xmllint — xmllint XML parsing, XPath queries, DTD/XSD validation, pretty-print. pwsh Select-Xml + System.Xml alternatives. libxml2 vs Saxon for XSLT across all 5 shells

Equivalents in every shell

xmllint --xpath "//book[@id='1']/title/text()" books.xml

xmllint --xpath "//book[@id='1']/title/text()" books.xml

xmllint --xpath "//book[@id='1']/title/text()" books.xml

Select-Xml -Path books.xml -XPath "//book[@id='1']/title" | ForEach-Object { $_.Node.InnerText }

xmllint --xpath "//book[@id='1']/title/text()" books.xml

Worked examples

xmllint --xpath "//book[1]/title/text()" books.xml

xmllint --xpath "//book[1]/title/text()" books.xml

(Select-Xml -Path books.xml -XPath "//book[1]/title").Node.InnerText

xmllint --xpath "//book[1]/title/text()" books.xml

xmllint --format input.xml > output.xml

xmllint --format input.xml > output.xml

$xml = [xml](Get-Content input.xml); $sw = New-Object System.IO.StringWriter; $xw = New-Object System.Xml.XmlTextWriter($sw); $xw.Formatting = "Indented"; $xml.Save($xw); $sw.ToString() | Out-File output.xml

xmllint --format input.xml > output.xml

xmllint --schema schema.xsd document.xml --noout

xmllint --schema schema.xsd document.xml --noout

$xml = New-Object System.Xml.XmlDocument; $xml.Schemas.Add("", "schema.xsd") | Out-Null; $xml.Load("document.xml"); $xml.Validate($null)

xmllint --schema schema.xsd document.xml --noout

xmllint --noout document.xml && echo "valid"

xmllint --noout document.xml && echo "valid"

try { [xml](Get-Content document.xml) | Out-Null; "valid" } catch { "invalid: $_" }

xmllint --noout document.xml && echo valid

Gotchas

• **libxml2 is XPath 1.0 ONLY.** XPath 1.0 lacks: regex matching (`matches()`), higher-order functions, sequence operations, `let` bindings, native dates, `if/then/else`. For XPath 2.0+ features you need a different processor — `xmlstarlet` (similar CLI to xmllint, also 1.0) or Saxon-HE (XPath/XQuery 3.1, runs as `java -jar saxon-he.jar`). Common 1.0 workarounds: instead of `matches(@name, "^foo")`, use `starts-with(@name, "foo")`; instead of `current-date()`, pass as a parameter from the calling script. If your XPath is failing with cryptic errors, check: are you using a 2.0 feature in a 1.0 processor?
• **XML namespaces silently break unqualified XPath queries.** A document like `<root xmlns="http://example.com"><item>x</item></root>` looks unproblematic, but `xmllint --xpath "//item" file.xml` returns NOTHING — because `item` is in the `http://example.com` namespace, and unqualified XPath looks for the no-namespace `item`. Fix: bind a prefix and use it: `xmllint --xpath "//*[local-name()='item']" file.xml` (cheap workaround — ignores namespace), or properly use `--shell` and `setns ex http://example.com` then `xpath //ex:item`. For one-shot queries where namespace correctness doesn't matter, `local-name()='…'` is the fast escape hatch. For maintainable scripts, declare and bind the namespaces properly.
• **`--xpath` output is concatenated without separators.** Querying `//book/title` against 3 books returns `<title>A</title><title>B</title><title>C</title>` smashed together. To get one per line, use XPath's `string-join`: `--xpath "string-join(//book/title, '&#10;')"` (XPath 1.0 lacks `string-join` — for that you need libxml2 with EXSLT enabled, or a 2.0 processor). Workaround in 1.0: `xmllint --shell file.xml` and `cd //book` + iterate with `dir` / `cat`. Or pipe through `sed -e 's/></>\n</g'` to insert newlines between adjacent tags — fragile but works for simple cases.
• **XInclude and external entities — security implications.** xmllint resolves `<xi:include>` ELEMENTS by default (XInclude pulls in other files). It does NOT resolve external general entities (`<!ENTITY foo SYSTEM "file:///etc/passwd">`) unless `--loaddtd` or `--noent` is specified — disabled by default since CVE-2014-3660 (XXE). If you intentionally want entity expansion: `xmllint --noent --loaddtd file.xml`. For UNTRUSTED XML input (anything from a network / user upload), NEVER use `--noent`/`--loaddtd` — risk of XML eXternal Entity attacks reading server files. In .NET / pwsh, `XmlDocument.XmlResolver = $null` disables external entity resolution (default in modern .NET; older versions vulnerable).
• **XSLT support: libxslt + xsltproc, NOT xmllint.** xmllint can VALIDATE but cannot TRANSFORM via XSLT — that's `xsltproc` (sibling tool in libxslt). `xsltproc stylesheet.xsl input.xml > output.xml`. xsltproc supports XSLT 1.0 (libxslt is mature but 1.0-only). For XSLT 2.0/3.0: Saxon-HE. For RELAXED schema validation: `xmllint --relaxng schema.rng file.xml` (Relax NG is often more readable than XSD for hand-written schemas). For Schematron (rule-based validation): `xmllint` does NOT support it natively — use `iso-schematron-xslt2` from Schematron.com via xsltproc/Saxon.

WSL & PowerShell Core notes

Common tasks using xmllint

• Pretty-print an XML file

  Indent and format an XML document for human reading.