fix: simplify pages api a bit

playa/pdfdocument.py

@@ -800,6 +800,7 @@ class PDFDocument:
     """
 
     _fp: Union[BinaryIO, None] = None
+    _pages: Union[List[PDFPage], None] = None
 
     def __enter__(self) -> "PDFDocument":
         return self
@@ -1118,24 +1119,23 @@ def get_page_objects(self) -> Iterator[Tuple[int, PageType]]:
                 yield object_id, object_properties
 
     @property
-    def pages(self) -> Iterator[PDFPage]:
-        """Iterator over PDFPage objects, which contain
-        information about the pages in the document.
-        """
-        try:
-            page_labels: Iterator[Optional[str]] = self.page_labels
-        except PDFNoPageLabels:
-            page_labels = itertools.repeat(None)
-        try:
-            for page_number, ((objid, properties), label) in enumerate(
-                zip(self.get_page_objects(), page_labels)
-            ):
-                yield PDFPage(self, objid, properties, label, page_number + 1)
-        except PDFNoPageTree:
-            for page_number, ((objid, properties), label) in enumerate(
-                zip(self.get_pages_from_xrefs(), page_labels)
-            ):
-                yield PDFPage(self, objid, properties, label, page_number + 1)
+    def pages(self) -> List[PDFPage]:
+        if self._pages is None:
+            try:
+                page_labels: Iterator[Optional[str]] = self.page_labels
+            except PDFNoPageLabels:
+                page_labels = itertools.repeat(None)
+            try:
+                self._pages = [PDFPage(self, objid, properties, label, page_number + 1)
+                               for page_number, ((objid, properties), label) in enumerate(
+                                       zip(self.get_page_objects(), page_labels)
+                               )]
+            except PDFNoPageTree:
+                self._pages = [PDFPage(self, objid, properties, label, page_number + 1)
+                               for page_number, ((objid, properties), label) in enumerate(
+                                       zip(self.get_pages_from_xrefs(), page_labels)
+                               )]
+        return self._pages
 
     @property
     def names(self) -> Dict[str, Any]:

playa/pdfstructtree.py

@@ -164,7 +164,7 @@ class PDFStructTree(Findable):
     def __init__(
         self,
         doc: "PDFDocument",
-        pages: Union[Iterable[Tuple[Union[int, None], PDFPage]], None] = None,
+        pages: Union[Iterable[PDFPage], None] = None,
     ):
         if "StructTreeRoot" not in doc.catalog:
             raise PDFNoStructTree("Catalog has no 'StructTreeRoot' entry")
@@ -175,22 +175,18 @@ def __init__(
         self.page_dict: Dict[Any, Union[int, None]]
 
         if pages is None:
-            self.page_dict = {
-                page.pageid: idx + 1 for idx, page in enumerate(doc.pages)
-            }
+            self.page_dict = {page.pageid: page.page_number for page in doc.pages}
             self._parse_struct_tree()
         else:
             pagelist = list(pages)
-            self.page_dict = {
-                page.pageid: page_number for page_number, page in pagelist
-            }
+            self.page_dict = {page.pageid: page.page_number for page in pagelist}
             parent_tree_obj = self.root.get("ParentTree")
             # If we have a single page then we will work backwards from
             # its ParentTree - this is because structure elements could
             # span multiple pages, and the "Pg" attribute is *optional*,
             # so this is the approved way to get a page's structure...
             if len(pagelist) == 1 and parent_tree_obj is not None:
-                _, page = pagelist[0]
+                page = pagelist[0]
                 parent_tree = NumberTree(parent_tree_obj)
                 # If there is no marked content in the structure tree for
                 # this page (which can happen even when there is a

tests/test_open.py

@@ -46,13 +46,12 @@ def test_inline_data() -> None:
     # The necessary mocking would be useless considering that I will
     # shortly demolish these redundant and confusing APIs.
     with playa.open(TESTDIR / "contrib" / "issue-1008-inline-ascii85.pdf") as doc:
-        page = next(doc.pages)
-        _ = page.layout
+        _ = doc.pages[0].layout
 
 
 def test_multiple_contents() -> None:
     with playa.open(TESTDIR / "jo.pdf") as doc:
-        page = next(doc.pages)
+        page = doc.pages[0]
         assert len(page.contents) > 1
         _ = page.layout
 

tests/test_pdfstructtree.py

@@ -13,7 +13,7 @@ class TestClass(unittest.TestCase):
 
     def test_structure_tree_class(self) -> None:
         with playa.open(TESTDIR / "image_structure.pdf") as pdf:
-            stree = PDFStructTree(pdf, [(1, next(pdf.pages))])
+            stree = PDFStructTree(pdf, [pdf.pages[0]])
             doc_elem = next(iter(stree))
             assert [k.type for k in doc_elem] == ["P", "P", "Figure"]
 
@@ -22,7 +22,7 @@ def test_find_all_tree(self) -> None:
         Test find_all() and find() on trees
         """
         with playa.open(TESTDIR / "image_structure.pdf") as pdf:
-            stree = PDFStructTree(pdf, [(1, next(pdf.pages))])
+            stree = PDFStructTree(pdf, [pdf.pages[0]])
             figs = list(stree.find_all("Figure"))
             assert len(figs) == 1
             fig = stree.find("Figure")
@@ -68,8 +68,7 @@ def test_all_mcids(self) -> None:
             assert 1 in page_numbers
             assert 2 in page_numbers
 
-            pages = list(pdf.pages)
-            stree = PDFStructTree(pdf, [(2, pages[1])])
+            stree = PDFStructTree(pdf, [pdf.pages[1]])
             sect = next(stree.find_all("Sect"))
             mcids = list(sect.all_mcids())
             page_numbers = set(page for page, mcid in mcids)