fix: merge identifiers across a book's formats

Identifiers are a collection, but the whole dict was replaced per format, so
the last file to report won outright — an EPUB declaring an ASIN, a Google id
and a Calibre id kept none of them once a PDF contributed one ISBN. Accumulate
instead, letting a declared identifier outrank one scraped off a page.

backend/tests/unit/test_metadata_extractor.py

@@ -1,7 +1,7 @@
 import pytest
 from pathlib import Path
 from datetime import date
-from chitai.services.metadata_extractor import EpubExtractor
+from chitai.services.metadata_extractor import EpubExtractor, Extractor, PdfExtractor
 
 
 @pytest.mark.asyncio()
@@ -15,3 +15,62 @@ class TestEpubExtractor:
         assert metadata["authors"] == ["Herman Melville"]
         assert metadata["published_date"] == date(year=2001, month=7, day=1)
 
+
+
+EPUB = Path("tests/data_files/Metamorphosis - Franz Kafka.epub")
+PDF = Path("tests/data_files/Calculus Made Easy - Silvanus Thompson.pdf")
+
+
+@pytest.mark.asyncio()
+class TestIdentifierMerging:
+    """A book's formats each contribute identifiers; none of them replaces the rest."""
+
+    async def test_every_format_contributes(self, monkeypatch: pytest.MonkeyPatch) -> None:
+        """
+        Identifiers are a collection, not a single value.
+
+        Merging the whole dict meant the last format to report won outright: an EPUB
+        declaring an ASIN, a Google volume id and a Calibre id kept none of them once
+        a PDF contributed one ISBN.
+        """
+
+        async def epub(_file):
+            return {
+                "title": "How Linux Works",
+                "identifiers": {"isbn-13": "9781718500419", "asin": "1718500408"},
+            }
+
+        async def pdf(_file):
+            return {"identifiers": {"isbn-13": "9781718500402", "isbn-10": "1593270356"}}
+
+        monkeypatch.setattr(EpubExtractor, "extract_metadata", epub)
+        monkeypatch.setattr(PdfExtractor, "extract_metadata", pdf)
+
+        metadata = await Extractor.extract_metadata(
+            [Path("How Linux Works.epub"), Path("How Linux Works.pdf")]
+        )
+
+        assert metadata["identifiers"] == {
+            # Declared by the publisher's toolchain, so it outranks the PDF's, which
+            # was scraped off a copyright page that also prints the print edition's.
+            "isbn-13": "9781718500419",
+            "asin": "1718500408",
+            "isbn-10": "1593270356",
+        }
+
+    async def test_a_second_format_that_finds_nothing_erases_nothing(self) -> None:
+        """The PDF fixture carries no ISBN, so it must leave the EPUB's alone."""
+        metadata = await Extractor.extract_metadata([EPUB, PDF])
+
+        assert metadata["identifiers"] == {"id": "http://www.gutenberg.org/5200"}
+
+    async def test_one_format_on_its_own_is_unaffected(self) -> None:
+        metadata = await Extractor.extract_metadata([EPUB])
+
+        assert metadata["identifiers"] == {"id": "http://www.gutenberg.org/5200"}
+
+    async def test_no_identifiers_anywhere_leaves_the_field_absent(self) -> None:
+        """An empty dict would count as extracted metadata and overwrite nothing."""
+        metadata = await Extractor.extract_metadata([PDF])
+
+        assert "identifiers" not in metadata