Fernando created PDFBOX-5281:
--------------------------------
Summary: Error rendering png from PDF - Error at Type1Parser
parseBinary for Type 1 Font
Key: PDFBOX-5281
URL: https://issues.apache.org/jira/browse/PDFBOX-5281
Project: PDFBox
Issue Type: Bug
Components: FontBox
Affects Versions: 2.0.23
Environment: Component: Type1Parser
Environment: Java 8
Reporter: Fernando
When I try to convert a pdf page using "pdfRenderer.renderImageWithDPI" to a
png image using pdfbox version 2.0.23 I get the following error. The pdf is a
customer specific one, I can't share the original file here.
{code:java}
Error [PDType1Font] Can't read the embedded Type1 font
AAAAAB+NimbusMonoPS-Regular_00
java.io.IOException: Found Token[kind=NAME, text=readonly] but expected def
at org.apache.fontbox.type1.Type1Parser.read(Type1Parser.java:867)
at org.apache.fontbox.type1.Type1Parser.parseBinary(Type1Parser.java:610)
at org.apache.fontbox.type1.Type1Parser.parse(Type1Parser.java:64)
at org.apache.fontbox.type1.Type1Font.createWithSegments(Type1Font.java:85)
at org.apache.pdfbox.pdmodel.font.PDType1Font.<init>(PDType1Font.java:263)
at
org.apache.pdfbox.pdmodel.font.PDFontFactory.createFont(PDFontFactory.java:76)
at org.apache.pdfbox.pdmodel.PDResources.getFont(PDResources.java:146)
{code}
Using the PDFDebugger to have access to the FontFile of the page throwing the
error, "Save the Stream as PFB", and using the t1utils t1disasm to obtain the
readable equivalent of the binary private dictionary, which contains:
{code:java}
/Private 10 dict dup begin
/RD {string currentfile exch readstring pop} readonly def ←
/ND {noaccess def} readonly def
/NP {noaccess put} readonly def
/MinFeature {16 16} readonly def
{code}
And following the trace in PDFBox I can see it blowing up at "Type1Parser" in
the "parseBinary" method at line 602 they have the check for
""RD".equals(key)", where key is "readonly" and the last check from that list
"read(Token.NAME, "def");" has to be def, and because the key is "readonly" it
throws the error, even if it makes it pass "RD" it will have the same results
when parsing the contents for "ND" and "NP".
I have seen most of the Type 1 Font files have "executeonly" instead of
"readonly" in the /Private dict section, even this same file if I convert it
first to PS and back to PDF, extracting the font again I can see that the
instructions for RD, ND, and NP are rearranged to be "executeonly" using Mac
Preview or GS ps2pdf., and the PNG is generated without issue using the
PDFRenderer, I don't see a way to do this step programmatically at the moment.
{code:java}
dup /Private 20 dict dup begin
/-| {string currentfile exch readstring pop} executeonly def ←
/|- {noaccess def} executeonly def
/| {noaccess put} executeonly def
{code}
>From the Type 1 Font Spec, they don't provide a must follow receipt on what
>instructions can be appended after RD, ND, or NP, they state: "The RD, NP, and
>ND functions must be implemented by PostScript language procedures" , If we
>take a look to the PostScript Language Reference Manual:
??readonly: When an object is read-only, its value cannot be modified by
PostScript operators (an invalidaccess error will result), but it can still be
read by operators or executed by the PostScript interpreter.??
??executeonly: When an object is execute-only, its value cannot be read or
modified explicitly by PostScript operators (an invalidaccess error will
result), but it can still be executed by the PostScript interpreter—for
example, by invoking it with exec??
Both "readonly" and "executeonly" allows the instructions to be executed.
Suggestion:
# Would it be possible to add an optional "readMaybe(Token.NAME,
"readonly");" to the "parseBinary" RD, "ND", and "NP" keywords similar to how
it was done at PDFBOX-2202.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
