Merge branch 'release/0.1.2'

Author: ja573

src/db.py

@@ -104,6 +104,27 @@ def get_canonical_title(work) -> str:
 
         return getattr(work, "fullTitle", None)
 
+    @staticmethod
+    def get_page_range(work: Dict) -> str:
+        """Return a chapter page range from Thoth metadata."""
+        page_interval = work.get("pageInterval")
+        if page_interval:
+            return page_interval
+
+        first_page = work.get("firstPage")
+        last_page = work.get("lastPage")
+
+        if first_page and last_page:
+            return f"{first_page}-{last_page}"
+
+        if first_page:
+            return f"{first_page}-{first_page}"
+
+        if last_page:
+            return f"{last_page}-{last_page}"
+
+        return None
+
     def get_book(self) -> Dict:
         """Return book data"""
         work = self.db.work_by_doi(doi=self.doi)
@@ -124,6 +145,8 @@ def get_chapters(self, book: Dict) -> List:
                                         copyrightHolder
                                         longAbstract
                                         pageInterval
+                                        firstPage
+                                        lastPage
                                         doi
                                         license
                                         imprint {
@@ -150,7 +173,7 @@ def get_chapters(self, book: Dict) -> List:
                     "title":     work.get("fullTitle"),
                     "publisher": work.get("imprint", {}).get("imprintName"),
                     "abstract":  work.get("longAbstract"),
-                    "pages":     work.get("pageInterval"),
+                    "pages":     self.get_page_range(work),
                     "licence":   work.get("license"),
                     "workId":    work.get("workId")}
             chapters.append(data)

src/main.py

@@ -33,7 +33,14 @@ def run(input_file:    Path = typer.Option("./file.pdf",
 
         # Iterate over chapters metadata
         for chapter in metadata.get_chapters():
-            page_range = re.split('-|–', chapter.get("pages"))
+            pages = chapter.get("pages")
+            if not pages:
+                raise ValueError(
+                    "Missing page range for chapter "
+                    f"{chapter.get('doi') or chapter.get('title')}"
+                )
+
+            page_range = re.split('-|–', pages)
 
             doi_fragments = chapter.get("doi").split('/')
             output_file_name = doi_fragments[-1].lower() + '.pdf'

src/test_db.py

@@ -32,6 +32,18 @@ def work_by_doi(self, doi):
     }
 
 
+def test_thoth_get_page_range_falls_back_to_first_and_last_page():
+    assert Thoth.get_page_range(
+        {"pageInterval": None, "firstPage": "11", "lastPage": "20"}
+    ) == "11-20"
+
+
+def test_thoth_get_page_range_uses_single_page_when_needed():
+    assert Thoth.get_page_range(
+        {"pageInterval": None, "firstPage": "7", "lastPage": None}
+    ) == "7-7"
+
+
 def test_thoth_write_urls_uses_pat_and_preserves_payloads(monkeypatch):
     class MockDb:
         def __init__(self):

src/test_main.py

@@ -97,3 +97,56 @@ def write_urls(self, chapter):
             "workId": "work-1",
         }
     ]
+
+
+def test_cli_raises_clear_error_when_pages_missing(monkeypatch, tmp_path):
+    input_file = tmp_path / "book.pdf"
+    output_folder = tmp_path / "output"
+    input_file.write_bytes(b"%PDF-1.4\n")
+    output_folder.mkdir()
+
+    class MockPdf:
+        def __init__(self, source, tmp_dir):
+            self.tmp_dir = Path(tmp_dir)
+
+        def merge_pdfs(self, page_range, output_file_name):
+            raise AssertionError("merge_pdfs should not be called")
+
+    class MockMetadata:
+        def __init__(self, database, doi):
+            pass
+
+        def get_chapters(self):
+            return [
+                {
+                    "pages": None,
+                    "doi": "10.11647/obp.0309.01",
+                    "workId": "work-1",
+                }
+            ]
+
+        def write_metadata(self, chapter, output_file_path):
+            raise AssertionError("write_metadata should not be called")
+
+        def write_urls(self, chapter):
+            raise AssertionError("write_urls should not be called")
+
+    monkeypatch.setattr(main, "Pdf", MockPdf)
+    monkeypatch.setattr(main, "Metadata", MockMetadata)
+
+    result = runner.invoke(
+        main.app,
+        [
+            "--input-file",
+            str(input_file),
+            "--output-folder",
+            str(output_folder),
+            "10.11647/obp.0309",
+        ],
+    )
+
+    assert result.exit_code != 0
+    assert isinstance(result.exception, ValueError)
+    assert str(result.exception) == (
+        "Missing page range for chapter 10.11647/obp.0309.01"
+    )