Fix gallery resolution parsing for split archive paragraphs

src/main/java/com/lion/lionwebsite/Util/GalleryPageParser.java

@@ -11,6 +11,10 @@ import java.util.regex.Pattern;
 public final class GalleryPageParser {
     private GalleryPageParser() { }
 
+    /** 归档页 #db 单元格开头：分辨率标签 + 正数体积 + 受支持单位；后缀费用文本（Free、50GP）忽略。 */
+    private static final Pattern RESOLUTION_ENTRY =
+            Pattern.compile("^(Original|(\\d+)x)\\s+(\\d+(?:\\.\\d+)?)\\s*([A-Za-z]+)(?=\\s|$)", Pattern.CASE_INSENSITIVE);
+
     public static Gallery metadata(Document document, String link, int gid) throws IOException {
         String title = document.select("#gn").text();
         if (title.isBlank()) title = document.select("#gj").text();
@@ -52,16 +56,37 @@ public final class GalleryPageParser {
         throw new IOException("上游页面缺少归档入口");
     }
 
+    /**
+     * 解析归档页 #db 的可选分辨率。每个单元格合并后的文本形如 {@code Original 36.31 MiB Free}、
+     * {@code 800x 5.65 MiB Free}、{@code 2560x N/A N/A}，单段布局形如 {@code 1280x 12.5 MiB}，
+     * 都由同一条锚定规则覆盖；只看开头（分辨率 + 正数体积 + 单位），故 N/A 与费用文本不匹配，
+     * 配对也不会跨 {@code <td>} 借用相邻格。体积须能被 {@link CustomUtil#stringToFileSize(String)}
+     * 解析为正数，宽度须为正且不溢出。结果按宽度升序、{@code Original} 最后；无可用项抛 IOException。
+     */
     public static Map<String, String> resolutions(Document document) throws IOException {
-        var expression = Pattern.compile("^(\\d+x|Original)\\s+([\\d.]+)\\s*([A-Za-z]+)$", Pattern.CASE_INSENSITIVE);
         TreeMap<Integer, String> sizes = new TreeMap<>();
         String original = null;
-        for (Element paragraph : document.select("#db td p")) {
+        for (Element cell : document.select("#db td")) {
-            var matcher = expression.matcher(paragraph.text().strip());
+            var matcher = RESOLUTION_ENTRY.matcher(cell.text().replace('\u00a0', ' ').strip());
-            if (!matcher.matches()) continue;
+            if (!matcher.find()) continue;
-            String size = matcher.group(2) + matcher.group(3);
+            String size = matcher.group(3) + matcher.group(4);
-            if (matcher.group(1).equalsIgnoreCase("Original")) original = size;
+            try {
-            else sizes.put(Integer.parseInt(matcher.group(1).replace("x", "")), size);
+                if (CustomUtil.stringToFileSize(size) <= 0) continue;
+            } catch (RuntimeException unsupported) {
+                continue;
+            }
+            if (matcher.group(1).equalsIgnoreCase("Original")) {
+                original = size;
+                continue;
+            }
+            int width;
+            try {
+                width = Integer.parseInt(matcher.group(2));
+            } catch (NumberFormatException outOfRange) {
+                continue;
+            }
+            if (width <= 0) continue;
+            sizes.put(width, size);
         }
         Map<String, String> result = new LinkedHashMap<>();
         sizes.forEach((width, size) -> result.put(width + "x", size));

src/test/java/com/lion/lionwebsite/Util/UpstreamParserVariantsTest.java

@@ -3,6 +3,9 @@ package com.lion.lionwebsite.Util;
 import org.jsoup.Jsoup;
 import org.junit.jupiter.api.Test;
 import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
+import java.util.Set;
 import static org.junit.jupiter.api.Assertions.*;
 
 class UpstreamParserVariantsTest {
@@ -33,4 +36,119 @@ class UpstreamParserVariantsTest {
         assertEquals("value", JavaScriptAssignments.string("let mpvkey = 'value';", "mpvkey"));
         assertThrows(IOException.class, () -> JavaScriptAssignments.array("imagelist = [{\"k\":\"x\"}", "imagelist"));
     }
+
+    /** 线上真实结构：每个单元格拆成三个 <p>（分辨率/体积/费用）。旧实现逐段匹配导致全部漏掉。 */
+    @Test void resolutionsReadSplitParagraphCellsFromLiveLayout() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody><tr>"
+                + "<td><p>Original</p><p>36.31 MiB</p><p>Free</p></td>"
+                + "<td><p>800x</p><p>5.65 MiB</p><p>Free</p></td>"
+                + "<td><p>1280x</p><p>12.5 MiB</p><p>Free</p></td>"
+                + "<td><p>1920x</p><p>30.2 MiB</p><p>Free</p></td>"
+                + "<td><p>2560x</p><p>N/A</p><p>N/A</p></td>"
+                + "</tr></tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(List.of("800x", "1280x", "1920x", "Original"), new ArrayList<>(resolutions.keySet()));
+        assertEquals("5.65MiB", resolutions.get("800x"));
+        assertEquals("12.5MiB", resolutions.get("1280x"));
+        assertEquals("30.2MiB", resolutions.get("1920x"));
+        assertEquals("36.31MiB", resolutions.get("Original"), "Original 应排在最后");
+        assertFalse(resolutions.containsKey("2560x"), "N/A 的体积不得进结果");
+    }
+
+    /** 历史/单段布局：分辨率与体积在同一段 <p> 也必须继续兼容。 */
+    @Test void resolutionsKeepMergedParagraphLayoutCompatible() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody>"
+                + "<tr><td><p>1280x 12.5 MiB</p></td></tr>"
+                + "<tr><td><p>Original 30.2 MiB</p></td></tr>"
+                + "</tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(List.of("1280x", "Original"), new ArrayList<>(resolutions.keySet()));
+        assertEquals("12.5MiB", resolutions.get("1280x"));
+        assertEquals("30.2MiB", resolutions.get("Original"));
+    }
+
+    /** N/A 与体积分处两格时不得跨 td 拼接。 */
+    @Test void resolutionsNeverStitchAcrossCells() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody><tr>"
+                + "<td><p>2560x</p><p>N/A</p><p>N/A</p></td>"
+                + "<td><p>1920x</p><p>30.2 MiB</p></td>"
+                + "<td><p>800x</p><p>N/A</p></td>"
+                + "</tr></tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(Set.of("1920x"), resolutions.keySet());
+        assertEquals("30.2MiB", resolutions.get("1920x"));
+        assertTrue(resolutions.values().stream().noneMatch(size -> size.toUpperCase().contains("N/A")));
+    }
+
+    /** 单元格开头不是“分辨率标签 + 体积”时不得凭空造出条目。 */
+    @Test void resolutionsRequireResolutionLabelWithinSameCell() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody><tr>"
+                + "<td><p>800x</p><p>5.65 MiB</p></td>"
+                + "<td><p>12.5 MiB</p></td>"
+                + "<td><p>1920x</p><p>Costs 50 GP</p></td>"
+                + "</tr></tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(List.of("800x"), new ArrayList<>(resolutions.keySet()));
+        assertEquals("5.65MiB", resolutions.get("800x"));
+    }
+
+    /** 嵌套标签（strong/span/div）与 &nbsp; 由 text() 归并后仍应正确取到分辨率与体积。 */
+    @Test void resolutionsReadNestedTagsInsideCells() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody><tr>"
+                + "<td><div><p><strong>1280x</strong></p><div><p><span>12.5</span> <em>MiB</em></p></div></div></td>"
+                + "<td><p>Original</p><p>30.2&nbsp;MiB</p></td>"
+                + "</tr></tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(List.of("1280x", "Original"), new ArrayList<>(resolutions.keySet()));
+        assertEquals("12.5MiB", resolutions.get("1280x"));
+        assertEquals("30.2MiB", resolutions.get("Original"), "&nbsp; 分隔的体积也要认");
+    }
+
+    /** 未知单位、坏数字和费用文本都不得当作体积，只保留可解析的项。 */
+    @Test void resolutionsRejectUnknownUnitAndCostText() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody><tr>"
+                + "<td><p>1280x</p><p>12.5 ZiB</p></td>"
+                + "<td><p>1920x</p><p>Costs 50 GP</p></td>"
+                + "<td><p>2560x</p><p>.5 MiB</p></td>"
+                + "<td><p>800x</p><p>5.65 MiB</p></td>"
+                + "</tr></tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(List.of("800x"), new ArrayList<>(resolutions.keySet()));
+        assertEquals("5.65MiB", resolutions.get("800x"));
+    }
+
+    /** 完全取不到可用分辨率（空文档或只有 N/A）必须抛 IOException，而不是返回空表。 */
+    @Test void resolutionsThrowWhenNothingUsable() throws Exception {
+        assertThrows(IOException.class, () -> GalleryPageParser.resolutions(Jsoup.parse("<html/>")));
+        assertThrows(IOException.class, () -> GalleryPageParser.resolutions(Jsoup.parse(
+                "<div id=db><table><tbody><tr><td><p>2560x</p><p>N/A</p><p>N/A</p></td>"
+                        + "<td><p>N/A</p></td></tr></tbody></table></div>")));
+    }
+
+    /** 宽度为 0 或溢出、以及单位后粘连字符的项都要跳过，不能崩到调用方。 */
+    @Test void resolutionsSkipInvalidWidthAndGluedUnit() throws Exception {
+        var document = Jsoup.parse("<div id=db><table><tbody><tr>"
+                + "<td><p>0x</p><p>5.65 MiB</p><p>Free</p></td>"
+                + "<td><p>99999999999x</p><p>12.5 MiB</p><p>Free</p></td>"
+                + "<td><p>800x</p><p>5.65 MiB2</p><p>Free</p></td>"
+                + "<td><p>1280x</p><p>12.5 MiB</p><p>Free</p></td>"
+                + "</tr></tbody></table></div>");
+
+        var resolutions = GalleryPageParser.resolutions(document);
+
+        assertEquals(List.of("1280x"), new ArrayList<>(resolutions.keySet()));
+        assertEquals("12.5MiB", resolutions.get("1280x"));
+    }
 }