MAINT: Validate PDF magic byte in strict mode (#814)

Closes #626

PyPDF2/pdf.py

@@ -1831,6 +1831,12 @@ def read(self, stream):
         stream.seek(-1, 2)
         if not stream.tell():
             raise PdfReadError('Cannot read an empty file')
+        if self.strict:
+            stream.seek(0, 0)
+            header_byte = stream.read(5)
+            if header_byte != b"%PDF-":
+                raise PdfReadError("PDF starts with '{}', but '%PDF-' expected".format(header_byte.decode("utf8")))
+            stream.seek(-1, 2)
         last1M = stream.tell() - 1024 * 1024 + 1 # offset of last MB of stream
         line = b_('')
         while line[:5] != b_("%%EOF"):

Tests/test_papersizes.py

@@ -10,14 +10,13 @@ def test_din_a0():
     area_square_inch = area_square_pixels / 72**2
 
     # 25.4 millimeter is equal to 1 inches
-    area_square_mm = area_square_inch * (25.4)**2
+    area_square_mm = area_square_inch * (25.4) ** 2
     assert abs(area_square_mm - 999949) < 100
     conversion_factor = 72 / 25.4
     assert (dim.width - 841 * conversion_factor) < 1
     assert (dim.width - 1189 * conversion_factor) < 1
 
 
-
 @pytest.mark.parametrize("dimensions", papersizes._din_a)
 def test_din_a_ratio(dimensions):
     assert abs(dimensions.height - dimensions.width * 2**0.5) <= 2.5

Tests/test_reader.py

@@ -338,9 +338,15 @@ def test_read_empty():
     assert exc.value.args[0] == "Cannot read an empty file"
 
 
-def test_read_malformed():
+def test_read_malformed_header():
     with pytest.raises(PdfReadError) as exc:
         PdfFileReader(io.BytesIO(b"foo"))
+    assert exc.value.args[0] == "PDF starts with 'foo', but '%PDF-' expected"
+
+
+def test_read_malformed_body():
+    with pytest.raises(PdfReadError) as exc:
+        PdfFileReader(io.BytesIO(b"%PDF-"))
     assert exc.value.args[0] == "Could not read malformed PDF file"