zm/tougao
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Changes

Browse Source
This commit is contained in:
wyn
2026-05-21 11:30:46 +08:00
parent 4aab7f5b7e
commit 7e5a087a4e
3 changed files with 494 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files

184
application/common/ReferenceCheckService.php
View File

@@ -75,7 +75,10 @@ class ReferenceCheckService
// ->find();
$citations = $this->extractReferences((string)$main['content']);
// return $citations;
if (empty($citations)) {
$this->setAmRefCheckStatus($amId, self::AM_STATUS_NONE);
return;
}
$prod = Db::name('production_article')
->where('article_id', $main['article_id'])
->where('state', 0)
@@ -128,15 +131,10 @@ class ReferenceCheckService
$this->setAmRefCheckStatus($amId, self::AM_STATUS_RUNNING);
}
/**
* 按 article_id 扫描 t_article_main，为每个 blue 引用 × 文献号入队
*/
public function enqueueByArticle($articleId, $clearPrevious = true)
{
public function enqueueByArticle($articleId){
if ($articleId <= 0) {
throw new \InvalidArgumentException('article_id is required');
}
$prod = Db::name('production_article')
->where('article_id', $articleId)
->where('state', 0)
@@ -144,25 +142,18 @@ class ReferenceCheckService
if (empty($prod)) {
throw new \RuntimeException('production_article not found for article_id=' . $articleId);
}
$pArticleId = intval($prod['p_article_id']);
$referMap = $this->loadReferMapByPArticleId($pArticleId);
$mains = Db::name('article_main')
->field('am_id,content')
->field('am_id,content,article_id')
->where('article_id', $articleId)
->whereIn('state', [0, 2])
->order('sort asc')
->select();
if (empty($mains)) {
throw new \RuntimeException('article_main is empty');
}
if ($clearPrevious) {
$this->clearArticleChecks($articleId);
}
$queued = 0;
$skipped = 0;
$checkIds = [];
@@ -189,20 +180,16 @@ class ReferenceCheckService
$now = date('Y-m-d H:i:s');
// [70-73] 展开为 reference_no=70,71,72,73 共 4 条记录
$checkId = Db::name('article_reference_check_result')->insertGetId([
'article_id' => $articleId,
'am_id' => intval($main['am_id']),
'article_id' => $main['article_id'],
'p_article_id' => $pArticleId,
'p_refer_id' => intval($refer['p_refer_id']),
'refer_index' => $referIndex,
'am_id' => intval($main['am_id']),
'reference_no' => $refNo,
'reference_raw' => $cite['reference_raw'],
'cite_tag_start' => intval($cite['reference_start']),
'cite_tag_end' => intval($cite['reference_end']),
'text_start' => intval($cite['text_start']),
'text_end' => intval($cite['text_end']),
'content_a' => $cite['original_text'],
'content_b' => $referText,
'status' => 0,
'refer_index' => $refNo,
'origin_text' => $cite['original_text'],
'refer_text' => $referText,
'p_refer_id' => $referMap[$referIndex]['p_refer_id'],
'text_start' => $cite['text_start'],
'text_end' => $cite['text_end'],
'created_at' => $now,
'updated_at' => $now,
]);
@@ -658,12 +645,21 @@ class ReferenceCheckService
$referenceNumbers = $this->expandReferenceNumbers($rawRef);
$sentenceStart = $this->findSentenceStart($content, $tagStart);
$sentenceEnd = $this->findSentenceEnd($content, $tagEnd);
$originalText = mb_substr($content, $sentenceStart, $sentenceEnd - $sentenceStart);
$originalText = preg_replace('/<blue>\[[\d,\-\s]+\]<\/blue>/', '', $originalText);
$originalText = trim(strip_tags($originalText));
$sentenceEnd = $this->findSentenceEnd($content, $tagEnd, $tagEnd);
$originalText = $this->buildCitationContextText($content, $sentenceStart, $sentenceEnd);
if ($originalText === '' || empty($referenceNumbers)) {
if (!$this->isMeaningfulCitationContext($originalText)) {
list($sentenceStart, $sentenceEnd) = $this->widenCitationContextBounds(
$content,
$tagStart,
$tagEnd,
$sentenceStart,
$sentenceEnd
);
$originalText = $this->buildCitationContextText($content, $sentenceStart, $sentenceEnd);
}
if (!$this->isMeaningfulCitationContext($originalText) || empty($referenceNumbers)) {
continue;
}
@@ -707,29 +703,137 @@ class ReferenceCheckService
return array_values(array_unique($numbers));
}
private function buildCitationContextText($content, $start, $end)
{
$text = mb_substr($content, $start, max(0, $end - $start));
$text = preg_replace('/<blue>\[[\d,\-\s]+\]<\/blue>/', '', $text);
$text = trim(strip_tags($text));
$text = preg_replace('/\s+/u', ' ', $text);
return $text;
}
/**
* 过滤仅标点、过短或无字母/汉字的上下文（如去掉标签后只剩 "."）
*/
private function isMeaningfulCitationContext($text)
{
$text = trim($text);
if ($text === '') {
return false;
}
if ($this->isOnlyPunctuationOrSpace($text)) {
return false;
}
if (!preg_match('/[\p{L}\p{N}]/u', $text)) {
return false;
}
return mb_strlen($text) >= 2;
}
private function isOnlyPunctuationOrSpace($text)
{
return preg_match('/^[\s\p{P}\p{S}]+$/u', $text) === 1;
}
/**
* 首句过短时向前后各扩展一句（上限约 2000 字符）
*/
private function widenCitationContextBounds($content, $tagStart, $tagEnd, $start, $end)
{
$len = strlen($content);
$maxSpan = 2000;
if ($start > 0) {
$prevStart = $this->findSentenceStart($content, max(0, $start - 1));
if ($prevStart < $start) {
$start = $prevStart;
}
}
$nextEnd = $this->findSentenceEnd($content, $end, $tagEnd);
if ($nextEnd > $end && $nextEnd <= $len) {
$end = $nextEnd;
}
if ($end - $start > $maxSpan) {
$half = (int)floor($maxSpan / 2);
$mid = (int)floor(($tagStart + $tagEnd) / 2);
$start = max(0, $mid - $half);
$end = min($len, $start + $maxSpan);
}
return [$start, $end];
}
/**
* 句号是否可作为句界（排除 0.95、3.14 等小数点）
*/
private function isSentenceDelimiterAt($content, $pos, $delimiter)
{
$len = strlen($content);
if ($delimiter !== '.' || $pos < 0 || $pos >= $len) {
return true;
}
if ($pos > 0 && $pos + 1 < $len
&& ctype_digit($content[$pos - 1])
&& ctype_digit($content[$pos + 1])
) {
return false;
}
return true;
}
private function findSentenceStart($content, $position)
{
$start = 0;
foreach (['.', '。', '!', '?', "\n"] as $delimiter) {
$pos = strrpos(substr($content, 0, $position), $delimiter);
if ($pos !== false) {
if ($pos !== false && $this->isSentenceDelimiterAt($content, $pos, $delimiter)) {
$start = max($start, $pos + 1);
}
}
return $start;
}
private function findSentenceEnd($content, $position)
/**
* @param int $searchFrom 从该字节位置起查找句末
* @param int $tagEnd 引用标签结束位置；用于跳过 </blue> 后紧跟的孤立句号
*/
private function findSentenceEnd($content, $searchFrom, $tagEnd = 0)
{
$length = strlen($content);
$endPositions = [];
foreach (['.', '。', '!', '?', "\n"] as $delimiter) {
$pos = strpos($content, $delimiter, $position);
if ($pos !== false) {
$endPositions[] = $pos + 1;
$minPos = max(0, $searchFrom);
while ($minPos < $length) {
$endPositions = [];
foreach (['.', '。', '!', '?', "\n"] as $delimiter) {
$pos = strpos($content, $delimiter, $minPos);
if ($pos !== false && $this->isSentenceDelimiterAt($content, $pos, $delimiter)) {
$endPositions[] = $pos + 1;
}
}
if (empty($endPositions)) {
return $length;
}
$end = min($endPositions);
if ($tagEnd <= 0 || $end <= $tagEnd) {
return $end;
}
$gap = substr($content, $tagEnd, $end - $tagEnd);
$gapText = trim(strip_tags(preg_replace('/<blue>\[[\d,\-\s]+\]<\/blue>/', '', $gap)));
if ($gapText !== '' && !$this->isOnlyPunctuationOrSpace($gapText)) {
return $end;
}
$minPos = $end;
}
return empty($endPositions) ? $length : min($endPositions);
return $length;
}
private function pushJob($checkId, $delaySeconds = 0)