146
+ DocCfg(id='029176', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "U.S. China Tariffs - Trade War"', attached_to_email_id='029174'),
1147
+ DocCfg(id='032638', author=ROBERT_LAWRENCE_KUHN, description=f'{CNN} "Xi Jinping and the New Politburo Committee"', attached_to_email_id='032637'),
1148
+ DocCfg(id='023666', author=ROBERT_LAWRENCE_KUHN, description=f"sizzle reel / television appearances", date='2018-09-30', attached_to_email_id='033252'),
1053
1149
  DocCfg(id='016996', author=f'SciencExpress', description=f'article "Quantitative Analysis of Culture Using Millions of Digitized Books" by Jean-Baptiste Michel'),
1054
1150
  DocCfg(id='025104', author='SCMP', description=f"article about China and globalisation"),
1055
1151
  DocCfg(id='030030', author=SHIMON_POST, description=SHIMON_POST_ARTICLE, date='2011-03-29'),
@@ -1073,7 +1169,7 @@ OTHER_FILES_ARTICLES = [
1073
1169
  DocCfg(id='024997', author=SHIMON_POST, description=SHIMON_POST_ARTICLE, date='2011-09-08'),
1074
1170
  DocCfg(id='031941', author=SHIMON_POST, description=SHIMON_POST_ARTICLE, date='2011-11-17'),
1075
1171
  DocCfg(id='030829', author=f'South Florida Sun Sentinel', description=f'article about {BRAD_EDWARDS} and {JEFFREY_EPSTEIN}'),
1076
- DocCfg(id='021092', author='Tatler', description=f'single page of article about {GHISLAINE_MAXWELL} shredding documents', date='2019-08-15'),
1172
+ DocCfg(id='021092', author='Tatler', description=f'single page of article about {GHISLAINE_MAXWELL} shredding documents'),
1077
1173
  DocCfg(id='030333', author=f'The Independent', description=f'article about Prince Andrew, Epstein, and Epstein\'s butler who stole his address book'),
1078
1174
  DocCfg(id='010754', author=f'U.S. News', description=f"article about Yitzhak Rabin"),
1079
1175
  DocCfg(id='014498', author=VI_DAILY_NEWS, description='article', date='2016-12-13'),
@@ -1082,6 +1178,8 @@ OTHER_FILES_ARTICLES = [
1082
1178
  DocCfg(id='023046', author=VI_DAILY_NEWS, description='article', date='2019-02-27'),
1083
1179
  DocCfg(id='031170', author=VI_DAILY_NEWS, description='article', date='2019-03-06'),
1084
1180
  DocCfg(id='016506', author=VI_DAILY_NEWS, description='article', date='2019-02-28'),
1181
+ DocCfg(id='018862', author=VI_DAILY_NEWS, description='articles about Sen. Alvin Williams Jr. Fraud case, arson', date='2012-11-09'),
1182
+
1085
1183
  DocCfg(id='016507', author=VI_DAILY_NEWS, description=f'"Perversion of Justice" by {JULIE_K_BROWN}', date='2018-12-19'),
1086
1184
  DocCfg(id='019212', author=WAPO, description=f'and Times Tribune articles about Bannon, Trump, and healthcare execs'),
1087
1185
  DocCfg(id='033379', author=WAPO, description=f'"How Washington Pivoted From Finger-Wagging to Appeasement" (about Viktor Orban)', date='2018-05-25'),
@@ -1203,7 +1301,7 @@ OTHER_FILES_LEGAL = [
1203
1301
  DocCfg(id='013489', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f'Affidavit of {BRAD_EDWARDS}', date='2010-07-20'),
1204
1302
  DocCfg(id='029398', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f'article in Law.com'),
1205
1303
  DocCfg(id='026854', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f"Civil Docket"),
1206
- DocCfg(id='026384', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f"Complaint for rape and sexual abuse", date='2016-06-20'),
1304
+ DocCfg(id='026384', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f"Complaint for rape and sexual abuse", date='2016-06-20', attached_to_email_id='029837'),
1207
1305
  DocCfg(id='013463', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f'Deposition of Scott Rothstein', date='2010-03-23'),
1208
1306
  DocCfg(id='029257', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f'allegations and identity of plaintiff Katie Johnson', date='2016-04-26'),
1209
1307
  DocCfg(id='032321', author=JANE_DOE_V_EPSTEIN_TRUMP, description=f"Notice of Initial Conference", date='2016-10-04'),
@@ -1250,6 +1348,10 @@ OTHER_FILES_LEGAL = [
1250
1348
  DocCfg(id='028540', author='SCOTUS', description=f"decision in Budha Ismail Jam et al. v. INTERNATIONAL FINANCE CORP"),
1251
1349
  DocCfg(id='012197', author='SDFL', description=f"response to {JAY_LEFKOWITZ} on Epstein Plea Agreement Compliance"),
1252
1350
  DocCfg(id='022277', description=f"text of National Labour Relations Board (NLRB) law", is_interesting=False),
1351
+
1352
+ # DOJ files
1353
+ DocCfg(id='EFTA00007157', description='victim list and police log'),
1354
+ DocCfg(id='EFTA02730274', description='evidence inventory that appears to have since been deleted from the DOJ website'),
1253
1355
  ]
1254
1356
 
1255
1357
  OTHER_FILES_CONFERENCES = [
@@ -1263,7 +1365,7 @@ OTHER_FILES_CONFERENCES = [
1263
1365
  DocCfg(id='022267', author=SVETLANA_POZHIDAEVA, description=f'{WOMEN_EMPOWERMENT} founder essay about growing the seminar business'),
1264
1366
  DocCfg(id='022407', author=SVETLANA_POZHIDAEVA, description=f'{WOMEN_EMPOWERMENT} seminar pitch deck'),
1265
1367
  DocCfg(id='017524', author=SWEDISH_LIFE_SCIENCES_SUMMIT, description=f"2012 program", date='2012-08-18', attached_to_email_id='031226'),
1266
- DocCfg(id='026747', author=SWEDISH_LIFE_SCIENCES_SUMMIT, description=f"2017 program", date='2017-08-23'),
1368
+ DocCfg(id='026747', author=SWEDISH_LIFE_SCIENCES_SUMMIT, description=f"2017 program", date='2017-08-23', attached_to_email_id='031215'),
1267
1369
  DocCfg(id='014951', author='TED Talks', description=f"2017 program", date='2017-04-20'),
1268
1370
  DocCfg(id='024179', author=UN_GENERAL_ASSEMBLY, description=f'president and first lady schedule', date='2012-09-21'),
1269
1371
  DocCfg(
@@ -1317,12 +1419,12 @@ OTHER_FILES_FINANCE = [
1317
1419
  DocCfg(id='024302', author='Carvana', description=f"form 14A SEC filing proxy statement", date='2019-04-23'),
1318
1420
  DocCfg(id='029305', author='CCH Tax', description=f"Briefing on end of Defense of Marriage Act", date='2013-06-27'),
1319
1421
  DocCfg(id='026794', author=DEUTSCHE_BANK, description=f"Global Political and Regulatory Risk in 2015/2016"),
1320
- DocCfg(id='022361', author=DEUTSCHE_BANK_TAX_TOPICS, date='2013-05-01'),
1422
+ DocCfg(id='022361', author=DEUTSCHE_BANK_TAX_TOPICS, date='2013-05-01', attached_to_email_id='022359'),
1321
1423
  DocCfg(id='022325', author=DEUTSCHE_BANK_TAX_TOPICS, date='2013-12-20'),
1322
1424
  DocCfg(id='022330', author=DEUTSCHE_BANK_TAX_TOPICS, date='2013-12-20', description='table of contents'),
1323
1425
  DocCfg(id='019440', author=DEUTSCHE_BANK_TAX_TOPICS, date='2014-01-29'),
1324
1426
  DocCfg(id='024202', author=ELECTRON_CAPITAL_PARTNERS, description=f"Global Utility White Paper", date='2013-03-08'),
1325
- DocCfg(id='022372', author='Ernst & Young', description=f'2016 election report'),
1427
+ DocCfg(id='022372', author='Ernst & Young', date='2016-11-09', description=f'2016 election report'),
1326
1428
  DocCfg(
1327
1429
  id='025663',
1328
1430
  author=GOLDMAN_INVESTMENT_MGMT,
@@ -1331,7 +1433,13 @@ OTHER_FILES_FINANCE = [
1331
1433
  is_interesting=True,
1332
1434
  ),
1333
1435
  DocCfg(id='014532', author=GOLDMAN_INVESTMENT_MGMT, description=f"Outlook - Half Full", date='2017-01-01'),
1334
- DocCfg(id='026909', author=GOLDMAN_INVESTMENT_MGMT, description=f"The Unsteady Undertow Commands the Seas (Temporarily)", date='2018-10-14'),
1436
+ DocCfg(
1437
+ id='026909',
1438
+ attached_to_email_id='026893',
1439
+ author=GOLDMAN_INVESTMENT_MGMT,
1440
+ description=f"The Unsteady Undertow Commands the Seas (Temporarily)",
1441
+ date='2018-10-14',
1442
+ ),
1335
1443
  DocCfg(id='026944', author=GOLDMAN_INVESTMENT_MGMT, description=f"Risk of a US-Iran Military Conflict", date='2019-05-23'),
1336
1444
  DocCfg(id='018804', author='Integra Realty Resources', description=f"appraisal of going concern for IGY American Yacht Harbor Marina in {VIRGIN_ISLANDS}"),
1337
1445
  DocCfg(id='026679', author='Invesco', description=f"Global Sovereign Asset Management Study 2017"),
@@ -1346,7 +1454,7 @@ OTHER_FILES_FINANCE = [
1346
1454
  DocCfg(id='030840', author=JP_MORGAN, description=f"Market Thoughts"),
1347
1455
  DocCfg(id='022350', author=JP_MORGAN, description=f"tax efficiency of Intentionally Defective Grantor Trusts (IDGT)"),
1348
1456
  DocCfg(id='025242', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2012-04-09'),
1349
- DocCfg(id='030010', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2011-06-14'),
1457
+ DocCfg(id='030010', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, attached_to_email_id='030006', date='2011-06-14'),
1350
1458
  DocCfg(id='030808', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2011-07-11'),
1351
1459
  DocCfg(id='025221', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2011-07-25'),
1352
1460
  DocCfg(id='025229', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2011-08-04'),
@@ -1354,7 +1462,12 @@ OTHER_FILES_FINANCE = [
1354
1462
  DocCfg(id='024132', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2012-03-15'),
1355
1463
  DocCfg(id='024194', author=JP_MORGAN, description=JP_MORGAN_EYE_ON_THE_MARKET, date='2012-10-22'),
1356
1464
  DocCfg(id='025296', author='Laffer Associates', description=f'report predicting Trump win', date='2016-07-06'),
1357
- DocCfg(id='020824', author='Mary Meeker', description=f"USA Inc: A Basic Summary of America's Financial Statements compiled", date='2011-02-01'),
1465
+ DocCfg(
1466
+ id='020824',
1467
+ author='Mary Meeker',
1468
+ date='2011-02-01',
1469
+ description=f"USA Inc: A Basic Summary of America's Financial Statements compiled",
1470
+ ),
1358
1471
  DocCfg(id='025551', author='Morgan Stanley', description=f'report about alternative asset managers', date='2018-01-30'),
1359
1472
  DocCfg(id='019856', author='Sadis Goldberg LLP', description=f"report on SCOTUS ruling about insider trading", is_interesting=True),
1360
1473
  DocCfg(id='025763', author='S&P', description=f"Economic Research: How Increasing Income Inequality Is Dampening U.S. Growth", date='2014-08-05'),
@@ -1363,12 +1476,20 @@ OTHER_FILES_FINANCE = [
1363
1476
  DocCfg(id='026584', description=f"article about tax implications of disregarded entities", date='2009-07-01', is_interesting=True),
1364
1477
  DocCfg(
1365
1478
  id='024271',
1366
- description=f"Blockchain Capital and Brock Pierce pitch deck",
1367
1479
  date='2015-10-01',
1480
+ description=f"Blockchain Capital and Brock Pierce pitch deck",
1368
1481
  is_interesting=True,
1369
1482
  ),
1370
- DocCfg(id='024817', description=f"Cowen's Collective View of CBD / Cannabis report"),
1371
- DocCfg(id='012048', description=f"{PRESS_RELEASE} 'Rockefeller Partners with Gregory J. Fleming to Create Independent Financial Services Firm' and other articles"),
1483
+ DocCfg(
1484
+ id='024817',
1485
+ date='2019-02-25',
1486
+ description=f"Cowen's Collective View of CBD / Cannabis report",
1487
+ is_interesting=True
1488
+ ),
1489
+ DocCfg(
1490
+ id='012048',
1491
+ description=f"{PRESS_RELEASE} 'Rockefeller Partners with Gregory J. Fleming to Create Independent Financial Services Firm' and other articles"
1492
+ ),
1372
1493
 
1373
1494
  # private placement memoranda
1374
1495
  DocCfg(
@@ -1376,7 +1497,6 @@ OTHER_FILES_FINANCE = [
1376
1497
  date='2006-09-27',
1377
1498
  description=f"Michael Milken's Knowledge Universe Education (KUE) $1,000,000 corporate share placement notice (SEC filing?)"
1378
1499
  ),
1379
-
1380
1500
  DocCfg(id='024003', description=f"New Leaf Ventures ($375 million biotech fund) private placement memorandum"),
1381
1501
  ]
1382
1502
 
@@ -1438,6 +1558,11 @@ OTHER_FILES_PROPERTY = [
1438
1558
  description=f"{VIRGIN_ISLANDS} property deal pitch deck, building will be leased to the U.S. govt GSA",
1439
1559
  date='2014-06-01',
1440
1560
  ),
1561
+
1562
+ # DOJ files
1563
+ DocCfg(id='EFTA00001884', date='2019-03-14', description='photo of letter from Virgin Islands DOJ to Epstein'),
1564
+ DocCfg(id='EFTA00005783', date='2019-08-29', description='heavily redacted handwritten note and 30+ completely blacked out redacted pages'),
1565
+
1441
1566
  ]
1442
1567
 
1443
1568
  OTHER_FILES_REPUTATION = [
@@ -1455,12 +1580,12 @@ OTHER_FILES_SOCIAL = [
1455
1580
  DocCfg(id='017787', author=ALAN_DERSHOWITZ, description=DERSH_GIUFFRE_TWEET),
1456
1581
  DocCfg(id='033433', author=ALAN_DERSHOWITZ, description=f"{DERSH_GIUFFRE_TWEET} / David Boies", date='2019-03-02'),
1457
1582
  DocCfg(id='033432', author=ALAN_DERSHOWITZ, description=f"{DERSH_GIUFFRE_TWEET} / David Boies", date='2019-05-02'),
1458
- DocCfg(id='028815', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} business plan", date='2016-08-20'),
1459
- DocCfg(id='011170', author=ZUBAIR_AND_ANYA, description=f'{INSIGHTS_POD} collected tweets about #Brexit', date='2016-06-23'),
1460
- DocCfg(id='032324', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} election social media trend analysis", date='2016-11-05'),
1461
- DocCfg(id='032281', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} forecasting election for Trump", date='2016-10-25'),
1462
- DocCfg(id='028988', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} pitch deck", date='2016-08-20'),
1463
- DocCfg(id='026627', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} report on the presidential debate"),
1583
+ DocCfg(id='028815', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} business plan", date='2016-08-20', attached_to_email_id='033171'),
1584
+ DocCfg(id='011170', author=ZUBAIR_AND_ANYA, description=f'{INSIGHTS_POD} collected tweets from #Brexit case study', date='2016-06-23', attached_to_email_id='033171'),
1585
+ DocCfg(id='032324', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} election social media trend analysis", date='2016-11-05', attached_to_email_id='032323'),
1586
+ DocCfg(id='032281', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} eleciton social media trends report", date='2016-10-25', attached_to_email_id='032280'),
1587
+ DocCfg(id='028988', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} pitch deck", date='2016-08-20', attached_to_email_id='033171'),
1588
+ DocCfg(id='026627', author=ZUBAIR_AND_ANYA, description=f"{INSIGHTS_POD} report on the presidential debate", attached_to_email_id='026626'),
1464
1589
  DocCfg(id='022213', description=f"{SCREENSHOT} Facebook group called 'Shit Pilots Say' disparaging a 'global girl'"),
1465
1590
  DocCfg(id='030884', description=f"{TWEET} by Ed Krassenstein"),
1466
1591
  DocCfg(id='031546', description=f"{TWEET}s by Donald Trump about Russian collusion", date='2018-01-06'),
@@ -1488,7 +1613,7 @@ OTHER_FILES_POLITICS = [
1488
1613
  id='023133',
1489
1614
  author=f"{TERJE_ROD_LARSEN}, Nur Laiq, Fabrice Aidan",
1490
1615
  description=f'The Search for Peace in the Arab-Israeli Conflict',
1491
- date='2019-12-09',
1616
+ date='2014-12-09',
1492
1617
  ),
1493
1618
  DocCfg(id='033468', description=f'{ARTICLE_DRAFT} Rod Rosenstein', date='2018-09-24'),
1494
1619
  DocCfg(
@@ -1503,8 +1628,8 @@ OTHER_FILES_POLITICS = [
1503
1628
  date='2015-01-15', # TODO: this is just a guess
1504
1629
  duplicate_ids=['028887'],
1505
1630
  ),
1506
- DocCfg(id='010617', description=TRUMP_DISCLOSURES, date='2017-01-20', is_interesting=True),
1507
- DocCfg(id='016699', description=TRUMP_DISCLOSURES, date='2017-01-20', is_interesting=True),
1631
+ DocCfg(id='010617', description=TRUMP_DISCLOSURES, date='2017-01-20', is_interesting=True, attached_to_email_id='033091'),
1632
+ DocCfg(id='016699', description=TRUMP_DISCLOSURES, date='2017-01-20', is_interesting=True, attached_to_email_id='033091'),
1508
1633
  ]
1509
1634
 
1510
1635
  OTHER_FILES_ACADEMIA = [
@@ -1601,7 +1726,11 @@ OTHER_FILES_ARTS = [
1601
1726
  date='2010-02-01',
1602
1727
  duplicate_ids=['025210']
1603
1728
  ),
1604
- DocCfg(id='028281', description=f'art show flier for "The House Of The Nobleman" curated by Wolfe Von Lenkiewicz & Victoria Golembiovskaya'),
1729
+ DocCfg(
1730
+ id='028281',
1731
+ date='2010-10-13',
1732
+ description=f'art show flier for "The House Of The Nobleman" curated by Wolfe Von Lenkiewicz & Victoria Golembiovskaya',
1733
+ ),
1605
1734
  ]
1606
1735
 
1607
1736
  OTHER_FILES_MISC = [
@@ -1631,7 +1760,7 @@ OTHER_FILES_MISC = [
1631
1760
  DocCfg(id='031743', description=f'a few pages describing the internet as a "New Nation State" (Network State?)'),
1632
1761
  DocCfg(id='012718', description=f"{CVRA} congressional record", date='2011-06-17'),
1633
1762
  DocCfg(id='024117', description=f"FAQ about anti-money laundering (AML) and terrorist financing (CFT) law in the U.S."),
1634
- DocCfg(id='019448', description=f"Haitian business investment proposal called Jacmel", attached_to_email_id='019448'),
1763
+ DocCfg(id='019448', description=f"Haitian business investment proposal called Jacmel", attached_to_email_id='019446'),
1635
1764
  DocCfg(id='023644', description=f"interview with Mohammed bin Salman", date='2016-04-25'),
1636
1765
  DocCfg(
1637
1766
  id='030142',
@@ -1647,6 +1776,32 @@ OTHER_FILES_MISC = [
1647
1776
  DocCfg(id='033434', description=f"{SCREENSHOT} iPhone chat labeled 'Edwards' at the top"),
1648
1777
  DocCfg(id='029475', description=f'{VIRGIN_ISLANDS} Twin City Mobile Integrated Health Services (TCMIH) proposal/request for donation'),
1649
1778
  DocCfg(id='029448', description=f"weird short essay titled 'President Obama and Self-Deception'"),
1779
+
1780
+ # DOJ files
1781
+
1782
+ DocCfg(id='EFTA00007781', description='paychecks signed by Epstein deposited at Colonial Bank'),
1783
+ DocCfg(id='EFTA00009622', description='handwritten note transcribed Claude AI', date='2006-07-19', replace_text_with=EFTA00009622_TEXT),
1784
+ DocCfg(id='EFTA00039295', replace_text_with='Bureau of Prisons inmate telephone privileges Program Statement'),
1785
+ DocCfg(
1786
+ id='EFTA00004477',
1787
+ replace_text_with='Epstein 50th birthday photo book 12 "THAIS, MOSCOW GIRLS, AFRICA, HAWAII, [REDACTED] [REDACTED], Zorro, [REDACTED] [REDACTED] [REDACTED], CRACK WHOLE PROPOSAL, BALI/THAILAND/ASIA, RUSSIA, [REDACTED], [REDACTED], NUDES, YOGAL GIRLS',
1788
+ ),
1789
+ DocCfg(id='EFTA00008120', replace_text_with='"Part II: The Art of Receiving a Massage"'),
1790
+ DocCfg(id='EFTA00008020', replace_text_with='"Massage for Dummies"'),
1791
+ DocCfg(id='EFTA00008220', replace_text_with='"Massage book: Chapter 11: Putting the Moves Together"'),
1792
+ DocCfg(id='EFTA00008320', replace_text_with='"Massage for Dummies (???)"'),
1793
+ DocCfg(id='EFTA00000476', replace_text_with='photo of JEFFREY EPSTEIN CASH DISBURSEMENTS for the month 2006-09'),
1794
+ DocCfg(id='EFTA00039312', replace_text_with='Bureau of Prisons Program Statement / Memo about BOP Pharmacy Program'),
1795
+ # Phone bills TODO: Some kind of special handling?
1796
+ DocCfg(id='EFTA00006387', replace_text_with='T-Mobile phone bill covering 2006-06-15 to 2006-07-23'),
1797
+ DocCfg(id='EFTA00007501', replace_text_with='T-Mobile phone bill from 2005'),
1798
+ DocCfg(id='EFTA00006587', replace_text_with='T-Mobile phone bill from 2006-09-04 to 2016-10-15'),
1799
+ DocCfg(id='EFTA00006687', replace_text_with='T-Mobile phone bill from 2006-10-31 to 2006-12-25'),
1800
+ DocCfg(id='EFTA00007401', replace_text_with='T-Mobile phone bill from 2004-08-25 to 2005-07-13'),
1801
+ DocCfg(id='EFTA00007301', replace_text_with='T-Mobile response to subpoena March 23, 2007 - Blackberry phone logs for 2005'),
1802
+ DocCfg(id='EFTA00006487', replace_text_with='T-Mobile phone bill 2006-08-26'),
1803
+ DocCfg(id='EFTA00006100', replace_text_with='Palm Beach Police fax machine activity log 2005-12-28 to 2006-01-04'),
1804
+ DocCfg(id='EFTA00007253', replace_text_with='T-Mobile response to subpoena March 23, 2007 - phone bill '),
1650
1805
  ]
1651
1806
 
1652
1807
  OTHER_FILES_JUNK = [
@@ -1660,6 +1815,12 @@ OTHER_FILES_JUNK = [
1660
1815
  DocCfg(id='029351', description=OBAMA_JOKE, date='2013-07-26'),
1661
1816
  DocCfg(id='029354', description=OBAMA_JOKE, date='2013-07-26'),
1662
1817
  DocCfg(id='031293'),
1818
+
1819
+ # Completely redacted DOJ emails, no timestamp at all
1820
+ DocCfg(id='EFTA02731726'),
1821
+ DocCfg(id='EFTA02731728'),
1822
+ # Almost no timestamp
1823
+ DocCfg(id='EFTA00003154'),
1663
1824
  ]
1664
1825
 
1665
1826
  OTHER_FILES_CATEGORIES = [
@@ -1702,8 +1863,8 @@ for cfg in ALL_CONFIGS:
1702
1863
 
1703
1864
  # Add extra config objects for duplicate files that match the config of file they are duplicating
1704
1865
  for dupe_cfg in cfg.duplicate_cfgs():
1705
- if not isinstance(dupe_cfg, EmailCfg):
1706
- logger.debug(f"Generated synthetic config for dupe: {dupe_cfg}")
1866
+ # if not isinstance(dupe_cfg, EmailCfg):
1867
+ # logger.debug(f"Generated synthetic config for dupe: {dupe_cfg}")
1707
1868
 
1708
1869
  ALL_FILE_CONFIGS[dupe_cfg.id] = dupe_cfg
1709
1870
 
@@ -1716,17 +1877,17 @@ NORWEGAIN_REPLY_PATTERN = r"(Den .* folgende|(fre|lor|son)\. .* skrev .*):"
1716
1877
  REPLY_LINE_IN_A_MSG_PATTERN = r"In a message dated \d+/\d+/\d+.*writes:"
1717
1878
  REPLY_LINE_ENDING_PATTERN = r"[_ \n](AM|PM|[<_]|wrote:?)"
1718
1879
  REPLY_LINE_ON_NUMERIC_DATE_PATTERN = fr"On \d+/\d+/\d+[, ].*{REPLY_LINE_ENDING_PATTERN}"
1719
- REPLY_LINE_ON_DATE_PATTERN = fr"^On (\d+ )?((Mon|Tues?|Wed(nes)?|Thu(rs)?|Fri|Sat(ur)?|Sun)(day)?|(Jan|Feb|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w*)[, ].*{REPLY_LINE_ENDING_PATTERN}"
1880
+ REPLY_LINE_ON_DATE_PATTERN = fr"^[> •]*On (\d+ )?((Mon|Tues?|Wed(nes)?|Thu(rs)?|Fri|Sat(ur)?|Sun)(day)?|(Jan|Fe(b|vr\.)|Mar|Apr|May|Jun|Jul|Aug|Sep|Oct|Nov|Dec)\w*)[, ].*{REPLY_LINE_ENDING_PATTERN}"
1720
1881
  REPLY_LINE_PATTERN = rf"({FRENCH_REPLY_PATTERN}|{GERMAN_REPLY_PATTERN}|{NORWEGAIN_REPLY_PATTERN}|{REPLY_LINE_IN_A_MSG_PATTERN}|{REPLY_LINE_ON_NUMERIC_DATE_PATTERN}|{REPLY_LINE_ON_DATE_PATTERN}|{FORWARDED_LINE_PATTERN})"
1721
1882
  REPLY_REGEX = re.compile(REPLY_LINE_PATTERN, re.IGNORECASE | re.MULTILINE)
1722
1883
  SENT_FROM_REGEX = re.compile(r'^(?:(Please forgive|Sorry for all the) typos.{1,4})?((Envoyé de mon|Sent (from|via)).*(and string|AT&T|Droid|iPad|Phone|Mail|BlackBerry(.*(smartphone|device|Handheld|AT&T|T- ?Mobile))?)\.?)|Co-authored with iPhone auto-correct', re.M | re.I)
1723
1884
 
1724
-
1725
1885
  # No point in ever displaying these; their emails show up elsewhere because they're mostly CC recipients
1726
- UNINTERESTING_EMAILERS = FLIGHT_IN_2012_PEOPLE + IRAN_DEAL_RECIPIENTS + [
1886
+ UNINTERESTING_EMAILERS = FLIGHT_IN_2012_PEOPLE + IRAN_DEAL_RECIPIENTS + TRIVERS_CCS + [
1727
1887
  'Alan Dlugash', # CCed with Richard Kahn
1728
1888
  'Alan Rogers', # Random CC
1729
1889
  'Andrew Friendly', # Presumably some relation of Kelly Friendly
1890
+ 'Ariane Dwyer', # Sabba CC
1730
1891
  'BS Stern', # A random fwd of email we have
1731
1892
  'Cheryl Kleen', # Single email from Anne Boyles, displayed under Anne Boyles
1732
1893
  'Connie Zaguirre', # Random CC
@@ -1734,9 +1895,11 @@ UNINTERESTING_EMAILERS = FLIGHT_IN_2012_PEOPLE + IRAN_DEAL_RECIPIENTS + [
1734
1895
  'Danny Goldberg', # Random Paul Krassner emails
1735
1896
  GERALD_LEFCOURT, # Single CC
1736
1897
  GORDON_GETTY, # Random CC
1898
+ 'Grant J. Smith', # Ken Jenne CC
1737
1899
  JEFF_FULLER, # Random Jean Luc Brunel CC
1738
1900
  'Jojo Fontanilla', # Random CC
1739
1901
  'Joseph Vinciguerra', # Random CC
1902
+ 'Kirk Blouin', # John Page / Police Code Enforcement chain
1740
1903
  'Larry Cohen', # Random Bill Gates CC
1741
1904
  'Lyn Fontanilla', # Random CC
1742
1905
  'Mark Albert', # Random CC
@@ -1747,12 +1910,14 @@ UNINTERESTING_EMAILERS = FLIGHT_IN_2012_PEOPLE + IRAN_DEAL_RECIPIENTS + [
1747
1910
  'Nancy Portland', # Lawrence Krauss CC
1748
1911
  'Oliver Goodenough', # Robert Trivers CC
1749
1912
  'Peter Aldhous', # Lawrence Krauss CC
1913
+ 'Peter Green', # Farkas emailer
1750
1914
  'Players2', # Hoffenberg CC
1751
1915
  'Police Code Enforcement', # Kirk Blouin / John Page CC
1752
1916
  'Sam Harris', # Lawrence Krauss CC
1753
1917
  SAMUEL_LEFF, # Random CC
1754
1918
  'Sean T Lehane', # Random CC
1755
1919
  'Stephen Rubin', # Random CC
1920
+ THANU_BOONYAWATANA, # Eduardo Robles CC
1756
1921
  'Tim Kane', # Random CC
1757
1922
  'Travis Pangburn', # Random CC
1758
1923
  'Vahe Stepanian', # Random CC