357 lines
11 KiB
PHP
357 lines
11 KiB
PHP
<?php
|
||
|
||
namespace app\common;
|
||
|
||
use DOMDocument;
|
||
use DOMElement;
|
||
use DOMXPath;
|
||
use think\Exception;
|
||
use ZipArchive;
|
||
|
||
/**
|
||
* 从投稿 Word 生成「仅正文」docx：在 document.xml 上按块裁切，保留表格/图片/样式；
|
||
* 边界识别仅用可见文本（w:t），不读取域指令（Zotero/EndNote 的 JSON）。
|
||
*/
|
||
class ManuscriptBodyExtractor
|
||
{
|
||
const BODY_SUBDIR = 'public/plagiarism/body_only';
|
||
|
||
const W_NS = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
|
||
|
||
/** @var DOMDocument */
|
||
private $dom;
|
||
|
||
/** @var DOMElement */
|
||
private $bodyNode;
|
||
|
||
/** @var array<int,DOMElement> */
|
||
private $blocks = [];
|
||
|
||
/** @var array<int,string> */
|
||
private $blockTexts = [];
|
||
|
||
/**
|
||
* @return array{path:string, rel_path:string, line_count:int, ref_start:int, body_start:int, warnings:array}
|
||
*/
|
||
public function buildBodyOnlyDocx($sourcePath, $articleId = 0)
|
||
{
|
||
$sourcePath = trim((string) $sourcePath);
|
||
if (!is_file($sourcePath) || !is_readable($sourcePath)) {
|
||
throw new Exception('Manuscript not readable: ' . $sourcePath);
|
||
}
|
||
$ext = strtolower(pathinfo($sourcePath, PATHINFO_EXTENSION));
|
||
if ($ext !== 'docx') {
|
||
throw new Exception('body_only check requires DOCX manuscript, got: ' . $ext);
|
||
}
|
||
|
||
$this->loadDocumentBlocks($sourcePath);
|
||
if (empty($this->blocks)) {
|
||
throw new Exception('No content blocks in manuscript');
|
||
}
|
||
|
||
$refStart = $this->findReferenceStartIndex();
|
||
$bodyStart = $this->findBodyStartIndex();
|
||
$warnings = [];
|
||
|
||
if ($refStart < 0) {
|
||
$warnings[] = 'references_heading_not_found; using document end';
|
||
$refStart = count($this->blocks);
|
||
}
|
||
if ($bodyStart >= $refStart) {
|
||
throw new Exception('Could not locate main body (front matter may include entire document)');
|
||
}
|
||
|
||
$kept = 0;
|
||
for ($i = $bodyStart; $i < $refStart; $i++) {
|
||
if (trim($this->blockTexts[$i]) !== '') {
|
||
$kept++;
|
||
}
|
||
}
|
||
if ($kept < 3) {
|
||
throw new Exception('Body content too short after extraction (' . $kept . ' non-empty blocks)');
|
||
}
|
||
|
||
$relPath = $this->sliceDocxToNewFile($sourcePath, $articleId, $bodyStart, $refStart);
|
||
$rootDir = rtrim(ROOT_PATH ?: dirname(dirname(__DIR__)), '/\\');
|
||
$absPath = $rootDir . DIRECTORY_SEPARATOR . str_replace(['/', '\\'], DIRECTORY_SEPARATOR, $relPath);
|
||
|
||
return [
|
||
'path' => $absPath,
|
||
'rel_path' => $relPath,
|
||
'line_count' => $kept,
|
||
'ref_start' => $refStart,
|
||
'body_start' => $bodyStart,
|
||
'warnings' => $warnings,
|
||
];
|
||
}
|
||
|
||
private function loadDocumentBlocks($sourcePath)
|
||
{
|
||
$zip = new ZipArchive();
|
||
if ($zip->open($sourcePath) !== true) {
|
||
throw new Exception('Cannot open docx: ' . $sourcePath);
|
||
}
|
||
$xml = $zip->getFromName('word/document.xml');
|
||
$zip->close();
|
||
if ($xml === false || $xml === '') {
|
||
throw new Exception('word/document.xml missing in docx');
|
||
}
|
||
|
||
$this->dom = new DOMDocument();
|
||
$this->dom->preserveWhiteSpace = false;
|
||
$this->dom->formatOutput = false;
|
||
if (@$this->dom->loadXML($xml) === false) {
|
||
throw new Exception('Invalid word/document.xml');
|
||
}
|
||
|
||
$xpath = new DOMXPath($this->dom);
|
||
$xpath->registerNamespace('w', self::W_NS);
|
||
$body = $xpath->query('//w:body')->item(0);
|
||
if (!$body instanceof DOMElement) {
|
||
throw new Exception('w:body not found');
|
||
}
|
||
|
||
$this->bodyNode = $body;
|
||
$this->blocks = [];
|
||
$this->blockTexts = [];
|
||
|
||
foreach ($body->childNodes as $child) {
|
||
if ($child->nodeType !== XML_ELEMENT_NODE) {
|
||
continue;
|
||
}
|
||
/** @var DOMElement $child */
|
||
if ($child->localName === 'sectPr') {
|
||
continue;
|
||
}
|
||
$this->blocks[] = $child;
|
||
$this->blockTexts[] = $this->extractVisibleTextFromBlock($child);
|
||
}
|
||
}
|
||
|
||
/**
|
||
* 仅拼接 w:t 可见文本，忽略 w:instrText 等域指令（避免 Zotero JSON 参与裁切判断）。
|
||
*/
|
||
private function extractVisibleTextFromBlock(DOMElement $block)
|
||
{
|
||
$xpath = new DOMXPath($block->ownerDocument);
|
||
$xpath->registerNamespace('w', self::W_NS);
|
||
$nodes = $xpath->query('.//w:t', $block);
|
||
if (!$nodes || $nodes->length === 0) {
|
||
return '';
|
||
}
|
||
$parts = [];
|
||
foreach ($nodes as $node) {
|
||
$parts[] = $node->textContent;
|
||
}
|
||
$text = preg_replace('/\s+/u', ' ', implode('', $parts));
|
||
return trim((string) $text);
|
||
}
|
||
|
||
private function sliceDocxToNewFile($sourcePath, $articleId, $bodyStart, $refStart)
|
||
{
|
||
$rootDir = rtrim(ROOT_PATH ?: dirname(dirname(__DIR__)), '/\\');
|
||
$dir = $rootDir . DIRECTORY_SEPARATOR . self::BODY_SUBDIR;
|
||
if (!is_dir($dir)) {
|
||
@mkdir($dir, 0755, true);
|
||
}
|
||
|
||
$name = sprintf('body_article_%d_%s.docx', intval($articleId), date('Ymd_His'));
|
||
$absPath = $dir . DIRECTORY_SEPARATOR . $name;
|
||
|
||
if (!copy($sourcePath, $absPath)) {
|
||
throw new Exception('Failed to copy source docx');
|
||
}
|
||
|
||
$n = count($this->blocks);
|
||
|
||
$zip = new ZipArchive();
|
||
if ($zip->open($absPath) !== true) {
|
||
throw new Exception('Cannot open output docx');
|
||
}
|
||
|
||
$xml = $zip->getFromName('word/document.xml');
|
||
if ($xml === false) {
|
||
$zip->close();
|
||
throw new Exception('document.xml missing in output docx');
|
||
}
|
||
|
||
$outDom = new DOMDocument();
|
||
$outDom->preserveWhiteSpace = false;
|
||
$outDom->formatOutput = false;
|
||
if (@$outDom->loadXML($xml) === false) {
|
||
$zip->close();
|
||
throw new Exception('Invalid document.xml in output docx');
|
||
}
|
||
|
||
$xpath = new DOMXPath($outDom);
|
||
$xpath->registerNamespace('w', self::W_NS);
|
||
$body = $xpath->query('//w:body')->item(0);
|
||
if (!$body instanceof DOMElement) {
|
||
$zip->close();
|
||
throw new Exception('w:body not found in output docx');
|
||
}
|
||
|
||
$children = [];
|
||
foreach ($body->childNodes as $child) {
|
||
if ($child->nodeType === XML_ELEMENT_NODE) {
|
||
$children[] = $child;
|
||
}
|
||
}
|
||
|
||
$blockIdx = 0;
|
||
foreach ($children as $child) {
|
||
if (!($child instanceof DOMElement)) {
|
||
continue;
|
||
}
|
||
if ($child->localName === 'sectPr') {
|
||
continue;
|
||
}
|
||
if ($blockIdx < $bodyStart || $blockIdx >= $refStart) {
|
||
if ($child->parentNode) {
|
||
$child->parentNode->removeChild($child);
|
||
}
|
||
}
|
||
$blockIdx++;
|
||
}
|
||
|
||
if ($blockIdx !== $n) {
|
||
$zip->close();
|
||
@unlink($absPath);
|
||
throw new Exception('Document block count mismatch during slice');
|
||
}
|
||
|
||
$zip->addFromString('word/document.xml', $outDom->saveXML());
|
||
$zip->close();
|
||
|
||
if (!is_file($absPath) || filesize($absPath) < 200) {
|
||
throw new Exception('Failed to write body-only docx');
|
||
}
|
||
|
||
return self::BODY_SUBDIR . '/' . $name;
|
||
}
|
||
|
||
private function findReferenceStartIndex()
|
||
{
|
||
$stopKeywords = [
|
||
'acknowledgements', 'acknowledgments', 'funding', 'appendix', 'supplementary',
|
||
'conflict of interest', 'competing interests', 'author contributions',
|
||
'致谢', '基金', '附录', '补充材料', '利益冲突', '作者贡献',
|
||
];
|
||
|
||
foreach ($this->blockTexts as $i => $line) {
|
||
$t = trim($line);
|
||
if ($t === '') {
|
||
continue;
|
||
}
|
||
if (preg_match('/^\s*(references|reference|bibliography|参考文献|文献)\b\s*[:：]?\s*/iu', $t)) {
|
||
return $i;
|
||
}
|
||
$lower = strtolower($t);
|
||
foreach ($stopKeywords as $sk) {
|
||
$skLower = strtolower($sk);
|
||
if ($lower === $skLower || $lower === $skLower . ':' || $lower === $skLower . '：') {
|
||
if ($i > count($this->blockTexts) * 0.4) {
|
||
return $i;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
return -1;
|
||
}
|
||
|
||
private function findBodyStartIndex()
|
||
{
|
||
$n = count($this->blockTexts);
|
||
$introIdx = -1;
|
||
$keywordsIdx = -1;
|
||
|
||
for ($i = 0; $i < $n; $i++) {
|
||
$t = trim($this->blockTexts[$i]);
|
||
if ($t === '') {
|
||
continue;
|
||
}
|
||
if ($introIdx < 0 && $this->isIntroductionHeading($t)) {
|
||
$introIdx = $i;
|
||
}
|
||
if ($keywordsIdx < 0 && preg_match('/^\s*keywords?\b\s*[:：]?/iu', $t)) {
|
||
$keywordsIdx = $i;
|
||
}
|
||
}
|
||
|
||
if ($introIdx >= 0) {
|
||
return $introIdx;
|
||
}
|
||
|
||
if ($keywordsIdx >= 0) {
|
||
$afterKw = $this->indexAfterKeywordsBlock($keywordsIdx);
|
||
if ($afterKw < $n) {
|
||
return $afterKw;
|
||
}
|
||
}
|
||
|
||
return $this->indexAfterFrontMatterFallback();
|
||
}
|
||
|
||
private function isIntroductionHeading($t)
|
||
{
|
||
if (preg_match('/^\s*(introduction|background|materials and methods|materials & methods|methods and materials)\b\s*[:：]?/iu', $t)) {
|
||
return true;
|
||
}
|
||
if (preg_match('/^\s*(引言|前言|背景|材料与方法|资料与方法|研究方法)\b\s*[:：]?/iu', $t)) {
|
||
return true;
|
||
}
|
||
if (preg_match('/^\s*1[\.\s、]+(introduction|引言|前言)\b/iu', $t)) {
|
||
return true;
|
||
}
|
||
return false;
|
||
}
|
||
|
||
private function indexAfterKeywordsBlock($kwIdx)
|
||
{
|
||
$n = count($this->blockTexts);
|
||
for ($i = $kwIdx + 1; $i < $n; $i++) {
|
||
$t = trim($this->blockTexts[$i]);
|
||
if ($t === '') {
|
||
continue;
|
||
}
|
||
if ($this->isIntroductionHeading($t)) {
|
||
return $i;
|
||
}
|
||
if (preg_match('/^\s*abstract\b/iu', $t)) {
|
||
continue;
|
||
}
|
||
if (mb_strlen($t) >= 30 && !$this->looksLikeAffiliationLine($t)) {
|
||
return $i;
|
||
}
|
||
}
|
||
return min($kwIdx + 1, $n - 1);
|
||
}
|
||
|
||
private function indexAfterFrontMatterFallback()
|
||
{
|
||
$n = count($this->blockTexts);
|
||
$maxSkip = min(20, (int) floor($n * 0.15));
|
||
for ($i = 0; $i < $maxSkip && $i < $n; $i++) {
|
||
$t = trim($this->blockTexts[$i]);
|
||
if ($t === '') {
|
||
continue;
|
||
}
|
||
if ($this->isIntroductionHeading($t)) {
|
||
return $i;
|
||
}
|
||
}
|
||
return min(8, max(0, $n - 1));
|
||
}
|
||
|
||
private function looksLikeAffiliationLine($t)
|
||
{
|
||
if (preg_match('/@|mailto:|correspond|univ|university|hospital|institute|department|^\d+[\s,，]/iu', $t)) {
|
||
return true;
|
||
}
|
||
if (preg_match('/^\s*abstract\b/iu', $t) || preg_match('/^\s*keywords?\b/iu', $t)) {
|
||
return true;
|
||
}
|
||
return false;
|
||
}
|
||
}
|