ilnmors-homelab/docs/services/app/paperless-ngx.md

paperless

Prerequisite

Create database

• Create the password with openssl rand -base64 32
  • Save this value in secrets.yaml in postgresql.password.paperless
  • Access infra server to create paperless_db with podman exec -it postgresql psql -U postgres

CREATE USER paperless WITH PASSWORD 'postgresql.password.paperless';
CREATE DATABASE paperless_db;
ALTER DATABASE paperless_db OWNER TO paperless;

Create oidc secret and hash

• Create the secret with openssl rand -base64 32
• access to auth vm
  • podman exec -it authelia sh
  • authelia crypto hash generate pbkdf2 --password 'paperless.oidc.secret'
• Save this value in secrets.yaml in paperless.oidc.secret and paperless.oidc.hash
• Use client_secret_post, django encodes the secret value wrong frequently.

Create session secret value

• Create the secret with LC_ALL=C tr -dc 'A-Za-z0-9!#%&()*+,-./:;<=>?@[\]^_{|}~' </dev/urandom | head -c 32
  • Save this value in secrets.yaml in paperless.session_secret

Create admin password

• Create the secret with openssl rand -base64 32
• Save this value in secrets.yaml in paperless.il.password

Add postgresql dump backup list

• set_postgresql.yaml

- name: Set connected services list
  ansible.builtin.set_fact:
    connected_services:
      - ...
      - "paperless"

Configuration

Access to paperless

• https://paperless.ilnmors.com
  • name: il
  • E-mail: il@ilnmors.internal
  • password: paperless.il.password

Oauth configuration

• My Profiles: Connect new social account: Authelia
  • Continue
  • Login with Authelia

OCR configuration

• Configuration: OCR settings
  • Output Type: pdfa
  • Mode: skip
    • When the archive file has broken ocr text, then conduct replcae command manually
  • Skip archive File: never
  • Deskew: disable (toggle to enable and once more to active disable option)
  • rotate: disable (toggle to enable and once more to active disable option)

The non-standard pdf file

• Some pdf files doesn't follow the standard, for example korean court or government pdf files.
• Before upload this kind of non-standard pdf files, convert it first.
• This process uses ghostscript and powershell in Windows for console

# 1. The engine
$gsPath = "C:\Program Files\gs\gs10.07.0\bin\gswin64c.exe"

# 2. new folder which the converted file will be stored
$outputDirName = "converted_pdfs"
$outputDir = Join-Path (Get-Location) $outputDirName
if (!(Test-Path $outputDir)) { New-Item -ItemType Directory -Path $outputDir }

# 3. Find all pdf files
$files = Get-ChildItem -Filter *.pdf

foreach ($file in $files) {
    if ($file.FullName -like "*$outputDirName*") { continue }
    
    $inputPath = $file.FullName
    $outputPath = Join-Path $outputDir $file.Name
    
    Write-Host "convert: $($file.Name)" -ForegroundColor Cyan
    
    $gsArgs = @(
        "-sDEVICE=pdfwrite",
        "-dCompatibilityLevel=1.4",
        "-dPDFSETTINGS=/default",
        "-dNOPAUSE",
        "-dQUIET",
        "-dBATCH",
        "-dNoOutputFonts", # Change all text as image
        "-sOutputFile=$outputPath",
        "$inputPath"
    )
    
    # 실행
    & $gsPath @gsArgs
}

Write-Host "`n[Complete] All file is stored in '$outputDirName'." -ForegroundColor Green