316 lines
14 KiB
PHP
316 lines
14 KiB
PHP
<?php
|
|
/**
|
|
* Render scan (agent phase P2): learn which fields and lists a page really offers
|
|
* by rendering it as an editor and reading the data-* attributes of the HTML.
|
|
*
|
|
* Rendering never writes: scan mode (see content-loader.php) turns the editable
|
|
* helpers into editor markup without auto-writing missing fields, without a session
|
|
* and without analytics. Standard pages are rendered with renderAllSections() and can
|
|
* be scanned repeatedly. Custom templates are real PHP pages (header, footer, their own
|
|
* code), so they are rendered at most once per PHP process, in a sandbox:
|
|
* superglobals are swapped, output and warnings are captured, headers and the response
|
|
* code are restored, and a template that calls exit() is recorded instead of crashing
|
|
* the caller unnoticed.
|
|
*/
|
|
|
|
require_once dirname(__DIR__, 2) . '/page-path.php';
|
|
|
|
if (!class_exists('RaijuAgentScanDeferred', false)) {
|
|
/** A second template render in the same process: continue in the next step request. */
|
|
class RaijuAgentScanDeferred extends RuntimeException {
|
|
}
|
|
}
|
|
|
|
function raijuAgentSiteRoot(): string {
|
|
return dirname(__DIR__, 3);
|
|
}
|
|
|
|
// ---------------------------------------------------------------------------
|
|
// HTML -> structure
|
|
// ---------------------------------------------------------------------------
|
|
|
|
/**
|
|
* Read the editable structure out of rendered editor markup.
|
|
*
|
|
* @return array{fields:array<string,array>,lists:array<string,array>,groups:array<string,array>,binds:array<string,string>,sections:array}
|
|
* fields: path => {path,type,page,format?,altPath?,hidden,value,shared}
|
|
* lists: key => {key,page,defaults,count,indexes,hidden,item}
|
|
*/
|
|
function raijuAgentScanHtml(string $html, string $page): array {
|
|
$result = ['fields' => [], 'lists' => [], 'groups' => [], 'binds' => [], 'sections' => []];
|
|
if (trim($html) === '') {
|
|
return $result;
|
|
}
|
|
$previous = libxml_use_internal_errors(true);
|
|
$document = new DOMDocument();
|
|
$document->loadHTML('<?xml encoding="utf-8" ?>' . $html, LIBXML_NONET | LIBXML_NOERROR | LIBXML_NOWARNING);
|
|
libxml_clear_errors();
|
|
libxml_use_internal_errors($previous);
|
|
$xpath = new DOMXPath($document);
|
|
|
|
$innerHtml = static function (DOMElement $element) use ($document): string {
|
|
$out = '';
|
|
foreach ($element->childNodes as $child) {
|
|
$out .= $document->saveHTML($child);
|
|
}
|
|
return $out;
|
|
};
|
|
|
|
foreach ($xpath->query('//*[@data-field and @data-page]') as $element) {
|
|
/** @var DOMElement $element */
|
|
$path = trim($element->getAttribute('data-field'));
|
|
$owner = trim($element->getAttribute('data-page'));
|
|
if ($path === '' || $owner === '') continue;
|
|
$classes = ' ' . $element->getAttribute('class') . ' ';
|
|
$field = ['path' => $path, 'page' => $owner, 'hidden' => $element->getAttribute('data-hidden') === 'true', 'shared' => $owner !== $page];
|
|
|
|
if ($element->tagName === 'a' && $element->hasAttribute('data-editable-link')) {
|
|
$field['type'] = 'link';
|
|
$field['value'] = ['text' => trim($element->textContent), 'href' => $element->getAttribute('href')];
|
|
} elseif ($element->tagName === 'img' && $element->hasAttribute('data-editable-image')) {
|
|
$field['type'] = 'image';
|
|
$field['value'] = ['src' => $element->getAttribute('src'), 'alt' => $element->getAttribute('alt')];
|
|
if ($element->getAttribute('data-image-format') === 'split' && $element->getAttribute('data-alt-field') !== '') {
|
|
$field['format'] = 'split';
|
|
$field['altPath'] = trim($element->getAttribute('data-alt-field'));
|
|
}
|
|
} elseif ($element->hasAttribute('data-editable-icon')) {
|
|
$field['type'] = 'icon';
|
|
$field['value'] = $element->getAttribute('data-icon');
|
|
} elseif (str_contains($classes, ' editable-field-html ')) {
|
|
$field['type'] = 'html';
|
|
$field['value'] = trim($innerHtml($element));
|
|
} elseif (str_contains($classes, ' editable-field ')) {
|
|
$field['type'] = 'text';
|
|
$field['value'] = trim($element->textContent);
|
|
} else {
|
|
continue;
|
|
}
|
|
// The first occurrence wins; a path rendered twice is still one field.
|
|
$result['fields'][$owner . '|' . $path] ??= $field;
|
|
}
|
|
// Key by path for the page itself, keep foreign pages separate.
|
|
$keyed = [];
|
|
foreach ($result['fields'] as $compound => $field) {
|
|
$keyed[$field['shared'] ? $compound : $field['path']] = $field;
|
|
}
|
|
$result['fields'] = $keyed;
|
|
|
|
foreach ($xpath->query('//*[@data-editable-list]') as $element) {
|
|
$key = trim($element->getAttribute('data-list-key'));
|
|
if ($key === '') continue;
|
|
$defaults = json_decode($element->getAttribute('data-list-defaults'), true);
|
|
$result['lists'][$key] ??= [
|
|
'key' => $key,
|
|
'page' => trim($element->getAttribute('data-list-page')),
|
|
'defaults' => is_array($defaults) ? $defaults : [],
|
|
'count' => 0,
|
|
'indexes' => [],
|
|
'hidden' => [],
|
|
];
|
|
}
|
|
foreach ($xpath->query('//*[@data-list-index and @data-list-key]') as $element) {
|
|
$key = trim($element->getAttribute('data-list-key'));
|
|
$index = (int)$element->getAttribute('data-list-index');
|
|
if (!isset($result['lists'][$key])) {
|
|
$result['lists'][$key] = ['key' => $key, 'page' => trim($element->getAttribute('data-list-page')), 'defaults' => [], 'count' => 0, 'indexes' => [], 'hidden' => []];
|
|
}
|
|
if (!in_array($index, $result['lists'][$key]['indexes'], true)) {
|
|
$result['lists'][$key]['indexes'][] = $index;
|
|
if ($element->getAttribute('data-hidden') === 'true') $result['lists'][$key]['hidden'][] = $index;
|
|
}
|
|
}
|
|
foreach ($result['lists'] as $key => &$list) {
|
|
sort($list['indexes']);
|
|
$list['count'] = count($list['indexes']);
|
|
}
|
|
unset($list);
|
|
|
|
foreach ($xpath->query('//*[@data-editable-group and @data-group]') as $element) {
|
|
$group = trim($element->getAttribute('data-group'));
|
|
$schema = json_decode($element->getAttribute('data-group-schema'), true);
|
|
if ($group !== '' && is_array($schema)) {
|
|
$result['groups'][$group] ??= $schema;
|
|
}
|
|
}
|
|
foreach ($xpath->query('//*[@data-rj-bind-field]') as $element) {
|
|
$path = trim($element->getAttribute('data-rj-bind-field'));
|
|
if ($path !== '') $result['binds'][$path] = $element->getAttribute('data-rj-bind') ?: 'text';
|
|
}
|
|
foreach ($xpath->query('//*[@data-section-index]') as $element) {
|
|
$result['sections'][(int)$element->getAttribute('data-section-index')] = [
|
|
'id' => $element->getAttribute('data-section-id'),
|
|
'type' => $element->getAttribute('data-section-type'),
|
|
'hidden' => $element->getAttribute('data-hidden') === 'true',
|
|
];
|
|
}
|
|
ksort($result['sections']);
|
|
return $result;
|
|
}
|
|
|
|
// ---------------------------------------------------------------------------
|
|
// Rendering
|
|
// ---------------------------------------------------------------------------
|
|
|
|
/**
|
|
* Render a standard (sections) page as an editor. Repeatable.
|
|
*
|
|
* @return array{html:string,warnings:array,deps:array,exception:?string}
|
|
*/
|
|
function raijuAgentRenderStandard(string $contentKey): array {
|
|
require_once dirname(__DIR__, 2) . '/content-loader.php';
|
|
$warnings = [];
|
|
raijuContentScanBegin();
|
|
$handler = raijuAgentCaptureWarnings($warnings);
|
|
try {
|
|
$html = renderAllSections($contentKey);
|
|
$exception = null;
|
|
} catch (Throwable $error) {
|
|
$html = '';
|
|
$exception = get_class($error) . ': ' . $error->getMessage();
|
|
} finally {
|
|
restore_error_handler();
|
|
raijuContentScanEnd();
|
|
}
|
|
return ['html' => $html, 'warnings' => $warnings, 'deps' => raijuAgentStandardDeps(), 'exception' => $exception];
|
|
}
|
|
|
|
function raijuAgentStandardDeps(): array {
|
|
$root = raijuAgentSiteRoot() . '/includes';
|
|
return array_merge(
|
|
[$root . '/content-loader.php', $root . '/block-types.php', $root . '/html-sanitizer.php', $root . '/ai-labels.php'],
|
|
glob($root . '/block-renderers/*.php') ?: []
|
|
);
|
|
}
|
|
|
|
/** Collect PHP warnings/notices instead of printing them. Returns the previous handler. */
|
|
function raijuAgentCaptureWarnings(array &$warnings) {
|
|
return set_error_handler(static function (int $level, string $message, string $file, int $line) use (&$warnings): bool {
|
|
if (!(error_reporting() & $level)) return true;
|
|
if (count($warnings) < 20) {
|
|
$warnings[] = raijuAgentRelativePath($file) . ':' . $line . ' ' . $message;
|
|
}
|
|
return true;
|
|
});
|
|
}
|
|
|
|
function raijuAgentRelativePath(string $file): string {
|
|
$root = rtrim(raijuAgentSiteRoot(), '/') . '/';
|
|
return str_starts_with($file, $root) ? substr($file, strlen($root)) : basename($file);
|
|
}
|
|
|
|
/** Path of the PHP template behind a content key, or '' when the page has none. */
|
|
function raijuAgentTemplateFor(string $contentKey): string {
|
|
$parsed = raijuPageParseContentKey($contentKey);
|
|
if ($parsed === null) return '';
|
|
$root = raijuAgentSiteRoot();
|
|
$candidates = [raijuPageTemplatePath($root, $parsed['lang'], $parsed['path'])];
|
|
if ($parsed['path'] === 'home') {
|
|
array_unshift($candidates, $root . '/' . $parsed['lang'] . '/index.php');
|
|
}
|
|
foreach ($candidates as $candidate) {
|
|
if ($candidate !== '' && is_file($candidate)) return $candidate;
|
|
}
|
|
return '';
|
|
}
|
|
|
|
/**
|
|
* Render a custom template page as an editor, at most once per PHP process.
|
|
*
|
|
* @param array $options onExit: callable(string $reason) run from the shutdown handler when
|
|
* the template terminates the request (exit/die/fatal); the caller
|
|
* can then answer the client. The buffer is discarded first.
|
|
* @return array{html:string,warnings:array,deps:array,exception:?string}
|
|
* @throws RaijuAgentScanDeferred when a template was already rendered by this process
|
|
*/
|
|
function raijuAgentRenderTemplateOnce(string $templatePath, string $contentKey, array $options = []): array {
|
|
static $rendered = false;
|
|
if ($rendered) {
|
|
throw new RaijuAgentScanDeferred('Only one custom template can be rendered per request.');
|
|
}
|
|
$real = realpath($templatePath);
|
|
$root = realpath(raijuAgentSiteRoot());
|
|
if ($real === false || $root === false || !str_starts_with($real, $root . DIRECTORY_SEPARATOR) || pathinfo($real, PATHINFO_EXTENSION) !== 'php') {
|
|
throw new InvalidArgumentException('Template path is outside the site.');
|
|
}
|
|
$rendered = true;
|
|
require_once dirname(__DIR__, 2) . '/content-loader.php';
|
|
|
|
$parsed = raijuPageParseContentKey($contentKey) ?? ['lang' => 'en', 'path' => 'home'];
|
|
$savedGlobals = ['_GET' => $_GET, '_POST' => $_POST, '_FILES' => $_FILES, '_COOKIE' => $_COOKIE ?? []];
|
|
$savedServer = [];
|
|
foreach (['REQUEST_METHOD', 'REQUEST_URI', 'SCRIPT_NAME', 'PHP_SELF', 'QUERY_STRING', 'CONTENT_TYPE'] as $key) {
|
|
$savedServer[$key] = $_SERVER[$key] ?? null;
|
|
}
|
|
$publicPath = '/' . $parsed['lang'] . '/' . ($parsed['path'] === 'home' ? '' : $parsed['path']);
|
|
$_GET = [];
|
|
$_POST = [];
|
|
$_FILES = [];
|
|
$_SERVER['REQUEST_METHOD'] = 'GET';
|
|
$_SERVER['REQUEST_URI'] = $publicPath;
|
|
$_SERVER['SCRIPT_NAME'] = $_SERVER['PHP_SELF'] = $publicPath;
|
|
$_SERVER['QUERY_STRING'] = '';
|
|
unset($_SERVER['CONTENT_TYPE']);
|
|
|
|
$before = get_included_files();
|
|
$headersBefore = array_map(static fn($line) => strtolower(explode(':', $line, 2)[0]), headers_list());
|
|
$statusBefore = http_response_code();
|
|
$cwd = getcwd();
|
|
$bufferLevel = ob_get_level();
|
|
$warnings = [];
|
|
$state = ['finished' => false];
|
|
|
|
// exit()/die()/fatal errors inside the template skip everything below; this handler
|
|
// is the only code that still runs, so it records what happened.
|
|
register_shutdown_function(static function () use (&$state, $bufferLevel, $real, $options): void {
|
|
if ($state['finished']) return;
|
|
$fatal = error_get_last();
|
|
$reason = $fatal && in_array($fatal['type'], [E_ERROR, E_PARSE, E_CORE_ERROR, E_COMPILE_ERROR, E_RECOVERABLE_ERROR], true)
|
|
? 'fatal: ' . $fatal['message']
|
|
: 'the template ended the request (exit/die)';
|
|
while (ob_get_level() > $bufferLevel) @ob_end_clean();
|
|
foreach (array_diff(array_map(static fn($line) => strtolower(explode(':', $line, 2)[0]), headers_list()), $state['headersBefore'] ?? []) as $name) {
|
|
header_remove($name);
|
|
}
|
|
if (isset($state['status']) && $state['status']) http_response_code($state['status']);
|
|
raijuContentScanEnd();
|
|
if (function_exists('raijuAgentRecordScanFailure')) raijuAgentRecordScanFailure($real, $reason);
|
|
if (is_callable($options['onExit'] ?? null)) $options['onExit']($reason);
|
|
});
|
|
$state['headersBefore'] = $headersBefore;
|
|
$state['status'] = $statusBefore;
|
|
|
|
$html = '';
|
|
$exception = null;
|
|
ob_start();
|
|
raijuContentScanBegin();
|
|
raijuAgentCaptureWarnings($warnings);
|
|
try {
|
|
chdir(dirname($real)); // templates include '../includes/header.php' relative to their folder
|
|
(static function (string $__template): void {
|
|
include $__template;
|
|
})($real);
|
|
while (ob_get_level() > $bufferLevel + 1) ob_end_flush();
|
|
$html = (string)ob_get_clean();
|
|
} catch (Throwable $error) {
|
|
while (ob_get_level() > $bufferLevel) @ob_end_clean();
|
|
$exception = get_class($error) . ': ' . $error->getMessage();
|
|
} finally {
|
|
restore_error_handler();
|
|
raijuContentScanEnd();
|
|
if ($cwd !== false) chdir($cwd);
|
|
foreach (array_diff(array_map(static fn($line) => strtolower(explode(':', $line, 2)[0]), headers_list()), $headersBefore) as $name) {
|
|
header_remove($name);
|
|
}
|
|
if ($statusBefore) http_response_code($statusBefore);
|
|
foreach ($savedGlobals as $name => $value) $GLOBALS[$name] = $value;
|
|
foreach ($savedServer as $key => $value) {
|
|
if ($value === null) unset($_SERVER[$key]); else $_SERVER[$key] = $value;
|
|
}
|
|
$state['finished'] = true;
|
|
}
|
|
|
|
$deps = array_values(array_unique(array_merge([$real], array_diff(get_included_files(), $before), raijuAgentStandardDeps(), glob(raijuAgentSiteRoot() . '/includes/*.php') ?: [])));
|
|
return ['html' => $html, 'warnings' => $warnings, 'deps' => $deps, 'exception' => $exception];
|
|
}
|