feat(manifest): xpath support in checkver and autoupdate (#3458)

Author: Chawye Hsu

bin/checkver.ps1

@@ -112,6 +112,7 @@ $Queue | ForEach-Object {
     }
     $regex = ''
     $jsonpath = ''
+    $xpath = ''
     $replace = ''
 
     if ($json.checkver -eq 'github') {
@@ -140,12 +141,15 @@ $Queue | ForEach-Object {
     if ($json.checkver.jsonpath) {
         $jsonpath = $json.checkver.jsonpath
     }
+    if ($json.checkver.xpath) {
+        $xpath = $json.checkver.xpath
+    }
 
     if ($json.checkver.replace -and $json.checkver.replace.GetType() -eq [System.String]) {
         $replace = $json.checkver.replace
     }
 
-    if (!$jsonpath -and !$regex) {
+    if (!$jsonpath -and !$regex -and !$xpath) {
         $regex = $json.checkver
     }
 
@@ -159,6 +163,7 @@ $Queue | ForEach-Object {
         regex    = $regex;
         json     = $json;
         jsonpath = $jsonpath;
+        xpath    = $xpath;
         reverse  = $reverse;
         replace  = $replace;
     }
@@ -185,6 +190,7 @@ while ($in_progress -gt 0) {
     $url = $state.url
     $regexp = $state.regex
     $jsonpath = $state.jsonpath
+    $xpath = $state.xpath
     $reverse = $state.reverse
     $replace = $state.replace
     $expected_ver = $json.version
@@ -214,11 +220,25 @@ while ($in_progress -gt 0) {
         }
     }
 
+    if ($xpath) {
+        # Getting version from XML, using XPath
+        $ver = ([xml]$page).SelectSingleNode($xpath).'#text'
+        if (!$ver) {
+            next "couldn't find '$xpath' in $url"
+            continue
+        }
+    }
+
     if ($jsonpath -and $regexp) {
         $page = $ver
         $ver = ''
     }
 
+    if ($xpath -and $regexp) {
+        $page = $ver
+        $ver = ''
+    }
+
     if ($regexp) {
         $regex = New-Object System.Text.RegularExpressions.Regex($regexp)
         if ($reverse) {

lib/autoupdate.ps1

@@ -109,6 +109,29 @@ function find_hash_in_json([String] $url, [Hashtable] $substitutions, [String] $
     return format_hash $hash
 }
 
+function find_hash_in_xml([String] $url, [Hashtable] $substitutions, [String] $xpath) {
+    $xml = $null
+
+    try {
+        $wc = New-Object Net.Webclient
+        $wc.Headers.Add('Referer', (strip_filename $url))
+        $wc.Headers.Add('User-Agent', (Get-UserAgent))
+        $xml = [xml]$wc.downloadstring($url)
+    } catch [system.net.webexception] {
+        write-host -f darkred $_
+        write-host -f darkred "URL $url is not valid"
+        return
+    }
+
+    # Replace placeholders
+    if ($substitutions) {
+        $xpath = substitute $xpath $substitutions
+    }
+
+    $hash = $xml.SelectSingleNode($xpath).'#text'
+    return format_hash $hash
+}
+
 function find_hash_in_headers([String] $url) {
     $hash = $null
 
@@ -178,6 +201,12 @@ function get_hash_for_app([String] $app, $config, [String] $version, [String] $u
         $regex = $config.regex
     }
 
+    $xpath = ''
+    if ($config.xpath) {
+        $xpath = $config.xpath
+        $hashmode = 'xpath'
+    }
+
     if (!$hashfile_url -and $url -match "^(?:.*fosshub.com\/).*(?:\/|\?dwl=)(?<filename>.*)$") {
         $hashmode = 'fosshub'
     }
@@ -193,6 +222,9 @@ function get_hash_for_app([String] $app, $config, [String] $version, [String] $u
         'json' {
             $hash = find_hash_in_json $hashfile_url $substitutions $jsonpath
         }
+        'xpath' {
+            $hash = find_hash_in_xml $hashfile_url $substitutions $xpath
+        }
         'rdf' {
             $hash = find_hash_in_rdf $hashfile_url $basename
         }

schema.json

@@ -43,11 +43,15 @@
                     "pattern": "^\\$[.[].*$",
                     "type": "string"
                 },
+                "xpath": {
+                    "type": "string"
+                },
                 "mode": {
                     "enum": [
                         "download",
                         "extract",
                         "json",
+                        "xpath",
                         "rdf",
                         "metalink",
                         "fosshub",
@@ -237,6 +241,9 @@
                             "pattern": "^\\$[.[].*$",
                             "type": "string"
                         },
+                        "xpath": {
+                            "type": "string"
+                        },
                         "reverse": {
                             "description": "Reverse the order of regex matches",
                             "type": "boolean"