From 66c875209a9b85fd80875f44a3ae5b92548639ba Mon Sep 17 00:00:00 2001
From: Lars-Dominik Braun <lars@6xq.net>
Date: Sun, 3 Mar 2019 13:18:58 +0100
Subject: Fix Hintergrund extraction

The script added a bogus/empty page at offset 0x600, since only three
instead of four entries of the first page table were skipped.

To make sure offsets are correct, add the first page (not stored on
Hintergrund) to the output as well. FILE dataspace conversion can skip
this page with `--skip 1`.
---
 convertFileDs.py      | 7 +++++--
 extractHintergrund.py | 7 ++++++-
 2 files changed, 11 insertions(+), 3 deletions(-)

diff --git a/convertFileDs.py b/convertFileDs.py
index 89e0cab..d5c234b 100755
--- a/convertFileDs.py
+++ b/convertFileDs.py
@@ -9,7 +9,7 @@ datastructure here. See EUMEL packet “file handling”.
 
 import struct, copy
 from collections import namedtuple
-from eumel import Dataspace, DataspaceTypeMismatch, HeapReferenceUnresolved
+from eumel import Dataspace, DataspaceTypeMismatch, HeapReferenceUnresolved, pagesize
 
 Segment = namedtuple ('Segment', ['succ', 'pred', 'end'])
 Sequence = namedtuple ('Sequence', ['index', 'segmentbegin', 'segmentend', 'lineno', 'lines'])
@@ -77,12 +77,13 @@ class FileDataspace (Dataspace):
         Dataspace.__init__ (self, fd)
 
         # header of the BOUND LIST (aka TYPE FILE)
+        start = fd.tell ()
         self.used = self.parseSequence ()
         self.parseInt (2)
         self.parseSequence ()
         self.parseSequence ()
         self.parseInt (7)
-        assert self.fd.tell () == 0x38
+        assert self.fd.tell ()-start == 0x30
 
         rows = self.parseRows ()
 
@@ -150,6 +151,7 @@ if __name__ == '__main__':
     
     parser = argparse.ArgumentParser(description='Convert EUMEL FILE dataspace into plain text file.')
     parser.add_argument ('-v', '--verbose', help='Enable debugging messages', action='store_true')
+    parser.add_argument ('-s', '--skip', metavar='PAGES', type=int, default=0, help='Skip pages at the beginning of the file')
     parser.add_argument ('file', help='Input file')
     args = parser.parse_args ()
 
@@ -160,6 +162,7 @@ if __name__ == '__main__':
 
     with open (args.file, 'rb') as fd:
         try:
+            fd.seek (args.skip*pagesize)
             ds = FileDataspace (fd)
             linecount = len (ds.text.splitlines ())
             if linecount != ds.used.lines:
diff --git a/extractHintergrund.py b/extractHintergrund.py
index 8f178a0..5795d8d 100755
--- a/extractHintergrund.py
+++ b/extractHintergrund.py
@@ -44,6 +44,7 @@ anchor = Struct(
     Const(b"\xff"*4),
     ) * "System anchor block"
 
+assert pagesize//blockref.sizeof() == 128
 blockTable = Array(pagesize//blockref.sizeof(), blockref)
 
 # XXX: skip const
@@ -212,6 +213,10 @@ if __name__ == '__main__':
                         with open (f'{taskid:04d}_{dsid:04d}.ds', 'wb') as outfd:
                             os.ftruncate (outfd.fileno(), 0)
 
+                            # the first page of a dataspace is used by the OS
+                            # and not stored to the Hintergrund
+                            outfd.seek (pagesize)
+
                             # get the first three pages
                             for ref in d.blocks:
                                 copyblock (ref.value, fd, outfd)
@@ -219,7 +224,7 @@ if __name__ == '__main__':
                             # indirect block refs (level 4a)
                             assert len (d.blockTables) == 2
                             # first four entries of first table are empty and must not be written!
-                            copyBlockTable (d.blockTables[0].value, fd, outfd, 3)
+                            copyBlockTable (d.blockTables[0].value, fd, outfd, 4)
                             copyBlockTable (d.blockTables[1].value, fd, outfd)
 
                             # segment tables (level 4b)
-- 
cgit v1.2.3