From eecbd5106852746ec44f8b78f7b64fb944559f6e Mon Sep 17 00:00:00 2001
From: David Huggins-Daines <dhd@ecolingui.ca>
Date: Mon, 2 Dec 2024 10:04:42 -0500
Subject: [PATCH] fix: robustness to more pdf.js test cases

---
 playa/document.py |  2 +-
 playa/page.py     | 36 +++++++++++++++++++++---------------
 playa/utils.py    | 13 -------------
 3 files changed, 22 insertions(+), 29 deletions(-)

diff --git a/playa/document.py b/playa/document.py
index 307e038..171a892 100644
--- a/playa/document.py
+++ b/playa/document.py
@@ -853,7 +853,7 @@ def __init__(
         try:
             pos = self._find_xref()
             self._read_xref_from(pos, self.xrefs)
-        except (ValueError, IndexError) as e:
+        except (ValueError, IndexError, PDFSyntaxError) as e:
             log.debug("Using fallback XRef parsing: %s", e)
             newxref = XRefFallback(self.parser)
             self.xrefs.append(newxref)
diff --git a/playa/page.py b/playa/page.py
index 75a86a4..c6cf6d5 100644
--- a/playa/page.py
+++ b/playa/page.py
@@ -36,6 +36,7 @@
 from playa.exceptions import (
     PDFInterpreterError,
     PDFUnicodeNotDefined,
+    PDFSyntaxError,
 )
 from playa.font import Font
 
@@ -67,7 +68,6 @@
     get_transformed_bound,
     make_compat_bytes,
     mult_matrix,
-    parse_rect,
     normalize_rect,
     translate_matrix,
 )
@@ -98,6 +98,17 @@ def Object(*args, **kwargs): ...
 DeviceSpace = Literal["page", "screen", "user"]
 
 
+# FIXME: This should go in utils/pdftypes but there are circular imports
+def parse_rect(o: PDFObject) -> Rect:
+    try:
+        (x0, y0, x1, y1) = (num_value(x) for x in list_value(o))
+        return x0, y0, x1, y1
+    except ValueError:
+        raise ValueError("Could not parse rectangle %r" % (o,))
+    except TypeError:
+        raise PDFSyntaxError("Rectangle contains non-numeric values")
+
+
 class Page:
     """An object that holds the information about a page.
 
@@ -141,9 +152,7 @@ def __init__(
             self.attrs.get("Resources", {})
         )
         if "MediaBox" in self.attrs:
-            self.mediabox = normalize_rect(
-                parse_rect(resolve1(val) for val in resolve1(self.attrs["MediaBox"]))
-            )
+            self.mediabox = normalize_rect(parse_rect(self.attrs["MediaBox"]))
         else:
             log.warning(
                 "MediaBox missing from /Page (and not inherited),"
@@ -153,9 +162,7 @@ def __init__(
         self.cropbox = self.mediabox
         if "CropBox" in self.attrs:
             try:
-                self.cropbox = normalize_rect(
-                    parse_rect(resolve1(val) for val in resolve1(self.attrs["CropBox"]))
-                )
+                self.cropbox = normalize_rect(parse_rect(self.attrs["CropBox"]))
             except ValueError:
                 log.warning("Invalid CropBox in /Page, defaulting to MediaBox")
 
@@ -1213,11 +1220,10 @@ def __iter__(self) -> Iterator[LayoutDict]:
                         if len(args) == nargs:
                             gen = method(*args)
                         else:
-                            error_msg = (
-                                "Insufficient arguments (%d) for operator: %r"
-                                % (len(args), obj)
+                            log.warning(
+                                "Insufficient arguments (%d) for operator: %r",
+                                len(args), obj
                             )
-                            raise PDFInterpreterError(error_msg)
                     else:
                         log.debug("exec: %r", obj)
                         gen = method()
@@ -2126,17 +2132,17 @@ def __iter__(self) -> Iterator[ContentObject]:
                         if len(args) == nargs:
                             gen = method(*args)
                         else:
-                            error_msg = (
-                                "Insufficient arguments (%d) for operator: %r"
-                                % (len(args), obj)
+                            log.warning(
+                                "Insufficient arguments (%d) for operator: %r",
+                                len(args), obj
                             )
-                            raise PDFInterpreterError(error_msg)
                     else:
                         log.debug("exec: %r", obj)
                         gen = method()
                     if gen is not None:
                         yield from gen
                 else:
+                    # TODO: This can get very verbose
                     log.warning("Unknown operator: %r", obj)
             else:
                 self.push(obj)
diff --git a/playa/utils.py b/playa/utils.py
index cb3b7f4..accc7a6 100644
--- a/playa/utils.py
+++ b/playa/utils.py
@@ -2,7 +2,6 @@
 
 import string
 from typing import (
-    Any,
     Iterable,
     Iterator,
     List,
@@ -11,8 +10,6 @@
     Union,
 )
 
-from playa.exceptions import PDFSyntaxError
-
 
 def make_compat_bytes(in_str: str) -> bytes:
     """Converts to bytes, encoding to unicode."""
@@ -180,16 +177,6 @@ def apply_png_predictor(
 MATRIX_IDENTITY: Matrix = (1, 0, 0, 1, 0, 0)
 
 
-def parse_rect(o: Any) -> Rect:
-    try:
-        (x0, y0, x1, y1) = o
-        return float(x0), float(y0), float(x1), float(y1)
-    except ValueError:
-        raise ValueError("Could not parse rectangle")
-    except TypeError:
-        raise PDFSyntaxError("Rectangle contains non-numeric values")
-
-
 def normalize_rect(r: Rect) -> Rect:
     (x0, y0, x1, y1) = r
     if x1 < x0: