Metadata: library-style PDF author 'Last, First, dates' becomes 'First Last'

_pipeline/meta.py

@@ -57,6 +57,16 @@ def junk_author(a):
     return len(re.sub(r"[^A-Za-z]", "", a)) < 3 or a.lower() in JUNK_AUTHORS or bool(re.search(r"\d{3,}", a))
 
 
+def info_author_name(a):
+    """PDF metadata authors often look like 'Lincoln, Abraham, 1809-1865': drop the dates and put the first name first."""
+    a = re.sub(r"[,\s]*\(?\b(?:c\.?\s*)?\d{3,4}\s*-\s*(?:\d{3,4})?\)?\.?\s*$", "", clean_text(a)).strip(" ,;")
+    parts = [p.strip() for p in a.split(",")]
+    if len(parts) == 2 and all(re.fullmatch(r"[A-Za-z.'’\- ]+", p) for p in parts) and 1 <= len(parts[0].split()) <= 2 \
+            and not any(w.lower().strip(".") in NOT_NAME_WORDS | {"inc", "ltd", "llc", "jr", "sr", "corp", "gmbh"} for w in (parts[0] + " " + parts[1]).split()):
+        a = parts[1] + " " + parts[0]
+    return a
+
+
 def display_name(name):
     """'YOGI RAMACHARAKA' / 'yogi ramacharaka' -> 'Yogi Ramacharaka'. Mixed case is kept (McDonald)."""
     out = []
@@ -259,7 +269,9 @@ def detect_from_pdf(cfg, pdf, hint=None):
     res = {"title": "", "title_source": "", "author": "", "author_source": "", "notes": notes}
     if info["title"] and not junk_title(info["title"]):
         res["title"], res["title_source"] = bookio.title_case(clean_text(info["title"])), "pdf-metadata"
-    info_author = info["author"] if info["author"] and not junk_author(info["author"]) else ""
+    info_author = info_author_name(info["author"]) if info["author"] else ""
+    if info_author and junk_author(info_author):
+        info_author = ""
 
     pages, blank = first_pages_text(cfg, pdf, n, notes)
     page = read_title_page([(p, t) for p, t, _ in pages], hint)