Just for fun——PHP框架之简单的路由器（2）

改进

紧接上一篇文章Just for fun——PHP框架之简单的路由器（1）。
代码下载

效率不高原因

对于以下合并的正则

~^(?:
    /user/([^/]+)/(\d+)
    | /user/(\d+)
    | /user/([^/]+)
)$~x

最终匹配的是分组中的某一个，我们需要的子匹配也是那个分组中的，然而从结果看

preg_match($regex, '/user/nikic', $matches);
=> [
    "/user/nikic",   # 完全匹配
    "", "",          # 第一个（空）
    "",              # 第二个（空）
    "nikic",         # 第三个（被使用）
]

这里是最后一个路由被匹配了，但是其他分组的子匹配也被填充了，这是多余的。

解决思路

PCRE正则里?|也是非捕获分组，那么?|和?:有什么区别呢？？
区别在于?|会组号重置，看以下几个例子就懂了

preg_match('~(?:(Sat)ur|(Sun))day~', 'Saturday', $matches)
=> ["Saturday", "Sat", ""]   # 最后一个""其实是不存在的，写在这里是为了阐释概念

preg_match('~(?:(Sat)ur|(Sun))day~', 'Sunday', $matches)
=> ["Sunday", "", "Sun"]

preg_match('~(?|(Sat)ur|(Sun))day~', 'Saturday', $matches)
=> ["Saturday", "Sat"]

preg_match('~(?|(Sat)ur|(Sun))day~', 'Sunday', $matches)
=> ["Sunday", "Sun"]

所有我们可以用?|来代替?:来减少多余的子匹配填充，但是这样一来的话，如何判断哪个分组被匹配了呢？？（因为之前的判断技巧就失效了）
我们可以这样,添加一些多余子匹配

~^(?|
    /user/([^/]+)/(\d+)
  | /user/(\d+)()()
  | /user/([^/]+)()()()
)$~x

实现

dispatcher.php

 1) {
                preg_match_all('~\{([a-zA-Z0-9_]+?)\}~', $route, $matchesVariables);
                return [
                    preg_replace('~{[a-zA-Z0-9_]+?}~', '([a-zA-Z0-9_]+)', $route),
                    $matchesVariables[1],
                ];
            } else {
                return [
                    $route,
                    [],
                ];
            }
        }
        throw new \LogicException('register route failed, pattern is illegal');
    }

    /**
     * 注册路由
     * @param $httpMethod string | string[]
     * @param $route
     * @param $handler
     */
    public function addRoute($httpMethod, $route, $handler) {
        $routeData = $this->parse($route);
        foreach ((array) $httpMethod as $method) {
            if ($this->isStaticRoute($routeData)) {
                $this->addStaticRoute($method, $routeData, $handler);
            } else {
                $this->addVariableRoute($method, $routeData, $handler);
            }
        }
    }


    private function isStaticRoute($routeData) {
        return count($routeData[1]) === 0;
    }

    private function addStaticRoute($httpMethod, $routeData, $handler) {
        $routeStr = $routeData[0];

        if (isset($this->staticRoutes[$httpMethod][$routeStr])) {
            throw new \LogicException(sprintf(
                'Cannot register two routes matching "%s" for method "%s"',
                $routeStr, $httpMethod
            ));
        }

        if (isset($this->methodToRegexToRoutesMap[$httpMethod])) {
            foreach ($this->methodToRegexToRoutesMap[$httpMethod] as $route) {
                if ($route->matches($routeStr)) {
                    throw new \LogicException(sprintf(
                        'Static route "%s" is shadowed by previously defined variable route "%s" for method "%s"',
                        $routeStr, $route->regex, $httpMethod
                    ));
                }
            }
        }

        $this->staticRoutes[$httpMethod][$routeStr] = $handler;
    }


    private function addVariableRoute($httpMethod, $routeData, $handler) {
        list($regex, $variables) = $routeData;

        if (isset($this->methodToRegexToRoutesMap[$httpMethod][$regex])) {
            throw new \LogicException(sprintf(
                'Cannot register two routes matching "%s" for method "%s"',
                $regex, $httpMethod
            ));
        }

        $this->methodToRegexToRoutesMap[$httpMethod][$regex] = new Route(
            $httpMethod, $handler, $regex, $variables
        );
    }


    public function get($route, $handler) {
        $this->addRoute('GET', $route, $handler);
    }

    public function post($route, $handler) {
        $this->addRoute('POST', $route, $handler);
    }

    public function put($route, $handler) {
        $this->addRoute('PUT', $route, $handler);
    }

    public function delete($route, $handler) {
        $this->addRoute('DELETE', $route, $handler);
    }

    public function patch($route, $handler) {
        $this->addRoute('PATCH', $route, $handler);
    }

    public function head($route, $handler) {
        $this->addRoute('HEAD', $route, $handler);
    }

    /**
     * 分发
     * @param $httpMethod
     * @param $uri
     */
    public function dispatch($httpMethod, $uri) {
        $staticRoutes = array_keys($this->staticRoutes[$httpMethod]);
        foreach ($staticRoutes as $staticRoute) {
            if($staticRoute === $uri) {
                return [self::FOUND, $this->staticRoutes[$httpMethod][$staticRoute], []];
            }
        }

        $routeLookup = [];
        $regexes = [];
        foreach ($this->methodToRegexToRoutesMap[$httpMethod] as $regex => $route) {
            $index = count($route->variables);
            if(array_key_exists($index, $routeLookup)) {
                $indexNear = $this->getArrNearEmptyEntry($routeLookup, $index);
                array_push($regexes, $regex . str_repeat('()', $indexNear - $index));
                $routeLookup[$indexNear] = [
                    $this->methodToRegexToRoutesMap[$httpMethod][$regex]->handler,
                    $this->methodToRegexToRoutesMap[$httpMethod][$regex]->variables,
                ];
            } else {
                $routeLookup[$index] = [
                    $this->methodToRegexToRoutesMap[$httpMethod][$regex]->handler,
                    $this->methodToRegexToRoutesMap[$httpMethod][$regex]->variables,
                ];
                array_push($regexes, $regex);
            }
        }
        $regexCombined = '~^(?|' . implode('|', $regexes) . ')$~';
        if(!preg_match($regexCombined, $uri, $matches)) {
            return [self::NOT_FOUND];
        }
        list($handler, $varNames) = $routeLookup[count($matches) - 1];
        $vars = [];
        $i = 0;
        foreach ($varNames as $varName) {
            $vars[$varName] = $matches[++$i];
        }
        return [self::FOUND, $handler, $vars];
    }

    private function getArrNearEmptyEntry(&$arr, $index) {
        while (array_key_exists(++$index, $arr));
        return $index;
    }
}