I have a PDF where the page contents directory contains two streams. The first one consists of
{code}
/Basemap_Form Do
{code}
the second one starts with
{code}
q
1.0 0.0 0.0 1.0 103.680000 51.840000 cm
/LGIT:W Do
Q
{code}

OContentParser#parse receives these streams as two byte[] objects which are then concatenated using a ByteDoubleArrayInputStream which presents the byte[]s as a single concatenated InputStream. This is causes incorrect parsing.

During parsing the parser sees the following sequence of tokens
{code}
/Basemap_Form
Doq
1.0
0.0
...
{code}

The Doq token is the result of concatenating the two streams without introducing a white-space character to separate the Do and q tokens. The PDF spec on Page/Contents states that the division between streams is always at a lexical token boundary, so the parser needs to insert a token boundary between the streams somehow.

Using org.icepdf.core.io.SequenceInputStream with a ' ' separator character resolves the parsing problem.