peaeater

<#
    Zip up textbase files in preparation for FTP sync to Andornot.
    Peter Tyrrell, ptyrrell@andornot.com

    -textbases c:\path\to\accessions.tba, c:\path\to\descriptions.tba
    -out c:\path\to\destination
#>

param (
    [string[]] $textbases,

peaeater

param (
    [Parameter(Mandatory = $true)]
    [string]$in,
    [string]$out = "d:\dev-archive"
)

$extract_output_dir = "$in\extract\extracted\output\*"
remove-item $extract_output_dir -Recurse
Write-Host "Removed $extract_output_dir"

peaeater

<#
    Create sitemap index with attendant sitemaps from a Solr query.
    A new sitemap is created every 50,000 rows.
#>

param (
    [string]$ChangeFrequency = "weekly",
    [string]$IndexBaseUrl = "http://andi.andornot.com/",
    [string]$Logsrc = "Andi Solr Update",
    [string]$OutDir = ".\",

peaeater

<#
    Downsample PDF and convert to gray if necessary.
    Requires Ghostscript (gswin64c).
#>

param (
    [string]$indir,
    [string]$outdir = $indir,
    [string]$gs = "gswin64c",
    [string]$dpi = "150"

peaeater

# all logging settins are here on top
$logFile = "log-$(gc env:computername).log"
$logLevel = "DEBUG" # ("DEBUG","INFO","WARN","ERROR","FATAL")
$logSize = 1mb # 30kb
$logCount = 10
# end of settings

function Write-Log-Line ($line) {
    Add-Content $logFile -Value $Line
    Write-Host $Line

peaeater

<#
    Rip Audio CD to mp3 files with VLC.
#>

param (
    [string]$vlc = "c:\program files\videolan\vlc\vlc.exe",
    [string]$cddrive = "E:",
    [int]$track = 0
)

peaeater

@ECHO OFF

    setlocal ENABLEDELAYEDEXPANSION

    SET /a x=0

    FOR /R E:\ %%G IN (*.cda) DO (CALL :SUB_VLC "%%G")
    GOTO :eof

    :SUB_VLC

peaeater

param (
  [string]$server,
  [string]$dbname,
  [int]$daysToKeep,
  [Parameter(Mandatory=$false)]
  [string]$logsrc = "Andi Solr Update"
)

function logError([string]$logsrc, [string]$msg) {
    # write error msg to Application EventLog

peaeater

param (
  [Parameter(Mandatory=$true)]
  [string]$server,
  [Parameter(Mandatory=$true)]
  [string]$dbname,
  [Parameter(Mandatory=$true)]
  [int]$daysToKeep,
  [Parameter(Mandatory=$false)]
  [string]$logsrc = "Andi Solr Update"
)

peaeater

<#
    1. Leaf
    Given a text file of PDF filenames, extract content from PDFs recursively
    and create mirror directory structure for text file outputs.

    * Handles filenames with entry separators.
    * Ignores PDF older than its text file mirror unless -force param is used.
    * Requires poppler pdftotext.exe

    .\main.ps1 -in C:\dev\abc\extract\extracted\pdfs\db-pdfs-1.txt -pathprefix C:\dev\abc\raw\documents\publications -outdir C:\dev\abc\raw\documents\publications\test-out -poppler C:\utils\poppler-0.51\bin\pdftotext.exe