Author: lehmi
Date: Sat Jan 4 16:50:31 2020
New Revision: 1872321
URL: http://svn.apache.org/viewvc?rev=1872321&view=rev
Log:
PDFBOX-4569: refactor preflight parser, split parseObjectDynamically
Modified:
pdfbox/branches/issue4569/preflight/src/main/java/org/apache/pdfbox/preflight/parser/PreflightParser.java
Modified:
pdfbox/branches/issue4569/preflight/src/main/java/org/apache/pdfbox/preflight/parser/PreflightParser.java
URL:
http://svn.apache.org/viewvc/pdfbox/branches/issue4569/preflight/src/main/java/org/apache/pdfbox/preflight/parser/PreflightParser.java?rev=1872321&r1=1872320&r2=1872321&view=diff
==============================================================================
---
pdfbox/branches/issue4569/preflight/src/main/java/org/apache/pdfbox/preflight/parser/PreflightParser.java
(original)
+++
pdfbox/branches/issue4569/preflight/src/main/java/org/apache/pdfbox/preflight/parser/PreflightParser.java
Sat Jan 4 16:50:31 2020
@@ -681,131 +681,146 @@ public class PreflightParser extends PDF
else if (offsetOrObjstmObNr > 0)
{
// offset of indirect object in file
- // ---- go to object start
- source.seek(offsetOrObjstmObNr);
- // ---- we must have an indirect object
- long readObjNr;
- int readObjGen;
-
- long offset = source.getPosition();
- String line = readLine();
- Pattern pattern = Pattern.compile("(\\d+)\\s(\\d+)\\sobj");
- Matcher matcher = pattern.matcher(line);
- if (matcher.matches())
- {
- readObjNr = Long.parseLong(matcher.group(1));
- readObjGen = Integer.parseInt(matcher.group(2));
- }
- else
- {
-
- addValidationError(new
ValidationError(ERROR_SYNTAX_OBJ_DELIMITER, "Single space expected
[offset="+offset+"; key="+offsetOrObjstmObNr.toString()+"; line="+line+";
object="+pdfObject.toString()+"]"));
+ referencedObject = parseFileObject(offsetOrObjstmObNr, objKey);
+ }
+ else
+ {
+ // xref value is object nr of object stream containing object
to be parsed
+ // since our object was not found it means object stream was
not parsed so far
+ referencedObject = parseObjectStreamObject((int)
-offsetOrObjstmObNr, objKey);
+ }
+ if (referencedObject != null && referencedObject != COSNull.NULL)
+ {
+ pdfObject.setObject(referencedObject);
+ }
+ else
+ {
+ pdfObject.setToNull();
+ }
+ }
+ return referencedObject;
+ }
- // reset source cursor to read object information
- source.seek(offset);
- readObjNr = readObjectNumber();
- readObjGen = readGenerationNumber();
- skipSpaces(); // skip spaces between Object Generation
number and the 'obj' keyword
- for (char c : OBJ_MARKER)
- {
- if (source.read() != c)
- {
- addValidationError(new
ValidationError(ERROR_SYNTAX_OBJ_DELIMITER, "Expected pattern '"
- + new String(OBJ_MARKER) + " but missed at
character '" + c + "'"));
- throw new SyntaxValidationException("Expected
pattern '" + new String(OBJ_MARKER)
- + " but missed at character '" + c
+ "'",
- validationResult);
- }
- }
- }
+ private COSBase parseFileObject(Long offsetOrObjstmObNr, final
COSObjectKey objKey)
+ throws IOException
+ {
+ // offset of indirect object in file
+ // ---- go to object start
+ source.seek(offsetOrObjstmObNr);
+ // ---- we must have an indirect object
+ long readObjNr;
+ int readObjGen;
+
+ long offset = source.getPosition();
+ String line = readLine();
+ Pattern pattern = Pattern.compile("(\\d+)\\s(\\d+)\\sobj");
+ Matcher matcher = pattern.matcher(line);
+ if (matcher.matches())
+ {
+ readObjNr = Long.parseLong(matcher.group(1));
+ readObjGen = Integer.parseInt(matcher.group(2));
+ }
+ else
+ {
- // ---- consistency check
- if ((readObjNr != objKey.getNumber()) || (readObjGen !=
objKey.getGeneration()))
+ addValidationError(new ValidationError(ERROR_SYNTAX_OBJ_DELIMITER,
+ "Single space expected [offset=" + offset + "; key="
+ + offsetOrObjstmObNr.toString() + "; line=" + line
+ "; object="
+ + objKey.getNumber() + " " +
objKey.getGeneration() + "]"));
+
+ // reset source cursor to read object information
+ source.seek(offset);
+ readObjNr = readObjectNumber();
+ readObjGen = readGenerationNumber();
+ skipSpaces(); // skip spaces between Object Generation number and
the 'obj' keyword
+ for (char c : OBJ_MARKER)
+ {
+ if (source.read() != c)
{
- throw new IOException("XREF for " + objKey.getNumber() +
":" + objKey.getGeneration()
- + " points to wrong object: " + readObjNr + ":" +
readObjGen);
+ addValidationError(new
ValidationError(ERROR_SYNTAX_OBJ_DELIMITER,
+ "Expected pattern '" + new String(OBJ_MARKER)
+ + " but missed at character '" + c + "'"));
+ throw new SyntaxValidationException("Expected pattern '"
+ + new String(OBJ_MARKER) + " but missed at
character '" + c + "'",
+ validationResult);
}
+ }
+ }
- skipSpaces();
- referencedObject = parseDirObject();
- skipSpaces();
- long endObjectOffset = source.getPosition();
- String endObjectKey = readString();
-
- if (endObjectKey.equals("stream"))
- {
- source.seek(endObjectOffset);
- if (referencedObject instanceof COSDictionary)
- {
- COSStream stream = parseCOSStream((COSDictionary)
referencedObject);
- if (securityHandler != null)
- {
- securityHandler.decryptStream(stream, objNr,
objGenNr);
- }
- referencedObject = stream;
- }
- else
- {
- // this is not legal
- // the combination of a dict and the stream/endstream
forms a complete stream object
- throw new IOException("Stream not preceded by
dictionary (offset: " + offsetOrObjstmObNr + ").");
- }
- skipSpaces();
- endObjectOffset = source.getPosition();
- endObjectKey = readString();
-
- // we have case with a second 'endstream' before endobj
- if (!endObjectKey.startsWith("endobj") &&
endObjectKey.startsWith("endstream"))
- {
- endObjectKey = endObjectKey.substring(9).trim();
- if (endObjectKey.length() == 0)
- {
- // no other characters in extra endstream line
- endObjectKey = readString(); // read next line
- }
- }
- }
- else if (securityHandler != null)
- {
- securityHandler.decrypt(referencedObject, objNr, objGenNr);
- }
- if (!endObjectKey.startsWith("endobj"))
- {
- throw new IOException("Object (" + readObjNr + ":" +
readObjGen + ") at offset "
- + offsetOrObjstmObNr + " does not end with
'endobj'.");
- }
- else
- {
- offset = source.getPosition();
- source.seek(endObjectOffset - 1);
- if (!nextIsEOL())
- {
- addValidationError(new
ValidationError(PreflightConstants.ERROR_SYNTAX_OBJ_DELIMITER,
- "EOL expected before the 'endobj' keyword at
offset "+source.getPosition()));
- }
- source.seek(offset);
- }
+ // ---- consistency check
+ if ((readObjNr != objKey.getNumber()) || (readObjGen !=
objKey.getGeneration()))
+ {
+ throw new IOException("XREF for " + objKey.getNumber() + ":" +
objKey.getGeneration()
+ + " points to wrong object: " + readObjNr + ":" +
readObjGen);
+ }
- if (!nextIsEOL())
+ skipSpaces();
+ COSBase referencedObject = parseDirObject();
+ skipSpaces();
+ long endObjectOffset = source.getPosition();
+ String endObjectKey = readString();
+
+ if (endObjectKey.equals("stream"))
+ {
+ source.seek(endObjectOffset);
+ if (referencedObject instanceof COSDictionary)
+ {
+ COSStream stream = parseCOSStream((COSDictionary)
referencedObject);
+ if (securityHandler != null)
{
- addValidationError(new
ValidationError(PreflightConstants.ERROR_SYNTAX_OBJ_DELIMITER,
- "EOL expected after the 'endobj' keyword at offset
"+source.getPosition()));
+ securityHandler.decryptStream(stream, readObjNr,
readObjGen);
}
+ referencedObject = stream;
}
else
{
- // xref value is object nr of object stream containing object
to be parsed
- // since our object was not found it means object stream was
not parsed so far
- referencedObject = parseObjectStreamObject((int)
-offsetOrObjstmObNr, objKey);
+ // this is not legal
+ // the combination of a dict and the stream/endstream forms a
complete stream object
+ throw new IOException(
+ "Stream not preceded by dictionary (offset: " +
offsetOrObjstmObNr + ").");
}
- if (referencedObject != null && referencedObject != COSNull.NULL)
+ skipSpaces();
+ endObjectOffset = source.getPosition();
+ endObjectKey = readString();
+
+ // we have case with a second 'endstream' before endobj
+ if (!endObjectKey.startsWith("endobj") &&
endObjectKey.startsWith("endstream"))
{
- pdfObject.setObject(referencedObject);
+ endObjectKey = endObjectKey.substring(9).trim();
+ if (endObjectKey.length() == 0)
+ {
+ // no other characters in extra endstream line
+ endObjectKey = readString(); // read next line
+ }
}
- else
+ }
+ else if (securityHandler != null)
+ {
+ securityHandler.decrypt(referencedObject, readObjNr, readObjGen);
+ }
+ if (!endObjectKey.startsWith("endobj"))
+ {
+ throw new IOException("Object (" + readObjNr + ":" + readObjGen +
") at offset "
+ + offsetOrObjstmObNr + " does not end with 'endobj'.");
+ }
+ else
+ {
+ offset = source.getPosition();
+ source.seek(endObjectOffset - 1);
+ if (!nextIsEOL())
{
- pdfObject.setToNull();
+ addValidationError(
+ new
ValidationError(PreflightConstants.ERROR_SYNTAX_OBJ_DELIMITER,
+ "EOL expected before the 'endobj' keyword at
offset "
+ + source.getPosition()));
}
+ source.seek(offset);
+ }
+
+ if (!nextIsEOL())
+ {
+ addValidationError(new
ValidationError(PreflightConstants.ERROR_SYNTAX_OBJ_DELIMITER,
+ "EOL expected after the 'endobj' keyword at offset " +
source.getPosition()));
}
return referencedObject;
}
