Opened 7 months ago

Closed 7 months ago

Last modified 6 weeks ago

#14707 closed enhancement (fixed)


Reported by: Bruce Dubbs Owned by: Douglas R. Reno
Priority: normal Milestone: 11.0
Component: BOOK Version: SVN
Severity: normal Keywords:


New minor version.

Change History (5)

comment:1 by Douglas R. Reno, 7 months ago

Owner: changed from blfs-book to Douglas R. Reno
Status: newassigned

comment:2 by Douglas R. Reno, 7 months ago

10.2.0: February 23, 2021

        CLI Behavior Changes

            Operations that work on combining pages are much better about protecting form fields. In particular, --split-pages and --pages now preserve interaction form functionality by copying the relevant form field information from the original files. Additionally, if you use --pages to select only some pages from the original input file, unused form fields are removed, which prevents lots of unused annotations from being retained.

            By default, qpdf no longer allows creation of encrypted PDF files whose user password is non-empty and owner password is empty when a 256-bit key is in use. The --allow-insecure option, specified inside the --encrypt options, allows creation of such files. Behavior changes in the CLI are avoided when possible, but an exception was made here because this is security-related. qpdf must always allow creation of weird files for testing purposes, but it should not default to letting users unknowingly create insecure files. 

        Library Behavior Changes

            Note: the changes in this section cause differences in output in some cases. These differences change the syntax of the PDF but do not change the semantics (meaning). I make a strong effort to avoid gratuitous changes in qpdf's output so that qpdf changes don't break people's tests. In this case, the changes significantly improve the readability of the generated PDF and don't affect any output that's generated by simple transformation. If you are annoyed by having to update test files, please rest assured that changes like this have been and will continue to be rare events.

            QPDFObjectHandle::newUnicodeString now uses whichever of ASCII, PDFDocEncoding, of UTF-16 is sufficient to encode all the characters in the string. This reduces needless encoding in UTF-16 of strings that can be encoded in ASCII. This change may cause qpdf to generate different output than before when form field values are set using QPDFFormFieldObjectHelper but does not change the meaning of the output.

            The code that places form XObjects and also the code that flattens rotations trim trailing zeroes from real numbers that they calculate. This causes slight (but semantically equivalent) differences in generated appearance streams and form XObject invocations in overlay/underlay code or in user code that calls the methods that place form XObjects on a page. 

        CLI Enhancements

            Add new command line options for listing, saving, adding, removing, and and copying file attachments. See Section 3.7, “Embedded Files/Attachments Options” for details.

            Page splitting and merging operations, as well as --flatten-rotation, are better behaved with respect to annotations and interactive form fields. In most cases, interactive form field functionality and proper formatting and functionality of annotations is preserved by these operations. There are still some cases that aren't perfect, such as when functionality of annotations depends on document-level data that qpdf doesn't yet understand or when there are problems with referential integrity among form fields and annotations (e.g., when a single form field object or its associated annotations are shared across multiple pages, a case that is out of spec but that works in most viewers anyway).

            The option --password-file=filename can now be used to read the decryption password from a file. You can use - as the file name to read the password from standard input. This is an easier/more obvious way to read passwords from files or standard input than using @file for this purpose.

            Add some information about attachments to the json output, and added attachments as an additional json key. The information included here is limited to the preferred name and content stream and a reference to the file spec object. This is enough detail for clients to avoid the hassle of navigating a name tree and provides what is needed for basic enumeration and extraction of attachments. More detailed information can be obtained by following the reference to the file spec object.

            Add numeric option to --collate. If --collate=n is given, take pages in groups of n from the given files.

            It is now valid to provide --rotate=0 to clear rotation from a page. 

        Library Enhancements

            This release includes numerous additions to the API. Not all changes are listed here. Please see the ChangeLog file in the source distribution for a comprehensive list. Highlights appear below.

            Add QPDFObjectHandle::ditems() and QPDFObjectHandle::aitems() that enable C++-style iteration, including range-for iteration, over dictionary and array QPDFObjectHandles. See comments in include/qpdf/QPDFObjectHandle.hh and examples/ for details.

            Add QPDFObjectHandle::copyStream for making a copy of a stream within the same QPDF instance.

            Add new helper classes for supporting file attachments, also known as embedded files. New classes are QPDFEmbeddedFileDocumentHelper, QPDFFileSpecObjectHelper, and QPDFEFStreamObjectHelper. See their respective headers for details and examples/ for an example.

            Add a version of QPDFObjectHandle::parse that takes a QPDF pointer as context so that it can parse strings containing indirect object references. This is illustrated in examples/

            Re-implement QPDFNameTreeObjectHelper and QPDFNumberTreeObjectHelper to be more efficient, add an iterator-based API, give them the capability to repair broken trees, and create methods for modifying the trees. With this change, qpdf has a robust read/write implementation of name and number trees.

            Add new versions of QPDFObjectHandle::replaceStreamData that take std::function objects for cases when you need something between a static string and a full-fledged StreamDataProvider. Using this with QUtil::file_provider is a very easy way to create a stream from the contents of a file.

            The QPDFMatrix class, formerly a private, internal class, has been added to the public API. See include/qpdf/QPDFMatrix.hh for details. This class is for working with transformation matrices. Some methods in QPDFPageObjectHelper make use of this to make information about transformation matrices available. For an example, see examples/

            Several new methods were added to QPDFAcroFormDocumentHelper for adding, removing, getting information about, and enumerating form fields.

            Add method QPDFAcroFormDocumentHelper::transformAnnotations, which applies a transformation to each annotation on a page.

            Add QPDFPageObjectHelper::copyAnnotations, which copies annotations and, if applicable, associated form fields, from one page to another, possibly transforming the rectangles. 

        Build Changes

            A C++-14 compiler is now required to build qpdf. There is no intention to require anything newer than that for a while. C++-14 includes modest enhancements to C++-11 and appears to be supported about as widely as C++-11. 

        Bug Fixes

            The --flatten-rotation option applies transformations to any annotations that may be on the page.

            If a form XObject lacks a resources dictionary, consider any names in that form XObject to be referenced from the containing page. This is compliant with older PDF versions. Also detect if any form XObjects have any unresolved names and, if so, don't remove unreferenced resources from them or from the page that contains them. Unfortunately this has the side effect of preventing removal of unreferenced resources in some cases where names appear that don't refer to resources, such as with tagged PDF. This is a bit of a corner case that is not likely to cause a significant problem in practice, but the only side effect would be lack of removal of shared resources. A future version of qpdf may be more sophisticated in its detection of names that refer to resources.

            Properly handle strings if they appear in inline image dictionaries while externalizing inline images. 

comment:3 by Douglas R. Reno, 7 months ago

If you're doing an update to this package and have qpdf installed already, remember to run '' first before running the test suite. Otherwise you'll get undefined symbol errors from libtool because the build process will attempt to use the old version.

comment:4 by Douglas R. Reno, 7 months ago

Resolution: fixed
Status: assignedclosed

Fixed at r24333

comment:5 by Bruce Dubbs, 6 weeks ago

Milestone: 10.211.0

Milestone renamed

Note: See TracTickets for help on using tickets.