parse.php

<?php

$pageTemplate = "";
loadHtmlTemplate();

$files = array(
	"rfc4918-manual.html",
	"rfc2616-sec10.html",
	"rfc-additions.html"
);
parseRfcHtmlDocuments($files);


// -------------------------------------------------------------------------

function parseRfcHtmlDocuments($files)
{
	$statuses = array();
	foreach($files as $filename)
	{
		$dom = new DOMDocument;
		$dom->loadHTMLFile($filename);

		// Just look for the <h3> elements, and then their peers
		$items = $dom->getElementsByTagName('h3');
		for ($i = 0; $i < $items->length; $i++)
		{
			$explanation = "";
		
			list($major, $minor, $rev, $status, $description) = sscanf($items->item($i)->nodeValue, "%d.%d.%d %d %[^$]s");

			// If we fail to parse, it's probably a section thing
			if($status == 0)
			{
				// TODO: could actually make a section feature for this
				// 10.1 Informational 1xx
				// 10.2 Successful 2xx
				// 10.3 Redirection 3xx
				// 10.4 Client Error 4xx
				// 10.5 Server Error 5xx
				//print("" . $items->item($i)->nodeValue . "\n");
				continue;
			}

			$p = $items->item($i)->nextSibling;
			while($p != null)
			{
				if(strlen($p->nodeValue) > 1)
				{
					// TODO: consider capturing the document as-is ("<p>...</p>", instead of text value)
					$explanation .= $p->nodeValue;
				}
				$p = $p->nextSibling;

				// HACK: if the next node is the same as the original sibling (h3), bail
				if($p->nodeName == $items->item($i)->nodeName)
					$p = null;
			}
	
			// tidy things up - removed for now, causes formatting issues on 200
			// $explanation = str_replace("   ", "", $explanation);
		
			$result = array("status" => $status, "description" => trim($description), "explanation" => $explanation);
			$statuses[] = $result;
		}
	}

	foreach($statuses as $status)
	{
		// Write the full page, including template/header
		generateHtmlPage($status);

		// Write the cURL version page
		generateCurlPage($status);
	}

	generateLimonadeRoutes($statuses);
}

// -------------------------------------------------------------------------

function generateHtmlPage($status)
{
	global $pageTemplate;

	$statusCode = $status["status"];
	$filename = sprintf("./web/pages/%d.html.php", $status["status"]);
	$fp = fopen($filename, "w");

	$contents = $pageTemplate;

	$externalSponsorFile = "../sponsor.php";
	if(file_exists($externalSponsorFile))
		require_once($externalSponsorFile);
	else
		require_once("sponsor.php");

	$sponsor = sponsorDetails();	
	$contents = str_replace(__PAGE_TITLE_TEXT__, $status["status"] . ": " . $status["description"], $contents); // 200: OK
	$contents = str_replace(__PAGE_HEADING_TEXT__, $status["status"] . ": " . $status["description"], $contents); // 200: OK
	$contents = str_replace(__PAGE_PRE_TEXT__, $status["explanation"], $contents);
	$contents = str_replace(__PAGE_STATUS_CODE__, $status["status"], $contents);
	$contents = str_replace(__PAGE_LEAD_TEXT__, $sponsor[$statusCode], $contents);
	$contents = str_replace(__PAGE_SPONSOR_TITLE__, $sponsor["title"], $contents);
	$contents = str_replace(__PAGE_SPONSOR_URL__, $sponsor["url"], $contents);
	fwrite($fp, $contents);
	
	fclose($fp);
}

// -------------------------------------------------------------------------

function generateCurlPage($status)
{
	$filename = sprintf("./web/pages/%d.curl.php", $status["status"]);
	$fp = fopen($filename, "w");
	fwrite($fp, sprintf("%d: %s\n%s", $status["status"], $status["description"], $status["explanation"]));
	fclose($fp);
}

// -------------------------------------------------------------------------

function generateLimonadeRoutes($statuses)
{
	// Create limonade_routes.php
	
	$filename = sprintf("./web/pages/limonade_routes.php", $status["status"]);
	$fp = fopen($filename, "w");
	fwrite($fp, "<?php\n\n/* THIS FILE IS GENERATED BY PARSE.PHP - DO NOT MODIFY! */\n\n");
	fwrite($fp, "$" . "pages = array(\n");
	foreach($statuses as $status)
	{
		fwrite($fp, sprintf("  %d,\n", $status["status"]));
		/*
			dispatch('/NNN', 'page_NNN');
		    function page_NNN()
		    {
		        return 'Hello world!';
		    }
		*/
		//$dispatch = sprintf("dispatch('/%d', 'page_%d');\n", $status["status"], $status["status"]);
		//fwrite($fp, $dispatch);
	}
	fwrite($fp, ");\n\n");
	fclose($fp);
}

function loadHtmlTemplate()
{
	global $pageTemplate;
	$pageTemplate = file_get_contents("./template_html.txt");
}