Workflow: sparse partial checkout (code and _incoming only), book lookup before processing, safe rebase retry; revert cancel-in-progress

_pipeline/process.py

@@ -112,6 +112,26 @@ def step(name):
     print("\n=== %s ===" % name, flush=True)
 
 
+def resolve_meta(pdf, title=None, author=None):
+    """(title, author, note). Order: explicit arguments, the .json next to the PDF, the PDF itself, the file name.
+    Raises Failed, telling the operator what to add, if none of them works."""
+    meta = sidecar(pdf)
+    title, author = title or meta.get("title"), author or meta.get("author")
+    try:
+        title, author = ingest.detect(pdf, title, author)
+    except Exception as e:
+        raise Failed("could not read the PDF: %s" % e)
+    note = None
+    if not (title and author):
+        t, a = guess_from_filename(pdf)
+        title, author = title or t, author or a
+        note = "title/author not found in the PDF; taking them from the file name: '%s' by '%s'" % (title, author)
+    if not (title and author):
+        raise Failed("ingest: could not find title and author in the PDF or its file name. Put a file with the same name and "
+                     '.json next to the PDF: {"title": "...", "author": "..."}')
+    return title, author, note
+
+
 def check_ocr(cfg, book_dir):
     """Pages with no text layer must have been OCRed. Fails the run if they were not (setting ocr.required)."""
     book = bookio.read_book(book_dir)
@@ -140,19 +160,16 @@ def check_ocr(cfg, book_dir):
 def run_text_stages(cfg, pdf, title, author):
     """Ingest through license for every chapter. Returns (book_dir, book)."""
     import cli
-    meta = sidecar(pdf)
-    title, author = title or meta.get("title"), author or meta.get("author")
+    title, author, note = resolve_meta(pdf, title, author)
     step("ingest")
+    if note:
+        print(note)
     try:
         book_dir = ingest.run(cfg, pdf, title, author)
     except SystemExit as e:
-        t, a = guess_from_filename(pdf)
-        if not (t and a):
-            raise Failed("ingest: %s. Put a file with the same name and .json next to the PDF: "
-                         '{"title": "...", "author": "..."}' % e)
-        print("ingest: title/author not found in the PDF; taking them from the file name: '%s' by '%s'" % (t, a))
-        book_dir = ingest.run(cfg, pdf, title or t, author or a)
-        bookio.write_warnings(book_dir, "ingest-metadata", [("metadata-from-filename", os.path.basename(pdf), "title '%s', author '%s'" % (t, a))])
+        raise Failed("ingest: %s" % e)
+    if note:
+        bookio.write_warnings(book_dir, "ingest-metadata", [("metadata-from-filename", os.path.basename(pdf), "title '%s', author '%s'" % (title, author))])
     print("ingest     done ->", book_dir)
     check_ocr(cfg, book_dir)
     for name in ("structure",):