AGENTS/CONTEXT: new sources (cogito person pages, livelib author pages), verified RU editions (Franz, Jaffé/Man&HisSymbols, Lüscher, Kandinsky, Kast, Schaverien)

tools/rsl.py

@@ -19,18 +19,28 @@ def fetch(url, tries=3):
 
 def main():
     q = sys.argv[1]
-    maxpages = int(sys.argv[2]) if len(sys.argv) > 2 else 4
+    maxpages = int(sys.argv[2]) if len(sys.argv) > 2 else 2
+    seen = set()
     for page in range(1, maxpages + 1):
         url = "https://libgen.vg/biblioservice.php?" + urllib.parse.urlencode(
             {"value": q, "type": "rsl", "format": "json", "page": page})
-        d = json.loads(fetch(url))
+        try:
+            d = json.loads(fetch(url))
+        except SystemExit:
+            print("(fetch failed)")
+            break
         recs = {k: v for k, v in d.items() if k != "error"}
         if not recs:
             if page == 1:
                 print("(no records)")
             break
+        new = 0
         for k in sorted(recs, key=int):
             r = recs[k]
+            if r["id"] in seen:
+                continue
+            seen.add(r["id"])
+            new += 1
             title = " ".join(r.get("title", [])) + (
                 (" : " + r["title_add"]) if r.get("title_add") else "")
             author = "; ".join(r.get("author", []))
@@ -43,6 +53,8 @@ def main():
             print(f"[{r['id']}] {title}")
             print(f"    {author}")
             print(f"    {city} : {pub}, {year} | {pages} | {isbn}" + (f" | серия: {series}" if series else ""))
+        if new == 0:
+            break
         time.sleep(3)
 
 if __name__ == "__main__":