如何理解knockout执行过程
knockout执行过程是怎么样的呢？你是不是还很迷惑。这里我就结合我自己学习knockout的总结和自己的开发经验给大家讲解一些，有源码哟。
一、执行流程
一、主要类分析
2.1. 在applyBindings中，创建bindingContext，然后执行applyBindingsToNodeAndDescendantsInternal方法
2.2. 在applyBindinsToNodeAndDescendantsInteranl方法，主要完成当前Node的绑定，以及子Node的绑定
function applyBindingsToNodeAndDescendantsInternal (bindingContext, nodeVerified, bindingContextMayDifferFromDomParentElement) {
var shouldBindDescendants = true;
// Perf optimisation: Apply bindings only if...
// (1) We need to store the binding context on this node (because it may differ from the DOM parent node's binding context)
// Note that we can't store binding contexts on non-elements (e.g., text nodes), as IE doesn't allow expando properties for those
// (2) It might have bindings (e.g., it has a data-bind attribute, or it's a marker for a containerless template)
var isElement = (nodeVerified.nodeType === 1);
if (isElement) // Workaround IE <= 8 HTML parsing weirdness ko.virtualElements.normaliseVirtualElementDomStructure(nodeVerified);
var shouldApplyBindings = (isElement && bindingContextMayDifferFromDomParentElement) // Case (1)
|| ko.bindingProvider['instance']['nodeHasBindings'](nodeVerified); // Case (2)
if (shouldApplyBindings)
shouldBindDescendants = applyBindingsToNodeInternal(nodeVerified, null, bindingContext, bindingContextMayDifferFromDomParentElement)['shouldBindDescendants'];
if (shouldBindDescendants && !bindingDoesNotRecurseIntoElementTypes[ko.utils.tagNameLower(nodeVerified)]) {
// We're recursing automatically into (real or virtual) child nodes without changing binding contexts. So,
// * For children of a *real* element, the binding context is certainly the same as on their DOM .parentNode,
// hence bindingContextsMayDifferFromDomParentElement is false
// * For children of a *virtual* element, we can't be sure. Evaluating .parentNode on those children may
// skip over any number of intermediate virtual elements, any of which might define a custom binding context,
// hence bindingContextsMayDifferFromDomParentElement is true
applyBindingsToDescendantsInternal(bindingContext, nodeVerified, /* bindingContextsMayDifferFromDomParentElement: */ !isElement);
}
}
2.3. 进入applyBindingsToNodeInternal方法，其中会调用bindingProvider的getBindingsAccessors方法（用于分析和获取bindings数据，主要分析data-bind属性）
2.4. 创建dependentObservable对象（依赖监控对象）
var bindings;
if (sourceBindings && typeof sourceBindings !== 'function') {
bindings = sourceBindings;
} else {
var provider = ko.bindingProvider['instance'],
getBindings = provider['getBindingAccessors'] || getBindingsAndMakeAccessors; //自定义BingindHandler
// Get the binding from the provider within a computed observable so that we can update the bindings whenever
// the binding context is updated or if the binding provider accesses observables.
var bindingsUpdater = ko.dependentObservable( //依赖监控对象
function() { //做了read、write处理，实现双向关联(只做了read)，默认会执行一次read的。
bindings = sourceBindings ? sourceBindings(bindingContext, node) : getBindings.call(provider, node, bindingContext);
// Register a dependency on the binding context to support observable view models.
if (bindings && bindingContext._subscribable)
bindingContext._subscribable();
return bindings;
},
null, { disposeWhenNodeIsRemoved: node }
);
if (!bindings || !bindingsUpdater.isActive())
bindingsUpdater = null;
}
2.5. 然后分析bindings中每个binding，并将init、update方法创建为一个dependentObservable对象（其中bindings的执行是有顺序的）。
二、BindingProvider分析
此类主要提供关于data-bind属性的解析，主要提供getBindings、getBindingsAccessors、parseBindingsString（内容使用）方法辅助binding过程。创建function对象：
function createBindingsStringEvaluator(bindingsString, options) {
// Build the source for a function that evaluates "expression"
// For each scope variable, add an extra level of "with" nesting
// Example result: with(sc1) { with(sc0) { return (expression) } }
var rewrittenBindings = ko.expressionRewriting.preProcessBindings(bindingsString, options),
functionBody = "with($context){with($data||{}){return{" + rewrittenBindings + "}}}"; //执行with表达式
return new Function("$context", "$element", functionBody);
}
1、在分析bindings时，会区分NodeType为1、8的类型。如果是8（注释）就会调用virtualElements类的virtualNodeBindingValue方法来分析binding结果。
四、bindings的排序技巧
查看自定义binding是否有after属性，如果存在则进行递归操作：
function topologicalSortBindings(bindings) {
// Depth-first sort
var result = [], // The list of key/handler pairs that we will return
bindingsConsidered = {}, // A temporary record of which bindings are already in 'result'
cyclicDependencyStack = []; // Keeps track of a depth-search so that, if there's a cycle, we know which bindings caused it
ko.utils.objectForEach(bindings, function pushBinding(bindingKey) {
if (!bindingsConsidered[bindingKey]) {
var binding = ko['getBindingHandler'](bindingKey);
if (binding) {
// First add dependencies (if any) of the current binding
if (binding['after']) { //依赖检测，将after的引用先添加到数组中，然后再添加当前项 cyclicDependencyStack.push(bindingKey);
ko.utils.arrayForEach(binding['after'], function(bindingDependencyKey) {
if (bindings[bindingDependencyKey]) {
if (ko.utils.arrayIndexOf(cyclicDependencyStack, bindingDependencyKey) !== -1) {
throw Error("Cannot combine the following bindings, because they have a cyclic dependency: " + cyclicDependencyStack.join(", "));
} else {
pushBinding(bindingDependencyKey);
}
}
});
cyclicDependencyStack.length--;
}
// Next add the current binding result.push({ key: bindingKey, handler: binding });
}
bindingsConsidered[bindingKey] = true;
}
});
五、注意
1.所有的dependentObservable对象，在创建的过程中都会默认执行一次readFunction方法。
原文来自：博客园/小龙女先生