Rebuild split-rulebook journal parsing (vault sync 2026-08-18 to 08-23)

The bab5a62 vault commit split 9 rulebook monoliths into hub/ToC pages plus per-topic-file subfolders (rules/NN. <Book>/<Chapter>/<Topic>.md). The existing chapter-slicing parser (parseJournalBook) relied on finding a duplicate TOC-vs-body heading pair per chapter; after the split each hub heading only occurs once (in the ToC), so it was silently treating the ToC's link-list body as if it were real chapter content. Added parseSplitBookFolder (journal-parser.ts): walks a book's split subfolder directly, one JournalDoc per Rozdzial/Aneks subfolder-or-file, one JournalPage per topic file (or a single page for a standalone chapter file). The hub's own Wstep intro paragraph is unaffected by the split (it's still real prose, not a link) and continues to come from the existing prefaceTitle preface-extraction logic. Wired into build-journal-packs.ts for the 6 affected packs: core-rules-lore, bestiary-lore, magic-book-lore, abyss, blood-magic-history, economy. crimson-cult's rules source was never split into a subfolder, so it's left on the old code path unchanged. Also fixed a pre-existing derivePageTitle bug (page titles kept a leading markdown heading marker, e.g. "## Wstep" instead of "Wstep") to match the already-correct behavior in topical-parser.ts. Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-08-23 01:15:51 +02:00
parent 447b90dbb7
commit 7dd2af1ec5
2 changed files with 201 additions and 15 deletions
@@ -22,7 +22,8 @@
 * Output: `JournalDoc[]` consumed by `build-journal-packs.ts`.
 */

-import { readFileSync } from 'node:fs';
+import { readFileSync, readdirSync, statSync } from 'node:fs';
+import { resolve, basename, extname } from 'node:path';

 export interface JournalDoc {
  /** Stable kebab-case slug. */
@@ -199,7 +200,7 @@ function trimEdges(arr: string[]): string[] {

 /** First non-empty line if it looks short and title-like, else "Część N". */
 function derivePageTitle(lines: string[], pageNo: number): string {
-  const first = lines.find((l) => l.trim().length > 0)?.trim() ?? '';
+  const first = (lines.find((l) => l.trim().length > 0)?.trim() ?? '').replace(/^#{1,6}\s+/, '');
  if (first.length > 0 && first.length <= 80 && !/[.!?]$/.test(first)) {
    return first;
  }
@@ -292,9 +293,151 @@ function slugify(s: string): string {
  return s
    .toLowerCase()
    .normalize('NFKD')
-    .replace(/[\u0300-\u036f]/g, '')
+    .replace(/[̀-ͯ]/g, '')
    .replace(/ł/g, 'l')
    .replace(/[^a-z0-9]+/g, '-')
    .replace(/^-|-$/g, '')
    .slice(0, 60);
 }
+
+// ---------------------------------------------------------------------------
+// Split-book folder parser
+// ---------------------------------------------------------------------------
+//
+// The `bab5a62` vault split turned each `rules/NN. <Book>.md` monolith into a
+// hub/ToC page plus a `rules/NN. <Book>/<Chapter>/<Topic>.md` folder tree.
+// `parseJournalBook`'s heading-slicing approach can no longer find real
+// chapter content in the hub (its former body text is now just links), so
+// split books read their chapters directly from that folder tree instead:
+// one `JournalDoc` per immediate child (a `Rozdział`/`Aneks` subfolder or a
+// standalone chapter file), one `JournalPage` per topic file inside it (or a
+// single page for a standalone chapter file). The hub's own `Wstęp` intro
+// (still real prose, not a link) is unaffected and continues to come from
+// `parseJournalBook`'s existing `prefaceTitle` handling.
+
+export interface SplitBookFolderConfig {
+  /** Absolute path to the book's split subfolder (e.g. `.../00. Podręcznik Gry`). */
+  folderPath: string;
+  pack: string;
+  sourceBook: string;
+  /** Skip chapters whose folder/file name OR extracted title matches this regex. */
+  chapterSkip?: RegExp;
+  /** Only include chapters whose folder/file name OR extracted title matches this regex. */
+  chapterFilter?: RegExp;
+}
+
+export function parseSplitBookFolder(cfg: SplitBookFolderConfig): JournalDoc[] {
+  let entries: string[];
+  try {
+    entries = readdirSync(cfg.folderPath);
+  } catch {
+    return [];
+  }
+
+  const chapters = entries
+    .filter((e) => !e.startsWith('.'))
+    .map((e) => ({ entry: e, full: resolve(cfg.folderPath, e) }))
+    .filter(({ full }) => {
+      let s;
+      try {
+        s = statSync(full);
+      } catch {
+        return false;
+      }
+      return s.isDirectory() || extname(full).toLowerCase() === '.md';
+    });
+
+  chapters.sort((a, b) => {
+    const ka = chapterSortKey(chapterBaseName(a.entry));
+    const kb = chapterSortKey(chapterBaseName(b.entry));
+    return ka[0] - kb[0] || ka[1] - kb[1] || ka[2].localeCompare(kb[2]);
+  });
+
+  const docs: JournalDoc[] = [];
+  for (const { entry, full } of chapters) {
+    const name = chapterBaseName(entry);
+    const title = chapterTitle(name);
+    if (cfg.chapterSkip && (cfg.chapterSkip.test(name) || cfg.chapterSkip.test(title))) continue;
+    if (cfg.chapterFilter && !(cfg.chapterFilter.test(name) || cfg.chapterFilter.test(title))) continue;
+
+    const isDir = statSync(full).isDirectory();
+    const pages = isDir ? topicFilesToPages(full) : singleFileToPages(full);
+    if (pages.length === 0) continue;
+
+    docs.push({
+      id: `${cfg.pack}-${slugify(name)}`,
+      name,
+      pack: cfg.pack,
+      source: { book: cfg.sourceBook, chapter: title, line: 0 },
+      pages,
+    });
+  }
+  return docs;
+}
+
+function chapterBaseName(entry: string): string {
+  return extname(entry).toLowerCase() === '.md' ? basename(entry, '.md') : entry;
+}
+
+/**
+ * Strips a `Rozdział <roman> - ` or `Aneks <letter> - ` prefix, e.g.
+ * "Rozdział II - Talenty" -> "Talenty", "Aneks B - Przedmioty" -> "Przedmioty".
+ */
+function chapterTitle(name: string): string {
+  const m = name.match(/^(?:Rozdzia[łl]\s+[IVXLC]+|Aneks\s+[A-Za-z])\s*[-–:.]\s*(.+)$/i);
+  return m ? m[1].trim() : name;
+}
+
+function romanToInt(r: string): number {
+  const vals: Record<string, number> = { I: 1, V: 5, X: 10, L: 50, C: 100 };
+  const up = r.toUpperCase();
+  let total = 0;
+  for (let i = 0; i < up.length; i++) {
+    const cur = vals[up[i]] ?? 0;
+    const next = vals[up[i + 1]] ?? 0;
+    total += cur < next ? -cur : cur;
+  }
+  return total;
+}
+
+/** Rozdział chapters first (by numeral), then Aneks appendices (by letter), then anything else. */
+function chapterSortKey(name: string): [number, number, string] {
+  const rz = name.match(/^Rozdzia[łl]\s+([IVXLC]+)/i);
+  if (rz) return [0, romanToInt(rz[1]), name];
+  const an = name.match(/^Aneks\s+([A-Za-z])/i);
+  if (an) return [1, an[1].toUpperCase().charCodeAt(0), name];
+  return [2, 0, name];
+}
+
+function fileBody(filePath: string): string[] {
+  const raw = readFileSync(filePath, 'utf8').split(/\r?\n/);
+  return raw.slice(skipFrontmatter(raw));
+}
+
+function cleanBody(lines: string[]): string[] {
+  return trimEdges(lines.filter((l) => !SEPARATOR.test(l.trim()) && !OBSIDIAN_TOC_BULLET.test(l.trim())));
+}
+
+/** One topic file per page, sorted by filename. */
+function topicFilesToPages(dirPath: string): JournalPage[] {
+  const files = readdirSync(dirPath)
+    .filter((f) => extname(f).toLowerCase() === '.md')
+    .sort((a, b) => a.localeCompare(b));
+
+  const pages: JournalPage[] = [];
+  let sort = 100;
+  for (const file of files) {
+    const cleaned = cleanBody(fileBody(resolve(dirPath, file)));
+    if (cleaned.length === 0) continue;
+    pages.push({ name: basename(file, '.md'), html: linesToHtml(cleaned), sort });
+    sort += 100;
+  }
+  return pages;
+}
+
+/** A standalone chapter file (no subfolder) becomes a single page. */
+function singleFileToPages(filePath: string): JournalPage[] {
+  const cleaned = cleanBody(fileBody(filePath));
+  if (cleaned.length === 0) return [];
+  return [{ name: basename(filePath, '.md'), html: linesToHtml(cleaned), sort: 100 }];
+}