参考文献格式--book
This commit is contained in:
wangjinlei
405
application/common/BookCitationParser.php
Normal file
405
application/common/BookCitationParser.php
Normal file
@@ -0,0 +1,405 @@
|
||||
<?php
|
||||
|
||||
namespace app\common;
|
||||
|
||||
use app\common\service\LLMService;
|
||||
|
||||
/**
|
||||
* 图书类参考文献著录解析：原文字符串 → author / title / publisher / year / edition …
|
||||
*
|
||||
* 注意 refer_content 存的是带序号的原文（"1. "、"[1] "、"1) "），
|
||||
* 直接按第一个句点切分会把序号当作者、把作者当标题，所以必须先剥序号再解析。
|
||||
*
|
||||
* 规则解析拿不准时（缺标题、或标题看着像作者名单）再交给大模型兜底。
|
||||
*/
|
||||
class BookCitationParser
|
||||
{
|
||||
/** @var bool */
|
||||
private $useLlm;
|
||||
|
||||
public function __construct(array $config = [])
|
||||
{
|
||||
$this->useLlm = isset($config['use_llm']) ? (bool)$config['use_llm'] : true;
|
||||
}
|
||||
|
||||
/**
|
||||
* @return array author/title/publisher/place/year/edition/pages/isbn/container
|
||||
*/
|
||||
public function parse($content)
|
||||
{
|
||||
$out = [
|
||||
'author' => '',
|
||||
'title' => '',
|
||||
'publisher' => '',
|
||||
'place' => '',
|
||||
'year' => '',
|
||||
'edition' => '',
|
||||
'pages' => '',
|
||||
'isbn' => '',
|
||||
'container' => '',
|
||||
];
|
||||
|
||||
$clean = $this->normalize($content);
|
||||
if ($clean === '') {
|
||||
return $out;
|
||||
}
|
||||
|
||||
$out['isbn'] = $this->matchIsbn($clean);
|
||||
$clean = $this->stripTail($clean);
|
||||
|
||||
// 章节引用："章节作者. 章节名. In: 编者. 书名. 版次. 地点: 出版社; 年. 页码"
|
||||
// 出版信息属于 In: 之后的那本书，必须分开解析，否则会把编者当成书名
|
||||
if (preg_match('/^(.*?)\bIn\s*:\s*(.+)$/isu', $clean, $m) && trim($m[1]) !== '' && trim($m[2]) !== '') {
|
||||
$chapter = trim($m[1]);
|
||||
$book = trim($m[2]);
|
||||
|
||||
$this->parseSegments($this->takeAuthor($chapter, $out['author']), $out);
|
||||
|
||||
$bookOut = $out;
|
||||
$bookOut['title'] = '';
|
||||
$editors = '';
|
||||
$this->parseSegments($this->takeAuthor($book, $editors), $bookOut);
|
||||
|
||||
$out['container'] = $bookOut['title'];
|
||||
foreach (['publisher', 'place', 'year', 'edition', 'pages'] as $field) {
|
||||
if ($out[$field] === '') {
|
||||
$out[$field] = $bookOut[$field];
|
||||
}
|
||||
}
|
||||
if ($out['author'] === '') {
|
||||
$out['author'] = $editors;
|
||||
}
|
||||
} else {
|
||||
$this->parseSegments($this->takeAuthor($clean, $out['author']), $out);
|
||||
}
|
||||
|
||||
if ($this->needsLlm($out)) {
|
||||
$out = $this->refineByLlm($clean, $out);
|
||||
}
|
||||
|
||||
return $out;
|
||||
}
|
||||
|
||||
/**
|
||||
* 去掉条目序号："1. " / "[1] " / "(1) " / "1) " / "1、"
|
||||
*/
|
||||
public function stripIndex($content)
|
||||
{
|
||||
return preg_replace('/^\s*(?:\[\d{1,4}\]|\(\d{1,4}\)|\d{1,4}\s*[\.\)\]、,])\s*/u', '', (string)$content);
|
||||
}
|
||||
|
||||
private function normalize($content)
|
||||
{
|
||||
$content = trim((string)$content);
|
||||
$content = str_replace(
|
||||
['：', '；', '，', '（', '）', '［', '］', ' ', '．'],
|
||||
[':', ';', ',', '(', ')', '[', ']', ' ', '.'],
|
||||
$content
|
||||
);
|
||||
$content = preg_replace('/\s+/u', ' ', $content);
|
||||
$content = $this->stripIndex($content);
|
||||
// 中文句号统一成英文句点，便于按同一套规则切分
|
||||
$content = str_replace('。', '. ', $content);
|
||||
|
||||
return trim(preg_replace('/\s+/u', ' ', $content));
|
||||
}
|
||||
|
||||
/**
|
||||
* 去掉尾部的获取途径（Available at / Accessed / URL / DOI），它们会干扰分段
|
||||
*/
|
||||
private function stripTail($text)
|
||||
{
|
||||
$text = preg_replace('/\s*(?:Available\s*(?:at|from)|Retrieved\s+from|Accessed\s+(?:on\s+)?)\s*:?.*$/iu', '', $text);
|
||||
$text = preg_replace('/\s*https?:\/\/\S+/iu', '', $text);
|
||||
$text = preg_replace('/\s*\bdoi\s*:\s*\S+/iu', '', $text);
|
||||
$text = preg_replace('/\s*\bISBN(?:-1[03])?\s*:?\s*[\d\-\sXx]{10,20}\.?/iu', '', $text);
|
||||
|
||||
return trim($text);
|
||||
}
|
||||
|
||||
private function matchIsbn($text)
|
||||
{
|
||||
if (!preg_match('/\bISBN(?:-1[03])?[:\s]*([0-9][0-9\-\s]{8,20}[0-9Xx])/i', $text, $m)) {
|
||||
return '';
|
||||
}
|
||||
|
||||
return preg_replace('/[^0-9Xx]/', '', $m[1]);
|
||||
}
|
||||
|
||||
// ------------------------------------------------------------------
|
||||
// 作者
|
||||
// ------------------------------------------------------------------
|
||||
|
||||
/**
|
||||
* 从头部切出作者，返回剩余部分
|
||||
*/
|
||||
private function takeAuthor($text, &$author)
|
||||
{
|
||||
$author = '';
|
||||
|
||||
// 温哥华格式：Surname AB, Surname CD, et al.（可带 editor/eds 标记）
|
||||
$name = '\p{Lu}[\p{L}\'\x{2019}\-]+(?:\s+\p{Lu}[\p{L}\'\x{2019}\-]+)*\s+\p{Lu}{1,4}';
|
||||
$pattern = '/^((?:' . $name . ')(?:\s*,\s*(?:' . $name . '))*'
|
||||
. '(?:\s*,?\s*et\s+al\.?)?)(?:\s*,\s*(?:eds?|editors?))?\s*[\.,]\s+/u';
|
||||
if (preg_match($pattern, $text, $m)) {
|
||||
$author = trim($m[1], " ,.");
|
||||
return trim(substr($text, strlen($m[0])));
|
||||
}
|
||||
|
||||
// 机构作者
|
||||
$org = '/^([\p{Lu}][\p{L}\s&\-,\.]{4,90}?(?:Organization|Organisation|Association|Society|Institute'
|
||||
. '|Institutes|Ministry|Committee|Council|Administration|Agency|Department|Bureau|Foundation'
|
||||
. '|Academy|Commission|Centers?|WHO|CDC|NIH|FDA|NICE|UNICEF|OECD))\.\s+/u';
|
||||
if (preg_match($org, $text, $m)) {
|
||||
$author = trim($m[1], " ,.");
|
||||
return trim(substr($text, strlen($m[0])));
|
||||
}
|
||||
|
||||
// 中文作者：张三, 李四, 等.
|
||||
$cn = '/^([\x{4e00}-\x{9fff}·]{2,10}(?:\s*[,、;]\s*[\x{4e00}-\x{9fff}·]{2,10})*(?:\s*,?\s*等)?)\s*[\.,]\s*/u';
|
||||
if (preg_match($cn, $text, $m)) {
|
||||
$author = trim($m[1], " ,.");
|
||||
return trim(substr($text, strlen($m[0])));
|
||||
}
|
||||
|
||||
// 兜底：第一个句点之前
|
||||
$parts = preg_split('/\.\s+/u', $text, 2);
|
||||
if (is_array($parts) && count($parts) === 2 && mb_strlen($parts[0], 'UTF-8') <= 120) {
|
||||
$author = trim($parts[0], " ,.");
|
||||
return trim($parts[1]);
|
||||
}
|
||||
|
||||
return $text;
|
||||
}
|
||||
|
||||
// ------------------------------------------------------------------
|
||||
// 分段与归类
|
||||
// ------------------------------------------------------------------
|
||||
|
||||
private function parseSegments($text, array &$out)
|
||||
{
|
||||
foreach ($this->splitSegments($text) as $seg) {
|
||||
$seg = trim($seg, " .,;");
|
||||
if ($seg === '') {
|
||||
continue;
|
||||
}
|
||||
|
||||
// "6th ed. New York: Garland Science; 2015" 里版次和出版信息粘在一段，
|
||||
// 摘掉版次再往下判，否则出版地会变成 "6th ed. New York"
|
||||
$edition = $this->matchEdition($seg);
|
||||
if ($edition !== '') {
|
||||
if ($out['edition'] === '') {
|
||||
$out['edition'] = $edition;
|
||||
}
|
||||
$seg = trim($this->stripEdition($seg), " .,;");
|
||||
if ($seg === '') {
|
||||
continue;
|
||||
}
|
||||
}
|
||||
|
||||
if ($out['pages'] === '' && preg_match('/^(?:p{1,2}\.?|pages?)\s*([\divxlcIVXLC][\divxlcIVXLC\-\x{2013}]*)$/iu', $seg, $m)) {
|
||||
$out['pages'] = $m[1];
|
||||
continue;
|
||||
}
|
||||
|
||||
if ($this->takePublication($seg, $out)) {
|
||||
continue;
|
||||
}
|
||||
|
||||
if ($out['title'] === '') {
|
||||
$out['title'] = $this->cleanTitle($seg);
|
||||
}
|
||||
}
|
||||
|
||||
if ($out['year'] === '' && preg_match_all('/\b(?:19|20)\d{2}\b/', $text, $ym)) {
|
||||
$out['year'] = end($ym[0]);
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* 缩写后的句点不是分段边界（ed. / vol. / p. / Inc. / 姓名缩写）
|
||||
*
|
||||
* @return string[]
|
||||
*/
|
||||
private function splitSegments($text)
|
||||
{
|
||||
$abbr = 'ed|eds|edn|vol|vols|no|nos|pp|p|st|mt|dr|prof|inc|ltd|co|jr|sr|al|rev|suppl|fig|ch|pt|approx';
|
||||
$masked = preg_replace('/\b(' . $abbr . ')\.(?=\s)/iu', '$1' . "\x01", $text);
|
||||
// "Kumar V. Abbas AK." 这类姓名缩写之间的点
|
||||
$masked = preg_replace('/\b(\p{Lu})\.(?=\s*\p{Lu}\b)/u', '$1' . "\x01", $masked);
|
||||
|
||||
$segments = preg_split('/\.\s+|\.$/u', $masked);
|
||||
$out = [];
|
||||
foreach ((array)$segments as $seg) {
|
||||
$out[] = str_replace("\x01", '.', $seg);
|
||||
}
|
||||
|
||||
return $out;
|
||||
}
|
||||
|
||||
private function matchEdition($seg)
|
||||
{
|
||||
if (preg_match('/\b(\d{1,2})\s*(?:st|nd|rd|th|d)?\s*(?:ed\.?|edn\.?|edition)\b/iu', $seg, $m)) {
|
||||
return $m[1];
|
||||
}
|
||||
if (preg_match('/\b(first|second|third|fourth|fifth|sixth|seventh|eighth|ninth|tenth)\s+(?:ed\.?|edn\.?|edition)\b/iu', $seg, $m)) {
|
||||
return strtolower($m[1]);
|
||||
}
|
||||
if (preg_match('/(?:第\s*)?(\d{1,2})\s*版/u', $seg, $m)) {
|
||||
return $m[1];
|
||||
}
|
||||
|
||||
return '';
|
||||
}
|
||||
|
||||
private function stripEdition($seg)
|
||||
{
|
||||
$seg = preg_replace('/\b\d{1,2}\s*(?:st|nd|rd|th|d)?\s*(?:ed\.?|edn\.?|edition)\b[\.,;]?/iu', '', $seg);
|
||||
$seg = preg_replace('/\b(?:first|second|third|fourth|fifth|sixth|seventh|eighth|ninth|tenth)\s+(?:ed\.?|edn\.?|edition)\b[\.,;]?/iu', '', $seg);
|
||||
$seg = preg_replace('/(?:第\s*)?\d{1,2}\s*版/u', '', $seg);
|
||||
|
||||
return $seg;
|
||||
}
|
||||
|
||||
/**
|
||||
* 出版信息段：Place: Publisher; Year / Publisher, Year / 北京: 人民卫生出版社, 2018
|
||||
*/
|
||||
private function takePublication($seg, array &$out)
|
||||
{
|
||||
$hasYear = preg_match('/\b((?:19|20)\d{2})\b/', $seg, $ym);
|
||||
$hasPlace = preg_match('/^([^:]{2,60}):\s*(.+)$/u', $seg, $pm);
|
||||
if (!$hasYear && !$hasPlace) {
|
||||
return false;
|
||||
}
|
||||
|
||||
// "Depression in adults: treatment and management" 是带副标题的书名，不是"地点: 出版社"
|
||||
if ($hasPlace && !$hasYear && ($out['title'] === '' || !$this->looksLikePublisher($pm[2]))) {
|
||||
return false;
|
||||
}
|
||||
|
||||
// 标题里出现年份（如 "Global tuberculosis report 2020"）不算出版信息段
|
||||
if ($hasYear && !$hasPlace && $out['title'] === '' && !preg_match('/^\W*(?:19|20)\d{2}\W*$/', $seg)) {
|
||||
$withoutYear = trim(preg_replace('/\b(?:19|20)\d{2}\b/', '', $seg), " .,;");
|
||||
if ($withoutYear !== '' && !$this->looksLikePublisher($withoutYear)) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
if ($hasYear && $out['year'] === '') {
|
||||
$out['year'] = $ym[1];
|
||||
}
|
||||
|
||||
$body = $hasPlace ? trim($pm[2]) : $seg;
|
||||
if ($hasPlace && $out['place'] === '') {
|
||||
$out['place'] = trim($pm[1], " .,;");
|
||||
}
|
||||
$body = trim(preg_replace('/\b(?:19|20)\d{2}\b/', '', $body), " .,;:");
|
||||
if ($body !== '' && $out['publisher'] === '') {
|
||||
$out['publisher'] = $body;
|
||||
}
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
private function looksLikePublisher($text)
|
||||
{
|
||||
return (bool)preg_match(
|
||||
'/(Press|Publish\w*|Books?|Elsevier|Springer|Wiley|Saunders|Mosby|Lippincott|Williams|Wilkins'
|
||||
. '|McGraw|Academic|University|Univ\b|出版社|书局|WHO|Organization)/iu',
|
||||
$text
|
||||
);
|
||||
}
|
||||
|
||||
private function cleanTitle($title)
|
||||
{
|
||||
$title = trim((string)$title);
|
||||
// GB/T 7714 的文献类型标识：[M] [M/OL] [C] 等
|
||||
$title = preg_replace('/\s*\[[A-Z]{1,2}(?:\/[A-Z]{1,2})?\]\s*/u', ' ', $title);
|
||||
$title = preg_replace('/\s+/u', ' ', $title);
|
||||
|
||||
return trim($title, " .,;:");
|
||||
}
|
||||
|
||||
// ------------------------------------------------------------------
|
||||
// 大模型兜底
|
||||
// ------------------------------------------------------------------
|
||||
|
||||
private function needsLlm(array $out)
|
||||
{
|
||||
if (!$this->useLlm) {
|
||||
return false;
|
||||
}
|
||||
if ($out['title'] === '' || $out['author'] === '') {
|
||||
return true;
|
||||
}
|
||||
// 分段错位时标题里会留下一串人名
|
||||
return $this->looksLikeAuthorList($out['title']);
|
||||
}
|
||||
|
||||
private function looksLikeAuthorList($text)
|
||||
{
|
||||
$text = trim((string)$text);
|
||||
if ($text === '') {
|
||||
return false;
|
||||
}
|
||||
$name = '\p{Lu}[\p{L}\'\-]+\s+\p{Lu}{1,4}';
|
||||
|
||||
return (bool)preg_match('/^(?:' . $name . ')(?:\s*,\s*(?:' . $name . '))*(?:\s*,?\s*et\s+al\.?)?$/u', $text);
|
||||
}
|
||||
|
||||
private function refineByLlm($content, array $out)
|
||||
{
|
||||
try {
|
||||
$llm = new LLMService();
|
||||
$system = 'You extract bibliographic fields from a single book (or book chapter) reference string. '
|
||||
. 'Return ONLY a JSON object with these keys: '
|
||||
. '{"author":"","title":"","publisher":"","place":"","year":"","edition":"","pages":""}. '
|
||||
. 'Rules: author = the author or editor list exactly as written, without a trailing period; '
|
||||
. 'title = the title of the cited work only, without edition, publisher, place or year; '
|
||||
. 'for a chapter reference, title = the chapter title; '
|
||||
. 'publisher = publisher name only, no place; year = 4-digit publication year; '
|
||||
. 'edition = the edition number as digits only (e.g. "6"), empty if not stated; '
|
||||
. 'pages = page range if stated. Use an empty string for anything not stated. '
|
||||
. 'Do not translate, do not invent, do not add explanation.';
|
||||
$user = "Reference:\n" . mb_substr($content, 0, 1500);
|
||||
|
||||
$reply = $llm->requestChat([
|
||||
['role' => 'system', 'content' => $system],
|
||||
['role' => 'user', 'content' => $user],
|
||||
], 0);
|
||||
if ($reply === null || $reply === '') {
|
||||
return $out;
|
||||
}
|
||||
|
||||
$parsed = $llm->parseJsonResponse($reply);
|
||||
if (!is_array($parsed)) {
|
||||
return $out;
|
||||
}
|
||||
|
||||
foreach (['author', 'title', 'publisher', 'place', 'year', 'edition', 'pages'] as $field) {
|
||||
$value = trim((string)($parsed[$field] ?? ''));
|
||||
if ($value === '') {
|
||||
continue;
|
||||
}
|
||||
// 模型不得改写原文里没有的内容
|
||||
if ($field === 'year' && !preg_match('/^(?:19|20)\d{2}$/', $value)) {
|
||||
continue;
|
||||
}
|
||||
if ($field === 'edition') {
|
||||
$value = preg_match('/\d{1,2}/', $value, $m) ? $m[0] : '';
|
||||
if ($value === '') {
|
||||
continue;
|
||||
}
|
||||
}
|
||||
if ($field === 'title' && $this->looksLikeAuthorList($value)) {
|
||||
continue;
|
||||
}
|
||||
$out[$field] = $field === 'title' ? $this->cleanTitle($value) : $value;
|
||||
}
|
||||
} catch (\Throwable $e) {
|
||||
\think\Log::write('book citation llm parse failed: ' . $e->getMessage(), 'error');
|
||||
}
|
||||
|
||||
return $out;
|
||||
}
|
||||
}
|
||||
@@ -120,29 +120,23 @@ class BookMetadataService
|
||||
return $hints;
|
||||
}
|
||||
|
||||
// 章节引用（In: 编者. 书名. ...）反查的是书，不是章节
|
||||
$bookPart = $content;
|
||||
if (preg_match('/\bIn:\s*(.+)$/i', $content, $m)) {
|
||||
$bookPart = trim($m[1]);
|
||||
$hints['title'] = '';
|
||||
// 反查用规则解析即可，不为了取个 ISBN 再多花一次大模型调用
|
||||
$parsed = (new BookCitationParser(['use_llm' => false]))->parse($content);
|
||||
|
||||
// 章节引用反查的是书，不是章节
|
||||
if ($parsed['container'] !== '') {
|
||||
$hints['title'] = $parsed['container'];
|
||||
} elseif ($hints['title'] === '') {
|
||||
$hints['title'] = $parsed['title'];
|
||||
}
|
||||
|
||||
$hints['edition'] = $this->parseEdition($content);
|
||||
|
||||
if ($hints['year'] === '' && preg_match_all('/\b(?:19|20)\d{2}\b/', $content, $ym)) {
|
||||
$hints['year'] = end($ym[0]);
|
||||
foreach (['author', 'publisher', 'edition'] as $field) {
|
||||
if ($hints[$field] === '' && $parsed[$field] !== '') {
|
||||
$hints[$field] = $parsed[$field];
|
||||
}
|
||||
|
||||
// 地点: 出版社; 年份
|
||||
if ($hints['publisher'] === '' && preg_match('/:\s*([^;:]{2,60}?)\s*[;,]\s*(?:19|20)\d{2}/u', $bookPart, $m)) {
|
||||
$hints['publisher'] = trim($m[1], " .,");
|
||||
}
|
||||
|
||||
if ($hints['title'] === '') {
|
||||
$hints['title'] = $this->parseTitle($bookPart);
|
||||
}
|
||||
if ($hints['author'] === '') {
|
||||
$hints['author'] = $this->parseAuthor($bookPart);
|
||||
if ($hints['year'] === '') {
|
||||
$hints['year'] = $parsed['year'];
|
||||
}
|
||||
|
||||
return $hints;
|
||||
@@ -432,38 +426,6 @@ class BookMetadataService
|
||||
return trim($title, " .,;:");
|
||||
}
|
||||
|
||||
private function parseTitle($text)
|
||||
{
|
||||
$parts = preg_split('/\.\s+/u', trim((string)$text), 3);
|
||||
if (is_array($parts) && count($parts) >= 2) {
|
||||
return $this->cleanTitle($parts[1]);
|
||||
}
|
||||
|
||||
return '';
|
||||
}
|
||||
|
||||
private function parseAuthor($text)
|
||||
{
|
||||
$parts = preg_split('/\.\s+/u', trim((string)$text), 2);
|
||||
|
||||
return is_array($parts) && count($parts) >= 2 ? trim($parts[0]) : '';
|
||||
}
|
||||
|
||||
private function parseEdition($text)
|
||||
{
|
||||
if (preg_match('/\b(\d{1,2})\s*(?:st|nd|rd|th|d)?\s*(?:ed\.?|edition)\b/iu', $text, $m)) {
|
||||
return $m[1];
|
||||
}
|
||||
if (preg_match('/\b(first|second|third|fourth|fifth|sixth|seventh|eighth|ninth|tenth)\s+(?:ed\.?|edition)\b/iu', $text, $m)) {
|
||||
return $m[1];
|
||||
}
|
||||
if (preg_match('/第\s*(\d{1,2})\s*版/u', $text, $m)) {
|
||||
return $m[1];
|
||||
}
|
||||
|
||||
return '';
|
||||
}
|
||||
|
||||
private function editionOrdinal($edition)
|
||||
{
|
||||
$edition = strtolower(trim((string)$edition));
|
||||
|
||||
@@ -24,11 +24,15 @@ class ReferenceDispatchService
|
||||
/** @var BookMetadataService */
|
||||
private $bookMetadata;
|
||||
|
||||
/** @var BookCitationParser */
|
||||
private $bookParser;
|
||||
|
||||
public function __construct()
|
||||
{
|
||||
$this->classifier = new ReferenceTypeClassifier(['use_llm' => true]);
|
||||
$this->metadata = new ReferenceMetadataService();
|
||||
$this->bookMetadata = new BookMetadataService();
|
||||
$this->bookParser = new BookCitationParser();
|
||||
}
|
||||
|
||||
/**
|
||||
@@ -181,7 +185,9 @@ class ReferenceDispatchService
|
||||
$summary = is_array($meta) ? $meta['crossref_summary'] : null;
|
||||
$raw = is_array($summary) ? ($summary['raw'] ?? []) : [];
|
||||
$hasMeta = is_array($meta)
|
||||
&& (trim((string)$meta['title']) !== '' || trim((string)$meta['author']) !== '');
|
||||
&& (trim((string)$meta['title']) !== '' || trim((string)$meta['author']) !== '')
|
||||
// 图书著录里挂的 DOI 常常是被错误关联的期刊文章，题名对不上就不能拿来覆盖
|
||||
&& $this->metaMatchesContent($meta, $content);
|
||||
|
||||
if ($hasMeta) {
|
||||
$authorCitation = trim((string)$meta['author']);
|
||||
@@ -217,6 +223,44 @@ class ReferenceDispatchService
|
||||
Db::name('production_article_refer')->where('p_refer_id', $pReferId)->update($update);
|
||||
}
|
||||
|
||||
/**
|
||||
* DOI 抓回来的题名是否确实是这条著录说的那本书
|
||||
*/
|
||||
private function metaMatchesContent($meta, $content)
|
||||
{
|
||||
$title = $this->normalizeForMatch((string)$meta['title']);
|
||||
if ($title === '') {
|
||||
return false;
|
||||
}
|
||||
$blob = $this->normalizeForMatch($content);
|
||||
if ($blob === '' || strpos($blob, $title) !== false) {
|
||||
return true;
|
||||
}
|
||||
|
||||
$words = array_filter(explode(' ', $title), function ($w) {
|
||||
return strlen($w) > 3;
|
||||
});
|
||||
if (empty($words)) {
|
||||
return false;
|
||||
}
|
||||
$hit = 0;
|
||||
foreach ($words as $word) {
|
||||
if (strpos($blob, $word) !== false) {
|
||||
$hit++;
|
||||
}
|
||||
}
|
||||
|
||||
return ($hit / count($words)) >= 0.6;
|
||||
}
|
||||
|
||||
private function normalizeForMatch($text)
|
||||
{
|
||||
$text = strtolower(trim((string)$text));
|
||||
$text = preg_replace('/[^a-z0-9\x{4e00}-\x{9fff}\s]+/u', ' ', $text);
|
||||
|
||||
return trim(preg_replace('/\s+/u', ' ', $text));
|
||||
}
|
||||
|
||||
/**
|
||||
* 按书名/作者/年份/版次反查 ISBN，顺带补齐缺失的出版社与出版年
|
||||
*
|
||||
@@ -299,50 +343,22 @@ class ReferenceDispatchService
|
||||
}
|
||||
|
||||
/**
|
||||
* 无 Crossref 时从原文尽量抽取 book 结构化字段
|
||||
* 无可用 DOI 元数据时，从原文著录抽取 book 结构化字段
|
||||
*/
|
||||
private function parseBookFromContent($content)
|
||||
{
|
||||
$content = trim((string)$content);
|
||||
$out = [
|
||||
'author' => '',
|
||||
'title' => '',
|
||||
'joura' => '',
|
||||
'dateno' => '',
|
||||
'isbn' => '',
|
||||
$parsed = $this->bookParser->parse($content);
|
||||
|
||||
$author = trim((string)$parsed['author']);
|
||||
|
||||
return [
|
||||
'author' => $author !== '' ? rtrim($author, '.') . '.' : '',
|
||||
'title' => trim((string)$parsed['title']),
|
||||
'joura' => trim((string)$parsed['publisher']),
|
||||
'dateno' => trim((string)$parsed['year']),
|
||||
'isbn' => trim((string)$parsed['isbn']),
|
||||
'is_ja' => 1,
|
||||
];
|
||||
|
||||
if ($content === '') {
|
||||
return $out;
|
||||
}
|
||||
|
||||
if (preg_match('/\bISBN[:\s]*([\d\-Xx\s]+)/i', $content, $m)) {
|
||||
$out['isbn'] = preg_replace('/\s+/', '-', trim($m[1]));
|
||||
}
|
||||
|
||||
if (preg_match('/\b(19|20)\d{2}\b/', $content, $m)) {
|
||||
$out['dateno'] = $m[0];
|
||||
}
|
||||
|
||||
// Place: Publisher; Year → joura 取 Publisher
|
||||
if (preg_match('/:\s*([^;]+);\s*(19|20)\d{2}/', $content, $m)) {
|
||||
$out['joura'] = trim($m[1]);
|
||||
} elseif (preg_match('/\b([A-Z][A-Za-z .&]+(?:Press|Publishing|Publisher|Books?))\b/i', $content, $m)) {
|
||||
$out['joura'] = trim($m[1]);
|
||||
}
|
||||
|
||||
// 作者. 标题. ... 简单拆分
|
||||
$parts = preg_split('/\.\s+/', $content, 3);
|
||||
if (is_array($parts) && count($parts) >= 2) {
|
||||
$out['author'] = trim($parts[0]);
|
||||
if (substr($out['author'], -1) !== '.') {
|
||||
$out['author'] .= '.';
|
||||
}
|
||||
$out['title'] = trim(rtrim($parts[1], '.'));
|
||||
}
|
||||
|
||||
return $out;
|
||||
}
|
||||
|
||||
private function normalizeDoi($doi)
|
||||
|
||||
Reference in New Issue
Block a user