Small utility application to update Saxon-HE from GitHub

saxonicaUpdater.cs

using System;
using System.IO;
using System.Net.Http;
using System.Security.Cryptography;
using System.Xml;

namespace SaxonicaDownloader
{
    class Program
    {
        static void Main(string[] args)
        {
            const string mavenMetadataUrl = "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/maven-metadata.xml";
            const string mavenRepositoryUrl = "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/";

            // Download and verify maven-metadata.xml
            string mavenMetadataPath = DownloadAndVerifyFile(mavenMetadataUrl, Path.GetTempPath());

            // Parse maven-metadata.xml to get the latest version
            XmlDocument xmlDoc = new XmlDocument();
            xmlDoc.Load(mavenMetadataPath);
            string latestVersion = xmlDoc.SelectSingleNode("/metadata/versioning/latest").InnerText;

            // Download and verify JAR files
            string versionUrl = $"{mavenRepositoryUrl}{latestVersion}/";
            HttpClient client = new HttpClient();
            HttpResponseMessage response = client.GetAsync(versionUrl).Result;
            response.EnsureSuccessStatusCode();
            string htmlContent = response.Content.ReadAsStringAsync().Result;

            // Parse HTML content to get file links
            HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
            htmlDoc.LoadHtml(htmlContent);
            var fileLinks = htmlDoc.DocumentNode.SelectNodes("//a[contains(@href, '.jar')]");

            foreach (HtmlAgilityPack.HtmlNode link in fileLinks)
            {
                string fileName = link.GetAttributeValue("href", string.Empty);
                string fileUrl = $"{versionUrl}{fileName}";

                // Download and verify file
                string filePath = DownloadAndVerifyFile(fileUrl, Path.GetTempPath());

                // Move the file to the destination folder
                string destinationFolder = Path.Combine(Environment.GetFolderPath(Environment.SpecialFolder.LocalApplicationData), "Programs", "Saxonica");
                Directory.CreateDirectory(destinationFolder);
                File.Move(filePath, Path.Combine(destinationFolder, fileName));
            }
        }

        static string DownloadAndVerifyFile(string fileUrl, string downloadFolder)
        {
            HttpClient client = new HttpClient();
            string fileName = Path.GetFileName(fileUrl);
            string filePath = Path.Combine(downloadFolder, fileName);

            // Download the file
            byte[] fileBytes = client.GetByteArrayAsync(fileUrl).Result;
            File.WriteAllBytes(filePath, fileBytes);

            // Verify file hashes
            VerifyFileHashes(fileUrl, filePath);

            return filePath;
        }

        static void VerifyFileHashes(string fileUrl, string filePath)
        {
            string[] hashAlgorithms = { "md5", "sha1", "sha256", "sha512" };

            foreach (string hashAlgorithm in hashAlgorithms)
            {
                string hashUrl = $"{fileUrl}.{hashAlgorithm}";
                HttpClient client = new HttpClient();
                string remoteHash = client.GetStringAsync(hashUrl).Result;

                using (var hashAlg = HashAlgorithm.Create(hashAlgorithm))
                {
                    byte[] fileBytes = File.ReadAllBytes(filePath);
                    string localHash = BitConverter.ToString(hashAlg.ComputeHash(fileBytes)).Replace("-", string.Empty).ToLowerInvariant();

                    if (remoteHash != localHash)
                    {
                        throw new Exception($"Hash mismatch for {hashAlgorithm}: Remote={remoteHash}, Local={localHash}");
                    }
                }
            }
        }
    }
}

saxonicaUpdater.ps1

#Requires -Modules PSParseHTML
$GpgExec = Get-Command -Name gpg -All -CommandType Application | Select-Object -First 1
Invoke-RestMethod -Uri 'https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/maven-metadata.xml' -Headers @{ Accept = 'application/xml' } -OutFile "$env:TEMP\maven-metadata.xml"
$HashesMatch = $true
$MavenHashAlgorithms = 'md5', 'sha1', 'sha256', 'sha512'
foreach ($HashAlgorithm in $MavenHashAlgorithms) {
    $RemoteHash = Invoke-RestMethod -Uri "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/maven-metadata.xml.${HashAlgorithm}"
    $LocalHash = Get-FileHash -Path "$env:TEMP\maven-metadata.xml" -Algorithm $HashAlgorithm | Select-Object -ExpandProperty Hash
    $HashesMatch = $HashesMatch -and $RemoteHash -ieq $LocalHash
    if(-not $HashesMatch){
        Write-Warning -Message "'maven-metadata.xml' hashes do not match for ${HashAlgorithm}"
        Write-Warning -Message "Remote: ${RemoteHash}"
        Write-Warning -Message "Local: ${LocalHash}"
        break
    }
}
if(-not $HashesMatch){
    throw "Hashes do not match"
}

[xml]$SaxonicaVersions = Get-Content -Path "$env:TEMP\maven-metadata.xml"
$LatestVersion = $SaxonicaVersions.SelectSingleNode('/metadata/versioning/latest/text()').InnerText
$FileListing = Invoke-WebRequest -Uri "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/${LatestVersion}/" -Headers @{ Accept = 'text/html,application/xhtml+xml,application/xml' } | ConvertFrom-HTML -Engine AngleSharp
$Links = $FileListing.QuerySelectorAll("#contents > a").InnerHtml
<# 
    Each file has a GPG signature and hashes for each algorithm.
    Each signature also has hashes for each algorithm.
    For example, 'Saxon-HE-12.4.pom' has 'Saxon-HE-12.4.pom.asc',
    'Saxon-HE-12.4.pom.asc.md5', 'Saxon-HE-12.4.pom.asc.sha1',
    'Saxon-HE-12.4.pom.asc.sha256', 'Saxon-HE-12.4.pom.asc.sha512',
    'Saxon-HE-12.4.pom.md5', 'Saxon-HE-12.4.pom.sha1',
    'Saxon-HE-12.4.pom.sha256', 'Saxon-HE-12.4.pom.sha512'
    Hence, we will iterate through the links and verify the signature hashes, the signature,
    then the file hashes in that order
#> 

foreach ($Link in $Links | Where-Object -FilterScript { $_ -imatch '\.jar$' }) {
    $HashesMatch = $true
    # Get the JAR file
    Invoke-WebRequest -Uri "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/${LatestVersion}/${Link}" -Headers @{ 'Content-Type' = 'application/octet-stream' } -OutFile "$env:TEMP\$Link"
    # Get the signature
    $Signature = "${Link}.asc"
    Invoke-RestMethod -Uri "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/${LatestVersion}/${Signature}" -OutFile "$env:TEMP\${Signature}"
    
    # Check file hash for the signature
    foreach($HashAlgorithm in $MavenHashAlgorithms){
        $HashFile = "${Signature}.${HashAlgorithm}"
        $RemoteHash = Invoke-RestMethod -Uri "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/${LatestVersion}/${HashFile}"
        $LocalHash = Get-FileHash -Path "$env:TEMP\${Signature}" -Algorithm $HashAlgorithm | Select-Object -ExpandProperty Hash
        $HashesMatch = $HashesMatch -and $RemoteHash -ieq $LocalHash
        if(-not $HashesMatch){
            Write-Warning -Message "'$Signature' hashes do not match for ${HashAlgorithm}"
            Write-Warning -Message "Remote: ${RemoteHash}"
            Write-Warning -Message "Local: ${LocalHash}"
            break
        }
    }
    if(-not $HashesMatch){
        throw "Hashes do not match"
    }
    # Check the file hash for the JAR
    foreach($HashAlgorithm in $MavenHashAlgorithms){
        $HashFile = "${Link}.${HashAlgorithm}"
        $RemoteHash = Invoke-RestMethod -Uri "https://repo1.maven.org/maven2/net/sf/saxon/Saxon-HE/${LatestVersion}/${HashFile}"
        $LocalHash = Get-FileHash -Path "$env:TEMP\$Link" -Algorithm $HashAlgorithm | Select-Object -ExpandProperty Hash
        $HashesMatch = $HashesMatch -and $RemoteHash -ieq $LocalHash
        if(-not $HashesMatch){
            Write-Warning -Message "'$Link' hashes do not match for ${HashAlgorithm}"
            Write-Warning -Message "Remote: ${RemoteHash}"
            Write-Warning -Message "Local: ${LocalHash}"
            break
        }
    }
    if(-not $HashesMatch){
        throw "Hashes do not match"
    }
    # Check the GPG signature
    if($GpgExec){
        # gpg --auto-key-locate hkps://keys.openpgp.org --keyserver hkps://keys.openpgp.org  --keyserver-options auto-key-retrieve --verify "$env:TEMP\${Signature}" "$env:TEMP\$Link"
        &$GpgExec --auto-key-locate keyserver --keyserver hkps://keys.openpgp.org  --keyserver-options auto-key-retrieve --verify "$env:TEMP\${Signature}" "$env:TEMP\$Link"
        if($LASTEXITCODE -ne 0){
            throw "GPG verification failed"
        }
    }
    # Move the JAR file to $env:LOCALAPPDATA\Programs\Saxonica
    Move-Item -Path "$env:TEMP\$Link" -Destination "$env:LOCALAPPDATA\Programs\Saxonica" -Force
}