Also removed the 'page number' filter - would have to reimplement it
with the new format, which seems like effort. Also, such features are
better suited for the app to implement directly.
* CloudRecess multisrc
* ScyllaScans: move to cloudrecess multisrc
* FleksyScans: new source
* remove baseUrl
* simplify popular
always use /manga which isn't technically popular but list of all manga which is good enough
* HentaiSlayer (ar): move to cloudrecess
* remove CloudRecess (en): site appears to be dead
* small change
* rename
* review changes
why do I forget headers
* add alternative titles to description
* parse filters from popular and search response
avoid extra call
* remove placeholder author/artist
* LelscanVF: move to FuzzyDoodle
also improve date parsing
* add icons to lelscanvf
* flexyScans: change icon
other one is of Aksy Scan which also posts on the site
* i hate this theme
* bump
* remove useless slug update
* lint
* Update series slug on chapter list update
This was made for sources that changed slugs constantly.
Currently no one uses it, but who knows if they enable that again
* what an unstable experience
* Remove empty lines
* Fix intl
* newline
Co-authored-by: AwkwardPeak7 <48650614+AwkwardPeak7@users.noreply.github.com>
* why my build took 5 minutes
* I hate iguanas
---------
Co-authored-by: AwkwardPeak7 <48650614+AwkwardPeak7@users.noreply.github.com>
* remove randomua from madara
* don't use page path for page=1
* add back `madara_load_more`
* cleanup i18n and filters
* load more in a new source
* move back the filters
not worth it
* fix build
* altname to i18n as well
* utf-8
* Revert "utf-8"
This reverts commit 1335bc1b478da54d3a5eb6333ac1a26e3ee2825b.
* utf-8
* autodetect load_more_request
* load genres in background
* make genre classes protected
remove unnecessary change
* fetch genres changes
* launchIO countviews
* don't explicitly optin
* cleanup some request overrides
* make `useLoadMoreRequest` enum to be able to disable autodection where necessary
* fix logic
bruh
* use state variables
* defer countViews in overrides as well
* lint
* select().first -> selectFirst
* `load_more` search as well
* detect in search as well
* remove slipped override
* move detection to the function
* remove fetchGenreFailed
* don't use GlobalScope
* tweak load_more_request parameters
* remove ancient connectTimeout/readTimeout
already present in the client provided by the app
* small cleanup
* AralosBD: Added support for latest, better support for search and popular, new icons
* AralosBD: used the right icon generator to follow the partern used by other extensions icons
* AralosBD: updated icon to use squared version
* Update Japscan.kt
I completely changed the way to find the keys allowing the decryption of the base64 which contains the information of the images to be recovered. Before there was a recovery by regex, but this was not reliable because the location of these strings changes places and the order is important because it allows to reconstitute the two decryption keys. To succeed in recovering them correctly and in the right order, I debugged the javascript, which allowed me to understand how obfuscation works.
I understood that the base64 encryption file, the zjs, was not the same file when we had a mobile useragent and a PC useragent, it is however the same key, just that they are placed at different other place
at the end of the file we have the JS which assembles the keys directly as parameters of a function:
a0_0x39cacb('0x13d') + a0_0x39cacb('0x130') + a0_0x39cacb('0x118') + '6N', a0_0x39cacb('0x145') + a0_0x39cacb('0x116') + a0_0x39cacb('0x10f') + 'oe'
we have another function here:
return a0_0x46c1 = function(_0x46c1ab, _0x5a3fac) {
_0x46c1ab = _0x46c1ab - 0x105;
let _0x5c76b5 = _0xd80153[_0x46c1ab];
return _0x5c76b5;
}
which call string array which contains the key chunks
0 : "laretiLesrap"
1 : "gnp.600/sgmi/"
2 : "atad"
3 : "daol"
4 : "8ceABpX"
..
_0x46c1ab = _0x46c1ab - 0x105;
this operation allows to know the shift in the table, for example
By subtracting 0x105 from '0x13d', we get:
317 - 261 = 56
at location 56, so at index 55 of the array we have: "ohcVEMDTYO7FpBA20zRg"
so here: a0_0x39cacb('0x13d') + a0_0x39cacb('0x130') + a0_0x39cacb('0x118') + '6N'
we replace a0_0x39cacb('0x13d')
by this chain
"ohcVEMDTYO7FpBA20zRg" + a0_0x39cacb('0x130') + a0_0x39cacb('0x118') + '6N'
and that gives us the beginning of the key
and then reverse it
and that gives us the right key
* Update build.gradle
Update build version
* Update Japscan.kt - Completely reverse JavaScript
My modification was missing part of the javascript code to reverse, this modification brings greater stability.
In the JS is decrypts the data, there is a table of character string, this table is the most important element of the decryption because it is used in almost all the functions.
What the algorithm does in a nutshell:
- he shuffles the board until he arrives at a precise position, he pushes all the elements backwards until he finds the right combination. The correct combination is determined by the sum of the digits in the character strings with a more or less complex calculation.
- It calls the table in a "hidden" way with a function that takes a hexadecimal as a parameter (which corresponds to the index in the table)
- once all the keys have been retrieved from the character string table, it assembles them, then reverses them to give the complete key
How I proceeded:
- For the correct position in the table I perform almost the same calculation as the JS but I simplify, this to a tiny margin of error. In the calculation in the JS it parses the INTs in the chain and makes calculations with it, to simplify my task I only check if ALL the elements used contain INTs, I did several debug tests and they are obliged to contain them when the position is good. So starting from this principle I would almost certainly fall on the right position.
- Once my array is correctly sorted, I can with a simple calculation managed to recover the requested element with the hexadecimal
- I reverse the keys and I test the two positions of the keys to decipher the base64
* Update Japscan.kt
Added the possibility to have the parseInt in negative
* Update Japscan.kt
Modification needed to take into account the wider possibility of script changes
* Update Japscan.kt - Add comments
Ajout des commentaires
* Add an UserAgent on headersBuilder() to bypass Cloudflare error
* Add choice to set a custom UA
* Update header for imageRequest function
* Fix headers setup + remove function on setupPreferenceScreen
KirinRaikage
Draff