参考文献校对升级

2026-06-29 10:23:27 +08:00
parent edb3c1b27b
commit 6fdc4efb6f
11 changed files with 2680 additions and 380 deletions
--- a/application/api/controller/Base.php
+++ b/application/api/controller/Base.php
@@ -1178,6 +1178,107 @@ class Base extends Controller
        return $ids;
    }

+    /**
+     * 解析方括号引用内层（如 1,2 / 3-5），展开为文献序号列表。
+     *
+     * @return int[]
+     */
+    protected function expandCitationBracketNumbers(string $referencePart): array
+    {
+        $referencePart = trim($referencePart);
+        if ($referencePart === '') {
+            return [];
+        }
+        $referencePart = str_replace(
+            ['，', '–', '—', '−', '‐', '‑'],
+            [',', '-', '-', '-', '-', '-'],
+            $referencePart
+        );
+        $out = [];
+        $segments = preg_split('/\s*,\s*/', $referencePart);
+        foreach ($segments as $seg) {
+            $seg = trim((string)$seg);
+            if ($seg === '') {
+                continue;
+            }
+            $seg = str_replace(['–', '—', '−', '‐', '‑'], '-', $seg);
+            if (preg_match('/^(\d+)\s*-\s*(\d+)$/', $seg, $m)) {
+                $a = intval($m[1]);
+                $b = intval($m[2]);
+                if ($a > $b) {
+                    $t = $a;
+                    $a = $b;
+                    $b = $t;
+                }
+                for ($i = $a; $i <= $b; $i++) {
+                    $out[] = $i;
+                }
+            } else {
+                $n = intval($seg);
+                if ($n > 0) {
+                    $out[] = $n;
+                }
+            }
+        }
+        return $out;
+    }
+
+    /**
+     * 从正文片段提取被引用的文献序号（reference_no = index+1）。
+     * 兼容 <mycite data-id="p_refer_id"> 与 <blue>[n]</blue> / [n] 两种形态。
+     *
+     * @return int[]
+     */
+    protected function extractCitationRefNosFromMainContent(string $text, int $pArticleId = 0): array
+    {
+        if ($text === '') {
+            return [];
+        }
+
+        $nos = [];
+
+        $pReferIds = $this->extractMyciteIds($text);
+        if (!empty($pReferIds) && $pArticleId > 0) {
+            $refers = Db::name('production_article_refer')
+                ->where('p_article_id', $pArticleId)
+                ->whereIn('p_refer_id', $pReferIds)
+                ->where('state', 0)
+                ->field('p_refer_id,index')
+                ->select();
+            $idToNo = [];
+            foreach ($refers as $row) {
+                $idToNo[intval($row['p_refer_id'])] = intval($row['index']) + 1;
+            }
+            foreach ($pReferIds as $pid) {
+                if (isset($idToNo[$pid])) {
+                    $nos[] = $idToNo[$pid];
+                }
+            }
+        }
+
+        if (preg_match_all('/(?:<\s*blue[^>]*>)?\[([^\]]+)\](?:<\/\s*blue\s*>)?/iu', $text, $m)) {
+            foreach ($m[1] as $inner) {
+                $innerNorm = str_replace(
+                    ['，', '–', '—', '−', '‐', '‑'],
+                    [',', '-', '-', '-', '-', '-'],
+                    trim((string)$inner)
+                );
+                if (!preg_match('/^[\d\s,\-]+$/u', $innerNorm)) {
+                    continue;
+                }
+                foreach ($this->expandCitationBracketNumbers($innerNorm) as $n) {
+                    if ($n > 0) {
+                        $nos[] = $n;
+                    }
+                }
+            }
+        }
+
+        $nos = array_values(array_unique($nos));
+        sort($nos, SORT_NUMERIC);
+        return $nos;
+    }
+
    /**
     * table_data：二维数组 JSON [[{text,colspan,rowspan},...],...]；支持双重 JSON 字符串编码。
     *