fix: mypy

playa/document.py

@@ -248,11 +248,15 @@ def _load(self, parser: IndirectObjectParser) -> None:
         for pos, line in reverse_iter_lines(parser.buffer):
             line = line.strip()
             if line == b"trailer":
-                _, trailer = next(ObjectParser(parser.buffer, doc, pos + len(b"trailer")))
+                _, trailer = next(
+                    ObjectParser(parser.buffer, doc, pos + len(b"trailer"))
+                )
+                if not isinstance(trailer, dict):
+                    break
                 self.trailer.update(trailer)
                 log.debug("trailer=%r", self.trailer)
                 return
-        log.warning("b'trailer' not found in document")
+        log.warning("b'trailer' not found in document or invalid")
 
     @property
     def objids(self) -> Iterable[int]:
@@ -1311,16 +1315,16 @@ def _read_xref_from(
         log.debug("read_xref_from: start=%d, token=%r", start, token)
         if token is KEYWORD_XREF:
             parser.nextline()
-            xref = XRefTable(parser)
+            xref: XRef = XRefTable(parser)
         else:
             # It might be an XRefStream, if this is an indirect object...
-            _, token2 = parser.nexttoken()
-            _, token3 = parser.nexttoken()
-            if token3 is KEYWORD_OBJ:
+            _, token = parser.nexttoken()
+            _, token = parser.nexttoken()
+            if token is KEYWORD_OBJ:
                 # XRefStream: PDF-1.5
                 self.parser.seek(pos)
                 self.parser.reset()
-                xref: XRef = XRefStream(self.parser)
+                xref = XRefStream(self.parser)
             else:
                 # Well, maybe it's an XRef table without "xref" (but
                 # probably not)
@@ -1354,7 +1358,9 @@ def __init__(self, doc: Document):
             page_objects = list(doc._get_page_objects())
         except (KeyError, IndexError):
             page_objects = list(doc._get_pages_from_xrefs())
-        for page_idx, ((objid, properties), label) in enumerate(zip(page_objects, page_labels)):
+        for page_idx, ((objid, properties), label) in enumerate(
+            zip(page_objects, page_labels)
+        ):
             page = Page(doc, objid, properties, label, page_idx, doc.space)
             self._pages.append(page)
             if label is not None:

playa/page.py

@@ -159,7 +159,8 @@ def __init__(
         else:
             log.warning(
                 "MediaBox missing from Page id %d (and not inherited),"
-                " defaulting to US Letter (612x792)", pageid
+                " defaulting to US Letter (612x792)",
+                pageid,
             )
             self.mediabox = (0, 0, 612, 792)
         self.cropbox = self.mediabox

playa/structtree.py

@@ -20,7 +20,7 @@
 )
 
 from playa.data_structures import NumberTree
-from playa.parser import KEYWORD_NULL, PSLiteral
+from playa.parser import KEYWORD_NULL, PSLiteral, PDFObject
 from playa.pdftypes import ObjRef, resolve1, dict_value, list_value
 from playa.utils import decode_text
 
@@ -190,20 +190,21 @@ def __init__(
                     return
                 parent_id = page.attrs["StructParents"]
                 parent_array = list_value(parent_tree[parent_id])
+                assert isinstance(parent_array, list)  # srsly
                 self._parse_parent_tree(parent_array)
             else:
                 # ...EXCEPT that the ParentTree is sometimes missing, in which
                 # case we fall back to the non-approved way.
                 self._parse_struct_tree()
 
     def _make_attributes(
-        self, obj: Dict[str, Any], revision: Union[int, None]
+        self, attrs: Dict[str, Any], revision: Union[int, None]
     ) -> Dict[str, Any]:
-        attr_obj_list = []
+        attr_obj_list: List[PDFObject] = []
         for key in "C", "A":
-            if key not in obj:
+            if key not in attrs:
                 continue
-            attr_obj = resolve1(obj[key])
+            attr_obj = resolve1(attrs[key])
             # It could be a list of attribute objects (why?)
             if isinstance(attr_obj, list):
                 attr_obj_list.extend(resolve1(val) for val in attr_obj)
@@ -221,10 +222,12 @@ def _make_attributes(
                 if aobj == revision and prev_obj is not None:
                     attr_objs.append(prev_obj)
                 prev_obj = None
-            else:
+            elif isinstance(aobj, dict):
                 if prev_obj is not None:
                     attr_objs.append(prev_obj)
                 prev_obj = aobj
+            else:
+                logger.warning("Structure attribute of unknown type: %r", aobj)
         if prev_obj is not None:
             attr_objs.append(prev_obj)
         # Now merge all the attribute objects in the collected to a

tests/test_open.py

@@ -13,7 +13,7 @@
 except ImportError:
     pdfminer = None  # type: ignore
 import playa
-from playa.exceptions import PDFEncryptionError, PDFSyntaxError
+from playa.exceptions import PDFEncryptionError
 from .data import TESTDIR, BASEPDFS, PASSWORDS, XFAILS, CONTRIB
 
 # We know pdfminer.six gives different output for these and we don't