Merge pull request #5 from muskit/no-listen-rewrite
Rewrite bot to not use filtered streams
This commit is contained in:
muskit
+1
-1
@@ -143,5 +143,5 @@ cython_debug/
|
||||
.vscode
|
||||
|
||||
# project-specific (secret.ini: can't ignore existing file?)
|
||||
secrets.ini
|
||||
*.png
|
||||
*.json
|
||||
@@ -5,11 +5,69 @@ Twitter bot that tracks cross-company interactions between the non-JP branches o
|
||||
|
||||
**This project was created to run [this account](https://twitter.com/NijiHolo_EN_ID).**
|
||||
|
||||
## Running modes
|
||||
## Running
|
||||
Install dependencies.
|
||||
```
|
||||
pip install -r requirements.txt
|
||||
```
|
||||
Setup the `.env` in the project root. Refer to the `.env` section for variables.
|
||||
|
||||
Run the program from project root (not in `src`). Refer to the following section for options.
|
||||
```
|
||||
python src/main.py
|
||||
```
|
||||
|
||||
## Modes & Options
|
||||
The bot may run in these modes:
|
||||
* Catch-up (`c`): intended to run only once, scan all accounts for cross-company tweets and post them. Terminate when done posting all.
|
||||
- use `--auto-listen` to switch to listen mode when finished
|
||||
* Listen (`l`): listens for tweets from list, sharing it if it's cross-company
|
||||
* Pass no argument to run in listen mode, which scrapes all accounts in the *list* folder at an interval.
|
||||
* Pass `--straight-to-queue` to process the locally-stored queue first before attempting to scrape.
|
||||
* Command-line (`cmd`): an interactive mode for manual control and debugging (drops into Python interpretor)
|
||||
|
||||
## `.env`
|
||||
These need to be defined in a `.env` file at the project root (outside of `src`):
|
||||
|
||||
### Scraper Credentials
|
||||
To get around rate limitations imposed on users, we scrape with multiple accounts. Each account is defined in the file using the following format:
|
||||
```
|
||||
scraper_usernameX=twitter_username
|
||||
scraper_passwordX=twitter_password
|
||||
```
|
||||
where `X` is a number starting from 0, increasing by 1 for each account added. For instance:
|
||||
```
|
||||
scraper_username0=
|
||||
scraper_password0=
|
||||
scraper_username1=
|
||||
scraper_password1=
|
||||
```
|
||||
The first account (`scraper_username0` and `scraper_password0`) **MUST be defined (`scraper_username` and `scraper_password` without number will not work!)** and will be used to attempt scraping private accounts. Make sure this account follows any private accounts that you want to scrape!
|
||||
### Twitter API Stuff
|
||||
The following keys/tokens are used for the official API via `tweepy`. We mainly use these to just post tweets.
|
||||
```
|
||||
app_key=
|
||||
app_secret=
|
||||
user_token=
|
||||
user_secret=
|
||||
```
|
||||
### Screenshot Cookie *(optional)*
|
||||
This is the authentication token obtained from a browser when signed in on the Twitter website. It's only needed if you want to screenshot tweets from privated accounts. Make sure the token belongs to an account that follows desired private accounts! Maybe have it belong to `scraper_username0`?
|
||||
```
|
||||
web_auth_token=
|
||||
```
|
||||
### Example contents of `.env` without values
|
||||
```
|
||||
scraper_username0=
|
||||
scraper_password0=
|
||||
scraper_username1=
|
||||
scraper_password1=
|
||||
scraper_username2=
|
||||
scraper_password2=
|
||||
scraper_username3=
|
||||
scraper_password3=
|
||||
web_auth_token=
|
||||
app_key=
|
||||
app_secret=
|
||||
user_token=
|
||||
user_secret=
|
||||
```
|
||||
|
||||
*Created for the spirit of entertainment and in the name of unity.* ❤
|
||||
|
||||
@@ -1,998 +0,0 @@
|
||||
1576430556933672960
|
||||
1576489483457134594
|
||||
1576509602732322819
|
||||
1576660900668317696
|
||||
1576716382024015872
|
||||
1576720425505415170
|
||||
1576755412824600576
|
||||
1576995410992041986
|
||||
1577051131339235328
|
||||
1577055035577417728
|
||||
1577068103791366145
|
||||
1577070764175466499
|
||||
1577083110805876738
|
||||
1577098034147573760
|
||||
1577193525720272897
|
||||
1577193763252076544
|
||||
1577100518446403584
|
||||
1577592730104254464
|
||||
1577681106971000832
|
||||
1577751763859521570
|
||||
1577814530729525248
|
||||
1577857217352863744
|
||||
1577858408144936960
|
||||
1577866370179805184
|
||||
1577940553081176065
|
||||
1577996126955122688
|
||||
1577998007018328064
|
||||
1577998613615345664
|
||||
1577998915366162433
|
||||
1578245815130210304
|
||||
1578516453434675200
|
||||
1578638137701838848
|
||||
1578973277539926017
|
||||
1577940553081176065
|
||||
1578245815130210304
|
||||
1578516453434675200
|
||||
1578638137701838848
|
||||
1578973277539926017
|
||||
1579547332571705345
|
||||
1579547617717293056
|
||||
1579548229431365634
|
||||
1579554638369796102
|
||||
1579608397468794880
|
||||
1579628879270281216
|
||||
1579629421064921088
|
||||
1579758548355276800
|
||||
1579758871589302272
|
||||
1579768241400197120
|
||||
1579789920960655360
|
||||
1579790235730575360
|
||||
1579900565424701441
|
||||
1579964993243873287
|
||||
1580181779868430336
|
||||
1580258975047888896
|
||||
1580259413297807363
|
||||
1580259612670177280
|
||||
1580260690962501632
|
||||
1580261950604222464
|
||||
1580262961058263040
|
||||
1580265275265495040
|
||||
1580268043279958016
|
||||
1580269274848849920
|
||||
1580269426082910215
|
||||
1580269963679477761
|
||||
1580276935921655808
|
||||
1580277066930434049
|
||||
1580279094205259776
|
||||
1580283694731776000
|
||||
1580381402767720448
|
||||
1580449719511572480
|
||||
1580449785135640576
|
||||
1580476694519246850
|
||||
1580480705733148672
|
||||
1580516570518159366
|
||||
1580517097708609538
|
||||
1580576986145038336
|
||||
1580664540152606721
|
||||
1580838449292406784
|
||||
1580838874200227841
|
||||
1580842852476915714
|
||||
1580842860622286848
|
||||
1580879439193595904
|
||||
1581315661242171392
|
||||
1581359254363148288
|
||||
1581359894082424834
|
||||
1581512593918685184
|
||||
1581628130418384896
|
||||
1581628264434782209
|
||||
1581639528053321730
|
||||
1581644625604845568
|
||||
1581655283448745985
|
||||
1581732410584371200
|
||||
1581945888465096704
|
||||
1581945996095160320
|
||||
1582035073573888000
|
||||
1582082039033262082
|
||||
1582135400147001344
|
||||
1582135466068828161
|
||||
1582142861604446209
|
||||
1582148227943436288
|
||||
1582167082997014528
|
||||
1582188441793548290
|
||||
1582191818430521344
|
||||
1582191849540026368
|
||||
1582280269469716481
|
||||
1582283272024834050
|
||||
1582371009587740672
|
||||
1582461770198761472
|
||||
1582555511949385728
|
||||
1582617083179044864
|
||||
1582617435416694784
|
||||
1582635977373544449
|
||||
1582636277605617664
|
||||
1582773035178024961
|
||||
1583173642166292480
|
||||
1583176372696252417
|
||||
1583204758495457280
|
||||
1583526503492440064
|
||||
1583563981108629504
|
||||
1583575990411747328
|
||||
1583576858133921793
|
||||
1583668922167234561
|
||||
1583669374397059072
|
||||
1583689284594245633
|
||||
1583863041493413888
|
||||
1583888933494358017
|
||||
1583908414102519812
|
||||
1583909513832574976
|
||||
1583968857554767872
|
||||
1584008637134970883
|
||||
1584441576205320195
|
||||
1584442364591939584
|
||||
1584480659199754242
|
||||
1584528264915083265
|
||||
1584528264965787649
|
||||
1584622108025712641
|
||||
1584628579677126657
|
||||
1584629534527488001
|
||||
1584629633735364608
|
||||
1584649537997533185
|
||||
1584728511952011266
|
||||
1584753182156414976
|
||||
1585073813905375233
|
||||
1585097154032435200
|
||||
1585098477557555200
|
||||
1585112474097176576
|
||||
1585144962085048321
|
||||
1585145288767201288
|
||||
1585145360313618432
|
||||
1585145559887020032
|
||||
1585145633530597376
|
||||
1585161614461849600
|
||||
1585199425206394880
|
||||
1585208233010290690
|
||||
1585277536187064322
|
||||
1585424318103945216
|
||||
1585449222635225088
|
||||
1585511782885380097
|
||||
1585553749199978497
|
||||
1585828601727156224
|
||||
1585841772332302339
|
||||
1585871064021671936
|
||||
1585872552513134592
|
||||
1585872662915559424
|
||||
1585875080315559936
|
||||
1586100052623171586
|
||||
1586179378089582592
|
||||
1586374470939254785
|
||||
1586376473119686656
|
||||
1586392482321158144
|
||||
1586643720878706689
|
||||
1586737628338077701
|
||||
1586840114868629504
|
||||
1587216332940247040
|
||||
1587235419241996288
|
||||
1587255460943511553
|
||||
1587280116069142528
|
||||
1587290714487943168
|
||||
1587336915048894466
|
||||
1587459847636779008
|
||||
1587670143840112640
|
||||
1587670772323155969
|
||||
1587828606583115776
|
||||
1587830007753105408
|
||||
1587830065223684098
|
||||
1587891436371861504
|
||||
1587931182284517381
|
||||
1587936783299321856
|
||||
1587947549658386433
|
||||
1587983746543894534
|
||||
1588044083666231298
|
||||
1588093544685719552
|
||||
1588095705070981120
|
||||
1588100836575440896
|
||||
1588101073910104066
|
||||
1588101210329927680
|
||||
1588128449490612225
|
||||
1588149163811913731
|
||||
1588293190322253827
|
||||
1588366787065675777
|
||||
1588678173138780160
|
||||
1588762792521498629
|
||||
1588796489224294400
|
||||
1588958260258025472
|
||||
1589150733659963393
|
||||
1589439728973217793
|
||||
1589529660429602816
|
||||
1589532085144150017
|
||||
1589532195844411393
|
||||
1589540519583289344
|
||||
1589570601567809538
|
||||
1589641589135413248
|
||||
1589669189765451776
|
||||
1589916075705085952
|
||||
1589916571518242816
|
||||
1589918062358450176
|
||||
1589977133857722368
|
||||
1590885589603479552
|
||||
1591306004666675201
|
||||
1591307219169148930
|
||||
1591336097216622594
|
||||
1591336705730437120
|
||||
1591730177310359553
|
||||
1591839086364741632
|
||||
1592029314119520256
|
||||
1592086362508840960
|
||||
1592189421897797633
|
||||
1592514737165971456
|
||||
1592635108510474240
|
||||
1592743719639592960
|
||||
1593330237609164800
|
||||
1593612237217964038
|
||||
1594025064478801920
|
||||
1594025676578824193
|
||||
1594025967428608000
|
||||
1594305241377284102
|
||||
1594336607661760512
|
||||
1594475762614423552
|
||||
1594499300364779523
|
||||
1594537154453147649
|
||||
1594543914186317825
|
||||
1594581129041915904
|
||||
1594595823748550657
|
||||
1594617557696348160
|
||||
1594648307338739712
|
||||
1594648462431498242
|
||||
1594650682220769280
|
||||
1594705822281789443
|
||||
1580268846790778881
|
||||
1581083174305599488
|
||||
1581083425401806848
|
||||
1581088983148560384
|
||||
1581096278469246978
|
||||
1581513143087681536
|
||||
1581514105894350849
|
||||
1583568304576565254
|
||||
1584600364783144960
|
||||
1584612840824049684
|
||||
1587558629489926145
|
||||
1587669760078188545
|
||||
1591359889011773440
|
||||
1592042710722314244
|
||||
1592082624558034946
|
||||
1594945258965848066
|
||||
1595094770539712512
|
||||
1595144029448339457
|
||||
1595159086265339904
|
||||
1595190586537283585
|
||||
1595209971486306304
|
||||
1595520675607113728
|
||||
1595531274474643469
|
||||
1595605679938707457
|
||||
1595677291300163584
|
||||
1596235440700407811
|
||||
1596630228289605633
|
||||
1596636538116382722
|
||||
1596733577076449280
|
||||
1596797256665292802
|
||||
1596833246561521664
|
||||
1596996552887013377
|
||||
1597029385625665536
|
||||
1597051834689941504
|
||||
1597053360393228288
|
||||
1597265217515581440
|
||||
1597265603529969664
|
||||
1597266128665214976
|
||||
1597270533850427392
|
||||
1597321808331825153
|
||||
1597325594341310464
|
||||
1597426336636669952
|
||||
1597636961694797829
|
||||
1597654482758045696
|
||||
1597698554688335872
|
||||
1597699060462673920
|
||||
1597845762322223105
|
||||
1598567014347542529
|
||||
1598586430795747328
|
||||
1598796336270217217
|
||||
1599022459029966848
|
||||
1599310792008269824
|
||||
1599324276540297216
|
||||
1599350192134033409
|
||||
1599694863800279041
|
||||
1599703777665503232
|
||||
1599996347658051584
|
||||
1600462906335690753
|
||||
1600496863278407691
|
||||
1600499056165404672
|
||||
1600628466977832961
|
||||
1600669875503972353
|
||||
1600671934315786240
|
||||
1600677118874353665
|
||||
1600700567948767233
|
||||
1600705712682053632
|
||||
1600706014537564160
|
||||
1600706374857822209
|
||||
1600709284786053120
|
||||
1600709816791334913
|
||||
1600752783845560321
|
||||
1600772713135554560
|
||||
1601053614755446785
|
||||
1595709206627246081
|
||||
1596251536841400320
|
||||
1596364591269023744
|
||||
1596365300026421248
|
||||
1596500447308939265
|
||||
1597857498903764995
|
||||
1598352981967376385
|
||||
1598477628943794176
|
||||
1598552959738343425
|
||||
1599118485363490817
|
||||
1599279465838018560
|
||||
1599291560507895810
|
||||
1599297564633083904
|
||||
1599337445199253504
|
||||
1599358415092191234
|
||||
1599423114018185216
|
||||
1600363418166648835
|
||||
1600369195241205760
|
||||
1601558689529593857
|
||||
1601595882763616257
|
||||
1601642608698482688
|
||||
1601749751745372162
|
||||
1601754145861484544
|
||||
1601768148621217792
|
||||
1601877383522713601
|
||||
1601878641620967424
|
||||
1601878682137931776
|
||||
1601879062560968705
|
||||
1601881127958253570
|
||||
1602009414990430208
|
||||
1602012523708010496
|
||||
1602205274168852480
|
||||
1602205385993101313
|
||||
1602275459697639424
|
||||
1602506093829099520
|
||||
1602508255062020096
|
||||
1602540444365508608
|
||||
1602658754297843712
|
||||
1602743376201359360
|
||||
1602753038909050880
|
||||
1602766954514767872
|
||||
1602768627534467073
|
||||
1602800460288647168
|
||||
1602804823887421444
|
||||
1602832095952658434
|
||||
1602847291362123777
|
||||
1602851330309357568
|
||||
1602853324667191297
|
||||
1602963752101580800
|
||||
1603073603364233217
|
||||
1603074898758631424
|
||||
1603287279975686144
|
||||
1603297578514546688
|
||||
1603822421190008833
|
||||
1603838388347584518
|
||||
1603840758783815680
|
||||
1602766894465175552
|
||||
1603867200204701696
|
||||
1604185149448634369
|
||||
1604237041977495553
|
||||
1604261358039711744
|
||||
1604403162504232961
|
||||
1604513581780844551
|
||||
1604516203908632576
|
||||
1604519045495234561
|
||||
1604522344550043648
|
||||
1604556203903709185
|
||||
1604726291093233664
|
||||
1604726760142262272
|
||||
1605034554594832385
|
||||
1605065307986735104
|
||||
1605083978847096832
|
||||
1605252604874874880
|
||||
1605264950347399168
|
||||
1605266135339134976
|
||||
1605266238443241474
|
||||
1605268175796445200
|
||||
1605268292112945153
|
||||
1605341177791684608
|
||||
1605354994127241217
|
||||
1605398103036948484
|
||||
1605415343358447616
|
||||
1605552893553016832
|
||||
1605563286728196098
|
||||
1605707013589151745
|
||||
1605763163391156226
|
||||
1605763733040599040
|
||||
1605978761127636992
|
||||
1606359300565237763
|
||||
1606359416999116813
|
||||
1606359600680271877
|
||||
1606359770969014284
|
||||
1606385110697316371
|
||||
1607544867357433859
|
||||
1607925564412932096
|
||||
1608271072562139136
|
||||
1608325890097782784
|
||||
1608536973727367170
|
||||
1608539465340915715
|
||||
1608780569676251140
|
||||
1608826609766891522
|
||||
1608827143257223169
|
||||
1608827350376140802
|
||||
1608827612788576256
|
||||
1608829041943138305
|
||||
1608888487436255232
|
||||
1609051353543761921
|
||||
1609272099276984320
|
||||
1609724860590329856
|
||||
1610129643067310081
|
||||
1610130224003559424
|
||||
1610710681975914496
|
||||
1610710768030486530
|
||||
1605758391506599936
|
||||
1606347441795731471
|
||||
1608806613812678660
|
||||
1610391363820081152
|
||||
1610883754435973120
|
||||
1611004280282157057
|
||||
1611004684344475650
|
||||
1611005195936489478
|
||||
1611144693231779843
|
||||
1611145010531160064
|
||||
1611145304161624064
|
||||
1611160944197132288
|
||||
1611175600126935041
|
||||
1611231349939240960
|
||||
1611594073231720449
|
||||
1611669467972644864
|
||||
1611954084751474688
|
||||
1611957866096889859
|
||||
1611976715802214401
|
||||
1611978377568874498
|
||||
1611978836383522816
|
||||
1611989935439118336
|
||||
1612041276584792070
|
||||
1612222955332984832
|
||||
1612223088833200132
|
||||
1612223456724279297
|
||||
1612486617561985025
|
||||
1612492903158267906
|
||||
1612559384906813445
|
||||
1612828971590193152
|
||||
1613178801374232578
|
||||
1613178992894398470
|
||||
1613179047114346496
|
||||
1613179141196767233
|
||||
1613179219256958983
|
||||
1613179256917352448
|
||||
1613179328854122502
|
||||
1613411736790437889
|
||||
1613760768641572869
|
||||
1613784820076929025
|
||||
1613792568772366336
|
||||
1613799407069323266
|
||||
1614026676589019138
|
||||
1614045021606666244
|
||||
1614052812585066496
|
||||
1612040906529456130
|
||||
1613408647937417216
|
||||
1613822548906774529
|
||||
1614451292163944448
|
||||
1614504913299714050
|
||||
1614514545644666881
|
||||
1614516595333959680
|
||||
1614620215467454504
|
||||
1614654062267617280
|
||||
1614666713181880325
|
||||
1614668312809738240
|
||||
1614677261537673216
|
||||
1614677382820134914
|
||||
1614706430682054657
|
||||
1614712147917041665
|
||||
1614713259600535552
|
||||
1614713633707274241
|
||||
1614713877807398912
|
||||
1614714298080854017
|
||||
1614714610170613761
|
||||
1614714833966104577
|
||||
1614715700198928386
|
||||
1614727456782712832
|
||||
1614744314927779841
|
||||
1614801067224748033
|
||||
1614801210724474880
|
||||
1614801331427966977
|
||||
1614831641935122434
|
||||
1614882693657034754
|
||||
1614892923488960512
|
||||
1614896348918321153
|
||||
1614897026814324736
|
||||
1614910639213871104
|
||||
1614928185006919680
|
||||
1615072566439288832
|
||||
1615160760321343488
|
||||
1615198160649060355
|
||||
1615744899655307266
|
||||
1616432895001772040
|
||||
1616436672723304453
|
||||
1616496843705442304
|
||||
1616501597244387331
|
||||
1616506713582235648
|
||||
1616507859944222721
|
||||
1616509728888324098
|
||||
1616537220919492608
|
||||
1616539191462223874
|
||||
1616564014129070080
|
||||
1616565464246718466
|
||||
1616568259515502592
|
||||
1616582644992548864
|
||||
1616591801372258309
|
||||
1616592777718165504
|
||||
1616593204870287360
|
||||
1616999858153422851
|
||||
1617004670865805312
|
||||
1617136121200377856
|
||||
1617158328022618117
|
||||
1617233807194705920
|
||||
1617272807666372608
|
||||
1617317197373911040
|
||||
1617317424969437187
|
||||
1617318052000940035
|
||||
1617318239603662848
|
||||
1617318880489398274
|
||||
1617318933903671297
|
||||
1617499924962369537
|
||||
1617541083508142080
|
||||
1617719234787962880
|
||||
1617732292319006722
|
||||
1617734154870980609
|
||||
1617737559546941441
|
||||
1617737633568010241
|
||||
1617743342670073856
|
||||
1617746760046346241
|
||||
1617757228794281985
|
||||
1617764829892349952
|
||||
1617862776046899202
|
||||
1617870239760535553
|
||||
1617886523063664641
|
||||
1617930583505801224
|
||||
1618423435964846081
|
||||
1618424133347594240
|
||||
1618542182117539842
|
||||
1618542328389697539
|
||||
1618548432167403520
|
||||
1618554017730396163
|
||||
1618567686174429184
|
||||
1618677052806434816
|
||||
1618830423236448256
|
||||
1618834876064747520
|
||||
1618844600894107648
|
||||
1618845420557590530
|
||||
1618845486580129792
|
||||
1618983950336282624
|
||||
1618988673227436033
|
||||
1618991676781137921
|
||||
1619066157964881921
|
||||
1619099894391738368
|
||||
1619180071893479424
|
||||
1619202780224446467
|
||||
1619305720683585538
|
||||
1619399152651358210
|
||||
1619654784725950464
|
||||
1619687684951384065
|
||||
1619774044869632001
|
||||
1619775039435587588
|
||||
1619778797368389633
|
||||
1619832253449912321
|
||||
1619856428969398272
|
||||
1619924833621671937
|
||||
1620003405958873092
|
||||
1620227328566792192
|
||||
1620346225886396421
|
||||
1620346703831502848
|
||||
1620664802241835008
|
||||
1620674202293653504
|
||||
1620684007473287169
|
||||
1620711599593119744
|
||||
1620849272018333696
|
||||
1620952817803612160
|
||||
1620953485851394049
|
||||
1620954399475658753
|
||||
1620955271974780929
|
||||
1620957482465230848
|
||||
1620958412438929409
|
||||
1621310782666416128
|
||||
1621311858329407488
|
||||
1621312731034030080
|
||||
1621314869487091712
|
||||
1621403652991721472
|
||||
1621425409157140484
|
||||
1621561581690826752
|
||||
1621574909116948480
|
||||
1621591321067622402
|
||||
1621741476412821504
|
||||
1621743668045676546
|
||||
1621744901338841088
|
||||
1621746092449878016
|
||||
1621750280693092352
|
||||
1621750707388055552
|
||||
1621750740724350977
|
||||
1621751603970842624
|
||||
1621755848568733696
|
||||
1621771326494244865
|
||||
1621797995678429184
|
||||
1621798650778206208
|
||||
1621799269496913920
|
||||
1622048615672733697
|
||||
1622519854560595969
|
||||
1622565675922501632
|
||||
1622893611955240961
|
||||
1622953785713913857
|
||||
1623465800836284416
|
||||
1623471458696728581
|
||||
1623472449961529346
|
||||
1623707075300986880
|
||||
1623709743264190464
|
||||
1624643824692908032
|
||||
1624646092510879745
|
||||
1624648616446152706
|
||||
1624648791034044416
|
||||
1624649128771739648
|
||||
1624659722954747906
|
||||
1624853328155779078
|
||||
1624860275831672834
|
||||
1625010090326761472
|
||||
1625058230807277568
|
||||
1625118424694247424
|
||||
1625734543016660992
|
||||
1624520599656624129
|
||||
1624521241901076485
|
||||
1627514900703584256
|
||||
1627522526527062016
|
||||
1627529002029297664
|
||||
1627529397296320513
|
||||
1627563980154671105
|
||||
1627566950267371520
|
||||
1627573853823307776
|
||||
1627574225199591425
|
||||
1627606044070277120
|
||||
1627606106720686080
|
||||
1627608118292287488
|
||||
1627635743697866752
|
||||
1627639742312468482
|
||||
1627654096378384384
|
||||
1627655361410011136
|
||||
1627655522840358912
|
||||
1627861956634746884
|
||||
1627864965053808640
|
||||
1627934959846707200
|
||||
1627953946789838848
|
||||
1627985560920006657
|
||||
1627985896279797761
|
||||
1628003892641566720
|
||||
1628004446310649856
|
||||
1628007913032753157
|
||||
1628008071048925184
|
||||
1628047652691775488
|
||||
1628269360111579137
|
||||
1628283406114230272
|
||||
1628283727582498816
|
||||
1628333043961430016
|
||||
1628335653841944578
|
||||
1628640208425197569
|
||||
1628652565088059392
|
||||
1628680578613780480
|
||||
1628680986346266624
|
||||
1628717287233421316
|
||||
1628786187233099780
|
||||
1628786323128741888
|
||||
1628786724317941762
|
||||
1628786919470407684
|
||||
1628795062472835073
|
||||
1628795184044728320
|
||||
1629272877060046850
|
||||
1629272892595634178
|
||||
1629272914330636289
|
||||
1629335290530676736
|
||||
1629341302524346368
|
||||
1629402458689794050
|
||||
1629454168845893633
|
||||
1629468461100589056
|
||||
1629758716751798272
|
||||
1630026264705896448
|
||||
1630026923303911425
|
||||
1630027564344569856
|
||||
1630032477103222784
|
||||
1630033474861056000
|
||||
1630035097461346304
|
||||
1631048890504867843
|
||||
1631542496814825473
|
||||
1631590820678860800
|
||||
1631676512029282307
|
||||
1632102469072297986
|
||||
1632283032207114251
|
||||
1632285887995449345
|
||||
1632414814692601856
|
||||
1632556021049856005
|
||||
1632560765424259072
|
||||
1632629409751416835
|
||||
1632641410230026240
|
||||
1632641678623522816
|
||||
1633423730654478337
|
||||
1633657001749397505
|
||||
1633659447934611457
|
||||
1633664285078421505
|
||||
1633664320792940544
|
||||
1633911083202117632
|
||||
1634528563344064512
|
||||
1634531188034654209
|
||||
1634539654916751363
|
||||
1634620065864728581
|
||||
1634683766235856896
|
||||
1634746599263043584
|
||||
1635151666445193217
|
||||
1635154154497916928
|
||||
1635154248769093640
|
||||
1635168017276633088
|
||||
1635253449913925632
|
||||
1635253930849599489
|
||||
1635259184194408453
|
||||
1635260955562237952
|
||||
1635263134352162817
|
||||
1635336166865248257
|
||||
1635366683052482560
|
||||
1635462585616314368
|
||||
1635474518184722432
|
||||
1635477005226618887
|
||||
1635511568556507136
|
||||
1635566086815875073
|
||||
1635745147181735948
|
||||
1635797049512951808
|
||||
1635798410837245953
|
||||
1635817346664153095
|
||||
1635891740640174081
|
||||
1635904748774887426
|
||||
1635973373384871936
|
||||
1635976579624095745
|
||||
1636360081670545411
|
||||
1636360302207078402
|
||||
1636370880241754112
|
||||
1636377928140468227
|
||||
1636418785006567427
|
||||
1636576201102721027
|
||||
1636599574197403648
|
||||
1636635893149876226
|
||||
1636729072788045824
|
||||
1636766609984352256
|
||||
1636767125640728577
|
||||
1636984749553254400
|
||||
1637021438090829825
|
||||
1637022826472574976
|
||||
1637022930097049600
|
||||
1637035673621528576
|
||||
1637058107233939456
|
||||
1637061108992442368
|
||||
1637061432104849408
|
||||
1637205007891066882
|
||||
1637210521941442561
|
||||
1637872529342869504
|
||||
1637924701573099520
|
||||
1637925400537108480
|
||||
1637925419323371522
|
||||
1638013762165788672
|
||||
1638017789326663681
|
||||
1638044945520132096
|
||||
1638093064064364544
|
||||
1638094749335707648
|
||||
1638174096126263303
|
||||
1638240669256024064
|
||||
1638250006443352064
|
||||
1638276318944022547
|
||||
1638288078056857606
|
||||
1638496316635357185
|
||||
1638515168941592581
|
||||
1638525170330599424
|
||||
1638564705341001730
|
||||
1638565298990383104
|
||||
1638565311057391616
|
||||
1638565357861609474
|
||||
1638565800150958081
|
||||
1638565817280503808
|
||||
1638694373612298240
|
||||
1638694824638398465
|
||||
1638977522862813204
|
||||
1639137601226522626
|
||||
1639150016693452801
|
||||
1639159907168718850
|
||||
1639159970733359104
|
||||
1639160062311796737
|
||||
1639161115199541250
|
||||
1639170191790596096
|
||||
1639179598956208129
|
||||
1639257674704453634
|
||||
1639347991990636544
|
||||
1639472078574481408
|
||||
1639703873286701056
|
||||
1639705899932557312
|
||||
1639766858822762503
|
||||
1639796727753457667
|
||||
1639796887795511296
|
||||
1639829562258325504
|
||||
1640367083598024706
|
||||
1640438349252141057
|
||||
1640728560267608067
|
||||
1640730592303108096
|
||||
1640733153290592262
|
||||
1640733678484574215
|
||||
1640733839935909892
|
||||
1640924027014332417
|
||||
1640924197797998592
|
||||
1641206319397969921
|
||||
1641206458531323907
|
||||
1641207736296677376
|
||||
1641213550248050689
|
||||
1641214619329916929
|
||||
1641214980564344833
|
||||
1641331011186466816
|
||||
1641527643592949760
|
||||
1641638401534476293
|
||||
1641682622626627584
|
||||
1641775144657117185
|
||||
1641842256020635657
|
||||
1641842492587819009
|
||||
1641843861730230273
|
||||
1641843911789367296
|
||||
1641848890776223750
|
||||
1641863954535178261
|
||||
1641874376877563905
|
||||
1641881631232499712
|
||||
1641905300813803520
|
||||
1641920742576070658
|
||||
1641931894211133442
|
||||
1641931924611448835
|
||||
1641968426985099264
|
||||
1642044327445639168
|
||||
1642052711259701248
|
||||
1642271925807329280
|
||||
1642363150442934274
|
||||
1642366098359328770
|
||||
1642386091146899457
|
||||
1642386251453202432
|
||||
1642386862714294276
|
||||
1642387403498459139
|
||||
1642389173641568257
|
||||
1642390508629807104
|
||||
1642391144947671043
|
||||
1642391409058787328
|
||||
1642392966391287808
|
||||
1642401063419662336
|
||||
1642404506716590081
|
||||
1642407503333044224
|
||||
1642442692985593857
|
||||
1642683998026563587
|
||||
1642874077491892226
|
||||
1642936452525834251
|
||||
1642965863111479296
|
||||
1642974611460505604
|
||||
1643151747874172930
|
||||
1643375613665787907
|
||||
1643383401980870658
|
||||
1643456057425948672
|
||||
1643456150652739584
|
||||
1643514992803696644
|
||||
1644101531111952384
|
||||
1644101652398612481
|
||||
1644228307607486464
|
||||
1644804036896641024
|
||||
1644808862627770368
|
||||
1644825182161256448
|
||||
1644835718345228288
|
||||
1644835937266827272
|
||||
1644836105416474624
|
||||
1644905879475752960
|
||||
1644947050709102592
|
||||
1645279524341563392
|
||||
1645279596139671552
|
||||
1645279816734883841
|
||||
1645693219554852864
|
||||
1645875484411015169
|
||||
1645877585795256321
|
||||
1646006703572283399
|
||||
1646010597840936960
|
||||
1646010746344538112
|
||||
1646011153129021441
|
||||
1646122631043883009
|
||||
1646131757320933377
|
||||
1646133175196991488
|
||||
1646135538792153089
|
||||
1646135600570040320
|
||||
1646136821154467843
|
||||
1646192373062569990
|
||||
1646560089539743744
|
||||
1646566407684132865
|
||||
1646747817753362433
|
||||
1646748169055924224
|
||||
1646751823666073600
|
||||
1646753010209681409
|
||||
1646928376093892609
|
||||
1646928741275439104
|
||||
1646932746030161921
|
||||
1647078104823922690
|
||||
1647103804893167616
|
||||
1647225532344352770
|
||||
1647227289950904320
|
||||
1647276137628696578
|
||||
1647438699347468288
|
||||
1647526954575486976
|
||||
1647527000519884800
|
||||
1647633398964408323
|
||||
1647707442921234434
|
||||
1648048413408399362
|
||||
1648049825445974017
|
||||
1648052454364413952
|
||||
1648053219787161601
|
||||
1648149958296588296
|
||||
1648237005309095937
|
||||
1648241034030370816
|
||||
1648259276589404161
|
||||
1648380069210107904
|
||||
1648411348370444288
|
||||
1648424997789265920
|
||||
1648520806639759361
|
||||
1648623694338486273
|
||||
1648701720887713793
|
||||
1648703217771880449
|
||||
1648726724874752002
|
||||
1648733920442322944
|
||||
1648768838614278187
|
||||
1649308567747407875
|
||||
1649308879073816576
|
||||
1649355368361762816
|
||||
1649476854321913860
|
||||
1649477152012640283
|
||||
1649649276077703168
|
||||
1649774499267747841
|
||||
1649777168883363841
|
||||
1649801910659022849
|
||||
1649804472330387457
|
||||
1649804636268773378
|
||||
1649805303737663488
|
||||
1649821444442845185
|
||||
1649870014605037568
|
||||
1649947340558442496
|
||||
1650001834134720514
|
||||
1650055872289329153
|
||||
1650059824422596609
|
||||
1650060926119460865
|
||||
1650111617710063621
|
||||
1650111986506801153
|
||||
1650260086818832386
|
||||
1650262244846968835
|
||||
1650561499893493763
|
||||
1650561606991097856
|
||||
1650561615300034561
|
||||
1650561616940003328
|
||||
1650573608186441729
|
||||
1650574520711798805
|
||||
1650574719639322631
|
||||
1650575471719899141
|
||||
1650575762586492956
|
||||
1650625322914664448
|
||||
1650626698029703168
|
||||
1650642367748411392
|
||||
1650682902362079232
|
||||
1650684291280699392
|
||||
1650685691016384512
|
||||
1650690513169948672
|
||||
1650812641500098560
|
||||
1650874430711648259
|
||||
1650950207087996932
|
||||
1651032960978567168
|
||||
1651034395397029890
|
||||
1651034863351283713
|
||||
1651035920840245249
|
||||
1651323872484945922
|
||||
1651327435076513795
|
||||
1651355062633918465
|
||||
+27
-21
@@ -3,36 +3,42 @@
|
||||
# ----- hololive EN -----
|
||||
|
||||
# --- [Myth] ---
|
||||
gawrgura 1283657064410017793
|
||||
watsonameliaen 1283656034305769472
|
||||
moricalliope 1283653858510598144
|
||||
ninomaeinanis 1283650008835743744
|
||||
takanashikiara 1283646922406760448
|
||||
1283657064410017793 gawrgura
|
||||
1283656034305769472 watsonameliaen
|
||||
1283653858510598144 moricalliope
|
||||
1283650008835743744 ninomaeinanis
|
||||
1283646922406760448 takanashikiara
|
||||
|
||||
# --- [HOPE] ---
|
||||
irys_en 1363705980261855232
|
||||
1363705980261855232 irys_en
|
||||
|
||||
# --- [Council] ---
|
||||
hakosbaelz 1409783149211443200
|
||||
ourokronii 1409817096523968513
|
||||
ceresfauna 1409784760805650436
|
||||
tsukumosana 1409819816194576394
|
||||
nanashimumei_en 1409817941705515015
|
||||
1409783149211443200 hakosbaelz
|
||||
1409817096523968513 ourokronii
|
||||
1409784760805650436 ceresfauna
|
||||
1409819816194576394 tsukumosana
|
||||
1409817941705515015 nanashimumei_en
|
||||
|
||||
# --- [Advent] ---
|
||||
1656536840240005121 shiorinovella
|
||||
1656528951303614464 nerissa_en
|
||||
1656889310472437761 fuwamoco_en
|
||||
1656531547279982593 kosekibijou
|
||||
|
||||
# ----- HOLOSTARS EN -----
|
||||
|
||||
# --- [TEMPUS] ---
|
||||
# Gen 1
|
||||
noirvesper_en 1536579341332516864
|
||||
axelsyrios 1536577295632441344
|
||||
magnidezmond 1536576325296996352
|
||||
regisaltare 1536575088996524032
|
||||
1536579341332516864 noirvesper_en
|
||||
1536577295632441344 axelsyrios
|
||||
1536576325296996352 magnidezmond
|
||||
1536575088996524032 regisaltare
|
||||
# Gen 2
|
||||
gavisbettel 1582926739684339712
|
||||
machinaxflayon 1582922712166825986
|
||||
banzoinhakka 1582927907206631425
|
||||
josuijishinri 1582925071546732544
|
||||
1582926739684339712 gavisbettel
|
||||
1582922712166825986 machinaxflayon
|
||||
1582927907206631425 banzoinhakka
|
||||
1582925071546732544 josuijishinri
|
||||
|
||||
# --- STAFF ---
|
||||
omegaalpha_en 1397148959798226945
|
||||
hololivepro_EN 1540204458042621952
|
||||
1397148959798226945 omegaalpha_en
|
||||
1540204458042621952 hololivepro_EN
|
||||
|
||||
+9
-9
@@ -3,16 +3,16 @@
|
||||
# ----- hololive ID -----
|
||||
|
||||
# --- [Gen 1] ---
|
||||
ayunda_risu 1234752200145899520
|
||||
moonahoshinova 1234753886520393729
|
||||
airaniiofifteen 1235180878449397764
|
||||
1234752200145899520 ayunda_risu
|
||||
1234753886520393729 moonahoshinova
|
||||
1235180878449397764 airaniiofifteen
|
||||
|
||||
# --- [Gen 2] ---
|
||||
pavoliareine 1328275136575799297
|
||||
kureijiollie 1328277233492844544
|
||||
anyamelfissa 1328277750000492545
|
||||
1328275136575799297 pavoliareine
|
||||
1328277233492844544 kureijiollie
|
||||
1328277750000492545 anyamelfissa
|
||||
|
||||
# --- [Gen 3] ---
|
||||
kobokanaeru 1486629076005634049
|
||||
vestiazeta 1486633489101307907
|
||||
kaelakovalskia 1486636197908602880
|
||||
1486629076005634049 kobokanaeru
|
||||
1486633489101307907 vestiazeta
|
||||
1486636197908602880 kaelakovalskia
|
||||
+35
-35
@@ -3,54 +3,54 @@
|
||||
# ----- [NIJISANJI EN] -----
|
||||
|
||||
# --- [Lazulight] ---
|
||||
PomuRainpuff 1390637197167038464
|
||||
EliraPendora 1390620618001838086
|
||||
FinanaRyugu 1390209302120394754
|
||||
1390637197167038464 PomuRainpuff
|
||||
1390620618001838086 EliraPendora
|
||||
1390209302120394754 FinanaRyugu
|
||||
|
||||
# --- [Obsydia] ---
|
||||
Petra_Gurin 1413339084076978179
|
||||
Selen_Tatsuki 1413318241804439552
|
||||
Rosemi_Lovelock 1413326894435602434
|
||||
1413339084076978179 Petra_Gurin
|
||||
1413318241804439552 Selen_Tatsuki
|
||||
1413326894435602434 Rosemi_Lovelock
|
||||
|
||||
# --- [Ethyria] ---
|
||||
MillieParfait 1437952405283426310
|
||||
EnnaAlouette 1437963160544284675
|
||||
NinaKosaka 1437959162651156484
|
||||
ReimuEndou 1437961007029227520
|
||||
1437952405283426310 MillieParfait
|
||||
1437963160544284675 EnnaAlouette
|
||||
1437959162651156484 NinaKosaka p
|
||||
1437961007029227520 ReimuEndou
|
||||
|
||||
# --- [Luxiem]---
|
||||
Vox_Akuma 1465851881180348425
|
||||
shu_amino 1465850835951357955
|
||||
ike_eveland 1465851188562345985
|
||||
Mysta_Rias 1465851243167895554
|
||||
luca_kaneshiro 1465858739970273281
|
||||
1465851881180348425 Vox_Akuma
|
||||
1465850835951357955 shu_amino
|
||||
1465851188562345985 ike_eveland
|
||||
1465851243167895554 Mysta_Rias
|
||||
1465858739970273281 luca_kaneshiro
|
||||
|
||||
# --- [Noctyx] ---
|
||||
alban_knox 1490867613915828224
|
||||
uki_violeta 1491195742123397124
|
||||
Yugo_Asuma 1492604168145539072
|
||||
Fulgur_Ovid 1493392149664219138
|
||||
sonny_brisko 1493394108014292993
|
||||
1490867613915828224 alban_knox
|
||||
1491195742123397124 uki_violeta
|
||||
1492604168145539072 Yugo_Asuma p
|
||||
1493392149664219138 Fulgur_Ovid
|
||||
1493394108014292993 sonny_brisko
|
||||
|
||||
# --- [ILUNA] ---
|
||||
MariaMari0nette 1545351225293426688
|
||||
AsterArcadia 1545352592884084736
|
||||
ScarleYonaguni 1545354510515654656
|
||||
KyoKanek0 1545552756773208066
|
||||
AiaAmare 1545562635650957312
|
||||
RenZott0 1546328834559340544
|
||||
1545351225293426688 MariaMari0nette
|
||||
1545352592884084736 AsterArcadia
|
||||
1545354510515654656 ScarleYonaguni
|
||||
1545552756773208066 KyoKanek0
|
||||
1545562635650957312 AiaAmare
|
||||
1546328834559340544 RenZott0
|
||||
|
||||
# --- [XSOLEIL] ---
|
||||
MelocoKyoran 1589536631324692480
|
||||
HexHaywire 1589524401170833409
|
||||
D_Dropscythe 1589531775058968576
|
||||
ZaionLanZa 1589539582399348738
|
||||
KotokaTorahime 1591995159901663232
|
||||
Ver_Vermillion 1589791076709171201
|
||||
1589536631324692480 MelocoKyoran
|
||||
1589524401170833409 HexHaywire
|
||||
1589531775058968576 D_Dropscythe
|
||||
1589539582399348738 ZaionLanZa p
|
||||
1591995159901663232 KotokaTorahime
|
||||
1589791076709171201 Ver_Vermillion
|
||||
|
||||
# --- [ALTS] ---
|
||||
3W1W4 1507066475638673422
|
||||
RyuguFinana 1506863869901168642
|
||||
1507066475638673422 3W1W4
|
||||
1506863869901168642 RyuguFinana
|
||||
|
||||
# --- [STAFF] ---
|
||||
NIJISANJI_World 1214737620749578240
|
||||
1214737620749578240 NIJISANJI_World
|
||||
+22
-22
@@ -2,31 +2,31 @@
|
||||
|
||||
# ----- [NIJISANJI ex-ID] -----
|
||||
|
||||
ZEA_Cornelia 1165866976192823297
|
||||
Hana_Macchia 1165866977472024576
|
||||
Taka_Radjiman 1165866977715347456
|
||||
# graduated MiyuOttavia 1205742630467801088
|
||||
RiksaDhirendra 1205743596785127424
|
||||
Rai_Galilei 1205744131294654466
|
||||
AmiciaMichella 1205744430386315265
|
||||
Azura_Cecillia 1237600624646078464
|
||||
Nara_Haramaung 1237603606448058371
|
||||
LaylaAlstro2434 1237613895675596800
|
||||
# elonmusk'd 1290243278814683137
|
||||
Bonnivier_2434 1587724357496815616
|
||||
Etna_Crimson 1290243331629318144
|
||||
SiskaLeontyne 1290243510193369089
|
||||
DeremKado 1323147415168323586
|
||||
NagisaArcinia 1323147843398324225
|
||||
RezaAvanluna 1323147856828510208
|
||||
HyonaElatiora 1414845791944937476
|
||||
Xia_Ekavira 1414845844692504611
|
||||
MikaMelatika 1414849131655450626
|
||||
1165866976192823297 ZEA_Cornelia
|
||||
1165866977472024576 Hana_Macchia
|
||||
1165866977715347456 Taka_Radjiman
|
||||
# graduated 1205742630467801088 MiyuOttavia
|
||||
1205743596785127424 RiksaDhirendra
|
||||
1205744131294654466 Rai_Galilei
|
||||
1205744430386315265 AmiciaMichella
|
||||
1237600624646078464 Azura_Cecillia
|
||||
1237603606448058371 Nara_Haramaung
|
||||
1237613895675596800 LaylaAlstro2434
|
||||
# elonmusk'd 1290243278814683137 bobon_pranaja
|
||||
1587724357496815616 Bonnivier_2434
|
||||
1290243331629318144 Etna_Crimson
|
||||
1290243510193369089 SiskaLeontyne
|
||||
1323147415168323586 DeremKado
|
||||
1323147843398324225 NagisaArcinia
|
||||
1323147856828510208 RezaAvanluna
|
||||
1414845791944937476 HyonaElatiora
|
||||
1414845844692504611 Xia_Ekavira
|
||||
1414849131655450626 MikaMelatika
|
||||
|
||||
# --- [ALTS] ---
|
||||
HanaMacchia2 1405494022446010375
|
||||
1405494022446010375 HanaMacchia2
|
||||
|
||||
# --- [STAFF] ---
|
||||
NIJISANJI_ID 1152523848060850177
|
||||
1152523848060850177 NIJISANJI_ID
|
||||
|
||||
|
||||
|
||||
@@ -0,0 +1,27 @@
|
||||
[scraper rate limitations]
|
||||
50 searches/pages every 15 minutes
|
||||
- max 20 tweets per search
|
||||
|
||||
[possible combinations which involve a "target cross-tweeter" B]
|
||||
A retweets B
|
||||
- B's tweet may have cross-mentions (B1, B2, etc.)
|
||||
- rt_author_id=B; rt_mentions=B1,B2,...
|
||||
A retweets tweet mentioning B
|
||||
- rt_author_id=...; rt_mentions=B...
|
||||
|
||||
A quotes a tweet from B
|
||||
- B's tweet may have cross-mentions (B1, B2, etc.)
|
||||
- quote_retweeted=B; rt_mentions=B1,B2,...
|
||||
A quotes a tweet mentioning B
|
||||
- quote_retweeted=...; rt_mentions=B...
|
||||
|
||||
A replies to B
|
||||
r = B
|
||||
A replies to a tweet mentioning B
|
||||
- r=...; rtm=B1,B2,...
|
||||
|
||||
-- NO --
|
||||
A retweets a tweet that quotes a tweet mentioning B?
|
||||
|
||||
[potential code change]
|
||||
rtm --> tgm (target tweet's mentions)
|
||||
@@ -1,88 +0,0 @@
|
||||
# 1283657064410017793 2023-03-20
|
||||
# 1283656034305769472 2023-03-20
|
||||
# 1283653858510598144 2023-03-20
|
||||
# 1283650008835743744 2023-03-20
|
||||
# 1283646922406760448 2023-03-20
|
||||
# 1363705980261855232 2023-03-20
|
||||
# 1409783149211443200 2023-03-20
|
||||
# 1409817096523968513 2023-03-20
|
||||
# 1409784760805650436 2023-03-20
|
||||
# 1409819816194576394 2023-03-20
|
||||
# 1409817941705515015 2023-03-20
|
||||
# 1536579341332516864 2023-03-20
|
||||
# 1536577295632441344 2023-03-20
|
||||
# 1536576325296996352 2023-03-20
|
||||
# 1536575088996524032 2023-03-20
|
||||
# 1397148959798226945 2023-03-20
|
||||
# 1540204458042621952 2023-03-20
|
||||
# 1234752200145899520 2023-03-20
|
||||
# 1234753886520393729 2023-03-20
|
||||
# 1235180878449397764 2023-03-20
|
||||
# 1328275136575799297 2023-03-20
|
||||
# 1328277233492844544 2023-03-20
|
||||
# 1328277750000492545 2023-03-20
|
||||
# 1486629076005634049 2023-03-20
|
||||
# 1486633489101307907 2023-03-20
|
||||
# 1486636197908602880 2023-03-20
|
||||
# 1390637197167038464 2023-03-20
|
||||
# 1390620618001838086 2023-03-20
|
||||
# 1390209302120394754 2023-03-20
|
||||
# 1413339084076978179 2023-03-20
|
||||
# 1413318241804439552 2023-03-20
|
||||
# 1413326894435602434 2023-03-20
|
||||
# 1437952405283426310 2023-03-20
|
||||
# 1437963160544284675 2023-03-20
|
||||
# 1437959162651156484 2023-03-20
|
||||
# 1437961007029227520 2023-03-20
|
||||
# 1465851881180348425 2023-03-20
|
||||
# 1465850835951357955 2023-03-20
|
||||
# 1465851188562345985 2023-03-20
|
||||
# 1465851243167895554 2023-03-20
|
||||
# 1465858739970273281 2023-03-20
|
||||
# 1490867613915828224 2023-03-20
|
||||
# 1491195742123397124 2023-03-20
|
||||
# 1492604168145539072 2023-03-20
|
||||
# 1493392149664219138 2023-03-20
|
||||
# 1493394108014292993 2023-03-20
|
||||
# 1545351225293426688 2023-03-20
|
||||
# 1545352592884084736 2023-03-20
|
||||
# 1545354510515654656 2023-03-20
|
||||
# 1545552756773208066 2023-03-20
|
||||
# 1545562635650957312 2023-03-20
|
||||
# 1546328834559340544 2023-03-20
|
||||
# 1507066475638673422 2023-03-20
|
||||
# 1506863869901168642 2023-03-20
|
||||
# 1214737620749578240 2023-03-20
|
||||
# 1165866976192823297 2023-03-20
|
||||
# 1165866977472024576 2023-03-20
|
||||
# 1165866977715347456 2023-03-20
|
||||
# 1205742630467801088 2023-01-01
|
||||
# 1205743596785127424 2023-03-20
|
||||
# 1205744131294654466 2023-03-20
|
||||
# 1205744430386315265 2023-03-20
|
||||
# 1237600624646078464 2023-03-20
|
||||
# 1237603606448058371 2023-03-20
|
||||
# 1237613895675596800 2023-03-20
|
||||
# 1290243278814683137 2023-01-01
|
||||
# 1290243331629318144 2023-03-20
|
||||
# 1290243510193369089 2023-03-20
|
||||
# 1323147415168323586 2023-03-20
|
||||
# 1323147843398324225 2023-03-20
|
||||
# 1323147856828510208 2023-03-20
|
||||
# 1414845791944937476 2023-03-20
|
||||
# 1414845844692504611 2023-03-20
|
||||
# 1414849131655450626 2023-03-20
|
||||
# 1405494022446010375 2023-03-20
|
||||
# 1152523848060850177 2023-03-20
|
||||
# 1587724357496815616 2023-03-20
|
||||
# 1589536631324692480 2023-03-20
|
||||
# 1589524401170833409 2023-03-20
|
||||
# 1589531775058968576 2023-03-20
|
||||
# 1589539582399348738 2023-03-20
|
||||
# 1591995159901663232 2023-03-20
|
||||
# 1589791076709171201 2023-03-20
|
||||
# 1582926739684339712 2023-03-20
|
||||
# 1582922712166825986 2023-03-20
|
||||
# 1582927907206631425 2023-03-20
|
||||
# 1582925071546732544 2023-03-20
|
||||
|
||||
File diff suppressed because it is too large
Load Diff
+3
-2
@@ -1,6 +1,7 @@
|
||||
python-dotenv
|
||||
nest-asyncio
|
||||
pytz
|
||||
tweet-capture
|
||||
git+https://github.com/muskit/tweety.git
|
||||
tweepy
|
||||
tweet-capture
|
||||
opencv-python
|
||||
git+https://github.com/muskit/twint_2022_fix.git
|
||||
-23
@@ -1,23 +0,0 @@
|
||||
## Twitter developer credentials.
|
||||
|
||||
# ---->> MAKE SURE YOUR VALUES AREN'T UPLOADED TO THE REPO! <<----
|
||||
#
|
||||
# This file should be added to .gitignore as a safeguard.
|
||||
#
|
||||
# If Git still wants to commit this file after changing its contents,
|
||||
# force Git to stop tracking the file's changes:
|
||||
# git update-index --assume-unchanged [<file> ...]
|
||||
#
|
||||
# To resume tracking its changes:
|
||||
# git update-index --no-assume-unchanged [<file> ...]
|
||||
#
|
||||
# https://stackoverflow.com/questions/10755655/git-ignore-tracked-files
|
||||
|
||||
# note: api_key/secret = consumer_key/secret
|
||||
|
||||
[Credentials]
|
||||
api_key=xxx
|
||||
api_secret=yyy
|
||||
bearer_token=zzz
|
||||
oauth1_access_token=x
|
||||
oauth1_access_secret=y
|
||||
@@ -0,0 +1,35 @@
|
||||
from dotenv import dotenv_values
|
||||
|
||||
## Track multiple accounts in a pool, cycling to the next one when requested.
|
||||
class AccountPool:
|
||||
def __init__(self):
|
||||
self.__accounts: list[tuple[str, str]] = list()
|
||||
self.__idx = -1
|
||||
creds = dotenv_values()
|
||||
i = 0
|
||||
while True:
|
||||
if f'scraper_username{i}' in creds \
|
||||
and f'scraper_password{i}' in creds:
|
||||
self.__accounts.append((
|
||||
creds[f'scraper_username{i}'],
|
||||
creds[f'scraper_password{i}']
|
||||
))
|
||||
i += 1
|
||||
else:
|
||||
break
|
||||
|
||||
def use_index(self, idx):
|
||||
self.__idx = idx
|
||||
return self.current()
|
||||
|
||||
def current(self):
|
||||
if 0 <= self.__idx < len(self.__accounts):
|
||||
return self.__accounts[self.__idx]
|
||||
return None
|
||||
|
||||
def next(self) -> tuple[str, str] | None:
|
||||
self.__idx += 1
|
||||
if self.__idx >= len(self.__accounts):
|
||||
self.__idx = -1
|
||||
return None
|
||||
return self.current()
|
||||
@@ -1,42 +0,0 @@
|
||||
## Twitter developer credentials management.
|
||||
|
||||
import os
|
||||
import configparser
|
||||
|
||||
import util
|
||||
|
||||
# returns dictionary of the Credentials section.
|
||||
# [NOT TO BE USED OUTSIDE OF THIS FILE.]
|
||||
def __get_ini_credentials():
|
||||
c = configparser.RawConfigParser()
|
||||
if len(c.read(os.path.join(util.get_project_dir(), 'secrets.ini'))) > 0 and c.has_section('Credentials'):
|
||||
return c['Credentials']
|
||||
return None
|
||||
|
||||
# returns the consumer api_key stored in secrets.ini
|
||||
def api_key():
|
||||
c = __get_ini_credentials()
|
||||
return c.get(option='api_key', fallback='xxx') if c is not None else 'xxx'
|
||||
|
||||
# returns the consumer api_secret stored in secrets.ini
|
||||
def api_secret():
|
||||
c = __get_ini_credentials()
|
||||
return c.get(option='api_secret', fallback='yyy') if c is not None else 'yyy'
|
||||
|
||||
# returns the bearer_token stored in secrets.ini
|
||||
def bearer_token():
|
||||
c = __get_ini_credentials()
|
||||
return c.get(option='bearer_token', fallback='zzz') if c is not None else 'zzz'
|
||||
|
||||
# returns the access_token stroed in secrets.ini
|
||||
def access_token():
|
||||
c = __get_ini_credentials()
|
||||
return c.get(option='oauth1_access_token', fallback='zzz') if c is not None else 'aaa'
|
||||
|
||||
# returns the access_secret stroed in secrets.ini
|
||||
def access_secret():
|
||||
c = __get_ini_credentials()
|
||||
return c.get(option='oauth1_access_secret', fallback='zzz') if c is not None else 'bbb'
|
||||
|
||||
def get_all_secrets():
|
||||
return f'api_key:{api_key()}\napi_secret:{api_secret()}\nbearer_token:{bearer_token()}\naccess_token:{access_token()}\naccess_secret:{access_secret()}'
|
||||
+52
-72
@@ -8,54 +8,24 @@ import traceback
|
||||
import datetime
|
||||
import asyncio
|
||||
import shutil
|
||||
from datetime import datetime
|
||||
|
||||
import twint
|
||||
|
||||
from scraper import Scraper
|
||||
from util import *
|
||||
from talent_lists import *
|
||||
from twapi import TwAPI
|
||||
import talenttweet as tt
|
||||
import ttweetqueue as ttq
|
||||
|
||||
PROGRAM_ARGS = None
|
||||
safe_to_post_tweets = True
|
||||
errored = False
|
||||
|
||||
## Returns the ID of all tweets (up to limit) from a user ID.
|
||||
def get_user_tweets(id, since_date=None, limit=None):
|
||||
global safe_to_post_tweets
|
||||
|
||||
qrt_count = 0
|
||||
tweets = list()
|
||||
c = twint.Config()
|
||||
c.User_id = id
|
||||
c.Limit = limit
|
||||
c.Store_object = True
|
||||
c.Store_object_tweets_list = tweets
|
||||
c.Hide_output = True
|
||||
c.Since = '' if since_date == None else f'{since_date} 00:00:00'
|
||||
|
||||
user_str = f'@{util.get_username_local(id)}'
|
||||
print(f'Scraping tweets from {user_str} since {"forever ago" if c.Since == "" else c.Since}...')
|
||||
try:
|
||||
twint.run.Search(c)
|
||||
except:
|
||||
print(f'Had trouble getting tweets from {user_str}')
|
||||
safe_to_post_tweets = False
|
||||
traceback.print_exc()
|
||||
|
||||
for twt in tweets:
|
||||
if type(twt.quote_url) is str and twt.quote_url != '':
|
||||
qrt_count += 1
|
||||
|
||||
print(f'Scraped {len(tweets)} tweets, {qrt_count} of which are quote tweets.')
|
||||
return tweets
|
||||
|
||||
# Returns a list of sorted and filtered TalentTweets (should
|
||||
# be equivalent to queue.txt)
|
||||
async def get_cross_talent_tweets():
|
||||
async def get_cross_tweets_online():
|
||||
global safe_to_post_tweets
|
||||
|
||||
scraper = Scraper()
|
||||
queue = ttq.TalentTweetQueue.instance
|
||||
|
||||
# Begin getting tweets from online
|
||||
@@ -64,19 +34,25 @@ async def get_cross_talent_tweets():
|
||||
for i, (talent_id, talent_username) in enumerate(talent_lists.talents.items()):
|
||||
print(f'[{i+1}/{len(talent_lists.talents)}] {talent_username}-----------------------------------')
|
||||
try:
|
||||
tweets = get_user_tweets(talent_id, since_date=queue.finished_user_dates.get(talent_id, None))
|
||||
for tweet in tweets:
|
||||
if tweet.id not in queue.ttweets_dict and tweet.id not in queue.finished_ttweets:
|
||||
ttweet = await tt.TalentTweet.create_from_twint_tweet(tweet)
|
||||
if ttweet.is_cross_company():
|
||||
queue.add_ttweet(ttweet)
|
||||
since_date = queue.finished_user_dates.get(talent_id, None)
|
||||
ttweets = scraper.get_cross_ttweets_from_user(talent_username, since_date=since_date)
|
||||
print(f'got {len(ttweets)} TalentTweets')
|
||||
for ttweet in ttweets:
|
||||
if ttweet.tweet_id not in queue.finished_ttweets \
|
||||
and ttweet.is_cross_company():
|
||||
queue.add_ttweet(ttweet)
|
||||
except KeyboardInterrupt as e:
|
||||
raise e
|
||||
except:
|
||||
print('Error occurred processing tweet data.')
|
||||
safe_to_post_tweets = False
|
||||
print(traceback.format_exc())
|
||||
queue.finished_user_dates[talent_id] = '2000-01-01'
|
||||
traceback.print_exc()
|
||||
else:
|
||||
queue.finished_user_dates[talent_id] = util.get_current_date()
|
||||
queue.save_file()
|
||||
except KeyboardInterrupt:
|
||||
print('Interrupting tweet pulling... NOTE: remaining dates in queue file will not be updated!')
|
||||
queue.save_file()
|
||||
except:
|
||||
print('Unhandled error occurred while pulling tweets.')
|
||||
traceback.print_exc()
|
||||
@@ -88,9 +64,9 @@ async def get_cross_talent_tweets():
|
||||
# return False = errored or we posted at least one ttweet
|
||||
# return True = we didn't post a single ttweet
|
||||
async def process_queue() -> bool:
|
||||
global PROGRAM_ARGS
|
||||
global errored
|
||||
WAIT_TIME = 60*3
|
||||
|
||||
WAIT_TIME = 60*15
|
||||
ttweets_posted = 0
|
||||
errored = False
|
||||
|
||||
@@ -101,13 +77,10 @@ async def process_queue() -> bool:
|
||||
print('Posting queue is empty!')
|
||||
return True
|
||||
|
||||
if PROGRAM_ARGS.announce_catchup:
|
||||
TwAPI.instance.post_tweet(text=f'Starting to catch up through {queued_ttweets_count} logged tweets.')
|
||||
|
||||
try:
|
||||
while not queue.is_empty():
|
||||
ttweet = queue.get_next_ttweet()
|
||||
tweet_was_successful = await TwAPI.instance.post_ttweet(ttweet, is_catchup=True)
|
||||
tweet_was_successful = await TwAPI.instance.post_ttweet(ttweet)
|
||||
|
||||
print('running queue.good()...')
|
||||
queue.good()
|
||||
@@ -123,9 +96,6 @@ async def process_queue() -> bool:
|
||||
print('Unhandled error occurred while posting tweets from queue.')
|
||||
errored = True
|
||||
traceback.print_exc()
|
||||
else:
|
||||
if PROGRAM_ARGS.announce_catchup:
|
||||
await TwAPI.instance.post_tweet('Finished with catch-up tweets!')
|
||||
|
||||
if errored or ttweets_posted > 0:
|
||||
return False
|
||||
@@ -133,29 +103,39 @@ async def process_queue() -> bool:
|
||||
|
||||
# return True = no problems
|
||||
# return False = issue occurred where we couldn't post all past tweets properly
|
||||
async def run(program_args):
|
||||
global PROGRAM_ARGS
|
||||
async def run(PROGRAM_ARGS):
|
||||
global errored
|
||||
global safe_to_post_tweets
|
||||
PROGRAM_ARGS = program_args
|
||||
|
||||
ret = None
|
||||
queue = ttq.TalentTweetQueue.instance
|
||||
while True:
|
||||
await get_cross_talent_tweets()
|
||||
print(f'{queue.get_count()} cross-company tweets to attempt sharing.')
|
||||
try:
|
||||
if safe_to_post_tweets:
|
||||
if await process_queue():
|
||||
print('Posted no new tweets; we\'re caught up!')
|
||||
return True
|
||||
else:
|
||||
print('Tweets were not retrieved cleanly.')
|
||||
return False
|
||||
except:
|
||||
print('Unhandled error occurred while running catch up in posting phase.')
|
||||
traceback.print_exc()
|
||||
return False
|
||||
|
||||
if errored:
|
||||
return False
|
||||
async def queue_loop():
|
||||
while True:
|
||||
print(f'{queue.get_count()} cross-company tweets to attempt sharing.')
|
||||
try:
|
||||
if safe_to_post_tweets:
|
||||
if await process_queue():
|
||||
print('Posted no new tweets; we\'re caught up!')
|
||||
return True
|
||||
else:
|
||||
print('Tweets were not retrieved cleanly.')
|
||||
return False
|
||||
except KeyboardInterrupt:
|
||||
print('Interrupting queue processing...')
|
||||
return False
|
||||
except:
|
||||
print('Unhandled error occurred while running catch up in posting phase.')
|
||||
traceback.print_exc()
|
||||
return False
|
||||
|
||||
if errored:
|
||||
return False
|
||||
|
||||
await get_cross_tweets_online()
|
||||
|
||||
if PROGRAM_ARGS.straight_to_queue:
|
||||
print('Processing queue first before pulling tweets...')
|
||||
return await queue_loop()
|
||||
else:
|
||||
await get_cross_tweets_online()
|
||||
return await queue_loop()
|
||||
|
||||
+6
-51
@@ -1,66 +1,21 @@
|
||||
## The bot's listen mode
|
||||
# Continuously listen for cross-company interactions.
|
||||
from time import sleep
|
||||
|
||||
import asyncio
|
||||
import traceback
|
||||
import tweepy
|
||||
from talenttweet import TalentTweet
|
||||
|
||||
from twapi import TwAPI
|
||||
import ttweetqueue as ttq
|
||||
import api_secrets
|
||||
import talent_lists as tl
|
||||
import util
|
||||
import catchup
|
||||
|
||||
errors_encountered = 0
|
||||
|
||||
def on_response(resp):
|
||||
ttweet = TalentTweet.create_from_v2api_response(resp)
|
||||
if ttweet is None:
|
||||
print('Couldn\'t create ttweet from the response:')
|
||||
print(resp)
|
||||
return
|
||||
|
||||
tweet_username = util.get_username(ttweet.author_id)
|
||||
|
||||
if ttweet.is_cross_company():
|
||||
print(f'Tweet {ttweet.tweet_id} is cross-company! Creating post...')
|
||||
is_successful = asyncio.run(TwAPI.instance.post_ttweet(ttweet))
|
||||
if is_successful:
|
||||
ttq.TalentTweetQueue.instance.add_finished_tweet(ttweet.tweet_id)
|
||||
else:
|
||||
print(f'[WARNING] Failed to post ttweet for {tweet_username}/{ttweet.tweet_id}!')
|
||||
else:
|
||||
print(f'Tweet {tweet_username}/{ttweet.tweet_id} is not cross-company.')
|
||||
|
||||
def run():
|
||||
def run(PROGRAM_ARGS):
|
||||
global errors_encountered
|
||||
while True:
|
||||
try:
|
||||
sc = tweepy.StreamingClient(api_secrets.bearer_token())
|
||||
|
||||
# clear rules
|
||||
print('Clearing streaming rules...')
|
||||
rules_resp = sc.get_rules()
|
||||
if rules_resp.data:
|
||||
print('Deleted a rule!')
|
||||
sc.delete_rules(rules_resp.data)
|
||||
|
||||
# create new rules
|
||||
print('Creating new streaming rules...')
|
||||
for rule in tl.get_twitter_rules():
|
||||
sc.add_rules(tweepy.StreamRule(rule))
|
||||
print('--------------------------------------------')
|
||||
print(sc.get_rules().data)
|
||||
print('--------------------------------------------')
|
||||
|
||||
sc.on_response=on_response
|
||||
print('Starting listening stream...')
|
||||
sc.filter(
|
||||
expansions=TwAPI.TWEET_EXPANSIONS,
|
||||
media_fields=TwAPI.TWEET_MEDIA_FIELDS,
|
||||
tweet_fields=TwAPI.TWEET_FIELDS
|
||||
)
|
||||
asyncio.run(catchup.run(PROGRAM_ARGS))
|
||||
print('Sleeping for 10 minutes...')
|
||||
sleep(60*10) # run every 10 minutes
|
||||
except KeyboardInterrupt:
|
||||
print('Interrupt signal received. Exiting listen mode.')
|
||||
print(f'{errors_encountered} errors encountered throughout session.')
|
||||
|
||||
+15
-56
@@ -8,7 +8,6 @@ import nest_asyncio
|
||||
|
||||
import talent_lists
|
||||
import ttweetqueue as ttq
|
||||
import api_secrets
|
||||
import catchup
|
||||
import listen
|
||||
from twapi import TwAPI
|
||||
@@ -16,69 +15,36 @@ from twapi import TwAPI
|
||||
PROGRAM_ARGS = None
|
||||
|
||||
MODES_HELP_STR = '''mode to run the bot at:
|
||||
l,listen: listen for new tweets from all accounts; will not terminate unless error occurs
|
||||
c,catchup: scan all tweets from all accounts; will terminate when done
|
||||
d,delete-all: delete all tweets on account provided by secrets.ini; make sure the function is uncommented in twapi.py'''
|
||||
<blank> scrape accounts in lists and post cross-company tweets if relevant
|
||||
cmd drop into Python interpretor with access to initialized variables'''
|
||||
|
||||
def init_argparse():
|
||||
p = argparse.ArgumentParser(description='Twitter bot that follows interactions between Nijisanji EN/ID and hololive EN/ID members.', formatter_class=RawTextHelpFormatter)
|
||||
p.add_argument('mode', nargs='?', \
|
||||
help=MODES_HELP_STR)
|
||||
p.add_argument('--show-tokens', action='store_true', help='[DO NOT USE IN PUBLIC SETTING] print stored tokens from secrets.ini')
|
||||
p.add_argument('--announce-catchup', action='store_true', help='In catch-up mode, post a tweet announcing catch-up mode.')
|
||||
p.add_argument('--auto-listen', action='store_true', help='In catch-up mode, transition to listen mode after successfuly catching up.')
|
||||
p.add_argument('--no-delay', action='store_true', help='In self-destruct mode, clear tweets without safety waiting.')
|
||||
p.add_argument('mode', nargs='?', help=MODES_HELP_STR)
|
||||
p.add_argument('--no-listen', action='store_true', help='Run one scraping-posting cycle without waiting to run again.')
|
||||
p.add_argument('--straight-to-queue', action='store_true', help='Go through queue first before attempting to pull tweets.')
|
||||
return p
|
||||
|
||||
def command_line():
|
||||
# TODO (extra): implement command line mode for manually controlling the bot
|
||||
print('Shell coming soon. For now, here\'s a Python interpretor.')
|
||||
print('Here\'s a Python interpretor.')
|
||||
code.interact(local=globals())
|
||||
pass
|
||||
|
||||
async def self_destruct():
|
||||
if not PROGRAM_ARGS.no_delay:
|
||||
print('\033[31;6m-----DELETING ALL TWEETS IN 10 SECONDS!! PRESS CTRL+C TO CANCEL.-----\033[0m')
|
||||
await asyncio.sleep(10)
|
||||
await TwAPI.instance.nuke_tweets()
|
||||
|
||||
async def async_main():
|
||||
global PROGRAM_ARGS
|
||||
|
||||
## Determine running mode
|
||||
# match PROGRAM_ARGS.mode.lower():
|
||||
# case 'l' | 'listen':
|
||||
# print('RUNNING IN LISTEN MODE\n')
|
||||
# await listen.run()
|
||||
# case 'c' | 'catchup':
|
||||
# print('RUNNING IN CATCH-UP MODE\n')
|
||||
# if await catchup.run(PROGRAM_ARGS) and PROGRAM_ARGS.auto_listen:
|
||||
# print('CATCH-UP MODE DONE, GOING INTO LISTEN MODE')
|
||||
# await listen.run()
|
||||
# case 'd' | 'delete-all':
|
||||
# print('WARNING: SELF-DESTRUCT MODE')
|
||||
# await self_destruct()
|
||||
# case 'cmd':
|
||||
# command_line()
|
||||
# case _:
|
||||
# print('\ninvalid mode. run with no arguments or "-h" for help page, including mode list.')
|
||||
# return
|
||||
if PROGRAM_ARGS.mode == None:
|
||||
if PROGRAM_ARGS.no_listen:
|
||||
await catchup.run(PROGRAM_ARGS)
|
||||
else:
|
||||
listen.run(PROGRAM_ARGS)
|
||||
return
|
||||
|
||||
mode = PROGRAM_ARGS.mode.lower()
|
||||
if mode in ['l', 'listen']:
|
||||
print('RUNNING IN LISTEN MODE')
|
||||
await listen.run()
|
||||
elif mode in ['c', 'catchup']:
|
||||
print('RUNNING IN CATCH UP MODE')
|
||||
if await catchup.run(PROGRAM_ARGS) and PROGRAM_ARGS.auto_listen:
|
||||
print('CATCH UP MODE DONE, GOING INTO LISTEN MODE')
|
||||
listen.run()
|
||||
elif mode in ['d', 'delete-all']:
|
||||
print('WARNING: SELF-DESTRUCT MODE')
|
||||
await self_destruct()
|
||||
elif mode == 'cmd':
|
||||
if mode == 'cmd':
|
||||
command_line()
|
||||
else:
|
||||
print('\ninvalid mode. run with no arguments or -h for help and modes')
|
||||
print('\nunknown mode. run with no arguments or -h for help and modes')
|
||||
|
||||
def main():
|
||||
global PROGRAM_ARGS
|
||||
@@ -90,13 +56,6 @@ def main():
|
||||
|
||||
PROGRAM_ARGS = parser.parse_args()
|
||||
|
||||
if PROGRAM_ARGS.show_tokens:
|
||||
print(api_secrets.get_all_secrets())
|
||||
|
||||
if PROGRAM_ARGS.mode is None: return
|
||||
|
||||
## We expect to run in some mode now.
|
||||
|
||||
# Initialize shared API instance
|
||||
TwAPI()
|
||||
|
||||
|
||||
+156
@@ -0,0 +1,156 @@
|
||||
from os.path import exists
|
||||
from time import sleep
|
||||
from datetime import datetime, timedelta
|
||||
|
||||
import pytz
|
||||
|
||||
from tweety import Twitter
|
||||
from tweety.types import *
|
||||
from tweety.exceptions_ import *
|
||||
from tweety.filters import SearchFilters
|
||||
|
||||
from account_pool import AccountPool
|
||||
from tweety_utils import *
|
||||
from talenttweet import *
|
||||
import talent_lists
|
||||
|
||||
class Scraper:
|
||||
def __init__(self):
|
||||
Scraper.instance = self
|
||||
self.__account = AccountPool()
|
||||
self.try_login()
|
||||
|
||||
def try_login(self, account_idx: int = None) -> bool:
|
||||
if account_idx is not None:
|
||||
acc = self.__account.use_index(account_idx)
|
||||
else:
|
||||
acc = self.__account.next()
|
||||
|
||||
if acc is not None:
|
||||
name = acc[0]
|
||||
print(f"using {name}")
|
||||
self.app = Twitter(name)
|
||||
if exists(f"{name}.json"):
|
||||
try:
|
||||
self.app.connect()
|
||||
except:
|
||||
self.app.sign_in(*acc)
|
||||
else:
|
||||
self.app.sign_in(*acc)
|
||||
return True
|
||||
print('exhausted all accounts!')
|
||||
return False
|
||||
|
||||
# since MUST BE TIMEZONE AWARE
|
||||
# usage example: since=datetime(2023, 8, 1).replace(tzinfo=pytz.utc)
|
||||
def get_tweets_from_user(self, username: str, since: datetime = None) -> list[Tweet]:
|
||||
reached_backdate = False
|
||||
tweets: list[Tweet] = []
|
||||
cur = None
|
||||
|
||||
if since == None:
|
||||
since = datetime.utcnow().replace(tzinfo=pytz.utc) - timedelta(days=7)
|
||||
print(f'falling back to grabbing tweets since 7 days ago ({since.date()})')
|
||||
else:
|
||||
print(f'grabbing tweets since {since.date()}')
|
||||
|
||||
uid = self.app._get_user_id(username)
|
||||
print(f"{username} = {uid}")
|
||||
|
||||
def add_tweet(tweet: Tweet):
|
||||
# malformed tweet check
|
||||
nonlocal reached_backdate
|
||||
try:
|
||||
tweet.author.id
|
||||
except:
|
||||
print(f"skipping malformed tweet: {tweet}")
|
||||
return
|
||||
|
||||
# recover lost info
|
||||
if tweet.is_retweet:
|
||||
if tweet.retweeted_tweet is None:
|
||||
print(f'{tweet.author.username}/{tweet.id} is missing the RT! It\'s probably nothing...')
|
||||
# tweet.retweeted_tweet = self.app.tweet_detail(str(tweet.id)).retweeted_tweet
|
||||
tweet.is_retweet = False
|
||||
elif tweet.retweeted_tweet.author is None:
|
||||
print(f'WARNING: {tweet.author.username}/{tweet.id} is missing the RT author! Recovering details...')
|
||||
tweet.retweeted_tweet = self.app.tweet_detail(tweet.retweeted_tweet.id)
|
||||
|
||||
if tweet.is_quoted:
|
||||
if tweet.quoted_tweet is None: # quoted tweet is deleted
|
||||
# print(f'{tweet.author.username}/{tweet.id} is missing the QRT! Recovering...')
|
||||
# tweet.quoted_tweet = self.app.tweet_detail(str(tweet.id)).quoted_tweet
|
||||
tweet.is_quoted = False
|
||||
elif tweet.quoted_tweet.author is None:
|
||||
print(f'WARNING: {tweet.author.username}/{tweet.id} is missing the QRT author! Recovering details...')
|
||||
tweet.quoted_tweet = self.app.tweet_detail(tweet.quoted_tweet.id)
|
||||
|
||||
# fix reply if it exists
|
||||
# if tweet.is_reply and tweet.replied_to is None:
|
||||
# tweet.replied_to = self.app.tweet_detail(tweet.original_tweet['in_reply_to_status_id_str'])
|
||||
tweets.append(tweet)
|
||||
|
||||
if not reached_backdate and int(tweet.author.id) == uid and tweet.date <= since:
|
||||
print("reached backdate")
|
||||
reached_backdate = True
|
||||
|
||||
if uid in talent_lists.privated_accounts:
|
||||
self.try_login(0)
|
||||
|
||||
while not reached_backdate:
|
||||
try:
|
||||
# uts = self.app.get_tweets(uid, replies=True, cursor=cur)
|
||||
search = self.app.search(f'from:{username}', filter_=SearchFilters.Latest(), cursor=cur)
|
||||
cur_page = search.tweets
|
||||
print(f'obtained {len(cur_page)} tweets')
|
||||
|
||||
if len(cur_page) == 0: break
|
||||
|
||||
for e in cur_page:
|
||||
if isinstance(e, Tweet):
|
||||
add_tweet(e)
|
||||
elif isinstance(e, TweetThread):
|
||||
# FIXME: rework when replied_to is fixed (currently populates user_mentions)
|
||||
# latest tweet in thread = og author's reply
|
||||
for t in e:
|
||||
add_tweet(t)
|
||||
|
||||
cur = search.cursor
|
||||
except UnknownError:
|
||||
print("UnknownError occurred, probably rate-limited")
|
||||
if uid in talent_lists.privated_accounts:
|
||||
print("sticking pvt-accessible account. sleeping for 2 minutes...")
|
||||
sleep(120)
|
||||
print()
|
||||
l = self.try_login(0)
|
||||
else:
|
||||
l = self.try_login()
|
||||
if not l:
|
||||
print("sleeping for 2 minutes...")
|
||||
sleep(120)
|
||||
print()
|
||||
self.try_login()
|
||||
|
||||
tweets.sort(key=lambda t: t.id)
|
||||
return tweets
|
||||
|
||||
def get_cross_ttweets_from_user(self, username: str, since_date: str = None) -> list[TalentTweet]:
|
||||
if since_date is not None:
|
||||
d = since_date.split('-')
|
||||
since = datetime(*[int(x) for x in d]).replace(tzinfo=pytz.utc)
|
||||
else:
|
||||
since = None
|
||||
tweets = self.get_tweets_from_user(username, since)
|
||||
# print_tweets(tweets)
|
||||
ret: list[TalentTweet] = []
|
||||
for t in tweets:
|
||||
tt = TalentTweet.create_from_tweety(t)
|
||||
if tt.is_cross_company():
|
||||
ret.append(tt)
|
||||
return ret
|
||||
|
||||
if __name__== '__main__':
|
||||
talent_lists.init()
|
||||
s = Scraper()
|
||||
ttweets = s.get_cross_ttweets_from_user("pomurainpuff", since=datetime(2023, 7, 30).replace(tzinfo=pytz.utc))
|
||||
print("\n".join([x.__repr__() for x in ttweets]))
|
||||
+28
-13
@@ -1,11 +1,12 @@
|
||||
import util
|
||||
|
||||
holo_en = dict()
|
||||
holo_id = dict()
|
||||
niji_en = dict()
|
||||
niji_exid = dict()
|
||||
talents = dict()
|
||||
talents_company = dict()
|
||||
holo_en: dict[int, str] = dict()
|
||||
holo_id: dict[int, str] = dict()
|
||||
niji_en: dict[int, str] = dict()
|
||||
niji_exid: dict[int, str] = dict()
|
||||
talents: dict[int, str] = dict()
|
||||
talents_company: dict[int, str] = dict()
|
||||
privated_accounts: dict[int, str] = dict()
|
||||
|
||||
test_talents = dict()
|
||||
|
||||
@@ -16,12 +17,15 @@ def __create_dict(file, _dict, company):
|
||||
with open(file, 'r') as f:
|
||||
for line in f:
|
||||
words = line.split()
|
||||
if len(words) == 2 and line[0] != '#':
|
||||
name, id = line.split()
|
||||
name = f'{util.get_username_online(id, default=name)}' # attempt to get updated name
|
||||
talents[int(id)] = name
|
||||
_dict[int(id)] = name
|
||||
talents_company[int(id)] = company
|
||||
if len(words) >= 2 and line[0] != '#':
|
||||
t = line.split()
|
||||
id, name = int(t[0]), t[1]
|
||||
# name = f'{util.get_username_online(id, default=name)}' # attempt to get updated name
|
||||
talents[id] = name
|
||||
_dict[id] = name
|
||||
talents_company[id] = company
|
||||
if len(words) > 2 and words[2] == 'p':
|
||||
privated_accounts[id] = name
|
||||
def init():
|
||||
global holo_en
|
||||
global holo_id
|
||||
@@ -36,11 +40,22 @@ def init():
|
||||
# nijiEN
|
||||
__create_dict(f'{util.get_project_dir()}/lists/nijien.txt', niji_en, 'nijiEN')
|
||||
# nijiexID
|
||||
__create_dict(f'{util.get_project_dir()}/lists/nijiexid.txt', niji_exid, 'nijiex-ID')
|
||||
__create_dict(f'{util.get_project_dir()}/lists/nijiexid.txt', niji_exid, 'nijiex\'ID')
|
||||
# TODO: nijiex-KR
|
||||
|
||||
test_talents = holo_en
|
||||
|
||||
def is_niji(id: int) -> bool:
|
||||
return id in niji_en or id in niji_exid
|
||||
|
||||
def is_holo(id: int) -> bool:
|
||||
return id in holo_en or id in holo_id
|
||||
|
||||
def is_cross_company(id1: int, id2: int):
|
||||
return (is_niji(id1) and is_holo(id2)) or (is_holo(id1) and is_niji(id2))
|
||||
|
||||
# For filtered stream
|
||||
# DEPRECATED: thx elon
|
||||
def get_twitter_rules():
|
||||
global talents
|
||||
rules = list()
|
||||
|
||||
+163
-113
@@ -1,26 +1,58 @@
|
||||
import datetime
|
||||
from datetime import datetime
|
||||
from zoneinfo import ZoneInfo
|
||||
import platform
|
||||
|
||||
import pytz
|
||||
from tweety.types import *
|
||||
|
||||
import twapi
|
||||
import talent_lists
|
||||
# from talent_lists import is_cross_company, talents
|
||||
import talent_lists as tl
|
||||
import util
|
||||
|
||||
class TalentTweet:
|
||||
# Serialized one-liner format:
|
||||
# {tweet} {author} {time in seconds since epoch UTC} m {mention set} r {reply to author} q {quote tweet author} rt {retweeted user's id} rtm {mentions in retweet}
|
||||
def serialize(self):
|
||||
s = f'{self.tweet_id} {self.author_id} {int(self.date_time.timestamp())} '
|
||||
if self.date_time.tzinfo is None:
|
||||
print(f'warning: serialized tweet {self.tweet_id} has a NAIVE timestamp!')
|
||||
|
||||
if len(self.rt_mentions) > 0:
|
||||
s += 'rtm '
|
||||
for n in self.rt_mentions:
|
||||
s += f'{n} '
|
||||
|
||||
if self.rt_author_id != None:
|
||||
s += f'rt {self.rt_author_id} '
|
||||
return s[:-1] # stop here since retweets can't have other info
|
||||
|
||||
if len(self.mentions) > 0:
|
||||
s += 'm '
|
||||
for id in self.mentions:
|
||||
s += f'{id} '
|
||||
if self.reply_to:
|
||||
s += f'r {self.reply_to} '
|
||||
if self.quote_tweeted:
|
||||
s += f'q {self.quote_tweeted} '
|
||||
|
||||
return s[:-1]
|
||||
|
||||
@staticmethod
|
||||
def deserialize(serialized_str: str):
|
||||
tokens = serialized_str.split()
|
||||
if len(tokens) < 3:
|
||||
token_check = serialized_str.split('#')[0]
|
||||
if len(token_check) < 3:
|
||||
raise ValueError('not enough tokens to reconstruct a TalentTweet')
|
||||
|
||||
tweet_id, author_id = int(tokens[0]), int(tokens[1])
|
||||
date_time = datetime.datetime.fromtimestamp(float(tokens[2]), tz=pytz.utc)
|
||||
tokens = serialized_str.split()
|
||||
|
||||
mentions = set()
|
||||
tweet_id, author_id = int(tokens[0]), int(tokens[1])
|
||||
date_time = datetime.fromtimestamp(float(tokens[2]), tz=pytz.utc)
|
||||
|
||||
mentions = list()
|
||||
reply_to = None
|
||||
quote_retweeted = None
|
||||
rt = None
|
||||
rtm = list()
|
||||
|
||||
mode = ''
|
||||
for i in range(3, len(tokens)):
|
||||
@@ -30,145 +62,100 @@ class TalentTweet:
|
||||
|
||||
if tokens[i].isnumeric():
|
||||
if mode == 'm': # mentions
|
||||
mentions.add(int(tokens[i]))
|
||||
mentions.append(int(tokens[i]))
|
||||
continue
|
||||
if mode == 'r': # reply_to
|
||||
reply_to = int(tokens[i])
|
||||
continue
|
||||
if mode == 'q': # quote_retweeted
|
||||
quote_retweeted = int(tokens[i])
|
||||
if mode == 'rt': # retweeted user
|
||||
rt = int(tokens[i])
|
||||
if mode == 'rtm': # retweet/qrt mentions
|
||||
rtm.append(int(tokens[i]))
|
||||
|
||||
return TalentTweet(
|
||||
tweet_id=tweet_id, author_id=author_id,
|
||||
date_time=date_time, mrq=(mentions, reply_to, quote_retweeted)
|
||||
date_time=date_time, mrq=(mentions, reply_to, quote_retweeted),
|
||||
rt_author_id=rt, rt_mentions=rtm
|
||||
)
|
||||
|
||||
## Creates a TalentTweet from a Tweety-library Tweet.
|
||||
@staticmethod
|
||||
async def create_from_twint_tweet(tweet):
|
||||
# MRQ
|
||||
mentions = set()
|
||||
reply_to = None
|
||||
quoted_id = None
|
||||
|
||||
# reply_to/mentions
|
||||
is_reply = tweet.id != int(tweet.conversation_id)
|
||||
mentions = set([x['id'] for x in tweet.mentions])
|
||||
if is_reply and len(tweet.reply_to) > 0:
|
||||
reply_to = tweet.reply_to[0]['id'] # FIXME: QRT = is_reply and len(tweet.reply_to) == 0?
|
||||
reply_others = [x['id'] for x in tweet.reply_to[1:]]
|
||||
mentions.update(reply_others)
|
||||
try: mentions.remove(reply_to)
|
||||
except: pass
|
||||
|
||||
# qrt
|
||||
if type(tweet.quote_url) == str:
|
||||
# print(f'url: {tweet.quote_url} ({type(tweet.quote_url)})')
|
||||
quote_tokens = tweet.quote_url.split('/')
|
||||
if len(quote_tokens) >= 2:
|
||||
quoted_username = quote_tokens[-2]
|
||||
quoted_id = util.get_user_id_local(quoted_username)
|
||||
if quoted_id == -1:
|
||||
quoted_id = util.get_user_id_online(quoted_username)
|
||||
|
||||
# NOTE: strptime doesn't attach timezone info.
|
||||
# tweet's datetime will be in local time
|
||||
date_time = datetime.datetime.strptime(tweet.datetime, '%Y-%m-%d %H:%M:%S %Z')
|
||||
LOCAL_TIMEZONE = datetime.datetime.now().astimezone().tzinfo
|
||||
date_time = date_time.replace(tzinfo=LOCAL_TIMEZONE) # attach system local timezone
|
||||
return TalentTweet(tweet_id=tweet.id, author_id=tweet.user_id, date_time=date_time, mrq=(mentions, reply_to, quoted_id))
|
||||
|
||||
@staticmethod
|
||||
def create_from_v2api_response(resp):
|
||||
tweet = resp.data
|
||||
if tweet is None: return None
|
||||
|
||||
mrq = twapi.TwAPI.get_mrq(resp)
|
||||
rt_target = None
|
||||
rt_author_id = None
|
||||
|
||||
# check if is RT
|
||||
if tweet.referenced_tweets is not None and len(tweet.referenced_tweets) > 0:
|
||||
for ref in tweet.referenced_tweets:
|
||||
if ref.type == 'retweeted':
|
||||
rt_target = ref.id
|
||||
for incl_tweet in resp.includes['tweets']:
|
||||
if incl_tweet.id == ref.id:
|
||||
rt_author_id = incl_tweet.author_id
|
||||
def create_from_tweety(tweety: Tweet):
|
||||
if tweety.is_retweet:
|
||||
rtm = [int(x.id) for x in tweety.retweeted_tweet.user_mentions]
|
||||
elif tweety.is_quoted:
|
||||
rtm = [int(x.id) for x in tweety.quoted_tweet.user_mentions]
|
||||
else:
|
||||
rtm = list()
|
||||
|
||||
return TalentTweet(
|
||||
tweet_id=tweet.id,
|
||||
author_id=tweet.author_id,
|
||||
date_time=tweet.created_at,
|
||||
mrq=mrq,
|
||||
rt_target=rt_target,
|
||||
rt_author_id=rt_author_id
|
||||
tweet_id=int(tweety.id), author_id=int(tweety.author.id),
|
||||
date_time=tweety.date, text=tweety.text,
|
||||
mrq=(
|
||||
[int(x.id) for x in tweety.user_mentions],
|
||||
int(tweety.original_tweet['in_reply_to_user_id_str']) if tweety.is_reply else None,
|
||||
int(tweety.quoted_tweet.author.id) if tweety.quoted_tweet is not None else None
|
||||
),
|
||||
rt_author_id=tweety.retweeted_tweet.author.id if tweety.is_retweet else None,
|
||||
rt_mentions=rtm
|
||||
)
|
||||
|
||||
@staticmethod
|
||||
async def create_from_id(id):
|
||||
resp = await twapi.TwAPI.instance.get_tweet_response(id)
|
||||
return TalentTweet.create_from_v2api_response(resp)
|
||||
|
||||
def __init__(self, tweet_id: int, author_id: int, date_time: datetime.datetime, mrq: tuple, rt_target: int=None, rt_author_id: int=None):
|
||||
def __init__(self, tweet_id: int, author_id: int, date_time: datetime, text: str = None, mrq: tuple[list[int], int|None, int|None]=None, rt_author_id: int=None, rt_mentions: list[int]=None):
|
||||
# basic information
|
||||
self.tweet_id, self.author_id = tweet_id, author_id
|
||||
self.username = util.get_username_local(self.author_id)
|
||||
self.date_time = date_time
|
||||
self.mentions = tuple(int(x) for x in mrq[0])
|
||||
self.reply_to = int(mrq[1]) if mrq[1] is not None else None
|
||||
self.quote_retweeted = int(mrq[2]) if mrq[2] is not None else None
|
||||
self.rt_target, self.rt_author_id = rt_target, rt_author_id
|
||||
self.text = text
|
||||
|
||||
# all users involved, except for the author
|
||||
self.all_parties = {self.reply_to, self.quote_retweeted}
|
||||
self.all_parties.update(self.mentions)
|
||||
try:
|
||||
self.all_parties.remove(None)
|
||||
# filter users to only be talents
|
||||
self.mentions = {x for x in mrq[0] if x in tl.talents}
|
||||
self.rt_mentions = {x for x in rt_mentions if x in tl.talents}
|
||||
|
||||
self.reply_to = mrq[1]
|
||||
self.quote_tweeted = mrq[2]
|
||||
self.rt_author_id = rt_author_id
|
||||
|
||||
try: self.mentions.remove(self.reply_to)
|
||||
except: pass
|
||||
try:
|
||||
self.all_parties.remove(self.author_id)
|
||||
|
||||
# -1 if user is not in company
|
||||
self.reply_to = self.reply_to if self.reply_to is None or self.reply_to in tl.talents else -1
|
||||
self.quote_tweeted = self.quote_tweeted if self.quote_tweeted is None or self.quote_tweeted in tl.talents else -1
|
||||
self.rt_author_id = self.rt_author_id if self.rt_author_id is None or self.rt_author_id in tl.talents else -1
|
||||
|
||||
# all users involved except for the author
|
||||
self.all_parties = {self.reply_to, self.quote_tweeted, rt_author_id}
|
||||
self.all_parties.update(self.mentions, self.rt_mentions)
|
||||
try: self.all_parties.remove(None)
|
||||
except: pass
|
||||
try: self.all_parties.remove(self.author_id)
|
||||
except: pass
|
||||
|
||||
|
||||
def __repr__(self) -> str:
|
||||
return (
|
||||
f'{self.tweet_id} from {util.get_username_local(self.author_id)}):\n'
|
||||
f'======================================================\n'
|
||||
f'{self.tweet_id} from {self.username}:\n'
|
||||
f'{self.get_datetime_str()}\n'
|
||||
f'{self.get_all_parties_usernames()}\n'
|
||||
f'parties: {self.get_all_parties_usernames()}\n'
|
||||
f'mentions: {self.mentions}\n'
|
||||
f'reply_to: {self.reply_to}\n'
|
||||
f'quote_retweeted: {self.quote_retweeted}\n'
|
||||
f'Cross-company: {self.is_cross_company()}\n'
|
||||
f'quote_retweeted: {self.quote_tweeted}\n'
|
||||
f'cross-company? {self.is_cross_company()}\n'
|
||||
f'{self.serialize()}\n'
|
||||
f'======================================================'
|
||||
f'----\n{self.announce_text()}\n----\n'
|
||||
f'{self.url()}'
|
||||
)
|
||||
|
||||
# Serialized one-liner format:
|
||||
# {tweet} {author} {time in seconds since epoch} m {mention_set} r {reply_to_author} q {quote_retweet_author}
|
||||
def serialize(self):
|
||||
s = f'{self.tweet_id} {self.author_id} {self.date_time.timestamp()} '
|
||||
if len(self.mentions) > 0:
|
||||
s += 'm '
|
||||
for id in self.mentions:
|
||||
s += f'{id} '
|
||||
if self.reply_to:
|
||||
s += f'r {self.reply_to} '
|
||||
if self.quote_retweeted:
|
||||
s += f'q {self.quote_retweeted} '
|
||||
return s[:-1]
|
||||
def url(self):
|
||||
return util.get_tweet_url(self.tweet_id, self.username)
|
||||
|
||||
def is_cross_company(self):
|
||||
for other_id in self.all_parties:
|
||||
if self.author_id in talent_lists.holo_en:
|
||||
if other_id in talent_lists.niji_en or other_id in talent_lists.niji_exid:
|
||||
return True
|
||||
if self.author_id in talent_lists.niji_en:
|
||||
if other_id in talent_lists.holo_en or other_id in talent_lists.holo_id:
|
||||
return True
|
||||
if self.author_id in talent_lists.holo_id:
|
||||
if other_id in talent_lists.niji_en or other_id in talent_lists.niji_exid:
|
||||
return True
|
||||
if self.author_id in talent_lists.niji_exid:
|
||||
if other_id in talent_lists.holo_en or other_id in talent_lists.holo_id:
|
||||
return True
|
||||
if tl.is_cross_company(self.author_id, other_id):
|
||||
return True
|
||||
return False
|
||||
|
||||
def get_all_parties_usernames(self):
|
||||
@@ -183,3 +170,66 @@ class TalentTweet:
|
||||
def get_datetime_str(self):
|
||||
unpad = '#' if platform.system() == 'Windows' else '-'
|
||||
return self.date_time.strftime(f'%b %{unpad}d %Y, %{unpad}I:%M%p (%Z)')
|
||||
|
||||
def announce_text(self):
|
||||
# templates
|
||||
TWEET = '{0} tweeted mentioning {1}!'
|
||||
REPLY = '{0} replied to {1}!'
|
||||
REPLY_TO_MENTION_B = '{0} replied to a tweet{1}mentioning {1}!' #########################
|
||||
RETWEET = '{0} retweeted {1}!'
|
||||
RETWEET_MENTIONS_B = '{0} shared a tweet{1}mentioning {2}!' #########################
|
||||
QUOTE_TWEET = '{0} quote tweeted {1}!'
|
||||
QUOTED_TWEET_MENTIONS_B = '{0} quoted a tweet{1}mentioning {2}!' #########################
|
||||
|
||||
author_username = f'@/{util.get_username_with_company(self.author_id)}'
|
||||
ret = str()
|
||||
|
||||
print_mention_ids = set(self.mentions)
|
||||
try: print_mention_ids.remove(None)
|
||||
except: pass
|
||||
mention_usernames = [f'@/{util.get_username_with_company(x)}' for x in print_mention_ids]
|
||||
|
||||
def rtm_msg(TEMPLATE: str, rtm_author_username: str):
|
||||
if self.rt_author_id != -1: # rtm tweet is from talent; rtm should be everyone
|
||||
rtm_names = [f'@/{util.get_username_with_company(x)}' for x in self.rt_mentions]
|
||||
between = f' from {rtm_author_username} '
|
||||
ret += TEMPLATE.format(author_username, between, ", ".join(rtm_names))
|
||||
else: # rtm tweet is not from a talent; rtm should just be cross company
|
||||
rtm_names = [f'@/{util.get_username_with_company(x)}' for x in self.rt_mentions if tl.is_cross_company(self.author_id, x)]
|
||||
ret += TEMPLATE.format(author_username, ' ', ", ".join(rtm_names))
|
||||
|
||||
# Tweet types
|
||||
if self.rt_author_id is not None: # retweet
|
||||
rt_username = f'@/{util.get_username_with_company(self.rt_author_id)}' if self.rt_author_id != -1 else None
|
||||
if len(self.rt_mentions) > 0:
|
||||
rtm_msg(RETWEET_MENTIONS_B, rt_username)
|
||||
else:
|
||||
ret += RETWEET.format(author_username, rt_username)
|
||||
elif self.reply_to is not None: # reply
|
||||
reply_username = f'@/{util.get_username_with_company(self.reply_to)}' if self.reply_to != -1 else None
|
||||
if len(self.rt_mentions) > 0:
|
||||
rtm_msg(REPLY_TO_MENTION_B, reply_username)
|
||||
else:
|
||||
ret += REPLY.format(author_username, reply_username)
|
||||
elif self.quote_tweeted is not None: # qrt
|
||||
quoted_username = f'@/{util.get_username_with_company(self.quote_tweeted)}' if self.quote_tweeted != -1 else None
|
||||
if len(self.rt_mentions) > 0:
|
||||
rtm_msg(QUOTED_TWEET_MENTIONS_B, quoted_username)
|
||||
else:
|
||||
ret += QUOTE_TWEET.format(author_username, quoted_username)
|
||||
elif len(self.mentions) > 0: # standalone tweet
|
||||
ret += TWEET.format(author_username, ", ".join(mention_usernames))
|
||||
f'[{self.get_datetime_str()}]\n'
|
||||
return ret
|
||||
else:
|
||||
raise ValueError(f'TalentTweet {self.tweet_id} has insufficient other parties')
|
||||
|
||||
# mention line
|
||||
if len(print_mention_ids) > 0:
|
||||
ret += (
|
||||
'\nMentioning '
|
||||
f'{", ".join(mention_usernames)}'
|
||||
)
|
||||
|
||||
ret += f'\n\n{self.get_datetime_str()}'
|
||||
return ret
|
||||
|
||||
+12
-3
@@ -1,6 +1,7 @@
|
||||
# TODO: move queue structures and file handling here
|
||||
import os
|
||||
import shutil
|
||||
import traceback
|
||||
|
||||
import util
|
||||
import talenttweet as tt
|
||||
@@ -55,9 +56,12 @@ class TalentTweetQueue:
|
||||
if len(tokens) == 0 or tokens[0][0] == '#':
|
||||
continue
|
||||
ttweet = tt.TalentTweet.deserialize(line)
|
||||
# print(f'{ttweet.tweet_id}:\n{ttweet}')
|
||||
self.ttweets_dict[ttweet.tweet_id] = ttweet
|
||||
print(f'Found {len(self.finished_user_dates)} scraped accounts and {len(self.ttweets_dict)} tweets in queue.')
|
||||
except: pass
|
||||
except:
|
||||
traceback.print_exc()
|
||||
pass
|
||||
# finished ttweets
|
||||
try:
|
||||
with open(self.finished_ttweets_path, 'r') as f:
|
||||
@@ -70,8 +74,8 @@ class TalentTweetQueue:
|
||||
return self.get_count() <= 0
|
||||
|
||||
def add_ttweet(self, ttweet):
|
||||
self.__sorted = False
|
||||
self.ttweets_dict[ttweet.tweet_id] = ttweet
|
||||
self.__sorted = False
|
||||
|
||||
def get_ttweet(self, id):
|
||||
return self.ttweets_dict[id]
|
||||
@@ -80,7 +84,10 @@ class TalentTweetQueue:
|
||||
self.is_good = False
|
||||
if os.path.exists(self.current_ttweet_path):
|
||||
with open(self.current_ttweet_path, 'r') as f:
|
||||
return tt.TalentTweet.deserialize(f.readline())
|
||||
ttweet = tt.TalentTweet.deserialize(f.readline())
|
||||
if ttweet.tweet_id in self.ttweets_dict:
|
||||
self.ttweets_dict.pop(ttweet.tweet_id)
|
||||
return ttweet
|
||||
|
||||
self.__sort_ttweets_dict()
|
||||
key = list(self.ttweets_dict.keys())[0]
|
||||
@@ -105,6 +112,7 @@ class TalentTweetQueue:
|
||||
|
||||
# overwrite queue.txt
|
||||
def save_file(self):
|
||||
print('saving file...', end='')
|
||||
shutil.copyfile(self.queue_path, self.queue_backup_path)
|
||||
self.__sort_ttweets_dict()
|
||||
with open(self.queue_path, 'w') as f:
|
||||
@@ -117,6 +125,7 @@ class TalentTweetQueue:
|
||||
# write sorted ttweets
|
||||
for ttweet in self.ttweets_dict.values():
|
||||
f.write(ttweet.serialize() + '\n')
|
||||
print('done')
|
||||
|
||||
def add_finished_tweet(self, id):
|
||||
self.finished_ttweets.append(id)
|
||||
|
||||
+17
-159
@@ -2,9 +2,9 @@ import datetime
|
||||
import traceback
|
||||
import asyncio
|
||||
|
||||
from dotenv import dotenv_values
|
||||
import tweepy
|
||||
|
||||
import api_secrets
|
||||
import talenttweet as tt
|
||||
import talent_lists as tl
|
||||
import ttweetqueue as ttq
|
||||
@@ -70,104 +70,24 @@ class TwAPI:
|
||||
|
||||
|
||||
def __init__(self):
|
||||
creds = dotenv_values()
|
||||
TwAPI.instance = self
|
||||
self.client = tweepy.Client(
|
||||
bearer_token=api_secrets.bearer_token(),
|
||||
consumer_key=api_secrets.api_key(), consumer_secret=api_secrets.api_secret(),
|
||||
access_token=api_secrets.access_token(), access_token_secret=api_secrets.access_secret()
|
||||
consumer_key=creds['app_key'], consumer_secret=creds['app_secret'],
|
||||
access_token=creds['user_token'], access_token_secret=creds['user_secret']
|
||||
)
|
||||
self.api = tweepy.API(
|
||||
auth=tweepy.OAuthHandler(
|
||||
consumer_key=api_secrets.api_key(), consumer_secret=api_secrets.api_secret(),
|
||||
access_token=api_secrets.access_token(), access_token_secret=api_secrets.access_secret()
|
||||
consumer_key=creds['app_key'], consumer_secret=creds['app_secret'],
|
||||
access_token=creds['user_token'], access_token_secret=creds['user_secret']
|
||||
)
|
||||
)
|
||||
|
||||
try:
|
||||
self.me = self.client.get_me().data
|
||||
except Exception as e:
|
||||
print('Did you setup secrets.ini?')
|
||||
raise e
|
||||
print(f'Assuming the account of @{self.me.data["username"]} ({self.me["id"]})')
|
||||
|
||||
## ---[COMMENT OUT WHEN NOT IN USE]---
|
||||
# async def nuke_tweets(self):
|
||||
# async def delete_tweet(id):
|
||||
# try:
|
||||
# self.client.delete_tweet(id)
|
||||
# except tweepy.TooManyRequests as e:
|
||||
# wait_for = float(e.response.headers["x-rate-limit-reset"]) - datetime.datetime.now().timestamp() + 1
|
||||
# print(f'\thit rate limit deleting {id}, retrying in {wait_for} seconds...')
|
||||
# await asyncio.sleep(wait_for)
|
||||
# print('continuing...')
|
||||
# await delete_tweet(id)
|
||||
|
||||
# print(f'Retrieving all of {self.me["username"]}\'s tweets...')
|
||||
# tweets = self.get_all_tweet_ids_from_user(self.me['id'])
|
||||
|
||||
# print(f'Retrieved {len(tweets)} tweets.')
|
||||
# if not len(tweets) > 0:
|
||||
# print('No tweets obtained. Make sure the profile is public.')
|
||||
# return
|
||||
|
||||
# print(f'Deleting {len(tweets)} tweets...')
|
||||
# deleted_count = 0
|
||||
# try:
|
||||
# for tweet in tweets:
|
||||
# print(f'deleted {deleted_count}/{len(tweets)}')
|
||||
# await delete_tweet(tweet.id)
|
||||
# await asyncio.sleep(0.5)
|
||||
# deleted_count += 1
|
||||
# except:
|
||||
# print('Unhandled error occurred while trying to delete tweets.')
|
||||
# traceback.print_exc()
|
||||
# print('Try running again.')
|
||||
# else:
|
||||
# print('Saul Gone')
|
||||
|
||||
def get_all_tweet_ids_from_user(self, user_id):
|
||||
next_page_token = None
|
||||
tokens_retrieved = 0
|
||||
tweets_retrieved = 0
|
||||
tweets = list()
|
||||
while True:
|
||||
print(f'Retrieved {tokens_retrieved} tokens so far...')
|
||||
resp = self.client.get_users_tweets(
|
||||
user_id, max_results=100, pagination_token=next_page_token,
|
||||
media_fields=TwAPI.TWEET_MEDIA_FIELDS,
|
||||
tweet_fields=TwAPI.TWEET_FIELDS,
|
||||
expansions=TwAPI.TWEET_EXPANSIONS
|
||||
)
|
||||
|
||||
for tweet in resp.data:
|
||||
tweets.append(tweet)
|
||||
|
||||
# update counters and pagination token
|
||||
tweets_retrieved += resp.meta['result_count']
|
||||
try:
|
||||
next_page_token = resp.meta['next_token']
|
||||
tokens_retrieved += 1
|
||||
except KeyError:
|
||||
print("next_token wasn't provided; we've reached the end!")
|
||||
break # reached end of user's tweets
|
||||
|
||||
print(f'Retrieved {tweets_retrieved} tweets using {tokens_retrieved} tokens.')
|
||||
return tweets
|
||||
|
||||
async def get_tweet_response(self, id, attempt = 0):
|
||||
try:
|
||||
twt = TwAPI.instance.client.get_tweet(
|
||||
id,
|
||||
media_fields=TwAPI.TWEET_MEDIA_FIELDS,
|
||||
tweet_fields=TwAPI.TWEET_FIELDS,
|
||||
expansions=TwAPI.TWEET_EXPANSIONS
|
||||
)
|
||||
TwAPI.tweets_fetched += 1
|
||||
return twt
|
||||
except tweepy.TooManyRequests as e:
|
||||
wait_for = float(e.response.headers["x-rate-limit-reset"]) - datetime.datetime.now().timestamp() + 1
|
||||
print(f'[{attempt}]\tget_tweet_response({id}):\n\thit rate limit after {TwAPI.tweets_fetched} fetches -- trying again in {wait_for} seconds...')
|
||||
await asyncio.sleep(wait_for)
|
||||
return await self.get_tweet_response(id, attempt=attempt+1)
|
||||
print(f'Assuming the account of @{self.me.data["username"]} ({self.me["id"]})')
|
||||
except:
|
||||
pass
|
||||
|
||||
async def post_tweet(self, text='', media_ids: list=None, reply_to_tweet: int=None, quote_tweet_id: int=None):
|
||||
try:
|
||||
@@ -186,78 +106,31 @@ class TwAPI:
|
||||
|
||||
# return True = successfully posted a single ttweet
|
||||
# return False = did not post ttweet (duplicate)
|
||||
async def post_ttweet(self, ttweet: tt.TalentTweet, is_catchup=False, dry_run=False):
|
||||
async def post_ttweet(self, ttweet: tt.TalentTweet, dry_run=False):
|
||||
print(f'------{ttweet.tweet_id} ({util.get_username_local(ttweet.author_id)})------')
|
||||
|
||||
REPLY = '{0} replied to {1}!\n'
|
||||
QUOTE_TWEET = '{0} quote tweeted {1}!\n'
|
||||
TWEET = '{0} tweeted!\n'
|
||||
RETWEET = '{0} retweeted {1}!\n'
|
||||
|
||||
def create_text():
|
||||
author_username = f'@/{util.get_username_with_company(ttweet.author_id)}'
|
||||
print_mention_ids = set(ttweet.mentions)
|
||||
ret = str()
|
||||
if is_catchup:
|
||||
ret += f'{ttweet.get_datetime_str()}\n'
|
||||
pass
|
||||
|
||||
# Tweet types
|
||||
if ttweet.rt_target is not None: # retweet
|
||||
ret += RETWEET.format(f'{author_username}', f'@/{util.get_username_with_company(ttweet.rt_author_id)}')
|
||||
elif ttweet.reply_to is not None: # reply
|
||||
reply_username = f'@/{util.get_username_with_company(ttweet.reply_to)}'
|
||||
ret += REPLY.format(author_username, reply_username)
|
||||
# if qrt, push id into mentions
|
||||
print_mention_ids.add(ttweet.quote_retweeted)
|
||||
elif ttweet.quote_retweeted is not None: # qrt
|
||||
quoted_username = f'@/{util.get_username_with_company(ttweet.quote_retweeted)}'
|
||||
ret += QUOTE_TWEET.format(author_username, quoted_username)
|
||||
elif len(ttweet.mentions) > 0: # standalone tweet
|
||||
ret += TWEET.format(author_username)
|
||||
else:
|
||||
raise ValueError(f'TalentTweet {ttweet.tweet_id} has insufficient other parties')
|
||||
|
||||
try: print_mention_ids.remove(None)
|
||||
except: pass
|
||||
|
||||
# mention line
|
||||
if len(print_mention_ids) > 0:
|
||||
mention_usernames = [f'@/{util.get_username_with_company(x)}' for x in print_mention_ids]
|
||||
ret += (
|
||||
'mentioning '
|
||||
f'{", ".join(mention_usernames)}\n'
|
||||
)
|
||||
ret += '\n'
|
||||
# ret += '(this is a missed tweet)\n' if is_catchup else ''
|
||||
return ret
|
||||
|
||||
text = create_text()
|
||||
ttweet_url = util.ttweet_to_url(ttweet)
|
||||
text = ttweet.announce_text()
|
||||
ttweet_url = ttweet.url()
|
||||
|
||||
if dry_run: print('-------------------- DRY RUN --------------------')
|
||||
print(text)
|
||||
print(ttweet)
|
||||
if dry_run: return False
|
||||
|
||||
# NO DRY-RUN: actually post tweet
|
||||
# main tweet: text + screenshot
|
||||
try:
|
||||
print('creating main QRT w/ screenshot...', end='')
|
||||
print('creating main QRT w/ screenshot...')
|
||||
media_ids = [await self.get_ttweet_image_media_id(ttweet)]
|
||||
twt_resp = await self.post_tweet(text, media_ids=media_ids, quote_tweet_id=ttweet.tweet_id)
|
||||
print('done')
|
||||
except:
|
||||
print('error occurred trying to create main tweet, falling back to URL-main + reply screencap format')
|
||||
traceback.print_exc()
|
||||
text += f"\n{ttweet_url}"
|
||||
try:
|
||||
print('posting main tweet...', end='')
|
||||
twt_resp = await self.post_tweet(text)
|
||||
print('posting main tweet...')
|
||||
twt_resp = await self.post_tweet(text, quote_tweet_id=ttweet.tweet_id)
|
||||
print('done')
|
||||
twt_id = twt_resp.data['id']
|
||||
# if ttweet.reply_to is not None:
|
||||
# re_ttweet = tt.TalentTweet(tweet_id=ttweet.reply_to, author_id=)
|
||||
# media_ids.insert(0, await self.get_ttweet_image_media_id())
|
||||
|
||||
try:
|
||||
print('creating reply img...', end='')
|
||||
@@ -275,18 +148,3 @@ class TwAPI:
|
||||
else:
|
||||
raise e
|
||||
return True
|
||||
|
||||
def post_ttweet_by_id(self, tweet_id, is_catchup=False, dry_run=False):
|
||||
ttweet = asyncio.run(tt.TalentTweet.create_from_id(tweet_id))
|
||||
print(f'm({ttweet.mentions}), r({ttweet.reply_to}), q({ttweet.quote_retweeted})')
|
||||
if ttweet.is_cross_company():
|
||||
print(f'Tweet {ttweet.tweet_id} is cross-company! Creating post...')
|
||||
asyncio.run(self.post_ttweet(ttweet, is_catchup=is_catchup, dry_run=dry_run))
|
||||
ttq.TalentTweetQueue.instance.add_finished_tweet(ttweet.tweet_id)
|
||||
else:
|
||||
print(f'Tweet {tweet_id} is not cross-company.')
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
@@ -0,0 +1,24 @@
|
||||
from tweety.types import *
|
||||
|
||||
def url(t: Tweet):
|
||||
return f'https://twitter.com/{t.author.username}/status/{t.id}'
|
||||
|
||||
def print_tweets(tweets: list[Tweet | TweetThread]):
|
||||
print(f'{len(tweets)} tweets:')
|
||||
for t in tweets:
|
||||
if isinstance(t, Tweet):
|
||||
print(f'{t.date} : {url(t)} :', end=' ')
|
||||
|
||||
if t.is_retweet:
|
||||
print(f'RT ({t.retweeted_tweet.author.username})', end=' ')
|
||||
|
||||
if t.is_reply:
|
||||
print(f'is reply!', end=' ')
|
||||
if t.replied_to is not None:
|
||||
print(f'reply to {t.replied_to.author.username}', end=' ')
|
||||
|
||||
print("m=" + ",".join([x.username for x in t.user_mentions]))
|
||||
elif isinstance(t, TweetThread):
|
||||
print('-----------TTd----------')
|
||||
print_tweets(t.tweets)
|
||||
print('-----------end----------')
|
||||
+16
-52
@@ -4,6 +4,7 @@ import os
|
||||
import sys
|
||||
import traceback
|
||||
from datetime import datetime
|
||||
from dotenv import dotenv_values
|
||||
|
||||
import tweepy
|
||||
import pytz
|
||||
@@ -45,7 +46,7 @@ def get_current_timestamp():
|
||||
def get_current_date():
|
||||
return datetime.today().strftime('%Y-%m-%d')
|
||||
|
||||
def get_key_from_value(d, val):
|
||||
def get_key_from_value(d: dict, val):
|
||||
keys = [k for k, v in d.items() if v == val]
|
||||
if keys:
|
||||
return keys[0]
|
||||
@@ -53,11 +54,12 @@ def get_key_from_value(d, val):
|
||||
|
||||
async def create_ttweet_image(ttweet):
|
||||
tc = TweetCapture()
|
||||
tc.cookies = [{'name': 'auth_token', 'value': dotenv_values()['web_auth_token']}]
|
||||
if 'linux' in sys.platform:
|
||||
# Linux chromedriver path
|
||||
tc.driver_path = '/usr/bin/chromedriver'
|
||||
filename = f'{get_project_dir()}/img.png'
|
||||
url = ttweet_to_url(ttweet)
|
||||
url = ttweet.url()
|
||||
img = None
|
||||
print(url)
|
||||
try: os.remove(filename)
|
||||
@@ -66,7 +68,7 @@ async def create_ttweet_image(ttweet):
|
||||
img = await tc.screenshot(
|
||||
url=url,
|
||||
path=filename,
|
||||
mode=4,
|
||||
mode=0,
|
||||
night_mode=1,
|
||||
show_parent_tweets=True
|
||||
)
|
||||
@@ -80,28 +82,20 @@ async def create_ttweet_image(ttweet):
|
||||
return img
|
||||
|
||||
def get_tweet_url(id, username):
|
||||
return f'https://twitter.com/{username}/status/{id}'
|
||||
return f'https://www.twitter.com/{username}/status/{id}'
|
||||
|
||||
def ttweet_to_url(ttweet):
|
||||
username = get_username(ttweet.author_id)
|
||||
return get_tweet_url(ttweet.tweet_id, username)
|
||||
## Attempt to pull username from local; pull from online if doesn't exist.
|
||||
def get_username(id):
|
||||
ret = talent_lists.talents.get(id, None)
|
||||
if ret == None:
|
||||
return get_username_online(id)
|
||||
return ret
|
||||
|
||||
# twint
|
||||
# May not work with short user IDs (ie. 1354241437)
|
||||
# def get_username_online(id, default=None):
|
||||
# c = twint.Config()
|
||||
# c.User_id = id
|
||||
# c.Store_object = True
|
||||
# c.Hide_output = True
|
||||
# try:
|
||||
# twint.output.users_list.clear()
|
||||
# twint.run.Lookup(c)
|
||||
# user = twint.output.users_list[0]
|
||||
# return user.username
|
||||
# except:
|
||||
# return str(default) if default is not None else f'{id}'
|
||||
def get_username_with_company(id):
|
||||
company = talent_lists.talents_company.get(id, None)
|
||||
return f'{get_username(id)} {f"({company})" if company is not None else ""}'
|
||||
|
||||
def get_username_local(id):
|
||||
def get_username_local(id: int):
|
||||
return talent_lists.talents.get(id, f'{id}')
|
||||
|
||||
# Retrieve username via API v2 (tweepy)
|
||||
@@ -115,33 +109,3 @@ def get_username_online(id, default=None):
|
||||
print(f'Unhandled error retrieving username for {id}!')
|
||||
traceback.print_exc()
|
||||
return str(default) if default is not None else f'id:{id}'
|
||||
|
||||
## Attempt to pull username from local; pull from online if doesn't exist.
|
||||
def get_username(id):
|
||||
ret = talent_lists.talents.get(id, None)
|
||||
if ret == None:
|
||||
return get_username_online(id)
|
||||
return ret
|
||||
|
||||
def get_username_with_company(id):
|
||||
company = talent_lists.talents_company.get(id, None)
|
||||
return f'{get_username(id)} {f"({company})" if company is not None else ""}'
|
||||
|
||||
def get_user_id_local(username) -> int:
|
||||
talent_usernames = list(talent_lists.talents.values())
|
||||
for i in range(0, len(talent_usernames)):
|
||||
if username.lower() == talent_usernames[i].lower():
|
||||
return list(talent_lists.talents)[i]
|
||||
|
||||
def get_user_id_online(username) -> int:
|
||||
c = twint.Config()
|
||||
c.Username = username
|
||||
c.Store_object = True
|
||||
c.Hide_output = True
|
||||
try:
|
||||
twint.output.users_list.clear()
|
||||
twint.run.Lookup(c)
|
||||
user = twint.output.users_list[0]
|
||||
return user.id
|
||||
except:
|
||||
return -1
|
||||
|
||||
Reference in New Issue
Block a user