Merge pull request #3 from hildjj/docs2

Some doc tweaks

README.md

@@ -1,4 +1,6 @@
-A simple DOM for [expat-wasm](https://github.com/hildjj/expat-wasm).
+# expat-wasm-dom
+
+A Document Object Model for [expat-wasm](https://github.com/hildjj/expat-wasm).
 
 To install:
 
@@ -10,11 +12,18 @@ To use:
 
 ```js
 import DomParser from 'expat-wasm-dom'
-const doc = DomParser.parse('<foo/>')
+const doc = DomParser.parse('<foo><bar/></foo>')
 console.log(doc.root.toString())
+console.log(doc.get('//bar'))
 ```
 
+Note that there is an implementation of XPath3 included.  It is not
+feature-complete yet, but it has a bunch of useful features already.
+
 Full [documentation](https://hildjj.github.io/expat-wasm-dom/) is available.
 
+A Command Line Interface is available in [expat-wasm-dom-cli](https://github.com/hildjj/expat-wasm-dom/tree/main/cli).
+
 --
 [![Tests](https://github.com/hildjj/expat-wasm-dom/actions/workflows/node.js.yml/badge.svg)](https://github.com/hildjj/expat-wasm-dom/actions/workflows/node.js.yml)
+[![codecov](https://codecov.io/gh/hildjj/expat-wasm-dom/branch/main/graph/badge.svg?token=MPGGvwbrqR)](https://codecov.io/gh/hildjj/expat-wasm-dom)

cli/README.md

@@ -0,0 +1,51 @@
+# expat-wasm-dom-cli
+
+Command line utilities for [expat-wasm-dom](https://github.com/hildjj/expat-wasm-dom).
+
+## ewlint
+
+```txt
+Usage: ewlint [options] [file...]
+
+Parse each input file.  If successful, print the contents of the file as
+reconstructed from the DOM.
+
+Arguments:
+  file                Files to parse, use "-" for stdin (default: "-")
+
+Options:
+  -b, --base          Apply xml:base attributes to entities from included files
+  -e,--expand <root>  Expand external entities, but only within the specified
+                      root directory
+  -E                  Expand external entities, but only within the current
+                      working directory
+  --encoding <enc>    Encoding to assume for the input files (choices:
+                      "US-ASCII", "UTF-8", "UTF-16", "ISO-8859-1", default:
+                      Sniff encoding heuristically)
+  -r,--refs           Keep entity references, rather than expanding them
+  -h, --help          display help for command
+```
+
+## ewq
+
+```txt
+Usage: ewq [options] <xpath> [files...]
+
+Parse each input file, then apply the given xpath to the generated documents.
+
+Arguments:
+  xpath
+  files               Files to parse, use "-" for stdin (default: "-")
+
+Options:
+  -b, --base          Apply xml:base attributes to entities from included files
+  -e,--expand <root>  Expand external entities, but only within the specified
+                      root directory
+  -E                  Expand external entities, but only within the current
+                      working directory
+  --encoding <enc>    Encoding to assume for the input files (choices:
+                      "US-ASCII", "UTF-8", "UTF-16", "ISO-8859-1", default:
+                      Sniff encoding heuristically)
+  -r,--refs           Keep entity references, rather than expanding them
+  -h, --help          display help for command
+```

cli/package.json

@@ -7,20 +7,29 @@
     "ewlint": "bin/ewlint.js",
     "ewq": "bin/ewq.js"
   },
-  "scripts": {
-    "test": "echo \"Error: no test specified\" && exit 1"
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/hildjj/expat-wasm-dom.git",
+    "directory": "cli"
   },
-  "keywords": [],
+  "keywords": [
+    "xml",
+    "parse",
+    "dom",
+    "xpath",
+    "wasm",
+    "cli"
+  ],
   "author": "Joe Hildebrand <joe-github@cursive.net>",
   "license": "MIT",
-  "engines": {
-    "node": ">=16"
-  },
   "workspaces": [
     ".."
   ],
   "dependencies": {
     "commander": "10.0.1",
     "expat-wasm-dom": "*"
+  },
+  "engines": {
+    "node": ">=16"
   }
 }

lib/dom.js

@@ -1,3 +1,4 @@
+/* eslint-disable no-use-before-define */
 import {XPath} from './xpath.js'
 import util from 'util'
 
@@ -191,7 +192,6 @@ export class Node {
       cur = cur.parent
     }
 
-    // eslint-disable-next-line no-use-before-define
     if (cur instanceof Document) {
       return cur
     }
@@ -208,7 +208,7 @@ export class Node {
    * @returns {string} The escaped string
    */
   static escape(str, attrib = false) {
-    const re = attrib ? /[<&"]/g : /[<&]/g
+    const re = attrib ? /[<&"']/g : /[<&]/g
     const ret = str.replace(re, c => {
       switch (c) {
         case '<': return '&lt;'
@@ -261,6 +261,17 @@ export class ParentNode extends Node {
    * @returns {Node} The added node
    */
   add(node) {
+    if (
+      (node instanceof Text) &&
+      (this.children.length > 0) &&
+      (this.children[this.children.length - 1] instanceof Text)
+    ) {
+      const prev = this.children[this.children.length - 1]
+      // @ts-ignore
+      prev.txt += node.txt
+      return prev
+    }
+
     node.parent = this
     this.children.push(node)
     return node
@@ -313,7 +324,6 @@ export class ParentNode extends Node {
    */
   *descendantElements() {
     for (const c of this.children) {
-      // eslint-disable-next-line no-use-before-define
       if (c instanceof Element) {
         yield c
         yield *c.descendantElements()
@@ -333,7 +343,6 @@ export class ParentNode extends Node {
    */
   *elements(local, ns) {
     for (const c of this.children) {
-      // eslint-disable-next-line no-use-before-define
       if ((c instanceof Element) &&
           (!ns || (ns === c.name.ns)) &&
           (!local || (local === c.name.local))) {
@@ -384,7 +393,6 @@ export class Document extends ParentNode {
    */
   get root() {
     return /** @type {Element|null} */ (
-      // eslint-disable-next-line no-use-before-define
       this.children.find(c => c instanceof Element)
     )
   }
@@ -443,9 +451,7 @@ export class Element extends ParentNode {
   constructor(name, attribs = [], ns = []) {
     super()
     this.name = (typeof name === 'string') ? {local: name} : name
-    // eslint-disable-next-line no-use-before-define
     this.att = attribs.map(([n, v]) => new Attribute(n, v))
-    // eslint-disable-next-line no-use-before-define
     this.ns = ns.map(([prefix, uri]) => new Namespace(prefix, uri))
   }
 
@@ -492,7 +498,6 @@ export class Element extends ParentNode {
     if (a) {
       a.value = value
     } else {
-      // eslint-disable-next-line no-use-before-define
       a = new Attribute(name, value)
       this.att.push(a)
     }
@@ -550,7 +555,7 @@ export class Element extends ParentNode {
     if (newText == null) {
       return super.text()
     }
-    // eslint-disable-next-line no-use-before-define
+
     const t = new Text(newText)
     t.parent = this
     this.children = [t]
@@ -721,7 +726,6 @@ export class Text extends Node {
    * @returns {string} The node, converted to a string
    */
   toString(options = TOSTRING_OPTS) {
-    // eslint-disable-next-line no-use-before-define
     if (this.parent instanceof CdataSection) {
       // TODO: escape "]]>" ?
       // replace ]]> by ]]]><![CDATA[]>
@@ -1189,7 +1193,6 @@ export class AttlistDecl extends ParentNode {
    * @returns {Node} The added node
    */
   add(node) {
-    // eslint-disable-next-line no-use-before-define
     if (node instanceof AttributeDecl || !this.parent) {
       return super.add(node)
     }

lib/xpath.js

@@ -291,6 +291,9 @@ export class XPath {
    */
   // eslint-disable-next-line class-methods-use-this
   _textTest(num, context) {
+    if (typeof context === 'string') {
+      return [context]
+    }
     if (typeof context.text !== 'function') {
       throw new Error('Cannot get text from this node')
     }
@@ -343,7 +346,7 @@ export class XPath {
     } else if (Array.isArray(expr)) {
       return (this._eval([context], expr).length > 0) ? [context] : []
     }
-    /* istanbul ignore next */
+
     throw new Error(`Unimplemented predicate in "${this.pattern}": "${util.inspect(expr)}"`)
   }
 
@@ -390,6 +393,12 @@ export class XPath {
           return [true]
         }
         break
+      case '!=':
+      case 'ne':
+        if (eLeft !== eRight) {
+          return [true]
+        }
+        break
       default:
         throw new Error(`Unimplemented op: "${op}"`)
     }