A command-line program to crop the margins of PDF files, with many options.
The pdfCropMargins program is a command-line application to automatically crop the margins of PDF files. Cropping the margins can make it easier to read the pages of a PDF document – whether the document is printed or displayed on a screen – because the display fonts are larger. This program is similar to the Perl script pdfcrop except with many more options.
Automatically detects the margins and can crop a given percentage of them.
Can crop all the pages to the same size to give a uniform appearance (such as in two-up).
Works on Linux, Windows, Cygwin, and OS X/Darwin.
Has an optional graphical user interface (GUI) for interactive cropping.
Works with either MuPDF, the pdftoppm program, or the Ghostscript program.
A version of pdftoppm for Windows is included.
Renders and analyzes page images to find the bounding boxes, which allows it to deal with noisy scanned PDFs.
Can automatically apply a Ghostscript repair operation to attempt to fix corrupt PDF files.
A rudimentary ‘undo’ capability is implemented by default.
Can crop pages uniformly based on the nth smallest crop values, which helps with noisy images or documents where a few pages have unwanted markings in their margins.
Can automatically run a document previewer on the output file.
The format of automatically-generated output-file names is easily modifiable.
Preserves document catalog information such as outlines if possible.
Crops rotated pages according to their appearance in the document viewer.
Can deal with at least simple cases of password-encrypted files.
Works with Python 3 and Python 2.
This GIF shows the optional GUI, before and after cropping a document:
See the CHANGELOG for recent changes and new features.
New in version 1.0.0
The new default method for rendering pages to calculate crops uses the Python PyMuPDF package, if it is detected. That package is included in the GUI install but not in the base install. This method works in-memory and tends to be faster. To get the old default behavior (i.e., searching first for pdftoppm and then for Ghostscript) use the option --calcbb o or the shortcut -c o.
The new preferred way to select the method of calculating bounding boxes is the option --calcbb or the shortcut -c. The options are:
-c d: The default, look for PyMuPDF, pdftoppm, and Ghostscript, in that order.
-c m: Force the use of MuPDF (PyMuPDF) rendering.
-c p: Force the use of pdftoppm rendering.
-c gr: Force the use of Ghostscript rendering (equivalent to --gsRender).
-c gb: Use Ghostscript to directly calculate the bounding boxes (equivalent to --gsBbox). This method does not work for scanned documents.
-c o: Revert to the old default behavior (pdftoppm then Ghostscript).
The older method-selection options still work, only the default has changed.
The default rendering resolution is now 72 dpi instead of 150 dpi. Resolution can still be set with the -x and -y options.
A new option flag --percentText which changes the interpretation of the percentage values passed to --percentRetain and --percentRetain4. With this flag the left and right margins are set to a percentage of the text width (bounding box width) and the top and bottom margins are set to a percentage of the text height.
The easiest way to install the pdfCropMargins program is by using pip.
The basic features work out of the box with the GUI install, but the default install and some of the options require either the external program pdftoppm or Ghostscript. For information on installing those programs on Linux and Windows, see: Installing pdftoppm and/or Ghostscript .
Note on Python 2 installs: The program still runs and installs on Python 2, but the Pillow dependency will not be automatically installed because the most recent version available for Python 2 has several moderate-severity security vulnerabilities. You can choose to install it yourself via pip install pillow --user; otherwise, only the --bbcalc gb (or --gsBbox) method of calculating crops will work (and Ghostscript is required). For the GUI, PySimpleGUI no longer supports 2.7 but the last version still seems to work.
If you are installing via pip with the --user option be sure $HOME/.local/bin is in your system PATH. (To install system-wide without --user the pip3 command below would also need to be run with sudo.)
This is the full install, with the GUI and external utility programs:
sudo apt install python3-pip ghostscript poppler-utils python3-setuptools python3-tk pip3 install pdfCropMargins[gui] --user --upgrade
Here is the plain install, without the GUI or MuPDF bounding-box calculations. It has fewer dependencies, but it requires that either pdftoppm or Ghostscript be installed. Not all options are available.
sudo apt install python3-pip ghostscript poppler-utils pip3 install pdfCropMargins --user --upgrade
To get the MuPDF rendering features without the GUI dependencies you can alternately use pdfCropMargins[mupdf] in the above commands.
Troubleshooting: If you have problems with the install of PyMuPDF, the first thing to try is to upgrade your version of the pip program and then retry:
pip3 install pip --upgrade --user
If you still get an error “Failed building wheel for pymupdf” with the GUI install you can try forcing a binary install of pyMuPDF:
pip3 install pdfCropMargins[gui] --user --upgrade --only-binary pymupdf
The pip program should be automatically installed along with Python. If you cannot find the pip executable you can usually run it like this:
python -m pip <arguments-to-pip>
Note that on some Windows installs the executable for Python is py rather than python.
To install with the GUI use:
pip install pdfCropMargins[gui] --upgrade
The plain install, without the GUI, has fewer dependencies but requires either Ghostscript or pdftoppm to be installed. A version of pdftoppm is supplied for Windows, which will be used as a fallback. To install without the GUI use:
pip install pdfCropMargins --upgrade
In order for the command pdf-crop-margins to work from the command line the bin directory that it is created in (by the pip command) must be on the Windows Path. The system-wide Python bin directory should already be on the path if you checked the box to modify Path when you installed Python; otherwise it should be added. (If you install with the --user option to pip then you need to be sure that the local Python bin directory is in the Windows Path.)
To get the MuPDF rendering features without the GUI dependencies you can alternately use pdfCropMargins[mupdf] in the plain install above.
The program can be run 1) from the command line, 2) from the command line with a GUI, 3) from a Python program, or 3) from the source repo.
Running from the command line
After installation via pip the program can be run with a command such as:
pdf-crop-margins -v -s -u your-file.pdf
That command prints verbose output, forces all pages to be the same size (-s) and then crops each page by the same amount (-u) for a uniform appearance, retaining the default of 10% of the margins. To run the same command with the GUI for fine-tuning, use:
pdf-crop-margins -v -s -u -gui your-file.pdf
For help and to see the many command-line options that are available, run:
pdf-crop-margins -h | more
The full output of that command is also listed below, at the bottom of this page. On Windows you may need to explicitly put the Python distribution’s Scripts directory into your environment PATH in order to avoid having to use the full pathname.
Running with the GUI
To run the GUI, assuming that version of pdfCropMargins has been installed, just pass the -gui flag in addition to any other flags. The program is still a command-line application, and still respects all the flags, but the GUI lets you fine-tune the values of some of the command-line arguments such as the percent to crop, etc. The output filenames, etc., are all the same as for the command-line version. Options modifiable from the GUI are initially set to any values passed in on the command line.
The graphical display shows the effect of each crop when you hit the ‘Crop’ button. Multiple cropping calls for the same document tend to be faster because the PDF pages usually only need to be rendered to images one time.
The program can also be called from a user’s Python program (when the pdfCropMargins package is discoverable in the Python path). Just import the crop function and then call it with a list containing the usual command-line arguments as separate items. For example:
from pdfCropMargins import crop crop(["-p", "20", "-u", "-s", "paper1.pdf"]) crop(["-p", "0", "-gui", "paper2.pdf"])
Any necessary exception handling should be performed by the calling code. The code may call sys.exit, so checking for SystemExit or BaseException may be required.
The crop function always returns four values, some of which may be set to None: the output file path, the exit code, the text written to standard input, and the text written to standard output. If the keyword argument string_io is set true then standard output and standard error streams are temporarily redirected to capture any output text as strings, which are returned as the final two arguments. Otherwise those values are set to None. The keyword argument quiet implies the string_io argument, but does not echo anything to the terminal while the crop function runs.
output_doc_pathname, exit_code, stdout_str, stderr_str = crop( ["-p4", "0", "10", "0", "10", "paper2.pdf"], string_io=True)
Running from the source distribution
The pdfCropMargins program can be run directly from the source code directory tree, provided the dependencies are installed. Just clone the repo and run the program pdfCropMargins.py located in the bin directory.
To pip install the program and its dependencies from the cloned repo rather than from PyPI just go to the root of the source directory and run pip install .[gui] for the GUI version or pip install . for the non-GUI version. (As usual, for code development use the -e option to make the code editable.)
Getting good crops
To diagnose unexpected crops, try running with the -v verbose argument. It will tell you the page with the smallest crop amount for a margin, among other data.
Running without the -u or -s options will crop each page separately, so you can see which pages might be causing problems (such as pages with noise near the borders or margin text).
To get cropped pages all the same size be sure to use both -s to first make pages the same size and -u to crop each page by the same amount.
Sometimes a small pre-crop is needed (-ap or -ap4) to get past small, unwanted markings near the edges of pages.
To see the documentation, run:
pdf-crop-margins -h | more
The output of that command follows:
Usage: pdf-crop-margins [-h] [-o OUTFILE_NAME] [-v] [-gui] [-p PCT] [-p4 PCT PCT PCT PCT] [-pt] [-a BP] [-a4 BP BP BP BP] [-ap BP] [-ap4 BP BP BP BP] [-u] [-m INT] [-m4 INT INT INT INT] [-mp INT] [-s] [-ms INT] [-e] [-g PAGESTR] [-c [d|m|p|gr|gb|o]] [-gs] [-gsr] [-t BYTEVAL] [-nb INT] [-ns INT] [-x DPI] [-y DPI] [-b [m|c|t|a|b]] [-f [m|c|t|a|b]] [-r] [-A] [-gsf] [-nc] [-pv PROG] [-mo] [-q] [-nco] [-pf] [-sc STR] [-su STR] [-ss STR] [-pw PASSWD] [-spr FLOAT:FLOAT] [-prw FLOAT FLOAT FLOAT FLOAT] [-dcb STR] [-dcw STR] [-i] [-pdl] [-gsp PATH] [-ppp PATH] [--version] [-wcdf FILEPATH] PDF_FILE [PDF_FILE ...] Description: A command-line application to crop the margins of PDF files. Cropping the margins can make it easier to read the pages of a PDF document -- whether the document is printed or displayed on a screen -- because the display fonts are larger. Margin-cropping is also sometimes useful when a PDF file is included in a document as a graphic. By default 10% of the existing margins will be retained; the rest will be eliminated. There are many options which can be set, however, including the percentage of existing margins to retain. Here is a simple example of cropping a file named document.pdf and writing the cropped output-document to a file named croppedDocument.pdf: pdf-crop-margins document.pdf -o croppedDocument.pdf If no destination is provided a filename will be automatically generated from the name of the source file (see below). The pdfCropMargins program works by changing the page sizes which are stored in the PDF file (and are interpreted by programs like Acrobat Reader). Both the CropBox and the MediaBox are set to the newly- computed cropped size. After this the view of the document in most programs will be the new, cropped view. In order to reduce the number of copies of a document which must be saved, a basic '--restore' option is provided. When cropping a file not produced by the pdfCropMargins program the default is to save the intersection of the MediaBox and any existing CropBox in the ArtBox. This saves the "usual" view of the original document in programs like Acrobat Reader. Subsequent crops of a file produced by pdfCropMargins do not by default alter the ArtBox. The restore option simply copies the saved values back to the MarginBox and CropBox. Note that this assumes the ArtBox is unused (it is rarely used, and this feature can be turned off with the -A option). So, for example, you can make annotations to a file with cropped margins and still produce a version with the annotations which viewers display as the original margins. Programs which change the "Producer" string in the PDF may interfere with this feature. Below are several examples using more of the command-line options, each applied to an input file called doc.pdf. The output filename is unspecified in most of these examples, so the program will automatically generate the filename (or an output filename can always be explicitly provided with '-o'): 1) Crop doc.pdf so that all the pages are set to the same size and the cropping amount is uniform across all the pages (this gives a nice two-up appearance). The default of retaining 10% of the existing margins is used. Note carefully that '-u' only makes the amount to be cropped uniform for each page; if the pages do not have the same size to begin with they will not have the same size afterward unless the '-s' option is also used. pdf-crop-margins -u -s doc.pdf 2) Crop each page of doc.pdf individually (i.e., not uniformly), keeping 50% of the existing margins. pdf-crop-margins -p 50 doc.pdf 3) Crop doc.pdf uniformly, keeping 50% of the left margin, 20% of the bottom margin, 40% of the right margin, and 10% of the top margin. pdf-crop-margins -u -p4 50 20 40 10 doc.pdf 4) Crop doc.pdf retaining 20% of the margins, and then reduce the right page margins only by an absolute 12 points. pdf-crop-margins -p 20 -a4 0 0 12 0 doc.pdf 5) Add a constant 5bp around the bare bounding boxes on all pages (note the negative value passed to the `-a` option, which adds space rather than removing it). pdf-crop-margins -p 0 -a -5 doc.pdf 6) Pre-crop the document by 5bp on each side before computing the bounding boxes. Then crop retaining 50% of the computed margins. This can be useful for difficult documents such as scanned books with page-edge noise or other "features" inside the current margins. pdf-crop-margins -ap 5 -p 50 doc.pdf 7) Crop doc.pdf, re-naming the cropped output file doc.pdf and backing up the original file in a file named backup_doc.pdf. pdf-crop-margins -mo -pf -su "backup" doc.pdf 8) Crop the margins of doc.pdf to 120% of their original size, increasing the margins. Use Ghostscript to find the bounding boxes without explicit rendering by pdfCropMargins. pdf-crop-margins -p 120 -c gb doc.pdf 9) Crop the margins of doc.pdf ignoring the 10 largest margins on each edge (over the whole document). This is especially good for noisy documents where all the pages have very similar margins, or when you want to ignore marginal annotations which only occur on a few pages. pdf-crop-margins -m 10 doc.pdf 10) Crop doc.pdf, launch the acroread viewer on the cropped output, and then query as to whether or not to rename the cropped file doc.pdf and back up the original file as doc_uncropped.pdf. pdf-crop-margins -mo -q doc.pdf 11) Crop pages 1-100 of doc.pdf, cropping all even pages uniformly and all odd pages uniformly. pdf-crop-margins -g 1-100 -e doc.pdf 12) Try to restore doc.pdf to its original margins, assuming it was cropped with pdfCropMargins previously. Note that the default output filename is still named doc_cropped.pdf, even though it is the recovered file. Use the '-mo' option to modify doc.pdf and backup the previous version. pdf-crop-margins -r doc.pdf There are many different ways to use this program. After finding a method which works well for a particular task or workflow pattern it is often convenient to make a simple shell script (batch file) which invokes the program with those particular options and settings. Simple template scripts for Bash and Windows are packaged with the program, in the bin directory. The program can also be called from a user's Python program (when discoverable in the Python path) by using code such as from pdfCropMargins import crop crop(["-p", "20", "-u", "-s", "paper.pdf"]) When printing a document with closely-cropped pages it may be necessary to use options such as "Fit to Printable Area". It may also be necessary to fine-tune the size of the retained margins if the edges of the text are being cut off. Sometimes a PDF file is corrupted or non-standard to the point where the routines used by this program raise an error and exit. In that case it can sometimes help to repair the PDF file before attempting to crop it. If it is readable by Ghostscript then the following command will often repair it sufficiently: gs -o repaired.pdf -sDEVICE=pdfwrite -dPDFSETTINGS=/prepress corrupted.pdf This command can also be used to convert some PostScript (.ps) files to PDF. The option '--gsFix' (or '-gsf') will automatically attempt to apply this fix, provided Ghostscript is available. See the description of that option for more information. The pdfCropMargins program handles rotated pages (such as pages in landscape mode versus portrait mode) as follows. All rotated pages are un-rotated as soon as they are read in. All the cropping is then calculated. Finally, as the crops are applied to the pages, the rotation is re-applied. This may give unexpected results in documents which mix pages at different rotations, especially with the '--uniform' or '--samePageSize' options. For rotated pages the arguments of all the options which take four arguments, one for each margin, are shifted so the left, bottom, right, and top margins correspond to the screen appearance (regardless of any internal rotation). All the command-line options to pdfCropMargins are described below. The following definition is useful in precisely defining what several of the options do. The "delta values" are the absolute reduction lengths, in points, which are applied to each original page to get the final cropped page. There is a delta value for each margin, on each page. In the usual case where all the margin sizes decrease, all the deltas are positive. A delta value can, however, be negative (e.g., when percentRetain > 100 or when a negative absolute offset is used). When a delta value is negative the corresponding margin size will increase. Positional arguments: PDF_FILE The pathname of the PDF file to crop. Use quotes around any file or directory name which contains a space. If no filename is given for the cropped PDF output file via the '-o' flag then a default output filename will be generated. By default it is the same as the source filename except that the suffix ".pdf" is replaced by "_cropped.pdf", overwriting by default if the file already exists. The file will be written to the working directory at the time when the program was run. If the input file has no extension or has an extension other than '.pdf' or '.PDF' then the suffix '.pdf' will be appended to the existing (possibly-null) extension. Globbing of wildcards is performed on Windows systems. Optional arguments: -h, --help Show this help message and exit. -o OUTFILE_NAME, --outfile OUTFILE_NAME An optional argument specifying the pathname of a file that the cropped output document should be written to. By default any existing file with the same name will be silently overwritten. If this option is not given the program will generate an output filename from the input filename. (By default "_cropped" is appended to the input filename before the file extension. If the extension is not '.pdf' or '.PDF' then '.pdf' is appended to the extension). Globbing of wildcards is performed on Windows systems. -v, --verbose Print more information about the program's actions and progress. Without this switch only warning and error messages are printed to the screen. -gui, --gui Run the graphical user interface. This mode allows you to interactively preview and test different cropping options without having to recalculate the bounding boxes each time (which can be slow). All the usual command-line options to the program are still respected. Clicking the 'Crop' button in the GUI crops with the current settings, writing out a cropped PDF file to the same filename that the command-line version would write to. Note that successive changes to the margins in the GUI are not cumulative: settings are always applied to the original document as it was passed in to the program. The 'Original' button reverts the display back to that original version. -p PCT, --percentRetain PCT Set the percent of margin space to retain in the image. This is a percentage of the original margin space. By default the percent value is set to 10. Setting the percentage to 0 gives a tight bounding box. Percent values greater than 100 increase the margin sizes from their original sizes, and negative values decrease the margins even more than a tight bounding box. -p4 PCT PCT PCT PCT, -pppp PCT PCT PCT PCT, --percentRetain4 PCT PCT PCT PCT Set the percent of margin space to retain in the image, individually for the left, bottom, right, and top margins, respectively. The four arguments should be percent values. -pt, --percentText Normally the percentage values passed to '--percentRetain' or ' --percentRetain4' define the percentage of existing margins to retain. This flag alters the interpretation of those percent values. The margins are instead set to the given percentage of the text width or height. The left and right margins are set to a percentage of the bounding box width and the bottom and top margins are set to a percentage of the bounding box height. -a BP, --absoluteOffset BP Decrease each margin size by an absolute floating point offset value, to be subtracted from each margin's size after the 'percentRetain' option is applied. The units are big points, bp, which is the unit used in PDF files. There are 72 bp in an inch. A single bp is approximately equal to a TeX point, pt (with 72.27pt in an inch). Negative values are allowed; positive numbers always decrease the margin size and negative numbers always increase it. Absolute offsets are always applied after any percentage change operations. -a4 BP BP BP BP, -aaaa BP BP BP BP, --absoluteOffset4 BP BP BP BP Decrease the margin sizes individually with four absolute offset values. The four floating point arguments should be the left, bottom, right, and top offset values, respectively. See the '--absoluteOffset' option for information on the units. -ap BP, --absolutePreCrop BP This option is like '--absoluteOffset' except that it is applied before any bounding box calculations (or any other operations). The argument is the same, in units of bp. All successive operations are then relative to this pre-crop box, considered to be the full-page box. Note that since this absolute crop is applied before any bounding boxes are computed it is relative to the original full-page boxes of the document (unlike 'absoluteOffset', which is a crop relative to the newly-cropped margin after 'percentRetain' is applied). As a consequence, the number of points may need to be larger than what would work for 'absoluteOffset'. This option can be used to ignore text and markings out at the edge of the margins by cropping it out before the bounding boxes are calculated. -ap4 BP BP BP BP, --absolutePreCrop4 BP BP BP BP This is the same as '--absolutePreCrop' except that four separate arguments can be given. The four floating point arguments should be the left, bottom, right, and top absolute pre-crop values, respectively. -u, --uniform Crop all the pages uniformly. This forces the magnitude of margin-cropping (absolute, not relative) to be the same on each page. This option is applied after all the delta values have been calculated for each page, individually. Then all the left- margin delta values, for each page, are set to the smallest left-margin delta value over every page. The bottom, right, and top margins are processed similarly. Note that this effectively adds some margin space (relative to the margins obtained by cropping pages individually) to some of the pages. If the pages of the original document are all the same size then the cropped pages will again all be the same size. The '--samePageSize' option can be used in combination with this option to force all pages to be the same size after cropping. -m INT, --uniformOrderStat INT Choosing this option implies the '--uniform' option, but the smallest delta value over all the pages is no longer chosen. Instead, for each margin the nth smallest delta value (with n numbered starting at zero) is chosen over all the pages. The argument is the integer n, for example '-m 4'. Choosing n to be half the number of pages gives the median delta value. This option is useful for cropping noisy scanned PDFs which have a common margin size on most of the pages, or for ignoring annotations which only appear in the margins of a few pages. This option essentially causes the program to ignores the n largest tight-crop margins when computing common delta values over all the pages. Increasing n always either increases the cropping amount or leaves it unchanged. Some trial-and-error may be needed to choose the best number. Using '-m 1' tends to work well with arXiv papers (which have a date in the margin of the first page). -m4 INT INT INT INT, -mmmm INT INT INT INT, --uniformOrderStat4 INT INT INT INT This option is the same as '--uniformOrderStat' (or '-m') except that separate values are specified for each margin individually. The margins are ordered as left, bottom, right, and top. -mp INT, --uniformOrderPercent INT This option is the same as '--uniformOrderStat' except that the order number n is automatically set to a given percentage of the number of pages which are set to be cropped (either the full number or the ones set with '--pages'). This option overrides '--uniformOrderStat' if both are set. The argument is a float percent value; rounding is done to get the final order- number. Setting the percent to 0 is equivalent to n=1, setting the percent to 100 is equivalent to setting n to the full number of pages, and setting the percent to 50 gives the median (for odd numbers of pages). -s, --samePageSize Set all the page sizes to be equal. This option only has an effect when the page sizes are different. The pages sizes are set to the size of the union of all the page regions, i.e., to the smallest bounding box which contains all the pages. This operation is always done before any others (except '-- absolutePreCrop'). The cropping is then done as usual, but note that any margin percentages (such as for '--percentRetain') are now relative to this new, possibly larger, page size. The resulting pages are still cropped independently by default, and will not necessarily all have the same size unless '--uniform' is also selected to force the cropping amounts to be the same for each page. If pages are selected with '--pages' then this option is only applied to those selected pages. -ms INT, --samePageSizeOrderStat INT Choosing this option implies the '--samePageSize' option, but the calculations for each edge of the smallest bounding box ignore the largest (or smallest for left and bottom edges) n values. The argument is the nonnegative number n. Each edge is calculated independently. This is an order statistic for selecting the uniform size to make the pages. Note that this will cut off parts of some pages if n>0. -e, --evenodd Crop all the odd pages uniformly, and all the even pages uniformly. The largest amount of cropping that works for all the pages in each group is chosen. If the '--uniform' ('-u') option is simultaneously set then the vertical cropping will be uniform over all the pages and only the horizontal cropping will differ between even and odd pages. -g PAGESTR, -pg PAGESTR, --pages PAGESTR Apply the cropping operation only to the selected pages. The argument should be a list of the usual form such as "2-4,5,9,20-30". The page-numbering is assumed to start at 1. Ordering in the argument list is unimportant, negative ranges are ignored, and pages falling outside the document are ignored. Note that restore information is always saved for all the pages (in the ArtBox) unless '--noundosave' is selected. -c [d|m|p|gr|gb|o], --calcbb [d|m|p|gr|gb|o] Choose the method to calculate bounding boxes (or to render the PDF pages in order to calculate the boxes). The default option 'd' will choose the MuPDF rendering option if the PyMuPDF dependency is installed, otherwise it will use pdftoppm rendering or Ghostscript rendering, in that order, if the external programs can be located. The options to force a particular method are MuPDF ('m'), pdftoppm ('p'), or Ghostscript ('gr') for rendering, or direct Ghostscript bounding-box calculation ('gb'). For pdftoppm or Ghostscript options the corresponding program must be installed and locatable (see the path-setting options below if the default locator fails). Only the explicit rendering methods will work for scanned pages (see '--gsBbox'). Choosing 'o' reverts to the old default behavior of first looking for pdftoppm and then looking for Ghostscript for rendering. -gs, --gsBbox This option is maintained for backward compatibility; using '-c gb' is now preferred. Use Ghostscript to directly find the bounding boxes for the pages, with no explict rendering of the pages. (The default is to explicitly render the PDF pages to image files and calculate bounding boxes from the images.) This method tends to be much faster, but it does not work with scanned PDF documents. It also does not allow for choosing the threshold value, applying blurs, etc. Any resolution options are passed to the Ghostscript bbox device. This option requires that Ghostscript be available in the PATH as "gswin32c.exe" or "gswin64c.exe" on Windows, or as "gs" on Linux. When this option is set the Pillow image library for Python is not required. -gsr, --gsRender This is maintained for backward compatibility; using '-c gr' is now preferred. Use Ghostscript to render the PDF pages to images. By default the pdftoppm program will be preferred for the rendering, if it is found. Note that this option has no effect if '--gsBbox' is chosen, since then no explicit rendering is done. -t BYTEVAL, --threshold BYTEVAL Set the threshold for determining what is background space (white). The value can be from 0 to 255, with 191 the default (75 percent). This option may not be available for some configurations since the PDF must be internally rendered as an image of pixels. In particular, it is ignored when '--gsBbox' is selected. Any pixel value over the threshold is considered to be background (white), and any value below it is considered to be text (black). Lowering the value should tend to make the bounding boxes smaller. The threshold may need to be lowered, for example, for scanned images with greyish backgrounds. For pages with dark backgrounds and light text a negative threshold value can be used. In that case the absolute value is used as the threshold but the test is reversed to consider pixel values greater than or equal to the threshold to be background. -nb INT, --numBlurs INT When PDF files are explicitly rendered to image files, apply a blur operation to the resulting images this many times. This can be useful for noisy images. -ns INT, --numSmooths INT When PDF files are explicitly rendered to image files, apply a smoothing operation to the resulting images this many times. This can be useful for noisy images. -x DPI, --resX DPI The x-resolution in dots per inch to use when the image is rendered to find the bounding boxes. The default is 150. Higher values produce more precise bounding boxes but require more time and memory. -y DPI, --resY DPI The y-resolution in dots per inch to use when the image is rendered to find the bounding boxes. The default is 150. Higher values produce more precise bounding boxes but require more time and memory. -b [m|c|t|a|b], --boxesToSet [m|c|t|a|b] By default the pdfCropMargins program sets both the MediaBox and the CropBox for each page of the cropped PDF document to the new, cropped page size. This default setting is usually sufficient, but this option can be used to select different PDF boxes to set. The option takes one argument, which is the first letter (lowercase) of a type of box. The choices are MediaBox (m), CropBox (c), TrimBox (t), ArtBox (a), and BleedBox (b). This option overrides the default and can be repeated multiple times to set several box types. -f [m|c|t|a|b], --fullPageBox [m|c|t|a|b] By default the program first (before any cropping is calculated) sets the MediaBox and CropBox of each page in (a copy of) the document to the intersection of its previous MediaBox and CropBox. This ensures that the cropping is relative to the usual document-view in programs like Acrobat Reader. This essentially defines what is assumed to be the full size of pages in the document, and all cropping is then performed relative to that full-page size. This option can be used to alternately use the MediaBox, the CropBox, the TrimBox, the ArtBox, or the BleedBox in defining the full-page size. The option takes one argument, which is the first letter (lowercase) of the type of box to use. If the option is repeated then the intersection of all the box arguments is used. Only one choice is allowed in combination with the '-gs' option since Ghostscript does its own internal rendering when finding bounding boxes. The default with '-gs' is the CropBox. -r, --restore This is a simple undo operation which essentially undoes all the crops ever made by pdfCropMargins and returns to the original margins (provided no other program modified the Producer metadata or ArtBoxes). By default, whenever this program crops a file for the first time it saves the MediaBox intersected with the CropBox as the new ArtBox (since the ArtBox is rarely used). The Producer metadata is checked to see if this was the first time. If so, the ArtBox for each page is simply copied to the MediaBox and the CropBox for the page. This restores the earlier view of the document, such as in Acrobat Reader (but does not completely restore the previous condition in cases where the MediaBox and CropBox differed or the ArtBox had a previous value). Any options such as '-u', '-p', and '-a' which do not make sense in a restore operation are ignored. Note that as far as default filenames the operation is treated as just another crop operation (the default-generated output filename still has a "_cropped.pdf" suffix). The '--modifyOriginal' option (or its query variant) can be used with this option. Saving in the ArtBoxes can be disabled by using the '--noundosave' option. -A, --noundosave Do not save any restore data in the ArtBox. This option will need to be selected if the document actually uses the ArtBox for anything important (which is rare). Note that the '-- restore' operation will not work correctly for the cropped document if this option is included in the cropping command. (The program does not currently check for this when doing a restore.) -gsf, --gsFix Attempt to repair the input PDF file with Ghostscript before it is read-in with PyPdf. This requires that Ghostscript be available. (See the general description text above for the actual command that is run.) This can also be used to automatically convert some PostScript files (.ps) to PDF for cropping. The repaired PDF is written to a temporary file; the original PDF file is not modified. The original filename is treated as usual as far as automatic name-generation, the '-- modify-original' option, and so forth. This option is often helpful if the program hangs or raises an error due to a corrupted PDF file. Note that when re-cropping a file already cropped by pdfCropMargins this option is probably not necessary, and if it is used in a re-crop (at least with current versions of Ghostscript) it will reset the Producer metadata which the pdfCropMargins program uses to tell if the file was already cropped by the program (so the '--restore' option will not work in combination with this option). This option is not recommended as something to use by default unless you encounter many corrupted PDF files and do not need to restore back to the original margins. -nc, --noclobber Never overwrite an existing file as the output file. -pv PROG, --preview PROG Run a PDF viewer on the cropped PDF output. The viewer process is run in the background. The viewer is launched after pdfCropMargins has finished all the other options. The only exception is when the '--queryModifyOriginal' option is also selected. In that case the viewer is launched before the query so that the user can look at the output before deciding whether or not to modify the original. (Note that answering 'y' will then move the file out from under the running viewer; close and re-open the file before adding annotations, highlighting, etc.) The single argument should be the path of the executable file or script to run the chosen viewer. The viewer is assumed to take exactly one argument, a PDF filename. For example, on Linux the Acrobat Reader could be chosen with /usr/bin/acroread or, if it is in the PATH, simply acroread. A shell script or batch file wrapper can be used to set any additional options for the viewer. -mo, --modifyOriginal This option moves (renames) the original file to a backup filename and then moves the cropped file to the original filename. Thus it effectively modifies the original file and makes a backup copy of the original, unmodified file. The backup filename for the original document is always generated from the original filename; any prefix or suffix which would be added by the program to generate a filename (by default a "_cropped" suffix) is modified accordingly (by default to "_uncropped"). The '--usePrefix', '--stringUncropped', and '-- stringSeparator' options can all be used to customize the generated backup filename. This operation is performed last, so if a previous operation fails the original document will be unchanged. Be warned that running pdfCropMargins twice on the same source filename will modify the original file; the '-noclobberOriginal' option can be used to avoid this. -q, --queryModifyOriginal This option selects the '--modifyOriginal' option, but queries the user about whether to actually do the final move operation. This works well with the '--preview' option: if the preview looks good you can opt to modify the original file (keeping a copy of the original). If you decline then the files are not swapped (and are just as if the '--modifyOriginal' option had not been set). -nco, --noclobberOriginal If the '--modifyOriginal' option is selected, do not ever overwrite an existing file as the backup copy for the original file. This essentially does the move operations for the '-- modifyOriginal' option in noclobber mode, and prints a warning if it fails. On failure the result is exactly as if the '-- modifyOriginal' option had not been selected. This option is redundant if the ordinary '--noclobber' option is also set. -pf, --usePrefix Prepend a prefix-string when generating default file names rather than appending a suffix-string. The same string value is used, either the default or the one set via the '-- stringCropped' or '--stringUncropped' option. With the default values for the other options and no output file specified, this option causes the cropped output for the input file "document.pdf" to be written to the file named "cropped_document.pdf" (instead of to the default filename "document_cropped.pdf"). -sc STR, --stringCropped STR This option can be used to set the string which will be appended (or prepended) to the document filename when automatically generating the output filename for a cropped file. The default value is "cropped". -su STR, --stringUncropped STR This option can be used to set the string which will be appended (or prepended) to the document filename when automatically generating the output filename for the original, uncropped file. The default value is "uncropped". -ss STR, --stringSeparator STR This option can be used to set the separator string which will be used when appending or prependeding string values to automatically generate filenames. The default value is "_". -pw PASSWD, --password PASSWD Specify a password to be used to decrypt an encrypted PDF file. Note that decrypting with an empty password is always tried, so this option is only needed for non-empty passwords. The resulting cropped file will not be encrypted, so use caution if important data is involved. -spr FLOAT:FLOAT, --setPageRatios FLOAT:FLOAT Force all the cropped page ratios to equal the given ratio. All crops are calculated and applied as usual, but either the left and right margins will be increased equally or else the top and bottom margins will be increased equally in order to make the ratio of width to height equal the set value. Margins are only ever increased. The format for the ratio is either a string width-to-height ratio such as '4.5:3' or else a floating point number like '0.75' which is the width divided by the height. This option can be useful in some PDF viewers. -prw FLOAT FLOAT FLOAT FLOAT, --pageRatioWeights FLOAT FLOAT FLOAT FLOAT This option weights any whitespace added by the '-- setPageRatios' argument. It takes four weight arguments, one per margin. The four floating point arguments should be the left, bottom, right, and top weights, respectively. The weights determine what proportion of the total height(width) increase necessary to achieve the target page ratio is added to the corresponding margin. All weights must be greater than zero. -dcb STR, --docCatBlacklist STR Data associated with the full document, such as outlines, bookmarks, and modes, is saved in the document catalog of the PDF file. By default it is all copied over to the cropped document if possible (with some exceptions, e.g., "/Pages" will be modified by cropping). If this is not wanted, or if it causes problems for a document, this option specifies a blacklist of document catalog items that will never be copied. The arguments should be passed as a single, quoted, whitespace- separated string, for example "/Outlines /PageMode /OpenAction". The special value "ALL" blacklists everything. An empty string blacklists nothing, and is the default. As an example, it can be useful to blacklist "/OpenAction" if on opening the PDF it does something like zooming which is not desired. Blacklisting `/PageMode` can also be useful to, for example, not open the outline by default. Running in verbose mode '-v' will show which document catalog items are and are not being copied for a document. -dcw STR, --docCatWhitelist STR See the '--docCatBlacklist' option. This is just a whitelist that essentially works the same way. The whitelist takes precedence over the blacklist. It specifies a list of items which will always be copied over even if they are in the blacklist. Useful combined with setting the blacklist to "ALL" if you only want one or two of the items. The default value is the empty string, which whitelists nothing and so only the blacklist is used. Setting to "ALL" guarantees that everything possible is copied over. -i, --showImages When explicitly rendering PDF files to image files, display the inverse image files that are used to find the bounding boxes. Useful for debugging and for choosing some of the other parameters (such as the threshold). This option requires a default external viewer program selected by the Pillow image manipulation package (xv on Unix, and usually Paint on Windows). -pdl, --pdftoppmLocal Use a locally-packaged pdftoppm executable rather than the system version. This option is only available on Windows machines; it is ignored otherwise. By default the first pdftoppm executable found in the directories in the PATH environment variable is used. On Windows the program will revert to this option if PDF image-rendering is required, PyMuPDF is not installed, and no system pdftoppm or Ghostscript executable can be found. -gsp PATH, --ghostscriptPath PATH Pass in a pathname to the ghostscript executable that the program should use. No globbing is done. Useful when the program is in a nonstandard location. -ppp PATH, --pdftoppmPath PATH Pass in a pathname to the pdftoppm executable that the program should use. No globbing is done. Useful when the program is in a nonstandard location. --version Return the pdfCropMargins version number and exit immediately. All other options are ignored. -wcdf FILEPATH, --writeCropDataToFile FILEPATH Write out the calculated list of crops to the file with the file pathname that is passed in and exit. Mostly used for automated testing and debugging. The pdfCropMargins program is Copyright (c) 2014 by Allen Barker. Released under the GNU GPL license, version 3 or later.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.