Recently I tried to use this package to extract images from a pdf file. I am on Windows and using mupdf version 2.1.1. I can get the list of images using the "$page images list" command and it seems to be correct. But wen it comes to saving them ($page images extract ...), the images are not extracted. I have tried different options available including with and without image id's, output directory, output image name, etc. In a few cases, the package actually segfaults. This was a single page pdf with a couple of images.
Is this an intended use case of the package? I wonder if I am completely off the mark here or if I am just not using the commands correctly.
On this PDF page the yellow annotations and the red seal were added with tclMuPDF
History
13-dec-2016 - Version 1.0b1 (beta) released. No support for MacOS
15-dec-2016 - Version 1.0 - Support for MacOS. API unchanged, but big internal optimization for reusing opened pages (read
28-jan-2017 - Version 1.1 (Withdrawn) - new commands: fields, field, anchor, mupdf::libinfo . Added package mupdf-notk for tcl-only usage. (read the documentation). Aligned with core library MuPDF v.1.10a
23-feb-2017 - Version 1.2 (Withdrawn) with a lot of new features:
commands for extracting and searching text
commands for extracting images from PDF pages (experimental)
first steps towards PDF manipulations: you can add new signature fields, then save your changes.
many minor auxiliary commands
26-feb-2017 - Version 1.2.1 - BugFixing - replaces Version 1.1 and 1.2
a bug related to the saveImage command, introduced in 1.1 was fixed here.
Versions 1.1 and 1.2 were withdrawn
29-sep-2017 - Version 1.3 - image extraction
subcommands for extracting images, previously released as experimental.images are now official, more powerful and deeply tested with a lot of different kinds of images.
Aligned with core library MuPDF v.1.11
29-oct-2017 - Version 1.4 - portfolio & password-protected files
commands for working with PDF-portfolio (embedded files)
ability to work with password-protected PDFs.
BUG-FIX removed limit to 130 char for extracting images (full path name length)
20-Dec-2018 - Version 1.5 - field get&set, new export options, graft & embed pages.
Working with fields: now you can also change the fields values (and bug-fixed support for Unicode characters)
Saving/exporting PDF: added -decrypt and -flatten option
Grafting & embedding pages: you can graft a page taken from a PDF, and put it over/below another pages. (watermarks, stamps,.. )
8-Jul-2020- Version 1.6 - passwords, forms&fields,annotations,blocks&lines ...
Password protected files
added commands for adding/changing passwords on a PDF. See methods [pdfHandleupwd] and [pdfHandleopwd]
Forms and Fields
added commands for changing field attributes. See method [pdfHandlefieldattrib]
added commands for flattening single fields. See method [pdfHandleflattenfield]
Annotations
added methods for listing, adding, changing, removing annotations. Currently only a limited set of annotation types are supported: highlight, underline, strikeout, squiggly. See methods [pageHandleannots] and [pageHandleannot ...]
Page and text structure
added methods for extracting the image/text "blocks" and the bounding box of each text line. These method are the basic building blocks - along with the new Annotation methods - for building an interactive PDF editor. See methods [pageHandleblocks] and [pageHandlelines]
11-mar-2021 - Version 2.0 New Major release (broken compatibility with 1.x)
10-aug-2021 - Version 2.1 (using MuPDF-core 1.18 features as of 2021-08-08)
text extraction
added options for extracting text and block/lines bounding-boxes.
BUG-FIX - fixed PDFColorToTkColor()
20-aug-2021 - Version 2.1.1
Better control on MUPDF warning messages and repaired PDF
BUGFIX: flattening annotations may produce cropped appearances. Now FIXED
BUGFIX: (documentation) FIXED some wrong info about the "embed" and the "annot create stamp" methods
BUGFIX: saving the same file (incrementally) more than once in a session, may produce malformed PDF . Now FIXED
28-jan-2023 - Version 2.2 - based on MuPDF 1.21.2
Option -overlay for rendering pages with a transparent background.
BUGFIX: Crash when closing some malformed PDF. Now FIXED
8-feb-2023 - Version 2.3 - based on MuPDF 1.21.2
Extended methods for text-searching and text-extraction.
added option '-mode' to 'text' and 'newsearch' for fine control over the text-extraction logic (ligatures, de-hyphenate, whitespaces ...)
14-jul-2023 - Version 2.4 - based on MuPDF 1.22.2
New command mupdf::new for creating a new empty PDF
New methods for adding/deleting/moving pages
New method addimage: insert a jpg/png image
BUGFIX: critical typo error in "graft" method (causing missing sync when a related doc is closed). NOW FIXED.
BUGFIX: saveImage : CRASH if "-from" is greater than the page size. Now FIXED
18-aug-2024 - Version 2.4.1 - based on MuPDF 1.24.8
Adapted for Tcl9 compatibility.
Binaries for a given platforms are delivered with two DLL (one Tor tcl8.6+ and one for Tcl9.0+)
Mupdf-core library upgraded to 1.24.8
Small internal changes for alignment with the new MuPdf-core API.
No changes in the tclMuPdf API ( reference manual for tclMuPdf 2.4 i s still valid)
31-aug-2024 - Version 2.4.2 - based on MuPDF 1.24.8
BUGFIX: interactive check of password protected PDFs, does not work with Tcl9. Now FIXED.
Added more tests for such cases
Code cleaning in XColors.c
Download
Starting from version 1.2.1, you can download tclMuPdf in a pre-built package with multi-platform support or, if you only need support for a single platform, you can download a lighter package.
Version 2.4.2 (31-Aug-2024)
[L1 ] FULL (Win x64, Linux x64, MacOS) (Warning: 39 MB)
Version 1.6 (July 2020) - Versions 1.x is no more maintained -
Note that specific platform support (e.g. "Linux 32") is not referred to the hosting O.S. architecture, but it's referred to the architecture of the TclTk interpreter. E.g. if you have a 32-bit TclTk interpreter running on a 64-bit Linux, you need the tclMuPdf package for linux-x32.
[L5 ] FULL (Win 32/64, Linux 32/64, MacOS) (Warning: 33MB)
[L13 ] tclMuPdf version 1.x Development-Kit. For developers/maintainers.
[L14 ] tclMuPdf version 2.x Development-Kit. For developers/maintainers.
Examples
set PDFfile /tmp/xxx.pdf
set pdfObj [mupdf::open $PDFfile]
set pageObj [$pdfObj getpage 0] ;# page 0 is the first page
$pageObj savePNG /tmp/img0.png -zoom 0.5
# ...
$pdfObj quit
# rendering just a part of a page (zoomed)
set PDFfile /tmp/zzz.pdf
set pdfObj [mupdf::open $PDFfile]
set pageObj [$pdfObj getpage 1]
$pageObj savePNG /tmp/img1.png -zoom 2.75 -from 50 450 200 700
Package mupdf integrates the MuPDF framework in Tcl. The focus of MuPDF is on speed, small code size, and high-quality anti-aliased rendering. The main goal of this integration is to generate images of the pdf pages, in a .png format, or directly in a Tk's photo image type. Thanks to its speed mupdf can be used for building interactive pdf-viewers with high-quality and real-time zooming. mupdf is a binary package, distributed in a multi-platform bundle, i.e. it can be used on
Windows 32/64 bit
Linux 32/64 bit
MacOS 64 bit
Just an example to get the flavor of how to use mupdf:
# open a file and save 1st page as a .png file
package require mupdf
set pdf [mupdf::open /mydir/sample.pdf]
set page [$pdf getpage 0] ;# 0 is the 1st page
$page savePNG /mydir/page0.png
$pdf quit
MuPDF with and without Tk
Starting from version 1.1, you can also run mupdf from a tclsh interpreter, without loading Tk. The following command
package require mupdf-notk
can be used in a tclsh interpreter to load the package without requiring Tk support. You will be still able to save images as PNG files, but of course, some subcommands related to Tk won't be available (e.g saveImage ) The command
package require mupdf
loads the full package (and requires Tk).
mupdf Commands
mupdf supports the following commands:
mupdf::openfilename ?-passwordpassword?
This is the main command: it opens the PDF-file filename and returns a pdfHandle to be used in subsequent operations. If filename is password-protected, you may specify a password adding the option -password; if option -password is not specified, mupdf will ask for a password. Read more about it in the section Working with Password Protected Files.
pdfHandleupwduser-password
pdfHandleopwdowner-password
add or change respectively the user-password or the owner-password. In order to reset the passwords,set them to "". Read more about it in the section Working with Password Protected Files.
pdfHandlequit
close and destroy the pdfHandle without saving the changes. All the related resources (e.g. opened pages) will be closed.
pdfHandleclose ?-flattenbool? ?-decryptbool?
close and destroy the pdfHandle and save the changes. All the related resources (e.g. opened pages) will be closed. See export for the meaning of the various options.
pdfHandlefullname
return the fully normalized pathname of the pdf-file.
pdfHandleauthentication
return the current authentication mode for pdfHandle. It may be none (no auth required), user (opened with user's password), owner (opened with owner's password).
pdfHandleversion
return the document's internal PDF-version.
pdfHandlenpages
return the number of pages.
pdfHandlegetpagen
return a pageHandle to be used in subsequent operations. Note that first page is page 0. Note that if the requested page is currently opened, getpage reuses the handle of the opened page.
pdfHandleanchoranchorName
return the location of anchorName as list of 3 numbers:
a page number ( -1 if anchorName is not found )
x displacement
y displacement x and y are hints for locating the anchor: they represent the displacement from the upper-left corner of the page (0,0).
pdfHandleopenedpages
return the list of all pageHandles currently opened related to pdfHandle
pdfHandlecloseallpages
close all currently opened pages related to pdfHandle
pdfHandlefields
return a list of field-records. Each field-record is a list of three elements:
the field-name
the field-type (button, radiobutton, checkbox, text, combobox, listbox, signature or unknown)
the field-value Note that for a signature field, if a signature is present the returned field-value is simply the fixed string <<signature>>.
pdfHandlefieldfieldname
return the field's value, or raise an error if fieldname is not a valid field. Warning: field-names with accented characters or in general characters from latin alphabets are accepted (even not encouraged). Fields-names with characters from non-latin alphabets (greek, russian, chinese, ...) are not supported.
pdfHandlefieldfieldnamevalue
set the field's value, or raise an error if fieldname is not a valid field. Warning: see limitations in Notes and limitations about field names and values for details.
pdfHandlefieldattribfieldname
get all the field's attributes, as a list of options/values, or raise an error if fieldname is not a valid field. Currently supported options are:
-readonly
pdfHandlefieldattribfieldname ?options?
set the field's attributes, or raise an error if fieldname is not a valid field. Currently supported options are:
-readonlybool
pdfHandleflattenfieldfieldname ?...?
flatten all the listed fields' appearances (and removes all the listed fields from the field-table). Note that flattening N fields with a single command is far more efficient than running N flattening commands; this is especially true for documents with many pages.
pdfHandlesignatures
return a list of signature field-records . Each field-record is a list of two elements:
the field-name
the field-value Empty signature fields (blanc signatures) have a field-value equal to "" (empty string). Currently, filled signature fields are simply denoted with the fixed string <<signature>>.
pdfHandlehaschanges
return 1 if pdfHandle has been changed, otherwise 0.
pdfHandlecanbesavedincrementally
return 1 if pdfHandle can be saved in incremental-mode, otherwise 0.
save the current document and its changes in an alternative filename. Valid options are:
-flattenbool - (**DEPRECATED**) flat all form's fields and annotations (default is false)
-decryptbool - ((**DEPRECATED**) if bool is true remove the password protection from an encrypted pdf. If bool is false keep the original password protection. Note that the target filename should be different from the original pdf-file related with pdfHandle and in general, different from the name of any pdf-file currently opened in this process. When a pdfHandle is closed (see mupdf::close or "pdfHandleclose"), all the changes will be saved to its original pdf-file. (see also mupdf::quit or "pdfHandlequit" for closing without saving changes). Saving with the -flatten option, is now (**DEPRECATED**). You should use "pdfHandleflattenfieldfieldname1 ..." and then save/export the document. Saving with the -decrypt option is now (**DEPRECATED**). The default is to remove the password protection, unless the opwd or upwd commands explicitely set new passwords.
extract the i-th embedded file. ( i >= 0 ) If option -dir is not specified, the embedded file is saved in the current directory. If option -as is not specified, embedded file is saved with its original name
search the string needle starting from page-number pagenum (default is page 0) and returning up to hits results (default is 10). The result of search subcommand is a list of page-positions. Each page-positions is a list of two elements:
the page-number
a list with the 4 coords of the box enclosing the searched needle Next results can be retrived with the search..more subcommand.
pdfHandlesearch..more ?-maxhits?
return a list of the next hits elements matching the last given needle. The result of search..more subcommand is a list of page-positions similar to the list returned by search.
pdfHandlegraftpageHandle
import in the currently open document referred by pdfHandle the full content of the page referred by pageHandle. The page contents are just imported, not displayed, and should be controlled via the next embed command. The result of graft is a graftID that should be used by the next embed ... command for displaying the image of the source page within one or more pages of the document. Note: this page-image is not a raster-image; it's vector based and hence remains accurate across zooming levels. Note: graftID is unique for each opened document pdfHandle and it's valid until pdfHandle is closed.
pageHandleembedgraftID ?...options...?
put a copy of a grafted page in the page referred by pageHandle. pageHandle should be a page of the same opened document who returned that graftID. Valid options are:
-overbool - displays the contents of the grafted page over (or below) the contents of the current page (default is true).
-angletheta - rotate the grafted imaged by theta degrees (-360.0 ..+360.0), Default is 0.0.
-zoomfactor - enlarge/shrink the grafted imaged by a factor of factor (default is 1.0).
-fromx0y0 ?x1y1? - just copy a rectangular sub-region of the grafted image. (x0,y0) and (x1,y1) specify diagonally opposite corners of the rectangle; (0,0) is the upper-left corner of the (visible portion of the) grafted page. If x1 and y1 are not specified, the default value is the bottom-right corner of the grafted page.
-tox0y0 ?x1y1? - place the grafted page in a rectangular subregion of the destination page. If x0 and y0 are not specifified, the grafted page is placed at (0,0) i.e. at the upper-left corner of the destination page. If x1 and y1 are not specified, the default value is the bottom-right corner of the destination page.
pageHandleannots
return the list of annotations-ID on pageHandle.
pageHandleannotannotID
return the list of the (so far recognized) annotIDs attributes. Two special read-only attributes -type and -rect' are provided for each annotation.
pageHandleannotannotIDattribute
get the value of attribute for the annotation annotID.
set the the value of one or more attributes of the annotation annotID.
pageHandleannotannotIDdelete
delete the annotation annotID
pageHandleannotcreatetype ?attributevalue ...?
create a new annotation. Currently supported types are: highlight, underline, strikeout, squiggly. Common attributes for each supported type are:
-color { R G B } - R,G,B must be decimal numers between 0.0 and 1.0 Specific options for highlight, underline, strikeout, squiggly are:
-vertices {x0y0x1y1 ... } - a list of 4x integers denoting the bounding boxes of the text to be marked. Note: these coordinates are not necessarily related to the text on a page. A new annotationID is returned.
pageHandleblocks
return the list of image/text blocks in pageHandle. A block is a list of 5 elements with the following format: textblock|imageblockx0y0x1y1
pageHandleclose
close and free all the resources of the page referred by pageHandle. the handle pdfHandle is destroyed.
pageHandlesize
return the physical size of the page as a list of two decimal numbers. Note that page size is expressed in points, i.e. 1/72 inch.
pageHandledocref
return a reference to the related pdf-document as a pdfHandle
pageHandlelines
return the list of the bbox of the text lines in pageHandle. A bbox is a list of 4 elements with the following format: x0y0x1y1
render the page in a .png file named filename. With a default -zoom factor equal to 1.0, a page whose size is W x H points is rendered as a raster image of W x H pixels. If -zoom is specified, the resulting image size is scaled by a factor of zoom. By default the whole page is rendered; the -from option, allows you to render only a given rectangular area of the page. x0y0 are the coords of the top-left corner and x1y1 are for the bottom-right corner. These coords must be expressed in terms of the physical size of the page, i.e in points Note that if these coords lies outside of the page, only the intersection of this area with the page area is rendered.
...
set page [$pdf getpage 0] ;# 0 is the 1st page
lassign [$page size] dx dy
# save just the upper half of the page
$page savePNG /mydir/page0.png -zoom 2.25 -from 0 0 $dx [expr $dy/2]
...
render the page in an existing Tk's photo image. The width and/or height of image are unchanged if the user has set on it an explicit image width or height (with the -width and/or -height configuration options, respectively). About the -zoom and -from options, the same rules for the savePNG apply. Option -to allows you to place the resulting raster image at the x0y0 coords of the destination image. By default, is -to0.00.0 NOTE: this command is not available with the package mupdf-notk.
pageHandleaddsigfieldfieldnamex0y0x1y1
add a blank signature field in a rectangular box at coords x0y0x1y1. fieldname must be unique among the existing field names.
search the string needle in the current page and return up to hitspositions (default is 10). By default search starts from top of the page. If you need to find the next hits, use option -fromtopfalse. The result of search subcommand is a list of positions. Each positions is a list with the 4 coords of the box enclosing the searched needle.
pageHandleimageslist ?-idimageID?
return a list of all the images contained in the page referred by pageHandle. The result of imageslist subcommand is a list of image-records. Each image-record is a list of six elements:
image's bit per component (number of color components may be inferred from colorspace)
mask-flag : 1 means that the image has a pixel-mask (i.e. some transparent pixels) If option -id is present, the resulting list is limited to the image-record for imageID.
if option -id is specified, extract and save the image referred by imageID (see imageslist). If option -id is missing, all the images contained in a page are extracted and saved. If option -dir is not specified, images are saved in the current directory. If option -as is not specified, images are saved with a name derived from the default mupdf::imagenamesformat (see below), otherwise images are saved as pattern (for pattern rules see below for mupdf::imagenamesformat). if option -transparency is true, save the (semi)transparent pixels (if any), otherwise transparent pixels (if any) are rendered as white pixels. extract returns a list of extracted-records. Each extracted-record is a list of three elements:
image-ID (unique for each page)
image-name (pdf page's internal name)
saved filename (empty string if the image was skipped (unknow format...))
mupdf::imagenamesformat ?pattern?
a pattern is a parametric filename specification similar to the format specification used by the C printf function. If pattern is not specified, return the currently defined default pattern. If pattern is specified, set the default pattern. When the "pageHandleimagesextract ..." command is called, all the images are saved with a filename based on a pattern (This pattern can be explicit, if option -as is present, or it can be implicit, based on the default mupdf::imagenamesformat). a pattern is a a simple filename specification (just the base-name, since the extension of the extracted images (png, jpg, ...) is automatically determined.) with zero or more special symbols like the following ones:
%p : page number (first page is 1)
%P : total number of pages
%i : image number - images in a page are numbered starting from 1
%I : total number of images (in the current page) Special symbols may also be written with a padding-specification like %5P; this notation means the the symbol %P should be padded as a 5-character string with leading '0's. If more than one image is extracted in a single operation, and pattern does not contain the %i symbol (the image number), then -%i is implicitely appended in order to avoid a filename collision. Assuming that the current page is the page 123 (i.e the 124th page), the following command
$pageHandle images extract -as "IMG_%5p"
will generate IMG_00124-1.jpg, IMG_00124-2.jpg ..... (note: the file extension may be different)
$pageHandle images extract -as "Z%5p(%2i)"
will generate Z00124(01).jpg, Z00124(02).jpg ..... (note: the file extension may be different)
mupdf::closehandle (**DEPRECATED**)
This command has been deprecated since mupdf 1.5. Use "pdfHandleclose...." or "pageHandleclose". if handle refers to a pageHandle, close the page. if handle refers to a pdfHandle, close the pdf (updating its original pdf-file) and all its opened pages.
mupdf::quitpdfHandle (**DEPRECATED**)
close and destroy the pdfHandle without saving the changes. This command has been deprecated since mupdf 1.5 . Use "pdfHandlequit" .
mupdf::isobjecthandle
return 1 if handle is a valid reference to a pdf or a page.
mupdf::typehandle
return document or page if handle is a valid reference, else raise an error.
mupdf::documents
return a list of the currently opened pdfHandles
mupdf::documentnames
return a list of pdf-filenames currently opened (fully normalized filenames).
mupdf::isopenfilename
check if filename is among the currently opened pdf-files.
mupdf::cli_passwordhelper ?helperProc?
get/set the helper procedure for shell-like applications. If helperProc is ', then the default helper is re-set. See section Working with Password Protected Files'.
mupdf::tk_passwordhelper ?helperProc?
get/set the helper procedure for applications with a Tk graphical interface. If helperProc is ', then the default helper is re-set. See section Working with Password Protected Files'.
mupdf::libinfo
return specific attributes of the underlying MuPdf libray as a list of keywords and their values. The provided keywords are:
version
The version of the underlying MuPDf library
..more to come ..
Working with Password Protected Files
You can open a password-protected PDF in a non-interactive or in an interactive way. With the first way, non-interactive way, you must provide in advance a password
set pdf [mupdf::open book.pdf -password "123open"]
If password is wrong then an error is raised; look at the error message and check for the errorcode: it should be MUPDF WRONGPASSWORD. Note that there's no distinction between owner's password and user's password; if you provide the right owner's password, the PDF is opened in owner-mode, if you provide the right user's password, the PDF is opened in user-mode. You can check if PDF has been opened in user-mode or owner-mode by calling the authentication command.
set pdf [mupdf::open book.pdf -password "123open"]
# if we are here, it means that password was OK
set mode [$pdf authentication]
# "none" means that book.pdf was not password-protected
# "user" means that the supplyed password was the user's password
# "owner" means ...
The second way , interactive way, does not require to provide an explicit password; if the PDF is password-protected, mupdf will ask for a password. Depending on the nature of your application (with or without Tk), mupdf will select and call a predefined (yet customizable) helper procedure. These predefined, built-in procedures can be changed with the mupdf::cli_passwordhelper or mupdf::tk_passwordhelper command.
# template for a (shell-like) password-helper procedure
proc mypswdhelper {filename} {
... ask for a password ....
.... get the password ....
....
return $password
}
mupdf::cli_passwordhelper mypswdhelper
# template for a (GUI) password-helper procedure
proc myGUIpswdhelper {filename} {
... raise some popup ...
... ask for a password ....
.... get the password ...
.... close the popup
return $password
}
mupdf::tk_passwordhelper myGUIpswdhelper
Removing protection from a password protected files
If you open a password-protected file and export it, by default the resulting PDF is saved without password. You can explicitely control this behaviour by appending the -decrypt option to the export or close commands (default is -decrypttrue).
Adding or changing a password
After opening a PDF, you can add or change the user/owner password with the opwd or the upwdcommands,like in the following example
set pdf [mupdf::open mybook.pdf]
# if mybook.pdf is proteccted, insert the password
$pdf upwd "my user-password"
$pdf opwd "my owner-password"
$pdf close ;# ..or $pdf export ...
Notes and limitations about field names and values
Due to current limitations of the underlying MuPdf library, field names and values should be limited to latin alphabets only. See the note enclosed with MuPdf's BUG 696478 (Nov 2018): '' ... "There is, however, a limitation with our form filling support that still
needs to be addressed: it only supports PDFDocEncoding, not Unicode.
... "We can currently fill out the form fields in any language, but the display code still only handles the Latin alphabet." '' This means that you can enter text in a form field in any alphabet (tested with Russian, Chinese, Greek, ...) but the rendered text for the non-latin alphabets will appear wrong. Another major limitation is the rendering of fields other that 'text' widgets, i.e. checkbox, radiobuttons... For these kind of widgets the support of the underlying library is still incomplete, in particular for values expressed in non-ASCII characters (e.g. accented letters ... )
Limitations
Full support for portfolio management is still incomplete; commands for adding/removing/reordering embedded files will require a more robust mupdf-core implementation.
Package MuPDF integrates the MuPDF framework in Tcl. The focus of MuPDF is on speed, small code size, and high-quality anti-aliased rendering. The main goal of this integration is to generate images of the pdf pages, in a .png format, or directly in a Tk's photo image type. MuPDF also provides a partial support for editing PDFs, allowing to fill form-fields and working with annotations. Thanks to its speed MuPDF can be used for building interactive pdf-viewers with high-quality and real-time zooming. MuPDF is a binary package, distributed in a multi-platform bundle, i.e. it can be used on
Windows 64 bit
Linux 64 bit
MacOS 64 bit
Just an example to get the flavor of how to use MuPDF:
# open a file and save 1st page as a .png file
package require tclMuPDF
set pdf [mupdf::open /mydir/sample.pdf]
set page [$pdf getpage 0] ;# 0 is the 1st page
$page savePNG /mydir/page0.png
$pdf quit
MuPDF 2.x and mupdf 1.x
Please note that MuPDF-2.x is a new major release breaking the backward compatibility with 1.x. In order to support these new features, and for the next planned features, MuPDF-2.x has been totally redesigned. In Appendix you can find a quick summary of the changes and how to adapt your old code. See From mupdf 1.x to MuPDF 2.x
MuPDF with and without Tk
MuPDF is provided in two variants:
tkMuPDF (or simply MuPDF) is the full package, it requires Tk
tclMuPDF does not require Tk. You will be still able to save images as PNG files, but of course some subcommands related to Tk won't be available (e.g saveImage )
MuPDF Commands
The MuPDF package is based on 2 main classes mupdf::Doc, Mupdf::Page whose instances correspond to a PDF-document and its pages, plus an auxiliary class mupdf::TextSearch whose purpose is to manage progressive searches across all the pages of a document
mupdf::Doc methods
mupdf::openfilename ?-passwordpassword?
This is the main command: it opens the PDF-file filename and returns a pdfObj to be used in subsequent operations.
If filename is password-protected, you may specify a password adding the option -password; if option -password is not specified, MuPDF will ask for a password. Read more about it in the section Working with Password Protected Files.
mupdf::newfilename
This command is for creating a new, empty document. It returns a pdfObj to be used in subsequent operations. Note that this document has no pages, so the first thing to do is to call addpage and add some contents (currently you can add only images and annotations).
Read more about it in the section Adding/removing pages.
pdfObjauthentication
return the current authentication mode for pdfObj. It may be none (no auth required), user (opened with user's password), owner (opened with owner's password).
pdfObjupwduser-password
pdfObjopwdowner-password
add or change respectively the user-password or the owner-password. In order to reset the passwords,set them to "".
Read more about it in the section Working with Password Protected Files.
pdfObjremovepassword
reset both the user-password and the owner-passwords.
pdfObjquit
close and destroy the pdfObj without saving the changes. All the related resources (e.g. opened pages ..) will be destroyed.
pdfObjclose
save the changes (if any) and then close and destroy the pdfObj. All the related resources (e.g. opened pages ..) will be destroyed.
pdfObjfullname
return the fully normalized pathname of the pdf-file.
pdfObjversion
return the document's internal PDF-version.
pdfObjaddpagepageNumber|end ?-sizedxdy?
add a new page before the page pageNumber, or at end, and return a new pageObj to be used in susequent operations.
See mupdf::Page methods. All the currently opened pages are properly renumbered.
pdfObjdeletepagepageNumber
delete the page at pageNumber. All the currently opened pages are properly renumbered.
pdfObjdeletepagespageNumApageNumB
delete all the pages from pageNumA to pageNumB. All the currently opened pages are properly renumbered.
pdfObjdeletepagespageNumApageNumB
delete all the pages from pageNumA to pageNumB. If pagNumA is less than pageNumB, then nothing is done, no error is raised. All the currently opened pages are properly renumbered.
pdfObjmovepagefromto
move the page from position from to position to. All the currently opened pages are properly renumbered.
pdfObjanchoranchorName
return the location of anchorName as list of 3 numbers:
a page number ( -1 if anchorName is not found )
x displacement
y displacement x and y are hints for locating the anchor: they represent the displacement from the upper-left corner of the page (0,0).
pdfObjnpages
return the number of pages.
pdfObjgetpagen
return a pageObj to be used in subsequent operations. See mupdf::Page methods. Note that first page is page 0. Note that if the requested page is currently opened, getpage reuses the handle of the opened page.
pdfObjopenedpages
return the list of all currently opened pages (as page-numbers) related to pdfObj
pdfObjispageopenedpageNum
return 1 if page pageNum is currently opened, else 0.
pdfObjclosepagepageNum
close the page referred bu pageNum. No error is returned if pageNum does not refer to an opened page.
pdfObjcloseallpages
close all currently opened pages related to pdfObj
pdfObjfields
return a list of field-records. Each field-record is a list of three elements:
the field-name
the field-type (button, radiobutton, checkbox, text, combobox, listbox, signature or unknown)
the field-value Note that for a signature field, if a signature is present the returned field-value is simply the fixed string <<signature>>.
pdfObjfieldfieldname
return the field's value, or raise an error if fieldname is not a valid field.
Warning: field-names with accented characters or in general characters from latin alphabets are accepted (even not encorauged). Fields-names with characters from non-latin alphabets (greek, russian, chinese, ...) are not supported.
pdfObjfieldfieldnamevalue
set the field's value, or raise an error if fieldname is not a valid field.
Warning: see limitations in Notes and limitations about field names and values for details.
pdfObjfieldattribfieldname
get all the field's attributes, as a list of options/values, or raise an error if fieldname is not a valid field.
Currently supported options are:
-readonly
pdfObjfieldattribfieldname ?options?
set the field's attributes, or raise an error if fieldname is not a valid field.
Currently supported options are:
-readonlybool
pdfObjflattenfieldfieldname ?...?
flatten all the listed fields appearances (and removes all the listed fields from the field-table).
Note that flattening N fields with a single command is far more efficient than running N flattening commands; this is especially true for documents with many pages.
pdfObjsignatures
return a list of signature field-records . Each field-record is a list of two elements:
the field-name
the field-value Empty signature fields (blanc signatures) have a field-value equal to "" (empty string). Currently, filled signature fields are simply denoted with the fixed string <<signature>>.
pdfObjaddsigfieldfieldnamepageNumberx0y0x1y1
add a blank signature field on the page number pageNumber in a rectangular box at coords x0y0x1y1.
fieldname must be unique among the existing field names.
pdfObjhaschanges
return 1 if pdfObj has been changed, otherwise 0.
pdfObjexportfilename
save the current document and its changes in an alternative filename. Note that in general, filename should be different from the name of any pdf-file currently opened in this process. The only exception is when filename is the same name of pdfObj, but in this case you MUST quit pdfObj just after export.
extract the i-th embedded file. ( i >= 0 ) If option -dir is not specified, the embedded file is saved in the current directory.
If option -as is not specified, embedded file is saved with its original name
pdfObjnewsearch ?-modelistOfFlags?
return a new searchObj object for performing text-search on the pdfObj document. See mupdf::TextSearch methods
pdfObjgraftpageObj
import in the currently open document referred by pdfObj the full content of the page referred by pageObj. The page contents are just imported, not displayed, and should be inserted in pdfObj via the next embed command. The result of graft is a graftID that should be used by the next embed ... command for inserting the source page within one or more pages of the document.
Note: this page-image is not a raster-image; it's vector based and hence remains accurate across zooming levels. Note: graftID is unique for each opened document pdfObj and it's valid until pdfObj is closed.
pdfObjgrafts
return the list of the currently grafted pages
pdfObjembedgraftIDpageNumber ?...options...?
embed a copy of a grafted page in the page referred by pageNumber.
Valid options are:
-overbool - displays the contents of the grafted page over (or below) the contents of the current page (default is true).
-angletheta - rotate the grafted imaged by theta degrees (-360.0 ..+360.0), Default is 0.0.
-zoomfactor - enlarge/shrink the grafted imaged by a factor of factor (default is 1.0).
-cropx0y0 ?x1y1? - just copy a rectangular sub-region of the grafted image. (x0,y0) and (x1,y1) specify diagonally opposite corners of the rectangle; (0,0) is the upper-left corner of the (visible portion of the) grafted page. If x1 and y1 are not specified, the default value is the bottom-right corner of the grafted page.
-tox0y0 - place the center of the grafted page at (x0, y0). If this option is not specified then the grafted page is placed in the center of the destination page.
Control on warning-messages
See the rationale for these commands in the section Better control for MuPDF internal errors/warnings and repaired PDFs
mupdf::printwarnings ?boolValue?
If boolVal is true, then all the next MUPDF warning-messages will be printed on stderr. If boolVal is false, printing on stderr is suppressed (but you can still inspect them with the "pdfObjwarnings"" command) If boolValue is missing, the current setting is returned (true/false) The default setting is false.
pdfObjwarnings
return a list with all the recorded MUPDF warning-messages.
pdfObjresetwarnings
reset the list of the warning-messages
pdfObjwasrepaired
returns true if the PDF was automatically repaired (and in this case you should find some details with the warnings method.)
mupdf::TextSearch methods
Instances of TextSearch objects allows to perform text-search on a given pdfObj document. TextSearch object can be created in two ways:
set searchObj [mupdf::TextSearch new ''pdfObj'' ?'''-mode''' ''listOfFlags''?]
or
set searchObj [''pdfObj'' newsearch ?'''-mode''' ''listOfFlags''?]
The -mode option can be used to control the quality of extracted text. Read more about it in the section Text-Extraction & Text-Search Flags. Note that when pdfObj is closed, all its TextSearch instances are automatically destroyed. TextSearch methods:
search for the string searchStr starting from the current search-position. Return a list of up to hits matches (default is 10); the current position is then accordling advanced. The result of the find subcommand is a list of page-positions. Each page-positions is a list of two elements:
the page-number
a list with the 4 coords of the box enclosing the searched searchStr If option -currpageonly is true, the search is limited to the page holding the current search-position.
searchObjcurrpage
return the current search-position as a page-number
searchObjcurrpagen
set the current search-position at the beginning of page n.
searchObjdocref
return a reference to the related pdf-document as a pdfObj
mupdf::Page methods
A Page object can be created starting from a pdfObj document with the following command:
set pageObj [$pdfObj getpage _n_]
If the requested page is currently opened, getpage reuses the handle of the opened page. Note that when pdfObj is closed, all its opened pages are automatically closed/destroyed.
pageObjannots
return the list of annotations-ID on pageObj. Annotations ID are unique within the whole PDF.
pageObjannotgetannotID
pageObjannotannotID
return the list of the annotIDs attributes. The special read-only attributes -type' is provided for each annotation.
pageObjannotgetannotIDattribute
pageObjannotannotIDattribute
get the value of attribute for the annotation annotID.
set the the value of one or more attributes of the annotation annotID.
pageObjannotdeleteannotID ?annotID ...?
delete one or more annotations
pageObjannotflattenannotID ?annotID ...?
flatten one or more annotations
pageObjannotcreatetype ?attributevalue ...?
create a new annotation and return a new annotationID.
Currently supported types are: highlight, underline, strikeout, squiggly and stamp. Common attributes for each supported type are:
-color tkcolor - color must be a simbolic name (lightblue) or #RRGGBB Specific options for highlight, underline, strikeout, squiggly are:
-vertices {x0y0x1y1 ... } - a list of 4x numbers denoting the bounding boxes of the text to be marked. Note: these coordinates are not necessarily related to the text on a page. Specific options for stamp are:
-xobjectxObjectID - the ID of a grafted page (see graft method )
-rotateangle - the rotation anngle (in degree)
-scales - the zoom factor
-to {x0y0} - a list of 2 numbers. Note that the center of the stamp will placed at the coords specified with the -to option. If this option is missing, the center of the stamp will be placed in the centre of the page.
pageObjblocks
return the list of image/text blocks in pageObj. A block is a list of 5 elements with the following format:
textblock|imageblockx0y0x1y1
pageObjclose
pageObjdestroy
close and free all the resources of the page referred by pageObj. the object pdfObj is destroyed.
pageObjsize
return the physical size of the page as a list of two decimal numbers. Note that page size is expressed in points, i.e. 1/72 inch.
pageObjdocref
return a reference to the related pdf-document as a pdfObj
pageObjpagenumber
return the pagenumber of pageObj
pageObjlines
return the list of the bbox of the text lines in pageObj. A bbox is a list of 4 elements with the following format:
extract the plain text from the page pageObj with the (optional) bounding-box of each block or line. The -mode option can be used to control the amount and the quality of extracted text. Read more about it in the section Text-Extraction & Text-Search Flags.
If -bbox is missing or equal to none, this method returns just the plain text.
If -bbox is equal to blocks, this method returns a list like the following
{bbox of 1st block} {plain text of the 1st block}
...
{bbox of the Nth block} {plain text of the Nth block}
If -bbox is equal to lines, this method returns a list like the following
{bbox of the 1st block} {
{bbox of the 1st line} {plain text of the 1st line}
...
{bbox of the Nth line} {plain text of the Nth line}
}
...
{bbox of the Nth block} {
{bbox of the 1st line} {plain text of the 1st line}
...
{bbox of the Nth line} {plain text of the Nth line}
}
With a default -zoom factor equal to 1.0, a page whose size is W x H points is rendered as a raster image of W x H pixels. If -zoom is specified, the resulting image size is scaled by a factor of zoom. By default the whole page is rendered; the -from option, allows you to render only a given rectangular area of the page. x0y0 are the coords of the top-left corner and x1y1 are for the bottom-right corner. These coords must be expressed in terms of the physical size of the page, i.e in points Note that if these coords lies outside of the page, only the intersection of this area with the page area is rendered. If the -overlay option is set to true then the page is rendered with a transparent background, else if it set to false (default) the page is rendered with a white background.
...
set page [$pdf getpage 0] ;# 0 is the 1st page
lassign [$page size] dx dy
# save just the upper half of the page
$page savePNG /mydir/page0.png -zoom 2.25 -from 0 0 $dx [expr $dy/2]
...
The width and/or height of image are unchanged if the user has set on it an explicit image width or height (with the -width and/or -height configuration options, respectively). About the -zoom, -from and -overlay options, the same rules for the savePNG apply. Option -to allows you to place the resulting raster image at the x0y0 coords of the destination image. By default, is -to0.00.0 NOTE: this command is not available with the package tclMuPDF.
pageObjimageslist ?-idimageID?
return a list of all the images contained in the page referred by pageObj.
The result of imageslist subcommand is a list of image-records. Each image-record is a list of six elements:
image's bit per component (number of color components may be inferred from colorspace)
mask-flag : 1 means that the image has a pixel-mask (i.e. some transparent pixels) If option -id is present, the resulting list is limited to the image-record for imageID.
if option -id is specified, extract and save the image referred by imageID (see imageslist). If option -id is missing, all the images contained in a page are extracted and saved.
If option -dir is not specified, images are saved in the current directory. If option -as is not specified, images are saved with a name derived from the default mupdf::imagenamesformat (see below), otherwise images are saved as pattern (for pattern rules see below for mupdf::imagenamesformat). if option -transparency is true, save the (semi)transparent pixels (if any), otherwise transparent pixels (if any) are rendered as white pixels. extract returns a list of extracted-records. Each extracted-record is a list of three elements:
image-ID (unique for each page)
image-name (pdf page's internal name)
saved filename (empty string if the image was skipped (unknow format...))
mupdf::imagenamesformat ?pattern?
a pattern is a parametric filename specification similar to the format specification used by the C printf function.
If pattern is not specified, return the currently defined default pattern. If pattern is specified, set the default pattern. When the "pageObjimagesextract ..." command is called, all the images are saved with a filename based on a pattern (This pattern can be explicit, if option -as is present, or it can be implicit, based on the default mupdf::imagenamesformat). a pattern is a a simple filename specification (just the base-name, since the extension of the extracted images (png, jpg, ...) is automatically determined.) with zero or more special symbols like the following ones:
%p : page number (first page is 1)
%P : total number of pages
%i : image number - images in a page are numbered starting from 1
%I : total number of images (in the current page) Special symbols may also be written with a padding-specification like %5P; this notation means the the symbol %P should be padded as a 5-character string with leading '0's.
If more than one image is extracted in a single operation, and pattern does not contain the %i symbol (the image number), then -%i is implicitely appended in order to avoid a filename collision. Assuming that the current page is the page 123 (i.e the 124th page), the following command
$pageObj images extract -as "IMG_%5p"
will generate IMG_00124-1.jpg, IMG_00124-2.jpg ..... (note: the file extension may be different)
$pageObj images extract -as "Z%5p(%2i)"
will generate Z00124(01).jpg, Z00124(02).jpg ..... (note: the file extension may be different)
add a new or a reused image; return the ID of the loaded image.
-filefilename - specify a PNG or JPEG filename.
-reuseID - ID must be 0 (this means that option -file should be considered), or an ID returned by a previous call of addimage. At least one of -reuse, -file must be present. If -reuse is present and its value ID is greater than 0, then option -file is simply ignored.
Other options are:
-tox0y0x1y1 - defines the rectangular area where to put the image. By default this area is the full page. x0y0 are the coords of the top-left corner and x1y1 are for the bottom-right corner. These coords must be expressed in terms of the physical size of the page, i.e in points
-resizeboolean - If value is true, then the image will be resized in order to fill the whole destination area (preserving the initial proportions). (Default is true).
-overboolean - if this option is set to true, then the image is displayed over the current contents of the page, else it is displayed under the current contents. Default value is true.
-marginxdx1 ?dx2? - specifies how much horizontal spacing should be reserved for left,right margins within the destination area. If dx2 is missing, it is assumed to be equal to dx1. Default value is 0.
-marginydy1 ?dy2? - specifies how much vertical spacing should be reserved for top,bottom margins within the destination area. If dy2 is missing, it is assumed to be equal to dy1. Default value is 0.
-anchorwhere - where specifies how to align the image within the page's destination area. Must be one of the values n, ne, e, se, s, sw, w, nw, or c. Deafault value is "c" (center) meaning that the image is centered in the destination area.
MuPDF introspection commands
mupdf::isobjectobj
return 1 if obj is an oo-object (not limited to MuPDF objects).
mupdf::classes
returns the list of the oo-classes of MuPDF
mupdf::classinfoobj
return the oo::class of obj (not limited to MuPDF objects)
mupdf::documents
return a list of the currently opened pdfObjs
mupdf::documentnames
return a list of pdf-filenames currently opened (fully normalized filenames).
mupdf::isopenfilename
check if filename is among the currently opened pdf-files.
mupdf::Docnames
mupdf::Pagenames
mupdf::TextSearchnames
return a list of all the instances of Doc, Page, TextSearch. Note that when a Doc is destroyed, all its related Pages and TextSearch are destroyed, too.
mupdf::cli_passwordhelper ?helperProc?
get/set the helper procedure for shell-like applications. If helperProc is ', then the default helper is re-set. See section Working with Password Protected Files'.
mupdf::tk_passwordhelper ?helperProc?
get/set the helper procedure for applications with a Tk graphical interface. If helperProc is ', then the default helper is re-set. See section Working with Password Protected Files'.
mupdf::libinfo
return specific attributes of the underlying MuPdf library as a list of keywords and their values. The provided keywords are:
version
The version of the underlying MuPDf library
..more to come ..
Working with Password Protected Files
You can open a password-protected PDF in a non-interactive or in an interactive way. In non-interactive mode, you must provide in advance a password
set pdf [mupdf::open book.pdf -password "123open"]
If password is wrong then an error is raised; look at the error message and check for the errorcode: it should be MUPDF WRONGPASSWORD. Note that there's no distinction between owner's password and user's password; if you provide the right owner's password, the PDF is opened in owner-mode, if you provide the right user's password, the PDF is opened in user-mode. You can check if PDF has been opened in user-mode or owner-mode by calling the authentication command.
set pdf [mupdf::open book.pdf -password "123open"]
# if we are here, it means that password was OK
set mode [$pdf authentication]
# "none" means that book.pdf was not password-protected
# "user" means that the supplied password was the user's password
# "owner" means ...
In interactive mode, you should not provide an explicit password; if the PDF is password-protected, MuPDF will ask for a password. Depending on the nature of your application (with or without Tk), mupdf will select and call a predefined (yet customizable) helper procedure. These predefined, built-in procedures can be changed with the mupdf::cli_passwordhelper or mupdf::tk_passwordhelper command.
# template for a (shell-like) password-helper procedure
proc mypswdhelper {filename} {
... ask for a password ....
.... get the password ....
....
return $password
}
mupdf::cli_passwordhelper mypswdhelper
# template for a (GUI) password-helper procedure
proc myGUIpswdhelper {filename} {
... raise some popup ...
... ask for a password ....
.... get the password ...
.... close the popup
return $password
}
mupdf::tk_passwordhelper myGUIpswdhelper
Removing protection from a password protected files
If you open a password-protected file and export it, by default the resulting PDF is saved without a password. You can explicitly control this behavior by appending the -decrypt option to the export or close commands (default is -decrypttrue).
Adding or changing a password
After opening a PDF, you can add or change the user/owner password with the opwd or the upwd commands, like in the following example
set pdf [mupdf::open mybook.pdf]
# if mybook.pdf is protected, insert the password
$pdf upwd "my user-password"
$pdf opwd "my owner-password"
$pdf close ;# ..or $pdf export ...
Better control for MuPDF internal errors/warnings and repaired PDFs
In many cases MuPDF (the core library) is able to work with bad/corrupted PDFs, and if the errors are not critical, the 'repaired' PDF is opened as a good PDF. By default MuPDF prints on stderr all these errors/warnings; if the error is critical MuPDF raises an error (and so tclMuPDF), but if the malformed PDF can be automatically repaired, no error is raised, just some messages are printed on stderr. With the previous tclMuPDF release (2.1) we completely suppressed all these (non-critical) warning-messages generated by the core-MuPDF. This was a useful feature because many of those messages were annoying and rarely useful, but there are still some cases that shouldn't be ignored because in some cases the automatically repaired PDFs may be still inconsistent or damaged in some other parts. Starting with tclMuPDf 2.1.1 we added some new commands and methods, for better control of these warning-messages:
mupdf::printwarnings
pdfObjwarnings
pdfObjresetwarnings
pdfObjwasrepaired By using "mupdf::printwarnings" tou can enable/disable the warning-messages. By default printing on stderr is disabled, but if you set "mupdf::printwarningstrue" then all the next errors/warnings will be printed on stderr.
Even if printing is globally disabled, all the MUPDF warning-messages are still recorded and they can be inspected with the command "pdfObjwarnings". Note that each opened document pdfObj has its own list of warnings. The command "pdfObjresetwarnings" resets all the warnings bound to the pdfObj document; of course when the document is closed/deleted, all its related warnings are deleted. Finally, the "pdfObjwasrepaired" command can be used for checking if pdfObj was automatically repaired, so you can decide to quit or keep on working with the repaired pdfObj.
Notes and limitations about field names and values
Due to current limitations of the underlying MuPdf library, field names and values should be limited to latin alphabets only. See the note enclosed with MuPdf's BUG 696478 (Nov 2018): '' ... "There is, however, a limitation with our form-filling support that still
needs to be addressed: it only supports PDFDocEncoding, not Unicode.
... "We can currently fill out the form fields in any language, but the display code still only handles the Latin alphabet." '' This means that you can enter text in a form field in any alphabet (tested with Russian, Chinese, Greek, ...) but the rendered text for the non-latin alphabets will appear wrong. Another major limitation is the rendering of fields other than 'text' widgets, i.e. checkbox, radiobuttons... For these kinds of widgets, the support of the underlying library is still incomplete, in particular for values expressed in non-ASCII characters (e.g. accented letters ... )
Text-Extraction & Text-Search Flags
Extracting text from a PDF is still a matter of magic art. Different PDF-tools provide different results, based on their own internal rules and assumptions. MuPDF text-extraction is far to be perfect, (MuPDF claims it's an experimental feature), but starting with release 2.3, tclMuPDF provides some explicit flags that can be used for tuning the text-extraction logic. They control how text should be handled with respect to white spaces, hyphenation and ligatures. The following basic text-extraction command
$pageObj text ..other options..
is equivalent to
$pageObj text -mode {preserve-ligatures preserve-whitespace} ..other options..
Similarly, the following command for creating a text-search object
set searchObj [$pdfObj newsearch]
is equivalent to
set searchObj [$pdfObj newsearch -mode {preserve-whitespace dehyphenate}]
In general, text-extraction/search flags can be combined in a list for the -mode option. Be aware that if you specify the -mode option, you must set all desired options. These are the currently available flag, both for text-extraction and text-search:
preserve-ligatures If set, ligatures are passed through to the application in their original form. Otherwise, ligatures are expanded into their constituent parts, e.g. the ligature “ffi” is expanded into three separate characters f, f and i. MuPDF supports the following 7 ligatures: “ff”, “fi”, “fl”, “ffi”, “ffl”, , “ft”, “st”.
preserve-whitespace If set, whitespace is passed through in its original form. Otherwise, any type of horizontal whitespace (including horizontal tabs) will be replaced with space characters of variable width
inhibit-spaces If set, Mupdf will not try to add missing space characters where there are large gaps between characters. In PDF, the creator often does not insert spaces to point to the next character’s position but will provide the direct location address.
dehyphenate Ignore hyphens at line ends and join with the next line. if set, text extractions will return joined text lines (or spans) with the ending hyphen of the first line removed. So two separate spans “first meth-” and “od leads to wrong results” on different lines will be joined to one span “first method leads to wrong results” and correspondingly updated bboxes: the characters of the resulting span will no longer have identical y-coordinates.
preserve-spans If this option is set, spans on the same line will not be merged. Each line will thus be a span of text with the same font, color, and size.
Adding/removing pages
tclMuPdf 2.4 introduced a few basic commands for adding/deleting/moving pages in a PDF document. These commands are:
pdfObjaddpage ..
pdfObjdeletepage ..
pdfObjdeletepages ..
pdfObjmovepage .. These commands can be used both on existing documents and on new documents created with mupdf::new.
Please, don't confuse the new deletepage method with the old closepage method. closepage simply 'closes' an opened page, whilst deletepage 'removes' a page from a loaded document. New pages (those created with the addpage method) are initially empty (blank pages). Currently, you can only add images and annotations (and empty fields for digital signature) to these pages. Here is an example of how to add a 'cover' to an existing document.
set doc [mupdf::open "mydoc.pdf"]
set page [$doc addpage 0] ;# A4 size
$page addimage "mycover.jpg"
$doc close ;# -- document saved
# if you want to do a "save as..", don't call "$doc close", but rather
# $doc export "mydoc-with-cover.pdf"
# $doc quit ;# -- quit without saving
Limitations
Full support for portfolio management is still incomplete; commands for adding/removing/reordering embedded files will require a more robust mupdf-core implementation.
From mupdf 1.x to MuPDF 2.x
New commands and methods
pdfObjremovepassword
pdfObjispageopened
pdfObjclosepage
pdfObjflatten
pdfObjgrafts
mupdf::classes
mupdf::classinfoobj
Changed commands and methods
package require mupdf
package require mupdf-notk
With the release 2.x these package names are no more valid. Use:
package require tclMuPDF ;# replacement for "mupdf-notk"
package require tkMuPDF ;# replacement for "mupdf"
package require MuPDF ;# replacemente for "mupdf"
pdfObjopenedpages
now returns a list of page-numbers. Formerly it returned a list of pageObj
pdfObjclose options
options -flatten and -decrypt are no more supported. In case you need to flatten the fields or remove the password, you should call the new "pdfObjflatten" or "pdfObjremovepassword" methods before pdfObjclose
pdfObjflattenfield ...
has been replaced by "pdfObjflatten ..."
pdfObjaddsigfieldfieldnamepageNumx0 y0 x1 y1
changed parameters order and signature
pdfObjembed ...
changed parameters order and signature
pdfexport ...
options -flatten and -decrypt have been removed. In case you need to flatten the fields or remove the password, you should call the new "pdfObjflatten"" or "pdfObjremovepassword" methods before pdfObjexport
pageObjannotdeleteannotID ...
replaces "pageObjannotannotIDdelete"
pageObjannot ...
Formerly colors were expressed as a triple {R G B}. (0<=R,G,B<=1.0) Now colors should be expressed as tk-color (e.g "lightblue" or #223312).
Removed commands and methods
32 bit
Removed support for Linux-32bit and Windows-32 bit
package require mupdf
package require mupdf-notk
pdfObjflattenfield ...
Renamed as "pdfObjflatten ..."
pdfObjcanbesavedincrementally
Still present but deprecated.
pdfObjsearch ...
Use the "pdfObjnewsearch" command and the mupdf::TextSearch methods.