Try to Use Docker Container Manual

[svalcken]

Hello,

i tryed to use the Docker Container for Amazon DE.

In Environment is AMAZON_USER, AMAZON_PASSWORD defined.
My Account is with OTP, so i must start manual, i have found in an issue.

Start Container with
docker run --rm --platform linux/amd64 \ -e DEBUG=true \ -e AMAZON_USER=$amazon_email \ -e AMAZON_PASSWORD=$amazon_password \ -e AMAZON_TLD='de' \ -v "./data/:/home/node/docudigger" \ -v "/data/amazon-export/:/data/:rw" \ -it ghcr.io/disane87/docudigger:2.0.7-dev.3 /bin/bash

I have also tryed with ghcr.io/disane87/docudigger this image.

After start, i have this command used:
docudigger scrape amazon -u $AMAZON_USER -p $AMAZON_PASSWORD --yearFilter=2025 --onlyNew

This is the output:
(node:9) Warning: SyntaxError module: @oclif/core@4.0.31 task: findCommand (scrape:all.d) plugin: @disane-dev/docudigger root: /home/node/.npm-global/lib/node_modules/@disane-dev/docudigger message: Cannot use import statement outside a module See more details with DEBUG=* (Use node --trace-warnings ... to show where the warning was created) (node:9) Warning: SyntaxError module: @oclif/core@4.0.31 task: findCommand (scrape:amazon:index.d) plugin: @disane-dev/docudigger root: /home/node/.npm-global/lib/node_modules/@disane-dev/docudigger message: Cannot use import statement outside a module See more details with DEBUG=* Platform: linux [info] [2025-12-20 16:12:11] [scrape:amazon]: OnlyNew activated. [info] [2025-12-20 16:12:11] [scrape:amazon]: Getting last run [warn] [2025-12-20 16:12:11] [scrape:amazon]: process.json not found. Full run needed. OnlyNew deactivated. [info] [2025-12-20 16:12:11] [scrape:amazon]: Getting last run. Got: undefined [info] [2025-12-20 16:12:11] [scrape:amazon]: No latest scrape found. Processing all years and pages. [error] [2025-12-20 16:12:42] [scrape:amazon]: unhandledRejection: Navigation timeout of 30000 ms exceeded TimeoutError: Navigation timeout of 30000 ms exceeded at new Deferred (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/puppeteer-core/lib/cjs/puppeteer/util/Deferred.js:60:34) at Deferred.create (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/puppeteer-core/lib/cjs/puppeteer/util/Deferred.js:21:16) at new LifecycleWatcher (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/puppeteer-core/lib/cjs/puppeteer/cdp/LifecycleWatcher.js:69:60) at CdpFrame.waitForNavigation (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/puppeteer-core/lib/cjs/puppeteer/cdp/Frame.js:197:29) at CdpFrame.<anonymous> (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/puppeteer-core/lib/cjs/puppeteer/util/decorators.js:109:27) at CdpPage.waitForNavigation (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/puppeteer-core/lib/cjs/puppeteer/api/Page.js:605:43) at login (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/dist/commands/scrape/amazon/helpers/auth.helper.js:36:20) at async Amazon.run (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/dist/commands/scrape/amazon/index.js:24:26) at async Amazon._run (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/@oclif/core/lib/command.js:302:22) at async Config.runCommand (/home/node/.npm-global/lib/node_modules/@disane-dev/docudigger/node_modules/@oclif/core/lib/config/config.js:425:25)

i have no idea, what i can do to run this.
The OTP is not asks.

Nameserver is checked, manual Download with wget is checkt, so the download must work, but i can't check this.

What can i do?

[dosubot]

Hi @svalcken! I'm Dosu and I’m helping the docudigger team.

The main reason you aren't prompted for your OTP is that docudigger's Docker container currently does not support entering OTP/2FA codes for Amazon scraping. The code detects when OTP is required, but the part that should ask you for the code is commented out and not active, so the login just fails and you get a navigation timeout error from Puppeteer. This is a known limitation and affects all users with OTP enabled on their Amazon accounts in Docker, regardless of interactive/manual mode or environment variables—there is no supported way to provide the OTP right now [source] [source].

The warning about "Cannot use import statement outside a module" is a known issue in the current Docker images and doesn't directly cause the OTP problem; it's a side effect of how the CLI and plugins are loaded [source].

Currently, there are no official workarounds for OTP in Docker. Some users have suggested running the container interactively and entering the OTP manually, but in practice, the code does not prompt for it, so this doesn't work either. Disabling OTP or using a session with a remembered device might help, but persistent login is not supported because the browser profile is wiped on each run [source].

There are proposals to add support for OTP via environment variables or external integrations (like password managers), but these are not implemented yet. The maintainer is aware of the issue and is planning a refactor, but there is no ETA for a fix [source].

For now, if you need to download your Amazon invoices and have OTP enabled, you will need to use alternative methods, such as Amazon's privacy export or manual download.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[svalcken]

@dosu, Amazon's privacy export is for personal data not for invoices. Thats is no way.

Can the timeout changed from 30 sec to more? i have installed on Linux host. But this host has no X11 interface. So i use remote X on MacOS. in the 30sec i cant type password and then OTP code. So the Browser will be closed, befor i can end the login.
Bye the way, the required Chrome Version is older than current version.

i must installed the needed libs step by step.

The chrome has also a dialog for passkey, so this an aditional step and i think, the password cant't be please per script.

Is a possible way to use a headless chrome? I have seen in issue, a variant with OTP in environment. For it will be a possible methode. But can i this use?

[dosubot]

You can't currently change the 30-second Puppeteer navigation timeout via CLI or environment variable in docudigger; it's hardcoded. To increase it, you need to edit the code: add page.setDefaultNavigationTimeout(60000) (or a higher value) after line 99 in the newPage function of src/classes/scrape-command.class.ts to allow more time for manual login and OTP entry [source].

Headless Chrome is always used when running in Docker, and on Linux hosts without X11, so you don't need to set anything extra for headless mode [source].

There is no working way to provide OTP via environment variable or script—the code to prompt for OTP is present but commented out, so neither interactive nor automated OTP entry is possible right now [source].

Chrome passkey dialogs and extra authentication steps can't be automated or bypassed by docudigger; if Amazon requires a passkey, the login will fail [source].

Chrome version mismatches can cause issues; docudigger tries to use a pinned version in Docker, but if your system Chrome is newer or older, you may need to install the required version manually [source].

In summary: to get more time for login, you must patch the code; OTP and passkey automation are not supported; headless mode is automatic in Docker/Linux; and Chrome version must match what Puppeteer expects.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[svalcken]

@dosu, i have debugged.

One of the problem is, the dialog "no passkey" in Chrome apears and block the Password.
Manuel Close the Dalog and then put manual the Password, works. for this as workaround. Not fine, i have current not found, how disable the no passkey dialog.
The next problem is, the Script detect the mfa. and mark the save client flag. and then submit the page, but the otp is not inserted.
The Script, it is logged in and find no documents.
For me as user is the first idea, it is the timeout. but it is not the timeout.

How can i set source, Disableling from passkey Dialog and then wait for the otp?

[dosubot]

There’s currently no way in docudigger to disable the Chrome “no passkey” dialog programmatically—the Puppeteer launch arguments are basic and don’t include any flags to suppress passkey or authentication dialogs. Manual intervention (closing the dialog yourself) is the only workaround right now. Some users have reported that disabling passkeys in their Amazon account settings or using an in-private browser session can help, but there’s no built-in solution in docudigger for this dialog source.

For OTP: the code detects MFA, but the part that should prompt for OTP input is commented out. That means the script never waits for you to enter the OTP—it just clicks “remember device” and submits, skipping the OTP step. If you want to experiment, you can try uncommenting the lines in src/commands/scrape/amazon/helpers/auth.helper.ts that prompt for and type the OTP (look for the commented-out ux.prompt and page.type for the OTP field) source.

There’s no environment variable or CLI flag for OTP input yet, and no official support for automating passkey dialogs or OTP entry. These are known limitations and are being discussed for future updates, but aren’t implemented as of now source.

If you want to try Chrome/Puppeteer flags yourself, you could experiment with things like --disable-webauthn or --disable-features=WebAuthentication by adding them to the Puppeteer args in the code, but these are not guaranteed to work and are not part of docudigger’s defaults source.

In summary: disabling the passkey dialog and waiting for OTP both require code changes or manual workarounds; there’s no built-in support for either in the current version.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[svalcken]

@dosu,

the flags --disable-webauthn or --disable-features=WebAuthentication does not prevent the no passkey dialog.

i have any more options tryed out, but nothins has worked to disable the dialog. so i think here ist the first problem. The Script has the password and it can't put in the Password field. First in the Username, this is accepted, then comes the Page with password, but the passkey dialog apears and block the input. so the password write into nirvana.

on the page with otp, the script don't wait for user input and goes to the next step, after Login). Her breaks, it give to documents.

The test with enable the code for otp, runs into the error. ux is not defined. i must search what is ux.

[dosubot]

The "ux is not defined" error happens because the OTP prompt code in the Amazon auth helper uses CLI prompt utilities from @oclif/core (like ux.prompt), but the ux import is missing. To fix this, add import { ux } from "@oclif/core"; at the top of the file where you enabled the OTP prompt code. This will let the script prompt for OTP in interactive CLI mode, but it still won't work for unattended Docker runs or via environment variables—those features aren't implemented yet (source).

About the Chrome passkey dialog: there is no way to disable it programmatically in docudigger or Puppeteer. The dialog blocks the password field, so Puppeteer can't type the password correctly, and login fails. Manual intervention (closing the dialog yourself) is the only workaround right now. Some users have had luck disabling passkeys in their Amazon account settings or using an in-private browser session, but there is no reliable, built-in solution (source).

In summary: import ux from @oclif/core to fix the OTP prompt error, but OTP and passkey dialogs are not fully supported in docudigger yet.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[svalcken]

@dosu, here i have my problem.

i have patches the installed files, this is not the source code, it's allready js Files with a little bit other syntax.
so i have not used directly the source.

For this i am current not the right know how for node and how it work.

i cant put in the js files the import ux line. this does not work and direct an error apeares.

I think current, without "no Passkey" Dialog the user and password can be used, and the otp can in environment set or give as parameter to use it during login. The Code is already in, but i can't set current any value, no ux current avaible and i don't know, how can i import this.

in the file classes/scrap-command.class i have extended pupeteersArgs with
// --window-size=1920,1080, --window-size=1024,800, --no-sandbox, --disable-setuid-sandbox, --disable-features=PasswordManager,PasswordManagerFrontend, --enable-automation, --disable-features=WebAuthn, --disable-features=WebAuthnConditionalUI, --disable-webauthn, --disable-features=PasswordManagerOnboarding, --disable-save-password-bubble, --disable-autofill,

Nothing about this parameters helps to disable passkey dialog.

i have any hope to use it, but i can see, this is for me a little bit to much, to make it working.

So i have also the Problem, i won't install an X-Server on my server. My vision was an Docker Contaier with scrapping the invoices from amazon, and later extendig the system for other invocies. But my time is limited and i see current no way to fix it.
My idea was, use it for the invoices with paperless. Every Day Download the invoices without manual steps. Saving from browser to prevent MFA is an possible way, but the Browserdata will not be saved and i have no chance to save this.

i need for this a little bit more help.

i you have a version to try this, i can try it.

[dosubot]

You're running into these issues because patching the installed JS files in Node.js projects (especially for things like import { ux } from "@oclif/core") doesn't work reliably. The compiled JS files use a different module system (often CommonJS, not ES modules), so adding ES-style imports directly causes syntax errors or runtime failures. To properly enable OTP prompts or other features, you would need to edit the TypeScript source, add the correct imports, then rebuild the project and use the rebuilt package—just editing the JS files won't work and usually breaks things source.

Currently, there is no way to automate OTP input or disable the Chrome passkey dialog in docudigger. The code to prompt for OTP is present but commented out, and even if you enable it, it only works in interactive CLI mode (not unattended Docker runs), and only if you import ux correctly in the TypeScript source source. There is no support for passing OTP via environment variable or parameter, and no way to suppress the passkey dialog programmatically—manual intervention is required if it appears source.

Persistent browser sessions (which could avoid repeated MFA) are not supported, as docudigger wipes the browser profile on each run. Some users have suggested disabling passkeys in Amazon account settings or using in-private browser sessions, but there is no built-in solution for this source.

Right now, the only reliable way to use docudigger for unattended Amazon invoice scraping is to disable OTP on your Amazon account or use a device/session that Amazon already trusts (which is not always possible). The maintainer is aware of these limitations and is working on a refactor, but there is no ETA for a fix source.

If you want to follow or contribute to the discussion, these threads are the most relevant:

• OTP/2FA and Docker limitations
• Module system and refactor plans

Let me know if you want more details on any workaround or technical aspect!

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[svalcken]

@dosu,

i have another way found. i can open in browser an other tab. so i can manual login and save the browser.
After this step, i go back to the Page, opend from Scraper and close the no-passkey dialog and insert the password.
The login works without otp no. but the script does not find any years.

I think the otp is here not the problem, but nothing will be found, and the script works not for my, i think so.

[dosubot]

The reason docudigger isn’t finding any years or invoices after login is that Amazon changed its frontend, so the selectors and URL parameters in the current docudigger code are outdated—not because of OTP or session issues. You need to manually update the compiled JS files to match Amazon’s new structure for the scraper to work again [source].

Here’s what to do:

1. Edit dist/commands/scrape/amazon/helpers/selectors.helper.js:

   • Change:

     orderCards: 'div.order.js-order-card'
     orderNr: '.yohtmlc-order-id span:nth-last-child(1) bdi'
     yearFilter: 'select[name="orderFilter"]#orderFilter'

   • To:

     orderCards: 'div.order-card.js-order-card'
     orderNr: '.yohtmlc-order-id span:nth-last-child(1)'
     yearFilter: 'select[name="timeFilter"]#time-filter'

2. Edit dist/commands/scrape/amazon/index.js:

   • Find the function that builds the pagination URL (look for orderFilter=year-${year}) and change:

     const nextPageUrl = new URL(`?ie=UTF8&orderFilter=year-${year}&search=&startIndex=${10 * (orderPage)}`, amazon.orderPage);

   • To:

     const nextPageUrl = new URL(`?ie=UTF8&timeFilter=year-${year}&search=&startIndex=${10 * (orderPage)}&ref_=ppx_yo2ov_dt_b_filter_all_y${year}`, amazon.orderPage);

3. (Optional) Increase selector wait timeout for slow pages:

   • In dist/classes/scrape-command.class.js, set:

     this.selectorWaitTimeout = 30000

After these changes, run with LOG_LEVEL=debug to check if years and orders are detected. If you see “No years determined. Exiting.” in the logs, double-check the selector edits.

These are temporary workarounds and need to be repeated after every update until an official fix is released. If you run into errors with canceled orders (null click), there are extra patches for that—let me know if you need those details.

Full details and examples are in this thread.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}