PDF::Builder::Basic::PDF::File - Holds the trailers and
cross-reference tables for a PDF file
$p = PDF::Builder::Basic::PDF::File->open("filename.pdf", 1);
$p->new_obj($obj_ref);
$p->free_obj($obj_ref);
$p->append_file();
$p->close_file();
$p->release(); # IMPORTANT!
This class keeps track of the directory aspects of a PDF file.
There are two parts to the directory: the main directory object, which is
the parent to all other objects, and a chain of cross-reference tables and
corresponding trailer dictionaries, starting with the main directory
object.
Within this class hierarchy, rather than making everything visible
via methods, which would be a lot of work, there are various instance
variables which are accessible via associative array referencing. To
distinguish instance variables from content variables (which may come from
the PDF content itself), each such variable name will start with a
space.
Variable names which do not start with a space directly reflect
elements in a PDF dictionary. In the case of a
"PDF::Builder::Basic::PDF::File", the
elements reflect those in the trailer dictionary.
Since some variables are not designed for class users to access,
variables are marked in the documentation with (R) to indicate that
such an entry should only be used as read-only information.
(P) indicates that the information is private, and not
designed for user use at all, but is included in the documentation for
completeness and to ensure that nobody else tries to use it.
- newroot
- This variable allows the user to create a new root entry to occur in the
trailer dictionary which is output when the file is written or appended.
If you wish to override the root element in the dictionary you have, use
this entry to indicate that without losing the current Root entry. Notice
that newroot should point to a PDF level object and not just to a
dictionary, which does not have object status.
- INFILE (R)
- Contains the filehandle used to read this information into this PDF
directory. It is an IO object.
- fname (R)
- This is the filename which is reflected by INFILE, or the original IO
object passed in.
- update (R)
- This indicates that the read file has been opened for update and that at
some point, "$p->appendfile()" can be
called to update the file with the changes that have been made to the
memory representation.
- maxobj (R)
- Contains the first usable object number above any that have already
appeared in the file so far.
- outlist (P)
- This is a list of Objind which are to be output when the next
"appendfile()" or
"outfile()" occurs.
- firstfree
(P)
- Contains the first free object in the free object list. Free objects are
removed from the front of the list and added to the end.
- lastfree (P)
- Contains the last free object in the free list. It may be the same as the
"firstfree" if there is only one free
object.
- objcache (P)
- All objects are held in the cache to ensure that a system only has one
occurrence of each object. In effect, the objind class acts as a container
type class to hold the PDF object structure, and it would be unfortunate
if there were two identical place-holders floating around a system.
- epos (P)
- The end location of the read-file.
Each trailer dictionary contains a number of private instance
variables which hold the chain together.
- loc (P)
- Contains the location of the start of the cross-reference table preceding
the trailer.
- xref (P)
- Contains an anonymous array of each cross-reference table entry.
- prev (P)
- A reference to the previous table. Note this differs from the Prev entry
which is in PDF, which contains the location of the previous
cross-reference table.
PDF::Builder::Basic::PDF::File->new()
Creates a new, empty file object which can act as the
host to other PDF objects. Since there is no file associated with this object,
it is assumed that the object is created in readiness for creating a new PDF
file.
$p = PDF::Builder::Basic::PDF::File->open($filename, $update, %options)
Opens the file and reads all the trailers and cross
reference tables to build a complete directory of objects.
$filename may be a string or an IO
object.
$update specifies whether this file is
being opened for updating and editing (TRUE value), or simply to be
read (FALSE or undefined value).
%options may include
- diags => 1
- If "diags" is set to 1, various warning
messages will be given if a suspicious PDF structure is found, and some
fixup may be attempted. There is no guarantee that any fixup will change
the PDF to legitimate, or that there won't be other problems found further
down the line. If this flag is not given, and a structural problem
is found, it is fairly likely that errors (and even a program
crash) may happen further along. If you experience crashes when
reading in a PDF file, try running with
"diags" and see what is reported.
There are many PDF files out "in the wild" which,
while failing to conform to Adobe's standards, appear to be tolerated by
PDF Readers. Thus, Builder will no longer fail on them, but merely
comment on their existence.
$new_version = $p->version($version, %opts) # Set
$ver = $p->version() # Get
Gets/sets the PDF version (e.g., 1.5). Setting sets both
the header and trailer versions. Getting returns the higher of header and
trailer versions.
For compatibility with earlier releases, if no decimal point is
given, assume "1." precedes the number given.
A warning message is given if you attempt to decrease the
PDF version, as you might have already read in a higher level file, or used
a higher level feature. This message is suppressed if the 'silent' option is
given with any value.
$new_version = $p->header_version($version, %opts) # Set
$version = $p->header_version() # Get
Gets/sets the PDF version stored in the file header.
For compatibility with earlier releases, if no decimal point is
given, assume "1." precedes the number given.
A warning message is given if you attempt to decrease the
PDF version, as you might have already read in a higher level file, or used
a higher level feature. This message is suppressed if the 'silent' option is
given with any value.
$new_version = $p->trailer_version($version, %opts) # Set
$version = $p->trailer_version() # Get
Gets/sets the PDF version stored in the document catalog.
Note that the minimum PDF level for a trailer version is 1.4. It
is not permitted to set a PDF level of 1.3 or lower. An existing PDF (read
in) of 1.3 or below returns undefined.
For compatibility with earlier releases, if no decimal point is
given, assume "1." precedes the number given.
A warning message is given if you attempt to decrease the
PDF version, as you might have already read in a higher level file, or used
a higher level feature. This message is suppressed if the 'silent' option is
given with any value.
$prev_version = $p->require_version($version)
Ensures that the PDF version is at least
$version. Silently sets the version to the higher
level.
$p->release()
Releases ALL of the memory used by the PDF document and
all of its component objects. After calling this method, do
NOT expect
to have anything left in the
"PDF::Builder::Basic::PDF::File" object (so
if you need to save, be sure to do it before calling this method).
NOTE, that it is important that you call this method on any
"PDF::Builder::Basic::PDF::File" object
when you wish to destroy it and free up its memory. Internally, PDF files
have an enormous number of cross-references, and this causes circular
references within the internal data structures. Calling
"release()" causes a brute-force cleanup
of the data structures, freeing up all of the memory. Once you've called
this method, though, don't expect to be able to do anything else with the
"PDF::Builder::Basic::PDF::File" object;
it'll have no internal state whatsoever.
$p->append_file()
Appends the objects for output to the read file and then
appends the appropriate table.
$p->out_file($fname)
Writes a PDF file to a file of the given filename, based
on the current list of objects to be output. It creates the trailer dictionary
based on information in
$self.
$fname may be a string or an IO
object.
$p->create_file($fname)
Creates a new output file (no check is made of an
existing open file) of the given filename or IO object. Note: make sure that
"$p->{' version'}" is set correctly
before calling this function.
$p->close_file()
Closes up the open file for output, by outputting the
trailer, etc.
($value, $str) = $p->readval($str, %opts)
Reads a PDF value from the current position in the file.
If
$str is too short, read some more from the current
location in the file until the whole object is read. This is a recursive call
which may slurp in a whole big stream (unprocessed).
Returns the recursive data structure read and also the current
$str that has been read from the file.
$ref = $p->read_obj($objind, %opts)
Given an indirect object reference, locate it and read
the object returning the read-in object.
$ref = $p->read_objnum($num, $gen, %opts)
Returns a fully read object of given number and
generation in this file
$objind = $p->new_obj($obj)
Creates a new, free object reference based on free space
in the cross reference chain. If nothing is free, then think up a new number.
If $obj, then turns that object into this new object
rather than returning a new object.
$p->out_obj($obj)
Indicates that the given object reference should appear
in the output xref table whether with data or freed.
$p->free_obj($obj)
Marks an object reference for output as being
freed.
$p->remove_obj($objind)
Removes the object from all places where we might
remember it.
$p->ship_out(@objects)
$p->ship_out()
Ships the given objects (or all objects for output if
@objects is empty) to the currently open output file
(assuming there is one). Freed objects are not shipped, and once an object is
shipped it is switched such that this file becomes its source and it will not
be shipped again unless out_obj is called again. Notice that a shipped out
object can be re-output or even freed, but that it will not cause the data
already output to be changed.
$p->copy($outpdf, \&filter)
Iterates over every object in the file reading the
object, calling "filter" with the object,
and outputting the result. If "filter" is
not defined, just copies input to output.
The following methods and functions are considered private
to this class. This does not mean you cannot use them if you have a need,
just that they aren't really designed for users of this class.
$offset = $p->locate_obj($num, $gen)
Returns a file offset to the object asked for by
following the chain of cross reference tables until it finds the one you
want.
update($fh, $str, $instream)
Keeps reading $fh for more data
to ensure that $str has at least a line full for
"readval" to work on. At this point we also
take the opportunity to ignore comments.
$objind = $p->test_obj($num, $gen)
Tests the cache to see whether an object reference (which
may or may not have been getobj()ed) has been cached. Returns it if it
has.
$p->add_obj($objind)
Adds the given object to the internal object cache.
$tdict = $p->readxrtr($xpos, %options)
Recursive function which reads each of the
cross-reference and trailer tables in turn until there are no more.
Returns a dictionary corresponding to the trailer chain. Each
trailer also includes the corresponding cross-reference table.
The structure of the xref private element in a trailer dictionary
is of an anonymous hash of cross reference elements by object number. Each
element consists of an array of 3 elements corresponding to the three
elements read in [location, generation number, free or used]. See the PDF
specification for details.
See "open" for options
allowed.
$p->out_trailer($tdict, $update)
$p->out_trailer($tdict)
Outputs the body and trailer for a PDF file by outputting
all the objects in the ' outlist' and then outputting a xref table for those
objects and any freed ones. It then outputs the trailing dictionary and the
trailer code.
PDF::Builder::Basic::PDF::File->_new()
Creates a very empty PDF file object (used by
new() and open())
Martin Hosken Martin_Hosken@sil.org
Copyright Martin Hosken 1999
No warranty or expression of effectiveness, least of all regarding
anyone's safety, is implied in this software or documentation.