Release v0.13: Typography V3 (Safe mode) & Endnote Extraction Fix

2026-02-01 11:51:36 +08:00
parent 8e58415173
commit 40491e6386
95 changed files with 8493 additions and 11 deletions
@@ -0,0 +1,36 @@
+import json
+from bs4 import BeautifulSoup
+
+path = ".work/Gambling Man/book_structure.json"
+target_file = "e9781668070765/xhtml/ch08.xhtml"
+target_id = "uuid-100309c1-0643-4aa1-8ffb-1d2bfbd0e376"
+
+try:
+    with open(path, 'r') as f:
+        data = json.load(f)
+    
+    target_res = None
+    for item_id, res in data['resources'].items():
+        if "ch08.xhtml" in res.get('href', ''):
+            print(f"Found resource with ID: {item_id}, href: {res.get('href')}")
+            target_res = res
+            break
+            
+    if not target_res:
+        print(f"Resource {target_file} not found by href search.")
+        exit(1)
+            
+    content = target_res['content']
+    soup = BeautifulSoup(content, 'html.parser')
+    element = soup.find(id=target_id)
+    
+    if element:
+        print(f"--- HTML for {target_id} ---")
+        print(element.prettify())
+        print("--- Raw ---")
+        print(str(element))
+    else:
+        print("Element not found")
+
+except Exception as e:
+    print(e)