fix pagecount

scripts/patch_ninemanga_niadd.py

@@ -406,22 +406,57 @@ replace_exact(
                 val href = a.attrAsRelativeUrl("href")
                     .replace("%20", " ")
 
+                val anchorText = a.text().trim()
+                val rowText = a.parent()?.text().orEmpty().trim()
+
                 val rawTitle = a.ownText()
                     .trim()
                     .ifEmpty {
-                        a.text().trim()
+                        anchorText
                     }
 
                 val anchorDate = chapterDateRegex
-                    .find(a.text())
+                    .find(anchorText)
                     ?.value
 
                 val rowDate = chapterDateRegex
-                    .find(a.parent()?.text().orEmpty())
+                    .find(rowText)
                     ?.value
 
                 val dateText = anchorDate ?: rowDate
 
+                val metadataText = when {
+                    anchorDate != null -> anchorText
+                    rowDate != null -> rowText
+                    else -> ""
+                }
+
+                val textBeforeDate = if (dateText != null) {
+                    metadataText
+                        .substringBefore(dateText)
+                        .trim()
+                } else {
+                    metadataText
+                }
+
+                /*
+                 * Niadd exposes the page count immediately before
+                 * the chapter date, for example:
+                 *
+                 *   Падение русала 1 - 42 41 Aug 02, 2024
+                 *                           ^^
+                 *
+                 * Store that number in a URL fragment. Fragments
+                 * are local metadata and are not sent to the server.
+                 */
+                val pageCount = Regex(
+                    """\\s+(\\d+)\\s*$"""
+                )
+                    .find(textBeforeDate)
+                    ?.groupValues
+                    ?.getOrNull(1)
+                    ?.toIntOrNull()
+
                 val chapterTitle = if (
                     dateText != null &&
                     rawTitle.contains(dateText)
@@ -437,6 +472,15 @@ replace_exact(
                     rawTitle
                 }
 
+                val chapterUrl = if (
+                    pageCount != null &&
+                    pageCount > 0
+                ) {
+                    "$href#niadd-pages=$pageCount"
+                } else {
+                    href
+                }
+
                 MangaChapter(
                     id = generateUid(href),
                     title = chapterTitle.takeIf {
@@ -444,7 +488,7 @@ replace_exact(
                     },
                     number = i + 1f,
                     volume = 0,
-                    url = href,
+                    url = chapterUrl,
                     uploadDate = parseChapterDateByLang(
                         dateText.orEmpty()
                     ),
@@ -597,42 +641,24 @@ replace_exact(
     '''    override suspend fun getPages(
         chapter: MangaChapter,
     ): List<MangaPage> {
-        val doc = webClient
-            .httpGet(
-                chapter.url.toAbsoluteUrl(domain)
-            )
-            .parseHtml()
-
         if (domain.endsWith("niadd.com")) {
-            val pageLabelRegex = Regex(
-                """^\\s*(\\d+)\\s*/\\s*(\\d+)\\s*$"""
-            )
-
-            val totalPages = doc
-                .select("option")
-                .mapNotNull { option ->
-                    pageLabelRegex
-                        .matchEntire(
-                            option.text().trim()
-                        )
-                        ?.groupValues
-                        ?.getOrNull(2)
-                        ?.toIntOrNull()
-                }
-                .maxOrNull()
-                ?: error(
-                    "Cannot find Niadd page count for ${chapter.url}"
-                )
-
-            if (totalPages <= 0) {
-                error(
-                    "Invalid Niadd page count $totalPages for ${chapter.url}"
-                )
-            }
-
             val originalUrl = chapter.url
+                .substringBefore("#")
                 .substringBefore("?")
 
+            val totalPages = chapter.url
+                .substringAfter(
+                    "#niadd-pages=",
+                    "",
+                )
+                .substringBefore("&")
+                .toIntOrNull()
+                ?.takeIf { it > 0 }
+                ?: error(
+                    "Niadd page count is missing for ${chapter.url}. " +
+                        "Refresh manga details and try again."
+                )
+
             val normalizedUrl = originalUrl
                 .removeSuffix("/")
 
@@ -660,6 +686,12 @@ replace_exact(
             }
         }
 
+        val doc = webClient
+            .httpGet(
+                chapter.url.toAbsoluteUrl(domain)
+            )
+            .parseHtml()
+
         return doc
             .body()
             .requireElementById("page")