Merge pull request #176 from qaiu/copilot/optimize-pandomain-template

fix(PanDomainTemplate): 修复现有网盘域名模板正则表达式中的多处缺陷

parser/src/main/java/cn/qaiu/parser/PanDomainTemplate.java

@@ -68,8 +68,8 @@ public enum PanDomainTemplate {
         t-is.cn
      */
     LZ("蓝奏云",
-            compile("https://(?:[a-zA-Z\\d-]+\\.)?(" +
-                    "lanzoul|" +
+            compile("https://(?:[a-zA-Z\\d-]+\\.)?(?:" +
+                    "(?:lanzoul|" +
                     "lanzouh|" +
                     "lanosso|" +
                     "lanpv|" +
@@ -95,14 +95,16 @@ public enum PanDomainTemplate {
                     "lanzv|" +
                     "dmpdmp|" +
                     "lanrar|" +
+                    "webgetstore|" +
                     "lanzb|" +
                     "lanzoux|" +
                     "lanzout|" +
                     "lanzouc|" +
                     "lanzoui|" +
                     "lanzoug|" +
-                    "lanzoum" +
-                    ")\\.com/(?<KEY>.+)"),
+                    "lanzoum)\\.com" +
+                    "|t-is\\.cn" +
+                    ")/(?<KEY>.+)"),
             "https://w1.lanzn.com/{shareKey}",
             LzTool.class),
 
@@ -113,9 +115,9 @@ public enum PanDomainTemplate {
             "https://www.feijix.com/s/{shareKey}",
             FjTool.class),
 
-    // https://lecloud.lenovo.com/share/  https://lecloud.lenovo.com/mshare/
+    // https://lecloud.lenovo.com/share/
     LE("联想乐云",
-            compile("https://lecloud?\\.lenovo\\.com/m?share/(?<KEY>.+)"),
+            compile("https://lecloud\\.lenovo\\.com/share/(?<KEY>.+)"),
             "https://lecloud.lenovo.com/share/{shareKey}",
             LeTool.class),
 
@@ -241,7 +243,7 @@ public enum PanDomainTemplate {
             EcTool.class),
     // https://cowtransfer.com/s/
     COW("奶牛快传",
-            compile("https://(.*)cowtransfer\\.com/s/(?<KEY>.+)"),
+            compile("https://(?:[a-zA-Z\\d-]+\\.)?cowtransfer\\.com/s/(?<KEY>.+)"),
             "https://cowtransfer.com/s/{shareKey}",
             CowTool.class),
     CT("城通网盘",
@@ -264,7 +266,7 @@ public enum PanDomainTemplate {
             PodTool.class),
     // 404网盘 https://drive.google.com/file/d/xxx/view?usp=sharing
     PGD("GoogleDrive",
-            compile("https://drive\\.google\\.com/file/d/(?<KEY>.+)/view(\\?usp=(sharing|drive_link))?"),
+            compile("https://(?:[a-zA-Z\\d-]+\\.)?drive\\.google\\.com/file/d/(?<KEY>.+)/view(\\?usp=(sharing|drive_link))?"),
             "https://drive.google.com/file/d/{shareKey}/view?usp=sharing",
             PgdTool.class),
     // iCloud https://www.icloud.com.cn/iclouddrive/xxx#fonts
@@ -274,11 +276,11 @@ public enum PanDomainTemplate {
             PicTool.class),
     // https://www.dropbox.com/scl/fi/cwnbms1yn8u6rcatzyta7/emqx-5.0.26-el7-amd64.tar.gz?rlkey=3uoi4bxz5mv93jmlaws0nlol1&e=8&st=fe0lclc2&dl=0
     PDB("dropbox",
-            compile("https://www.dropbox.com/scl/fi/(?<KEY>\\w+)/.+?rlkey=(?<PWD>\\w+).*"),
+            compile("https://www\\.dropbox\\.com/scl/fi/(?<KEY>\\w+)/.+?rlkey=(?<PWD>\\w+).*"),
             "https://www.dropbox.com/scl/fi/{shareKey}/?rlkey={pwd}&dl=0",
             PdbTool.class),
     P115("115网盘",
-            compile("https://(115|anxia).com/s/(?<KEY>\\w+)(\\?password=(?<PWD>\\w+))?([&#].*)?"),
+            compile("https://(115|anxia)\\.com/s/(?<KEY>\\w+)(\\?password=(?<PWD>\\w+))?([&#].*)?"),
             "https://115.com/s/{shareKey}?password={pwd}",
             P115Tool.class),
     // 链接：https://www.yunpan.com/surl_yD7wz4VgU9v（提取码：fc70）
@@ -319,7 +321,7 @@ public enum PanDomainTemplate {
             MnesTool.class),
     // https://music.163.com/#/song?id=xxx
     MNE("网易云音乐歌曲详情",
-            compile("https://(y.)?music\\.163\\.com/(#|m/)?song\\?id=(?<KEY>.+)(&.*)?"),
+            compile("https://(y\\.)?music\\.163\\.com/(?:#/|m/)?song\\?id=(?<KEY>.+)(&.*)?"),
             "https://music.163.com/#/song?id={shareKey}",
             MnesTool.MneTool.class),
     // https://c6.y.qq.com/base/fcgi-bin/u?__=xxx
@@ -340,7 +342,7 @@ public enum PanDomainTemplate {
             MkgsTool.class),
     // https://www.kugou.com/share/2bi8Fe9CSV3.html?id=2bi8Fe9CSV3#6ed9gna4"
     MKGS2("酷狗音乐分享2",
-            compile("https://(?:[a-zA-Z\\d-]+\\.)?kugou\\.com/share/(?<KEY>.+).html.*"),
+            compile("https://(?:[a-zA-Z\\d-]+\\.)?kugou\\.com/share/(?<KEY>.+)\\.html.*"),
             "https://www.kugou.com/share/{shareKey}.html",
             MkgsTool.Mkgs2Tool.class),
     // https://www.kugou.com/mixsong/2bi8Fe9CSV3

parser/src/test/java/cn/qaiu/parser/PanDomainTemplateTest.java

@@ -130,33 +130,129 @@ public class PanDomainTemplateTest {
     }
 
     @Test
-    public void testLePatternMatching() {
+    public void testLzPatternWebgetstore() {
+        Pattern lzPattern = PanDomainTemplate.LZ.getPattern();
+
+        // webgetstore.com 以前遗漏，现已补入
+        Matcher m1 = lzPattern.matcher("https://webgetstore.com/somekey");
+        assertTrue("LZ should match webgetstore.com", m1.find());
+        assertEquals("somekey", m1.group("KEY"));
+
+        Matcher m2 = lzPattern.matcher("https://www.webgetstore.com/somekey");
+        assertTrue("LZ should match www.webgetstore.com", m2.find());
+        assertEquals("somekey", m2.group("KEY"));
+
+        // t-is.cn 以前遗漏，现已补入
+        Matcher m3 = lzPattern.matcher("https://t-is.cn/somekey");
+        assertTrue("LZ should match t-is.cn", m3.find());
+        assertEquals("somekey", m3.group("KEY"));
+
+        Matcher m4 = lzPattern.matcher("https://www.t-is.cn/somekey");
+        assertTrue("LZ should match www.t-is.cn", m4.find());
+        assertEquals("somekey", m4.group("KEY"));
+
+        // 已有域名仍然正常匹配
+        Matcher m5 = lzPattern.matcher("https://www.lanzoul.com/somekey");
+        assertTrue("LZ should match existing domain lanzoul.com", m5.find());
+        assertEquals("somekey", m5.group("KEY"));
+    }
+
+    @Test
+    public void testLePatternFix() {
         Pattern lePattern = PanDomainTemplate.LE.getPattern();
 
-        // /share/ 格式
+        // lecloud.lenovo.com 应匹配
         Matcher m1 = lePattern.matcher("https://lecloud.lenovo.com/share/abc123");
-        assertTrue("LE pattern should match /share/ format", m1.matches());
+        assertTrue("LE should match lecloud.lenovo.com", m1.find());
         assertEquals("abc123", m1.group("KEY"));
 
-        // /mshare/ 格式
-        Matcher m2 = lePattern.matcher("https://lecloud.lenovo.com/mshare/xyz789");
-        assertTrue("LE pattern should match /mshare/ format", m2.matches());
-        assertEquals("xyz789", m2.group("KEY"));
+        // leclou.lenovo.com (去掉'd') 不应匹配（原 lecloud? 的 bug）
+        assertFalse("LE should NOT match leclou.lenovo.com",
+                lePattern.matcher("https://leclou.lenovo.com/share/abc123").find());
+    }
 
-        // 负例：错误路径不匹配
-        assertFalse("Wrong path should not match",
-                lePattern.matcher("https://lecloud.lenovo.com/s/abc123").matches());
+    @Test
+    public void testCowPatternFix() {
+        Pattern cowPattern = PanDomainTemplate.COW.getPattern();
+
+        // 正常域名
+        Matcher m1 = cowPattern.matcher("https://cowtransfer.com/s/abc123");
+        assertTrue("COW should match cowtransfer.com", m1.find());
+        assertEquals("abc123", m1.group("KEY"));
+
+        Matcher m2 = cowPattern.matcher("https://share.cowtransfer.com/s/abc123");
+        assertTrue("COW should match share.cowtransfer.com", m2.find());
+        assertEquals("abc123", m2.group("KEY"));
+
+        // 潜在的URL注入：`(.*)` 是贪婪捕获组，可匹配 `evil.com/redirect/` 等前缀，
+        // 使形如 `https://evil.com/redirect/cowtransfer.com/s/key` 的 URL 被误识别。
+        // 修复后改为 `(?:[a-zA-Z\d-]+\.)?` 仅匹配一级合法子域名（可选），消除误匹配。
+        assertFalse("COW should NOT match redirect URLs containing cowtransfer.com in path",
+                cowPattern.matcher("https://evil.com/redirect/cowtransfer.com/s/abc").find());
+    }
+
+    @Test
+    public void testMnePatternFix() {
+        Pattern mnePattern = PanDomainTemplate.MNE.getPattern();
+
+        // 带 #/ 前缀的完整网页链接（修复前因 (y.) 未转义而存在 bug）
+        Matcher m1 = mnePattern.matcher("https://music.163.com/#/song?id=12345");
+        assertTrue("MNE should match #/song format", m1.find());
+        assertEquals("12345", m1.group("KEY"));
+
+        // 带 m/ 前缀的移动端链接
+        Matcher m2 = mnePattern.matcher("https://music.163.com/m/song?id=12345");
+        assertTrue("MNE should match m/song format", m2.find());
+        assertEquals("12345", m2.group("KEY"));
+
+        // y.music.163.com 子域名
+        Matcher m3 = mnePattern.matcher("https://y.music.163.com/song?id=12345");
+        assertTrue("MNE should match y.music.163.com", m3.find());
+        assertEquals("12345", m3.group("KEY"));
+
+        // 原 (y.) 中 `.` 未转义（`.` 匹配任意字符）：对于 `yXmusic.163.com`，
+        // `(y.)` 会消费 `yX`（y + 任意字符），剩余 `music.163.com` 再被 `music\.163\.com` 匹配，导致误匹配。
+        // 修复后 `(y\.)` 要求字面 `.`，`yX` 中 X ≠ `.` 无法匹配，不再误匹配。
+        assertFalse("MNE should NOT match yXmusic.163.com (old (y.) could erroneously match via backtracking)",
+                mnePattern.matcher("https://yXmusic.163.com/song?id=12345").find());
+    }
+
+    @Test
+    public void testP115PatternFix() {
+        Pattern p115Pattern = PanDomainTemplate.P115.getPattern();
+
+        // 正常匹配
+        Matcher m1 = p115Pattern.matcher("https://115.com/s/abc123");
+        assertTrue("P115 should match 115.com", m1.find());
+        assertEquals("abc123", m1.group("KEY"));
+
+        Matcher m2 = p115Pattern.matcher("https://anxia.com/s/abc123");
+        assertTrue("P115 should match anxia.com", m2.find());
+        assertEquals("abc123", m2.group("KEY"));
+
+        // 原 .com 未转义时 115Xcom 会被误匹配（现已修复）
+        assertFalse("P115 should NOT match 115Xcom",
+                p115Pattern.matcher("https://115Xcom/s/abc123").find());
+    }
+
+    @Test
+    public void testPgdSubdomain() {
+        Pattern pgdPattern = PanDomainTemplate.PGD.getPattern();
+
+        // 标准链接
+        Matcher m1 = pgdPattern.matcher("https://drive.google.com/file/d/abc123/view?usp=sharing");
+        assertTrue("PGD should match standard drive.google.com", m1.find());
+        assertEquals("abc123", m1.group("KEY"));
+
+        // 带子域名的链接（修复后支持）
+        Matcher m2 = pgdPattern.matcher("https://adsd.drive.google.com/file/d/151bR-nk-tOBm9QAFaozJIVt2WYyCMkoz/view");
+        assertTrue("PGD should match subdomain.drive.google.com", m2.find());
+        assertEquals("151bR-nk-tOBm9QAFaozJIVt2WYyCMkoz", m2.group("KEY"));
     }
 
     @Test
     public void verifyDuplicates() {
 
-        Matcher matcher = compile("https://(?:[a-zA-Z\\d-]+\\.)?drive\\.google\\.com/file/d/(?<KEY>.+)/view(\\?usp=(sharing|drive_link))?")
-                .matcher("https://adsd.drive.google.com/file/d/151bR-nk-tOBm9QAFaozJIVt2WYyCMkoz/view");
-        if (matcher.find()) {
-            System.out.println(matcher.group());
-            System.out.println(matcher.group("KEY"));
-        }
         // 校验重复
         Set<String> collect =
                 Arrays.stream(PanDomainTemplate.values()).map(PanDomainTemplate::getRegex).collect(Collectors.toSet());