epstein-files 1.0.14__py3-none-any.whl → 1.0.16__py3-none-any.whl

epstein_files/util/highlighted_group.py

@@ -15,7 +15,7 @@ from epstein_files.util.data import extract_last_name, listify, without_falsey
 CIVIL_ATTORNEY = 'civil attorney'
 CRIMINAL_DEFENSE_ATTORNEY = 'criminal defense attorney'
 CRIMINAL_DEFENSE_2008 = f"{CRIMINAL_DEFENSE_ATTORNEY} on 2008 case"
-EPSTEIN_LAWYER = 'epstein_lawyer'
+EPSTEIN_LAWYER = 'Epstein lawyer'
 EPSTEIN_V_ROTHSTEIN_EDWARDS_ATTORNEY = f"{CIVIL_ATTORNEY} working on {EPSTEIN_V_ROTHSTEIN_EDWARDS}"
 ESTATE_EXECUTOR = 'estate executor'
 EPSTEIN_ESTATE_EXECUTOR = f"Epstein {ESTATE_EXECUTOR}"
@@ -103,7 +103,7 @@ class HighlightedNames(HighlightedText):
     def get_info(self, name: str) -> str | None:
         """Label and additional info for 'name' if 'name' is in self.emailers."""
         info_pieces = [
-            None if len(self.emailers) == 1 else (self.category or self.label.title()),
+            None if len(self.emailers) == 1 else (self.category or self.label.replace('_', ' ')),
             self.emailers.get(name),
         ]
 
@@ -111,7 +111,7 @@ class HighlightedNames(HighlightedText):
         return ', '.join(info_pieces) if info_pieces else None
 
     def _emailer_pattern(self, name: str) -> str:
-        """Pattern matching 'name'. Extends value in EMAILER_ID_REGEXES with last name if it exists."""
+        """Pattern matching 'name'. Extends value in EMAILER_ID_REGEXES with first/last name if it exists."""
         name = remove_question_marks(name)
         last_name = extract_last_name(name)
         first_name = name.removesuffix(f" {last_name}")
@@ -141,7 +141,7 @@ class HighlightedNames(HighlightedText):
 
 HIGHLIGHTED_NAMES = [
     HighlightedNames(
-        label='africa',
+        label='Africa',
         style='light_pink4',
         pattern=r'Econet(\s*Wireless)|Ghana(ian)?|Johannesburg|Kenya|Nigerian?|Senegal(ese)?|Serengeti|(South\s*)?African?|(Strive\s*)?Masiyiwa|Tanzania|Ugandan?|Zimbabwe(an)?',
         emailers={
@@ -183,12 +183,15 @@ HIGHLIGHTED_NAMES = [
         pattern=r"CBD|cannabis|marijuana|THC|WEED(guide|maps)?[^s]?",
     ),
     HighlightedNames(
-        label='china',
+        label='China',
         style='bright_red',
-        pattern=r"Ali.?baba|Beijing|CCP|Chin(a|e?se)(?! Daily)|DPRK|Gino\s+Yu|Global Times|Guo|Hong|Huaw[ae]i|Kim\s*Jong\s*Un|Kong|Jack\s+Ma|Kwok|Ministry\sof\sState\sSecurity|Mongolian?|MSS|North\s*Korea|Peking|PRC|SCMP|Tai(pei|wan)|Xi(aomi)?|Jinping",
+        pattern=r"Ali.?baba|Beijing|CCP|Chin(a|e?se)(?! Daily)|DPRK|Global Times|Guo|Hong|Huaw[ae]i|Kim\s*Jong\s*Un|Kong|Jack\s+Ma|Kwok|Ministry\sof\sState\sSecurity|Mongolian?|MSS|North\s*Korea|Peking|PRC|SCMP|Tai(pei|wan)|Xi(aomi)?|Jinping",
+        emailers={
+            'Gino Yu': 'professor / game designer in Hong Kong',
+        }
     ),
     HighlightedNames(
-        label='deepak_chopra',
+        label='Deepak Chopra',
         style='dark_sea_green4',
         emailers = {
             'Carolyn Rangel': 'assistant',
@@ -196,7 +199,7 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='democrats',
+        label='Democrats',
         style='sky_blue1',
         pattern=r'(Al\s*)?Franken|((Bill|Hillart?y)\s*)?Clinton|((Chuck|Charles)\s*)?S(ch|hc)umer|(Diana\s*)?DeGette|DNC|Elena\s*Kagan|(Eliott?\s*)?Spitzer(, Eliot)?|George\s*Mitchell|(George\s*)?Soros|Hill?ary|Dem(ocrat(ic)?)?|(Jo(e|seph)\s*)?Biden|(John\s*)?Kerry|Lisa Monaco|(Matteo\s*)?Salvini|Maxine\s*Waters|(Barac?k )?Obama|(Nancy )?Pelosi|Ron\s*Dellums|Schumer|(Tim\s*)?Geithner|Vernon\s*Jordan',
     ),
@@ -271,7 +274,7 @@ HIGHLIGHTED_NAMES = [
     HighlightedNames(
         label=ESTATE_EXECUTOR,
         style='purple3 bold',
-        category='epstein lawyer',
+        category='Epstein lawyer',
         emailers = {
             DARREN_INDYKE: EPSTEIN_ESTATE_EXECUTOR,
             RICHARD_KAHN: EPSTEIN_ESTATE_EXECUTOR,
@@ -280,7 +283,7 @@ HIGHLIGHTED_NAMES = [
     HighlightedNames(
         label='europe',
         style='light_sky_blue3',
-        pattern=r'(Angela )?Merk(el|le)|Austria|(Benjamin\s*)?Harnwell|Berlin|Borge|Boris\s*Johnson|Brexit(eers?)?|Brit(ain|ish)|Brussels|Cannes|(Caroline|Jack)?\s*Lang(, Caroline)?|Cypr(iot|us)|Davos|ECB|England|EU|Europe(an)?(\s*Union)?|Fr(ance|ench)|Geneva|Germany?|Gillard|Gree(ce|k)|Ital(ian|y)|Jacques|(Kevin\s*)?Rudd|Le\s*Pen|London|Macron|Melusine|Munich|(Natalia\s*)?Veselnitskaya|(Nicholas\s*)?Sarkozy|Nigel(\s*Farage)?|Norw(ay|egian)|Oslo|Paris|Polish|(Sebastian )?Kurz|(Vi(c|k)tor\s+)?Orbah?n|Edward Rod Larsen|Strasbourg|Strauss[- ]?Kahn|Swed(en|ish)(?![-\s]+America)|Switzerland|(Tony\s)?Blair|U\.?K\.?|Ukrain(e|ian)|Vienna|(Vitaly\s*)?Churkin|Zug',
+        pattern=r'(Angela )?Merk(el|le)|Austria|(Benjamin\s*)?Harnwell|Berlin|Borge|Boris\s*Johnson|Brexit(eers?)?|Brit(ain|ish)|Brussels|Cannes|(Caroline|Jack)?\s*Lang(, Caroline)?|Cypr(iot|us)|Davos|ECB|England|EU|Europe(an)?(\s*Union)?|Fr(ance|ench)|Geneva|Germany?|Gillard|Gree(ce|k)|Ital(ian|y)|Jacques|(Kevin\s*)?Rudd|Le\s*Pen|London|Macron|Melusine|Munich|(Natalia\s*)?Veselnitskaya|(Nicholas\s*)?Sarkozy|Nigel(\s*Farage)?|Norw(ay|egian)|Oslo|Paris|Polish|pope|(Sebastian )?Kurz|(Vi(c|k)tor\s+)?Orbah?n|Edward Rod Larsen|Strasbourg|Strauss[- ]?Kahn|Swed(en|ish)(?![-\s]+America)|Switzerland|(Tony\s)?Blair|U\.?K\.?|Ukrain(e|ian)|Vienna|(Vitaly\s*)?Churkin|Zug',
         emailers = {
             ANDRZEJ_DUDA: 'former president of Poland',
             MIROSLAV_LAJCAK: 'Russia-friendly Slovakian politician, friend of Steve Bannon',
@@ -290,9 +293,8 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='famous_lawyer',
+        label='famous lawyer',
         style='medium_purple3',
-        category='famous_lawyer',
         pattern=r'(David\s*)?Bo[il]es|dersh|(Gloria\s*)?Allred|(Mi(chael|ke)\s*)?Avenatti',
         emailers = {
             ALAN_DERSHOWITZ: 'Harvard Law School professor and all around (in)famous American lawyer',
@@ -302,7 +304,7 @@ HIGHLIGHTED_NAMES = [
     HighlightedNames(
         label=FINANCE,
         style='green',
-        pattern=r'Apollo|Ari\s*Glass|Bank|(Bernie\s*)?Madoff|Black(rock|stone)|B\s*of\s*A|Boothbay(\sFund\sManagement)?|Chase\s*Bank|Credit\s*Suisse|DB|Deutsche\s*(Asset|Bank)|Electron\s*Capital\s*(Partners)?|Fenner|FRBNY|Goldman(\s*Sachs)|HSBC|Invesco|(Janet\s*)?Yellen|(Jerome\s*)?Powell(?!M\. Cabot)|(Jimmy\s*)?Cayne|JPMC?|j\.?p\.?\s*morgan(\.?com|\s*Chase)?|Madoff|Merrill(\s*Lynch)?|(Michael\s*)?(Cembalest|Milken)|Mizrahi\s*Bank|MLPF&S|((anti.?)?money\s+)?launder(s?|ers?|ing)?(\s+money)?|Morgan Stanley|(Peter L. )?Scher|(Ray\s*)?Dalio|(Richard\s*)?LeFrak|Schwartz?man|Serageldin|UBS|us.gio@jpmorgan.com',
+        pattern=r'Apollo|Ari\s*Glass|Bank|(Bernie\s*)?Madoff|Black(rock|stone)|B\s*of\s*A|Boothbay(\sFund\sManagement)?|Chase\s*Bank|Credit\s*Suisse|DB|Deutsche?\s*(Asset|Bank)|Electron\s*Capital\s*(Partners)?|Fenner|FRBNY|Goldman(\s*Sachs)|HSBC|Invesco|(Janet\s*)?Yellen|(Jerome\s*)?Powell(?!M\. Cabot)|(Jimmy\s*)?Cayne|JPMC?|j\.?p\.?\s*morgan(\.?com|\s*Chase)?|Madoff|Merrill(\s*Lynch)?|(Michael\s*)?(Cembalest|Milken)|Mizrahi\s*Bank|MLPF&S|((anti.?)?money\s+)?launder(s?|ers?|ing)?(\s+money)?|Morgan Stanley|(Peter L. )?Scher|(Ray\s*)?Dalio|(Richard\s*)?LeFrak|Schwartz?man|Serageldin|UBS|us.gio@jpmorgan.com',
         emailers={
             AMANDA_ENS: 'Citigroup',
             BRAD_WECHSLER: f"head of {LEON_BLACK}'s personal investment vehicle according to FT",
@@ -316,7 +318,8 @@ HIGHLIGHTED_NAMES = [
             MELANIE_SPINELLA: f'representative of {LEON_BLACK}',
             MORTIMER_ZUCKERMAN: 'business partner of Epstein',
             PAUL_BARRETT: None,
-            PAUL_MORRIS: 'Deutsche Bank',
+            PAUL_MORRIS: DEUTSCHE_BANK,
+            'Steven Elkman': DEUTSCHE_BANK,
         }
     ),
     HighlightedNames(
@@ -334,7 +337,7 @@ HIGHLIGHTED_NAMES = [
         },
     ),
     HighlightedNames(
-        label=HARVARD.lower(),
+        label=HARVARD,
         style='light_goldenrod3',
         pattern=r'Cambridge|(Derek\s*)?Bok|Elisa(\s*New)?|Harvard(\s*(Business|Law|University)(\s*School)?)?|(Jonathan\s*)?Zittrain|(Stephen\s*)?Kosslyn',
         emailers = {
@@ -349,7 +352,7 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='india',
+        label='India',
         style='bright_green',
         pattern=r'Abraaj|Anna\s*Hazare|(Arif\s*)?Naqvi|(Arvind\s*)?Kejriwal|Hardeep( Pur[ei]e)?|Indian?|InsightsPod|Modi|Mumbai|Tranchulas',
         emailers = {
@@ -359,7 +362,7 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='israel',
+        label='Israel',
         style='dodger_blue2',
         pattern=r"AIPAC|Bibi|(eh|(Ehud|Nili Priell) )?barak|Ehud\s*Barack|Israeli?|Jerusalem|J\s*Street|Mossad|Netanyahu|(Sheldon\s*)?Adelson|Tel\s*Aviv|(The\s*)?Shimon\s*Post|Yitzhak|Rabin|YIVO|zionist",
         emailers={
@@ -369,7 +372,7 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='japan',
+        label='Japan',
         style='color(168)',
         pattern=r'BOJ|(Bank\s+of\s+)?Japan(ese)?|jpy?(?! Morgan)|SG|Singapore|Toky[op]',
     ),
@@ -389,14 +392,14 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='latin america',
+        label='Latin America',
         style='yellow',
         pattern=r'Argentin(a|ian)|Bolsonar[aio]|Bra[sz]il(ian)?|Bukele|Caracas|Castro|Colombian?|Cuban?|El\s*Salvador|((Enrique )?Pena )?Nieto|LatAm|Lula|Mexic(an|o)|(Nicolas\s+)?Maduro|Panama( Papers)?|Peru|Venezuelan?|Zambrano',
     ),
     HighlightedNames(
         label='law enforcement',
         style='color(24) bold',
-        pattern=r'ag|(Alicia\s*)?Valle|AML|(Andrew\s*)?McCabe|attorney|((Bob|Robert)\s*)?Mueller|(Byung\s)?Pak|CFTC?|CIA|CIS|CVRA|Dep(artmen)?t\.?\s*of\s*(the\s*)?(Justice|Treasury)|DHS|DOJ|FBI|FCPA|FDIC|Federal\s*Bureau\s*of\s*Investigation|FinCEN|FINRA|FOIA|FTC|IRS|(James\s*)?Comey|(Jennifer\s*Shasky\s*)?Calvery|((Judge|Mark)\s*)?(Carney|Filip)|(Kirk )?Blouin|KYC|NIH|NS(A|C)|OCC|OFAC|(Lann?a\s*)?Belohlavek|lawyer|(Michael\s*)?Reiter|OGE|Office\s*of\s*Government\s*Ethics|Police Code Enforcement|(Preet\s*)?Bharara|SCOTUS|SD(FL|NY)|SEC|Secret\s*Service|Securities\s*and\s*Exchange\s*Commission|Southern\s*District\s*of\s*(Florida|New\s*York)|State\s*Dep(artmen)?t|Strzok|Supreme\s*Court|Treasury\s*(Dep(artmen)?t|Secretary)|TSA|USAID|(William\s*J\.?\s*)?Zloch',
+        pattern=r'ag|(Alicia\s*)?Valle|AML|(Andrew\s*)?McCabe|((Bob|Robert)\s*)?Mueller|(Byung\s)?Pak|CFTC?|CIA|CIS|CVRA|Dep(artmen)?t\.?\s*of\s*(the\s*)?(Justice|Treasury)|DHS|DOJ|FBI|FCPA|FDIC|Federal\s*Bureau\s*of\s*Investigation|FinCEN|FINRA|FOIA|FTC|IRS|(James\s*)?Comey|(Jennifer\s*Shasky\s*)?Calvery|((Judge|Mark)\s*)?(Carney|Filip)|(Kirk )?Blouin|KYC|NIH|NS(A|C)|OCC|OFAC|(Lann?a\s*)?Belohlavek|(Michael\s*)?Reiter|OGE|Office\s*of\s*Government\s*Ethics|Police Code Enforcement|(Preet\s*)?Bharara|SCOTUS|SD(FL|NY)|SEC|Secret\s*Service|Securities\s*and\s*Exchange\s*Commission|Southern\s*District\s*of\s*(Florida|New\s*York)|State\s*Dep(artmen)?t|Strzok|Supreme\s*Court|Treasury\s*(Dep(artmen)?t|Secretary)|TSA|USAID|(William\s*J\.?\s*)?Zloch',
         emailers = {
             ANN_MARIE_VILLAFANA: 'southern district of Florida U.S. Attorney',
             DANNY_FROST: 'Director of Communications at Manhattan DA',
@@ -419,8 +422,8 @@ HIGHLIGHTED_NAMES = [
     HighlightedNames(
         label='mideast',
         style='dark_sea_green4',
-        # this won't match ever because of word boundary: [-\s]9/11[\s.]
-        pattern=r"Abdulmalik Al-Makhlafi|Abdullah|Abu\s+Dhabi|Afghanistan|Al[-\s]?Qa[ei]da|Ahmadinejad|Arab|Aramco|Assad|Bahrain|Basiji?|Benghazi|Cairo|Chagoury|Dj[iu]bo?uti|Doha|Dubai|Egypt(ian)?|Emir(at(es?|i))?|Erdogan|Fashi|Gaddafi|(Hamid\s*)?Karzai|Hamad\s*bin\s*Jassim|HBJ|Houthi|Imran\s+Khan|Iran(ian)?|Isi[ls]|Islam(abad|ic|ist)?|Istanbul|Kh?ashoggi|(Kairat\s*)?Kelimbetov|kasshohgi|Kaz(akh|ich)stan|Kazakh?|Kh[ao]menei|Khalid\s*Sheikh\s*Mohammed|KSA|Leban(ese|on)|Libyan?|Mahmoud|Marra[hk]e[cs]h|MB(N|S|Z)|Mohammed\s+bin\s+Salman|Morocco|Mubarak|Muslim|Nayaf|Pakistani?|Omar|(Osama\s*)?Bin\s*Laden|Osama(?! al)|Palestin(e|ian)|Persian?|Riya(dh|nd)|Saddam|Salman|Saudi(\s+Arabian?)?|Shariah?|SHC|sheikh|shia|(Sultan\s*)?Yacoub|Syrian?|(Tarek\s*)?El\s*Sayed|Tehran|Tunisian?|Turk(ey|ish)|UAE|((Iraq|Iran|Kuwait|Qatar|Yemen)i?)",
+        # something like this won't match ever because of word boundary: [-\s]9/11[\s.]
+        pattern=r"Abdulmalik Al-Makhlafi|Abdullah|Abu\s+Dhabi|Afghanistan|Al[-\s]?Qa[ei]da|Ahmadinejad|Arab|Aramco|Assad|Bahrain|Basiji?|Benghazi|Cairo|Chagoury|Dj[iu]bo?uti|Doha|Dubai|Egypt(ian)?|Emir(at(es?|i))?|Erdogan|Fashi|Gaddafi|(Hamid\s*)?Karzai|Hamad\s*bin\s*Jassim|HBJ|Houthi|Imran\s+Khan|Iran(ian)?|Isi[ls]|Islam(abad|ic|ist)?|Istanbul|Kh?ashoggi|(Kairat\s*)?Kelimbetov|kasshohgi|Kaz(akh|ich)stan|Kazakh?|Kh[ao]menei|Khalid\s*Sheikh\s*Mohammed|KSA|Leban(ese|on)|Libyan?|Mahmoud|Marra[hk]e[cs]h|MB(N|S|Z)|Mid(dle)?\s*East|Mohammed\s+bin\s+Salman|Morocco|Mubarak|Muslim|Nayaf|Pakistani?|Omar|(Osama\s*)?Bin\s*Laden|Osama(?! al)|Palestin(e|ian)|Persian?|Riya(dh|nd)|Saddam|Salman|Saudi(\s+Arabian?)?|Shariah?|SHC|sheikh|shia|(Sultan\s*)?Yacoub|Syrian?|(Tarek\s*)?El\s*Sayed|Tehran|Tunisian?|Turk(ey|ish)|UAE|((Iraq|Iran|Kuwait|Qatar|Yemen)i?)",
         emailers = {
             ANAS_ALRASHEED: f'former information minister of Kuwait {QUESTION_MARKS}',
             AZIZA_ALAHMADI: 'Abu Dhabi Department of Culture & Tourism',
@@ -463,7 +466,7 @@ HIGHLIGHTED_NAMES = [
         }
     ),
     HighlightedNames(
-        label='republicans',
+        label='Republicans',
         style='bold dark_red',
         pattern=r'Alberto\sGonzale[sz]|(Alex\s*)?Acosta|(Bill\s*)?Barr|Bill\s*Shine|(Bob\s*)?Corker|(John\s*(R.?\s*)?)Bolton|Broidy|(Chris\s)?Christie|Devin\s*Nunes|(Don\s*)?McGa[hn]n|McMaster|(George\s*)?Nader|GOP|(Brett\s*)?Kavanaugh|Kissinger|Kobach|Koch\s*Brothers|Kolfage|Kudlow|Lewandowski|(Marco\s)?Rubio|(Mark\s*)Meadows|Mattis|McCain|(?<!Merwin Dela )Cruz|(Michael\s)?Hayden|((General|Mike)\s*)?(Flynn|Pence)|(Mitt\s*)?Romney|Mnuchin|Nikki|Haley|(Paul\s+)?(Manafort|Volcker)|(Peter\s)?Navarro|Pompeo|Reagan|Reince|Priebus|Republican|(Rex\s*)?Tillerson|(?<!Cynthia )(Richard\s*)?Nixon|Sasse|Tea\s*Party',
         # There's no emails from these people, they're just here to automate the regex creation for both first + last names
@@ -481,7 +484,7 @@ HIGHLIGHTED_NAMES = [
         },
     ),
     HighlightedNames(
-        label='russia',
+        label='Russia',
         style='red bold',
         pattern=r'Alfa\s*Bank|Anya\s*Rasulova|Chernobyl|Day\s+One\s+Ventures|(Dmitry\s)?(Kiselyov|(Lana\s*)?Pozhidaeva|Medvedev|Rybolo(o?l?ev|vlev))|Dmitry|FSB|GRU|KGB|Kislyak|Kremlin|Kuznetsova|Lavrov|Lukoil|Moscow|(Oleg\s*)?Deripaska|Oleksandr Vilkul|Rosneft|RT|St.?\s*?Petersburg|Russian?|Sberbank|Soviet(\s*Union)?|USSR|Vladimir|(Vladimir\s*)?(Putin|Yudashkin)|Women\s*Empowerment|Xitrans',
         emailers = {
@@ -511,7 +514,7 @@ HIGHLIGHTED_NAMES = [
         },
     ),
     HighlightedNames(
-        label='southeast_asia',
+        label='southeast Asia',
         style='light_salmon3 bold',
         pattern=r'Bangkok|Burm(a|ese)|Cambodian?|Laos|Malaysian?|Myan?mar|Thai(land)?|Vietnam(ese)?',
     ),
@@ -541,7 +544,7 @@ HIGHLIGHTED_NAMES = [
         pattern=r'BVI|(Jane|Tiffany)\s*Doe|Katie\s*Johnson|(Virginia\s+((L\.?|Roberts)\s+)?)?Giuffre|Virginia\s+Roberts',
     ),
     HighlightedNames(
-        label='victim_lawyer',
+        label='victim lawyer',
         style='dark_magenta bold',
         pattern=r'(Alan(\s*P.)?|MINTZ)\s*FRAADE|Paul\s*(G.\s*)?Cassell|Rothstein\s*Rosenfeldt\s*Adler|(Scott\s*)?Rothstein|(J\.?\s*)?(Stan(ley)?\s*)?Pottinger',
         emailers = {
@@ -554,7 +557,7 @@ HIGHLIGHTED_NAMES = [
         style='sea_green1',
         pattern=r'Antigua|Bahamas|Caribb?ean|Dominican\s*Republic|(Great|Little)\s*St.?\s*James|Haiti(an)?|(John\s*)deJongh(\s*Jr\.?)|(Kenneth E\. )?Mapp|Palm\s*Beach(?!\s*Post)|PBI|S(ain)?t.?\s*Thomas|USVI|(?<!Epstein )VI|(The\s*)?Virgin\s*Islands(\s*Daily\s*News)?',  # TODO: VI Daily News should be yellow but it's hard bc Daily News xists
         emailers = {
-            CECILE_DE_JONGH: f'First lady 2007-2015',
+            CECILE_DE_JONGH: f'first lady 2007-2015',
             STACEY_PLASKETT: 'non-voting member of Congress',
             KENNETH_E_MAPP: 'Governor',
         },
@@ -573,6 +576,9 @@ HIGHLIGHTED_NAMES = [
         label=STEVE_BANNON,
         style='color(58)',
         pattern=r'((Steve|Sean)\s*)?Bannon?|(American\s*)?Dharma',
+        emailers = {
+            STEVE_BANNON: 'Trump campaign manager, memecoin grifter, convicted criminal',
+        }
     ),
     HighlightedNames(
         emailers={STEVEN_HOFFENBERG: HEADER_ABBREVIATIONS['Hoffenberg']},
@@ -646,7 +652,7 @@ ALL_HIGHLIGHTS = HIGHLIGHTED_NAMES + HIGHLIGHTED_TEXTS
 
 
 class EpsteinHighlighter(RegexHighlighter):
-    """rich.highlighter that finds and colors interesting keywords based on the above config."""
+    """Finds and colors interesting keywords based on the above config."""
     base_style = f"{REGEX_STYLE_PREFIX}."
     highlights = [highlight_group.regex for highlight_group in ALL_HIGHLIGHTS]
 
@@ -666,10 +672,8 @@ def get_style_for_category(category: str) -> str | None:
     elif category == SOCIAL:
         return get_style_for_category(PUBLICIST)
 
-    category = CATEGORY_STYLE_MAPPING.get(category, category)
-
     for highlight_group in HIGHLIGHTED_NAMES:
-        if highlight_group.label == category:
+        if highlight_group.label == CATEGORY_STYLE_MAPPING.get(category, category):
             return highlight_group.style
 
 

epstein_files/util/output.py

@@ -4,13 +4,14 @@ from rich.padding import Padding
 
 from epstein_files.documents.email import Email
 from epstein_files.documents.messenger_log import MessengerLog
+from epstein_files.documents.other_file import FIRST_FEW_LINES, OtherFile
 from epstein_files.epstein_files import EpsteinFiles, count_by_month
 from epstein_files.util.constant import output_files
 from epstein_files.util.constant.html import *
 from epstein_files.util.constant.names import *
 from epstein_files.util.constant.output_files import JSON_FILES_JSON_PATH, JSON_METADATA_PATH
 from epstein_files.util.data import dict_sets_to_lists
-from epstein_files.util.env import args, specified_names
+from epstein_files.util.env import args
 from epstein_files.util.file_helper import log_file_write
 from epstein_files.util.logging import logger
 from epstein_files.util.rich import *
@@ -20,34 +21,33 @@ PRINT_COLOR_KEY_EVERY_N_EMAILS = 150
 # Order matters. Default names to print emails for.
 DEFAULT_EMAILERS = [
     JEREMY_RUBIN,
-    AL_SECKEL,
     JOI_ITO,
     JABOR_Y,
     STEVEN_SINOFSKY,
+    AL_SECKEL,
     DANIEL_SIAD,
     JEAN_LUC_BRUNEL,
     STEVEN_HOFFENBERG,
+    RENATA_BOLOTOVA,
+    MASHA_DROKOVA,
     EHUD_BARAK,
     MARTIN_NOWAK,
-    MASHA_DROKOVA,
-    RENATA_BOLOTOVA,
     STEVE_BANNON,
-    PRINCE_ANDREW,
     JIDE_ZEITLIN,
     DAVID_STERN,
     MOHAMED_WAHEED_HASSAN,
     JENNIFER_JACQUET,
     TYLER_SHEARS,
     CHRISTINA_GALBRAITH,
+    ZUBAIR_KHAN,
     None,
 ]
 
 # Order matters. Default names to print tables w/email subject, timestamp, etc for. # TODO: get rid of this ?
 DEFAULT_EMAILER_TABLES: list[str | None] = [
     GHISLAINE_MAXWELL,
-    LEON_BLACK,
+    PRINCE_ANDREW,
     SULTAN_BIN_SULAYEM,
-    DEEPAK_CHOPRA,
     ARIANE_DE_ROTHSCHILD,
 ]
 
@@ -55,17 +55,17 @@ if len(set(DEFAULT_EMAILERS).intersection(set(DEFAULT_EMAILER_TABLES))) > 0:
     raise RuntimeError(f"Some names appear in both DEFAULT_EMAILERS and DEFAULT_EMAILER_TABLES")
 
 
-def print_emails(epstein_files: EpsteinFiles) -> int:
-    """Returns number of emails printed."""
+def print_emails_section(epstein_files: EpsteinFiles) -> list[Email]:
+    """Returns emails that were printed (may contain dupes if printed for both author and recipient)."""
     print_section_header(('Selections from ' if not args.all_emails else '') + 'His Emails')
-    print_all_files_page_link(epstein_files)
+    print_other_page_link(epstein_files)
     emailers_to_print: list[str | None]
     emailer_tables: list[str | None] = []
     already_printed_emails: list[Email] = []
     num_emails_printed_since_last_color_key = 0
 
-    if specified_names:
-        emailers_to_print = specified_names
+    if args.names:
+        emailers_to_print = args.names
     else:
         print_centered(Padding(epstein_files.table_of_emailers(), (2, 0)))
 
@@ -97,7 +97,7 @@ def print_emails(epstein_files: EpsteinFiles) -> int:
         for name in DEFAULT_EMAILER_TABLES:
             epstein_files.print_emails_table_for(name)
 
-    if not specified_names:
+    if not args.names:
         epstein_files.print_email_device_info()
 
     if args.all_emails:
@@ -106,7 +106,7 @@ def print_emails(epstein_files: EpsteinFiles) -> int:
     fwded_articles = [e for e in already_printed_emails if e.config and e.is_fwded_article()]
     log_msg = f"Rewrote {len(Email.rewritten_header_ids)} of {len(already_printed_emails)} email headers"
     logger.warning(f"{log_msg}, {len(fwded_articles)} of the emails were forwarded articles.")
-    return len(already_printed_emails)
+    return already_printed_emails
 
 
 def print_json_files(epstein_files: EpsteinFiles):
@@ -136,6 +136,39 @@ def print_json_stats(epstein_files: EpsteinFiles) -> None:
     print_json("count_by_month", count_by_month(epstein_files.all_documents()))
 
 
+def print_other_files_section(files: list[OtherFile], epstein_files: EpsteinFiles) -> None:
+    """Returns the OtherFile objects that were interesting enough to print."""
+    category_table = OtherFile.count_by_category_table(files)
+    other_files_preview_table = OtherFile.files_preview_table(files)
+    header_pfx = '' if args.all_other_files else 'Selected '
+    print_section_header(f"{FIRST_FEW_LINES} of {len(files)} {header_pfx}Files That Are Neither Emails Nor Text Messages")
+
+    if args.all_other_files:
+        console.line(1)
+    else:
+        print_other_page_link(epstein_files)
+        console.line(2)
+
+        for table in [category_table, other_files_preview_table]:
+            table.title = f"{header_pfx}{table.title}"
+
+    print_centered(category_table)
+    console.line(2)
+    console.print(other_files_preview_table)
+
+
+def print_text_messages_section(epstein_files: EpsteinFiles) -> None:
+    """Print summary table and stats for text messages."""
+    print_section_header('All of His Text Messages')
+    print_centered("(conversations are sorted chronologically based on timestamp of first message)\n", style='gray30')
+
+    for log_file in epstein_files.imessage_logs:
+        console.print(Padding(log_file))
+        console.line(2)
+
+    print_centered(MessengerLog.summary_table(epstein_files.imessage_logs))
+
+
 def write_json_metadata(epstein_files: EpsteinFiles) -> None:
     json_str = epstein_files.json_metadata()
 

epstein_files/util/rich.py

@@ -143,13 +143,6 @@ def parenthesize(msg: str | Text, style: str = '') -> Text:
     return Text('(', style=style).append(txt).append(')')
 
 
-def print_all_files_page_link(epstein_files: 'EpsteinFiles') -> None:
-    markup_msg = link_markup(SITE_URLS[EMAIL], 'the other page', style='light_slate_grey bold')
-    txt = Text.from_markup(markup_msg).append(f' is uncurated and has all {len(epstein_files.other_files)}')
-    txt.append(f" unclassifiable files and {len(epstein_files.emails):,} emails")
-    print_centered(parenthesize(txt), style='dim')
-
-
 def print_author_header(msg: str, color: str | None, footer: str | None = None) -> None:
     txt = Text(msg, justify='center')
     color = color or 'white'
@@ -195,14 +188,12 @@ def print_header(epstein_files: 'EpsteinFiles') -> None:
 
     console.print(f"{not_optimized_msg}.\n", style='dim')
     print_page_title(width=TITLE_WIDTH)
-    site_type: SiteType = EMAIL if args.all_emails else TEXT_MESSAGE
+    site_type = EMAIL if args.all_emails else TEXT_MESSAGE
     print_starred_header(f"This is the Epstein {site_type.title()}s site", num_spaces=4, num_stars=14)
-    other_site_type: SiteType = TEXT_MESSAGE if site_type == EMAIL else EMAIL
-    other_site_msg = "another site with" + (' all of' if other_site_type == EMAIL else '')
+    other_site_msg = "another site with" + (' all of' if other_site_type() == EMAIL else '')
     other_site_msg += f" Epstein's {other_site_type}s also generated by this code"
-
-    markup_msg = link_markup(SITE_URLS[other_site_type], other_site_msg, OTHER_SITE_LINK_STYLE)
-    print_centered(parenthesize(Text.from_markup(markup_msg)), style='bold')
+    other_site_link_markup = link_markup(other_site_url(), other_site_msg, OTHER_SITE_LINK_STYLE)
+    print_centered(parenthesize(Text.from_markup(other_site_link_markup)), style='bold')
     word_count_link = link_text_obj(WORD_COUNT_URL, 'most frequently used words in the emails and texts', AUX_SITE_LINK_STYLE)
     print_centered(parenthesize(word_count_link))
     metadata_link = link_text_obj(JSON_METADATA_URL, 'author attribution explanations', AUX_SITE_LINK_STYLE)
@@ -272,6 +263,19 @@ def print_numbered_list_of_emailers(_list: list[str | None], epstein_files = Non
     console.line()
 
 
+def print_other_page_link(epstein_files: 'EpsteinFiles') -> None:
+    markup_msg = link_markup(other_site_url(), 'the other page', style='light_slate_grey bold')
+
+    if other_site_type() == EMAIL:
+        txt = Text.from_markup(markup_msg).append(f' is uncurated and has all {len(epstein_files.other_files)}')
+        txt.append(f" unclassifiable files and {len(epstein_files.emails):,} emails")
+    else:
+        txt = Text.from_markup(markup_msg).append(f' displays only a small collection of emails and')
+        txt.append(" unclassifiable files of particular interest")
+
+    print_centered(parenthesize(txt), style='dim')
+
+
 def print_page_title(expand: bool = True, width: int | None = None) -> None:
     title_panel = Panel(Text(PAGE_TITLE, justify='center'), expand=expand, style=TITLE_STYLE, width=width)
     console.print(Align.center(vertically_pad(title_panel)))

epstein_files/util/word_count.py

@@ -14,7 +14,7 @@ from epstein_files.util.constant.common_words import COMMON_WORDS_LIST, COMMON_W
 from epstein_files.util.constant.names import OTHER_NAMES
 from epstein_files.util.constant.output_files import WORD_COUNT_HTML_PATH
 from epstein_files.util.data import ALL_NAMES, flatten, sort_dict
-from epstein_files.util.env import args, specified_names
+from epstein_files.util.env import args
 from epstein_files.util.logging import logger
 from epstein_files.util.rich import (console, highlighter, print_centered, print_color_key, print_page_title,
      print_panel, print_starred_header, write_html)
@@ -201,7 +201,7 @@ def write_word_counts_html() -> None:
     emails = [e for e in epstein_files.non_duplicate_emails() if not (e.is_junk_mail() or e.is_fwded_article())]
 
     for email in emails:
-        if specified_names and email.author not in specified_names:
+        if args.names and email.author not in args.names:
             continue
 
         logger.info(f"Counting words in {email}\n  [SUBJECT] {email.subject()}")
@@ -218,14 +218,12 @@ def write_word_counts_html() -> None:
             for word in line.split():
                 word_count.tally_word(word, SearchResult(email, [MatchedLine(line, i)]))
 
-    # Add in iMessage conversation words
-    imessage_logs = epstein_files.imessage_logs_for(specified_names) if specified_names else epstein_files.imessage_logs
-
-    for imessage_log in imessage_logs:
+    # Add in iMessage conversations
+    for imessage_log in epstein_files.imessage_logs:
         logger.info(f"Counting words in {imessage_log}")
 
         for i, msg in enumerate(imessage_log.messages):
-            if specified_names and msg.author not in specified_names:
+            if args.names and msg.author not in args.names:
                 continue
             elif HTML_REGEX.search(line):
                 continue
@@ -234,7 +232,7 @@ def write_word_counts_html() -> None:
                 word_count.tally_word(word, SearchResult(imessage_log, [MatchedLine(msg.text, i)]))
 
     print_page_title(expand=False)
-    print_starred_header(f"Most Common Words in {len(emails):,} Emails and {len(imessage_logs)} iMessage Logs")
+    print_starred_header(f"Most Common Words in {len(emails):,} Emails and {len(epstein_files.imessage_logs)} iMessage Logs")
     print_centered(f"(excluding {len(COMMON_WORDS_LIST)} particularly common words at bottom)", style='dim')
     console.line()
     print_color_key()

{epstein_files-1.0.14.dist-info → epstein_files-1.0.16.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: epstein-files
-Version: 1.0.14
+Version: 1.0.16
 Summary: Tools for working with the Jeffrey Epstein documents released in November 2025.
 Home-page: https://michelcrypt4d4mus.github.io/epstein_text_messages/
 License: GPL-3.0-or-later

epstein_files-1.0.16.dist-info/RECORD

@@ -0,0 +1,33 @@
+epstein_files/__init__.py,sha256=nOJ26LghS0sCdQ4qGwklVm18kbuFRn6Uu1-JDE8jwd4,5050
+epstein_files/documents/communication.py,sha256=oqNsSDWe-N0jSmchIHxpihihzIWha-foFqMwKZlxyng,2057
+epstein_files/documents/document.py,sha256=5GdM8grhaaefmXeV26XwYRZvMNqFETFnE7mFpVGGNik,17249
+epstein_files/documents/email.py,sha256=fp8hwXoBvc0xMXoddOBFQZLjzUNwKxHdnRnk0htsQ-g,41186
+epstein_files/documents/emails/email_header.py,sha256=wkPfSLbmzkAeQwvhf0bAeFDLPbQT-EeG0v8vNNLYktM,7502
+epstein_files/documents/imessage/text_message.py,sha256=icIiKRRuZapkV9r_PID_7hEfy7YvPrIm9Emc4QiYxbw,2806
+epstein_files/documents/json_file.py,sha256=WcZW5NNqA67rHTdopbOGtup00muNaLlvrNgKb-K4zO8,1504
+epstein_files/documents/messenger_log.py,sha256=zRVVe82cNmNmg50kqScH38yGGoSx8NjCWfLwFGUb0rs,6501
+epstein_files/documents/other_file.py,sha256=CCwOYsipYTWZj8pSTl0kgUy_LRu7Z5ZuWygQEYhilNA,9778
+epstein_files/epstein_files.py,sha256=LHApXzJBKauK4Z8Bn00FPr4A83VTdITnGqLn0W_vpnQ,17883
+epstein_files/util/constant/common_words.py,sha256=aR0UjoWmxyR49XS-DtHECQ1CiA_bK8hNP6CQ1TS9yZA,3696
+epstein_files/util/constant/html.py,sha256=9U098TGzlghGg4WfxLYHyub5JGR17Dv7VP5i2MSu8Kk,1415
+epstein_files/util/constant/names.py,sha256=M9sPZwjK1G4Jii4HJpWpUa5Y6lZ7CSJAMqRdpg5YdQg,10505
+epstein_files/util/constant/output_files.py,sha256=BkV4_gmdj46RfGy5SFYp6dgTty3FtlBth5YGmaGutls,1700
+epstein_files/util/constant/strings.py,sha256=rOKmYrGIZFPQbhCgd8z7mUF0CbQQBO4ij8kXvGTDkoE,1857
+epstein_files/util/constant/urls.py,sha256=-cylNL7xJi18a6fawVTF8MO3oEdjMvo9Hg1MwC-7ydI,5098
+epstein_files/util/constants.py,sha256=4WABkEBshduSV4cZkxv0MwgJci9g6MxjhtJfPjersEo,112495
+epstein_files/util/data.py,sha256=JccGFZGiCGm7XtwpQTocIjGYOr6hTUpEPwHhjyW9Xnc,3164
+epstein_files/util/doc_cfg.py,sha256=aBIm0hyxf-aeMsb8ZUNiQFVsPFimjVUIkrVdDrg1iQU,9105
+epstein_files/util/env.py,sha256=aSUC_TwdtKB8z6PCDaLeu3k5qZE3AgcsP9jU2p9eaP0,5640
+epstein_files/util/file_helper.py,sha256=uJzNHihuAi2-XMNbUdjdovgppcIt-fVIIrXHXVShA6Q,2904
+epstein_files/util/highlighted_group.py,sha256=Tm4Fy7oZv3PLWW1yCaRCY2NDfr2RM8zB9sudvqw11lw,36746
+epstein_files/util/logging.py,sha256=fuREq06xUUI3DfCV2JE-8QM-sQKxpLDj0_AYFO6qR1M,1983
+epstein_files/util/output.py,sha256=hnWK0YHEGC5GSlcQA80ZpxUnXV42S0ytKDARl1qk-0E,9012
+epstein_files/util/rich.py,sha256=13sPSve9UEqo51t_5hf86Z492ju4OCHWMxrz6r1PnNQ,15338
+epstein_files/util/search_result.py,sha256=1fxe0KPBQXBk4dLfu6m0QXIzYfZCzvaSkWqvghJGzxY,567
+epstein_files/util/timer.py,sha256=8hxW4Y1JcTUfnBrHh7sL2pM9xu1sL4HFQM4CmmzTarU,837
+epstein_files/util/word_count.py,sha256=o_-HnfzHdPDPR8oA_dv6fjy1dbsHee8p_aoe62PEQHw,9213
+epstein_files-1.0.16.dist-info/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
+epstein_files-1.0.16.dist-info/METADATA,sha256=_tOJD6MZxx4QAja1cVCNOn61_5675M6DTtvFoXQx9UI,5867
+epstein_files-1.0.16.dist-info/WHEEL,sha256=d2fvjOD7sXsVzChCqf0Ty0JbHKBaLYwDbGQDwQTnJ50,88
+epstein_files-1.0.16.dist-info/entry_points.txt,sha256=5qYgwAXpxegeAicD_rzda_trDRnUC51F5UVDpcZ7j6Q,240
+epstein_files-1.0.16.dist-info/RECORD,,

epstein_files-1.0.14.dist-info/RECORD

@@ -1,33 +0,0 @@
-epstein_files/__init__.py,sha256=9mClHe3YkTWHulLOZIO2uaxYcff-krpEsc0vphYlwps,4975
-epstein_files/documents/communication.py,sha256=oqNsSDWe-N0jSmchIHxpihihzIWha-foFqMwKZlxyng,2057
-epstein_files/documents/document.py,sha256=1qL1V5VrrtI7uuuLvOw-JU83khRzzEURduuoh_pfghI,17114
-epstein_files/documents/email.py,sha256=eKfDGwbP1nYVfhozM0bSVf-5T-_l_k-0D_HVrQldm38,40904
-epstein_files/documents/emails/email_header.py,sha256=wkPfSLbmzkAeQwvhf0bAeFDLPbQT-EeG0v8vNNLYktM,7502
-epstein_files/documents/imessage/text_message.py,sha256=JPHlO-5Xgt5V5E9fDsgCEr3XOp7FyzlhIHmZhFNZcmg,2806
-epstein_files/documents/json_file.py,sha256=WcZW5NNqA67rHTdopbOGtup00muNaLlvrNgKb-K4zO8,1504
-epstein_files/documents/messenger_log.py,sha256=le-cj_ZBDUb_jnxP_F05KlnPjh0nDbgpPSWl4-HL-J4,6502
-epstein_files/documents/other_file.py,sha256=tJmEH3cGeEcRIZW5F-RuSJprYM_2r1jEbsF_kjSsA3E,9788
-epstein_files/epstein_files.py,sha256=aO2vntbA7DonSk96xB5us5VaDSpp79Mpu8eYJ_g_8Qo,20025
-epstein_files/util/constant/common_words.py,sha256=aR0UjoWmxyR49XS-DtHECQ1CiA_bK8hNP6CQ1TS9yZA,3696
-epstein_files/util/constant/html.py,sha256=9U098TGzlghGg4WfxLYHyub5JGR17Dv7VP5i2MSu8Kk,1415
-epstein_files/util/constant/names.py,sha256=smC9xsR9E9WECXi_ygK_6lrcEPZWZINLZrp_-BthUtA,10377
-epstein_files/util/constant/output_files.py,sha256=BkV4_gmdj46RfGy5SFYp6dgTty3FtlBth5YGmaGutls,1700
-epstein_files/util/constant/strings.py,sha256=etCmLZPuriVYDap4iG_0Nfd92P1cvRpK2R7ZpVAGDrw,1942
-epstein_files/util/constant/urls.py,sha256=EKzpL1np3cdyYvUlgv1zl9IJU8Vrg8jXpAu6YpM0_Bk,5283
-epstein_files/util/constants.py,sha256=W-4ibPdv8I62uNyoBUR7sWWiqUWWr2n414WnZNA_Who,112605
-epstein_files/util/data.py,sha256=JccGFZGiCGm7XtwpQTocIjGYOr6hTUpEPwHhjyW9Xnc,3164
-epstein_files/util/doc_cfg.py,sha256=uVUgZF7p5QMhbPEVauQWlRWQRhP0Imr5XaNQtMk0E7c,9053
-epstein_files/util/env.py,sha256=zg-TJ8ea6nTa0PyYgY97Zksre8au1ELv2i8vOizZt7Y,5645
-epstein_files/util/file_helper.py,sha256=tacTe1GcAnckPFvjMgxRRSLnFgr2aVIYsgfDR_C9uXk,2780
-epstein_files/util/highlighted_group.py,sha256=mNcdML0h88e2voEiUJxpgAVXHh3jQ8f0XV8qVdZMkBM,36550
-epstein_files/util/logging.py,sha256=fuREq06xUUI3DfCV2JE-8QM-sQKxpLDj0_AYFO6qR1M,1983
-epstein_files/util/output.py,sha256=bb_-jRoFNNMHXPSaptEXdkbqNDJYOpNSs6dkJKvi_HQ,7581
-epstein_files/util/rich.py,sha256=bQPlOEZ6qTpVF4yPnwmlBk4K11zH5RJjwbWeBIOUUGE,15191
-epstein_files/util/search_result.py,sha256=1fxe0KPBQXBk4dLfu6m0QXIzYfZCzvaSkWqvghJGzxY,567
-epstein_files/util/timer.py,sha256=8hxW4Y1JcTUfnBrHh7sL2pM9xu1sL4HFQM4CmmzTarU,837
-epstein_files/util/word_count.py,sha256=dTBNqX98sj7W46O0uit76I4hYBhm0bsY1Q7RHwWehiU,9349
-epstein_files-1.0.14.dist-info/LICENSE,sha256=OXLcl0T2SZ8Pmy2_dmlvKuetivmyPd5m1q-Gyd-zaYY,35149
-epstein_files-1.0.14.dist-info/METADATA,sha256=AUw8p6SQuvkJONsHErig0L--FyAU2y892DYpdpimqtg,5867
-epstein_files-1.0.14.dist-info/WHEEL,sha256=d2fvjOD7sXsVzChCqf0Ty0JbHKBaLYwDbGQDwQTnJ50,88
-epstein_files-1.0.14.dist-info/entry_points.txt,sha256=5qYgwAXpxegeAicD_rzda_trDRnUC51F5UVDpcZ7j6Q,240
-epstein_files-1.0.14.dist-info/RECORD,,