From aa3a42f94e0b3148ef391e05129d11a7fa2525b6 Mon Sep 17 00:00:00 2001
From: Alexander Rose <alex.rose@rcsb.org>
Date: Fri, 20 Sep 2019 17:13:13 -0700
Subject: [PATCH] added simple xml parser
---
src/mol-util/data-source.ts | 17 +++--
src/mol-util/xml-parser.ts | 134 ++++++++++++++++++++++++++++++++++++
2 files changed, 144 insertions(+), 7 deletions(-)
create mode 100644 src/mol-util/xml-parser.ts
diff --git a/src/mol-util/data-source.ts b/src/mol-util/data-source.ts
index 99c54d8d6..42de77608 100644
--- a/src/mol-util/data-source.ts
+++ b/src/mol-util/data-source.ts
@@ -9,6 +9,7 @@
import { Task, RuntimeContext } from '../mol-task';
import { utf8Read } from '../mol-io/common/utf8';
+import { parseXml } from './xml-parser';
// polyfill XMLHttpRequest in node.js
const XHR = typeof document === 'undefined' ? require('xhr2') as {
prototype: XMLHttpRequest;
@@ -25,7 +26,7 @@ const XHR = typeof document === 'undefined' ? require('xhr2') as {
// Gzip
// }
-export interface AjaxGetParams<T extends 'string' | 'binary' | 'json' = 'string'> {
+export interface AjaxGetParams<T extends 'string' | 'binary' | 'json' | 'xml' = 'string'> {
url: string,
type?: T,
title?: string,
@@ -49,10 +50,10 @@ export function readFromFile(file: File, type: 'string' | 'binary') {
export function ajaxGet(url: string): Task<string>
export function ajaxGet(params: AjaxGetParams<'string'>): Task<string>
export function ajaxGet(params: AjaxGetParams<'binary'>): Task<Uint8Array>
-export function ajaxGet<T = any>(params: AjaxGetParams<'json'>): Task<T>
+export function ajaxGet<T = any>(params: AjaxGetParams<'json' | 'xml'>): Task<T>
export function ajaxGet(params: AjaxGetParams<'string' | 'binary'>): Task<string | Uint8Array>
-export function ajaxGet(params: AjaxGetParams<'string' | 'binary' | 'json'>): Task<string | Uint8Array | object>
-export function ajaxGet(params: AjaxGetParams<'string' | 'binary' | 'json'> | string) {
+export function ajaxGet(params: AjaxGetParams<'string' | 'binary' | 'json' | 'xml'>): Task<string | Uint8Array | object>
+export function ajaxGet(params: AjaxGetParams<'string' | 'binary' | 'json' | 'xml'> | string) {
if (typeof params === 'string') return ajaxGetInternal(params, params, 'string', false);
return ajaxGetInternal(params.title, params.url, params.type || 'string', false /* params.compression === DataCompressionMethod.Gzip */, params.body);
}
@@ -174,7 +175,7 @@ async function processAjax(ctx: RuntimeContext, asUint8Array: boolean, decompres
}
}
-function ajaxGetInternal(title: string | undefined, url: string, type: 'json' | 'string' | 'binary', decompressGzip: boolean, body?: string): Task<string | Uint8Array> {
+function ajaxGetInternal(title: string | undefined, url: string, type: 'json' | 'xml' | 'string' | 'binary', decompressGzip: boolean, body?: string): Task<string | Uint8Array> {
let xhttp: XMLHttpRequest | undefined = void 0;
return Task.create(title ? title : 'Download', async ctx => {
try {
@@ -195,8 +196,10 @@ function ajaxGetInternal(title: string | undefined, url: string, type: 'json' |
if (type === 'json') {
ctx.update({ message: 'Parsing JSON...', canAbort: false });
- const data = JSON.parse(result);
- return data;
+ return JSON.parse(result);
+ } else if (type === 'xml') {
+ ctx.update({ message: 'Parsing XML...', canAbort: false });
+ return parseXml(result);
}
return result;
diff --git a/src/mol-util/xml-parser.ts b/src/mol-util/xml-parser.ts
new file mode 100644
index 000000000..841296298
--- /dev/null
+++ b/src/mol-util/xml-parser.ts
@@ -0,0 +1,134 @@
+/**
+ * Copyright (c) 2019 mol* contributors, licensed under MIT, See LICENSE file for more info.
+ *
+ * @author Alexander Rose <alexander.rose@weirdbyte.de>
+ */
+
+export type XMLNodeAttributes = { [k: string]: any }
+export interface XMLNode {
+ name?: string
+ content?: string
+ attributes: XMLNodeAttributes
+ children?: XMLNode[]
+}
+export interface XMLDocument {
+ declaration?: XMLNode,
+ root?: XMLNode
+}
+
+export function getXMLNodeByName(name: string, children: XMLNode[]) {
+ for (let i = 0, il = children.length; i < il; ++i) {
+ if (children[i].name === name) return children[i]
+ }
+}
+
+const reStrip = /^['"]|['"]$/g
+const reTag = /^<([\w-:.]+)\s*/
+const reContent = /^([^<]*)/
+const reAttr = /([\w:-]+)\s*=\s*("[^"]*"|'[^']*'|\w+)\s*/
+
+function strip (val: string) {
+ return val.replace(reStrip, '')
+}
+
+/**
+ * Simple XML parser
+ * adapted from https://github.com/segmentio/xml-parser (MIT license)
+ */
+export function parseXml (xml: string): XMLDocument {
+ // trim and strip comments
+ xml = xml.trim().replace(/<!--[\s\S]*?-->/g, '')
+
+ return document()
+
+ function document () {
+ return {
+ declaration: declaration(),
+ root: tag()
+ }
+ }
+
+ function declaration () {
+ const m = match(/^<\?xml\s*/)
+ if (!m) return
+
+ // tag
+ const node: XMLNode = {
+ attributes: {}
+ }
+
+ // attributes
+ while (!(eos() || is('?>'))) {
+ const attr = attribute()
+ if (!attr) return node
+ node.attributes[attr.name] = attr.value
+ }
+ match(/\?>\s*/)
+ return node
+ }
+
+ function tag () {
+ const m = match(reTag)
+ if (!m) return
+
+ // name
+ const node: XMLNode = {
+ name: m[1],
+ attributes: {},
+ children: []
+ }
+
+ // attributes
+ while (!(eos() || is('>') || is('?>') || is('/>'))) {
+ const attr = attribute()
+ if (!attr) return node
+ node.attributes[attr.name] = attr.value
+ }
+
+ // self closing tag
+ if (match(/^\s*\/>\s*/)) {
+ return node
+ }
+ match(/\??>\s*/)
+
+ // content
+ node.content = content()
+
+ // children
+ let child
+ while ((child = tag())) {
+ node.children!.push(child)
+ }
+
+ // closing
+ match(/^<\/[\w-:.]+>\s*/)
+ return node
+ }
+
+ function content () {
+ const m = match(reContent)
+ if (m) return m[1]
+ return ''
+ }
+
+ function attribute () {
+ const m = match(reAttr)
+ if (!m) return
+ return { name: m[1], value: strip(m[2]) }
+ }
+
+ function match (re: RegExp) {
+ const m = xml.match(re)
+ if (!m) return
+ xml = xml.slice(m[0].length)
+ return m
+ }
+
+ function eos () {
+ return xml.length === 0
+ }
+
+ function is (prefix: string) {
+ return xml.indexOf(prefix) === 0
+ }
+}
\ No newline at end of file
--
GitLab