yegor256 / Xembly

Xembly is an Assembly-like imperative programming language for data manipulation in XML documents. It is a much simplier alternative to XSLT and XQuery. Read this blog post for a more detailed explanation: Xembly, an Assembly for XML. You may also want to watch this webinar.

You need this dependency:

<dependency>
  <groupId>com.jcabi.incubator</groupId>
  <artifactId>xembly</artifactId>
  <version>0.22</version>
</dependency>

Here is a command line implementation (as Ruby gem): xembly-gem

For example, you have an XML document:

<orders>
  <order id="553">
    <amount>$45.00</amount>
  </order>
</orders>

And you want to change the amount of the order #553 from $45.00 to $140.00. Xembly script would look like:

XPATH "orders/order[@id=553]";
SET "$140.00";

It is much simpler and compact than XSLT or XQuery.

This Java package implements Xembly:

Document document = DocumentBuilderFactory.newInstance()
  .newDocumentBuilder().newDocument();
new Xembler(
  new Directives(
    "ADD 'orders'; ADD 'order'; ATTR 'id', '553'; SET '$140.00';"
  )
).apply(document);

Since version 0.9 you can directly transform directives to XML:

String xml = new Xembler(
  new Directives()
    .add("root")
    .add("order")
    .attr("id", "553")
    .set("$140.00")
).xml();

This code will produce this XML document:

<root>
  <order id="553">$140</order>
</root>

Directives

Full list of supported directives in the current version:

• ADD: adds new node to all current nodes
• ADDIF: adds new node, if it's absent
• SET: sets text value of current node
• XSET: sets text value, calculating it with XPath
• CDATA: same as SET, but makes CDATA
• UP: moves cursor one node up
• XPATH: moves cursor to the nodes found by XPath
• REMOVE: removes all current nodes
• STRICT: throws an exception if cursor is missing nodes
• PI: adds processing instruction
• PUSH: saves cursor in stack
• POP: retrieves cursor from stack
• NS: sets namespace of all current nodes
• COMMENT: adds XML comment

"Cursor" or "current nodes" is where we're currently located in the XML document. When Xembly script starts, the cursor is empty and simply points to the highest level in the XML hierarchy. Pay attention, it doesn't point to the root node. It points to one level above the root. Remember, when document is empty, there is no root.

Then, we start executing directives one by one. After each directive cursor is moving somewhere. There may be many nodes under the cursor, or just one, or none. For example, let's assume we're starting with this simple document <car/>:

ADD 'hello';        // nothing happens, since cursor is empty
XPATH '/car';       // there is one node <car> under the cursor
ADD 'make';         // the result is "<car><make/></car>",
                    // cursor has one node "<make/>"
ATTR 'name', 'BMW'; // the result is "<car><make name='BMW'/></car>"
                    // cursor still has one node "<make/>"
UP;                 // cursor has one node "<car>"
ADD 'mileage';      // the result is "<car><make name='BMW'/><mileage/></car>"
                    // cursor still has one node "<car>"
XPATH '*';          // cursor has two nodes "<make name='BMW'/>"
                    // and "<mileage/>"
REMOVE;             // the result is "<car/>", since all nodes under
                    // the cursor are removed

You can create a collection of directives either from text or via supplementary methods, one per each directive. In both cases, you need to use class Directives:

import org.xembly.Directives;
new Directives("XPATH '//car'; REMOVE;");
new Directives().xpath("//car").remove();

The second option is preferable, because it is faster - there is no parsing involved.

ADD

ADD directive adds a new node to every node in the current node set. ADD expects exactly one mandatory argument, which is the name of a new node to be added (case sensitive):

ADD 'orders';
ADD 'order';

Even if the node with the same name already exists, a new node will be added. Use ADDIF if you need to add only if the same-name node is absent.

After execution, ADD directive moves the cursor to the nodes just added.

ADDIF

ADDIF directive adds a new node to every node of the current set, only if it's absent. ADDIF expects exactly one argument, which is the name of the node to be added (case sensitive):

ADD 'orders';
ADDIF 'order';

After execution, ADDIF directive moves the cursor to the nodes just added.

SET

SET changes text content of all current nodes, and expects exactly one argument, which is the text content to set:

ADD "employee";
SET "John Smith";

SET doesn't move the cursor anywhere.

XSET

XSET changes text content of all current nodes to a value calculated with XPath expression:

ADD "product-1";
ADD "price";
XSET "sum(/products/price) div count(/products)";

XSET doesn't move the cursor anywhere.

UP

UP moves all current nodes to their parents.

XPATH

XPATH changes current nodes to the all found by XPath expression:

XPATH "//employee[@id='234' and name='John Smith']/name";
SET "John R. Smith";

REMOVE

REMOVE removes current nodes under the cursor and moves the cursor to their parents:

ADD "employee";
REMOVE;

STRICT

STRICT checks that there is certain number of current nodes:

XPATH "//employee[name='John Doe']";  // move cursor to the employee
STRICT "1";                           // throw an exception if there
                                      // is not exactly one node under
                                      // the cursor

This is a very effective mechanism of validation of your script, in production mode. It is similar to assert statement in Java. It is recommended to use STRICT regularly, to make sure your cursor has correct amount of nodes, to avoid unexpected modifications.

STRICT doesn't move the cursor anywhere.

PI

PI directive add a new processing directive to the XML:

PI "xsl-stylesheet" "href='http://example.com'";

PI doesn't move the cursor anywhere.

PUSH and POP

PUSH and POP directives saves current DOM position to stack and restores it from there.

Let's say you start your Xembly manipulations from a place in DOM, which location is not determined for you. After your manipulations are done, you want to get back to exactly the same place. You should use PUSH to save your current location and POP to restore it back, when manipulations are finished, for example:

PUSH;                        // doesn't matter where we are
                             // we just save the location to stack
XPATH '//user[@id="123"]';   // move the cursor to a completely
                             // different location in the XML
ADD 'name';                  // add "<name/>" to all nodes under the cursor
SET 'Jeff';                  // set text value to the nodes
POP;                         // get back to where we were before the PUSH

PUSH basically saves the cursor into stack and POP restores it from there. This is a very similar technique to PUSH/POP directives in Assembly. The stack has no limits, you can push multiple times and pop them back. It is a stack, that's why it is First-In-Last-Out (FILO).

This operation is fast and it is highly recommended to use it everywhere, to be sure you're not making unexpected changes to the XML document. Every time you're not sure where your

NS

NS adds a namespace attribute to a node:

XPATH '/garage/car';                // move cursor to "<car/>" node(s)
NS "http://www.w3.org/TR/html4/";   // set namespace there

If original document was like this:

<garage>
  <car>BMW</car>
  <car>Toyota</car>
</garage>

After applying that two directives it will look like this:

<garage xmlns:a="http://www.w3.org/TR/html4/">
  <a:car>BMW</a:car>
  <a:car>Toyota</a:car>
</garage>

The namspace prefix may no necessarily be a:, but it doesn't really matter.

NS doesn't move the cursor anywhere.

XML Collections

Let's say you want to build an XML document with a collection of names:

package org.xembly.example;
import org.xembly.Directives;
import org.xembly.Xembler;
public class XemblyExample {
  public static void main(String[] args) throws Exception {
    String[] names = new String[] {
      "Jeffrey Lebowski",
      "Walter Sobchak",
      "Theodore Donald 'Donny' Kerabatsos",
    };
    Directives directives = new Directives().add("actors");
    for (String name : names) {
      directives.add("actor").set(name).up();
    }
    System.out.println(new Xembler(directives).xml());
  }
}

Standard output will contain this text:

<?xml version="1.0" encoding="UTF-8"?>
<actors>
  <actor>Jeffrey Lebowski</actor>
  <actor>Walter Sobchak</actor>
  <actor>Theodore Donald &apos;Donny&apos; Kerabatsos</actor>
</actors>

Merging Documents

When you need to add an entire XML document, you can convert it first into Xembly directives and then add them all together:

Iterable<Iterable> dirs = new Directives()
  .add("garage")
  .append(Directives.copyOf(node))
  .add("something-else");

This static utility method copyOf() converts an instance of class org.w3c.dom.Node into a collection of Xembly directives. Then, method append() adds them all together to the main list.

Unfortunately, not every valid XML document can be parsed by copyOf(). For example, this one will lead to a runtime exception: <car>2015<name>BMW</name></car>. Read more about Xembly limitations, a few paragraphs below.

Escaping Invalid XML Text

XML, as standard, doesn't allow certain characters in its body. For example, this code will throw an exception:

String xml = new Xembler(
  new Directives().add("car").set("\u00")
).xml();

Character \u00 is not allowed in XML. Actually, these ranges are not allowed: \u00..\u08, \u0B..\u0C, \u0E..\u1F, \u7F..\u84, and \u86..u9F.

This means that you should validate everything and make sure you're setting only "valid" text values to XML nodes. Sometimes, it's not feasible to always check them. Sometimes you may simply need to save whatever is possible and call it a day. There a utility static method Xembler.escape(), to help you do that:

String xml = new Xembler(
  new Directives().add("car").set(Xembler.escape("\u00"))
).xml();

This code won't throw an exception. Method Xembler.escape() will conver "\u00" to "\u0000". It is recommended to use this method everywhere, if you are not sure about the quality of the content.

Shaded Xembly JAR With Dependencies

Usually, you're supposed to use this dependency in your pom.xml:

<dependency>
  <groupId>com.jcabi.incubator</groupId>
  <artifactId>xembly</artifactId>
</dependency>

However, if you have conflicts between dependencies, you can use our "shaded" JAR, that includes all dependencies:

<dependency>
  <groupId>com.jcabi.incubator</groupId>
  <artifactId>xembly</artifactId>
  <classifier>jar-with-dependencies</classifier>
</dependency>

Known Limitations

Xembly is not intended to be a replacement of XSL or XQuery. It is a lightweight (!) instrument for XML manipulations. There are a few things that can't be done by means of Xembly:

• You can't add, remove, or modify XML comments (but you can find them with XPath)

• DTD section can't be modified

• Elements and text content can't be mixed, e.g. this structure is not supported: <test>hello <b>friend</a></test>

Some of these limitations may be removed in the next versions. Please, submit an issue.

How To Contribute

Fork repository, make changes, send us a pull request. We will review your changes and apply them to the master branch shortly, provided they don't violate our quality standards. To avoid frustration, before sending us your pull request, please run full Maven build:

$ mvn clean install -Pqulice

You must fix all static analysis issues, otherwise we won't be able to merge your pull request. The build must be "clean".

Delivery Pipeline

Git master branch is our cutting edge of development. It always contains the latest version of the product, always in -SNAPSHOT suffixed version. Nobody is allowed to commit directly to master — this branch is basically read-only. Everybody contributes changes via pull requrests. We are using rultor, a hosted chatbot, in order to merge pull requests into master. Only our architect is allowed to send pull requests to @rultor for merge, using merge command. Before it happens, a mandatory code review must be performed for a pull request.

After each successful merge of a pull request, our project manager gives deploy command to @rultor. The code from master branch is tested, packaged, and deployed to Sonatype, in version *-SNAPSHOT.

Every once in a while, the architect may decide that it's time to release a new minor/major version of the product. When it happens, he gives release command to @rultor. The code from master branch is tested, versioned, packaged, and deployed to Sonatype and Maven Central. A new Git tag is created. A new GitHub release is created and briefly documented. All this is done automatically by @rultor.

Got questions?

If you have questions or general suggestions, don't hesitate to submit a new Github issue. But keep these Five Principles of Bug Tracking in mind.