Using Spec Files

When you execute

pyinstaller options..

the first thing PyInstaller does is to build a spec (specification) file myscript.spec. That file is stored in the --specpath directory, by default the current directory.

The spec file tells PyInstaller how to process your script. It encodes the script names and most of the options you give to the pyinstaller command. The spec file is actually executable Python code. PyInstaller builds the app by executing the contents of the spec file.

For many uses of PyInstaller you do not need to examine or modify the spec file. It is usually enough to give all the needed information (such as hidden imports) as options to the pyinstaller command and let it run.

There are four cases where it is useful to modify the spec file:

  • When you want to bundle data files with the app.

  • When you want to include run-time libraries (.dll or .so files) that PyInstaller does not know about from any other source.

  • When you want to add Python run-time options to the executable.

  • When you want to create a multiprogram bundle with merged common modules.

These uses are covered in topics below.

You create a spec file using this command:

pyi-makespec options [other scripts …]

The options are the same options documented above for the pyinstaller command. This command creates the name.spec file but does not go on to build the executable.

After you have created a spec file and modified it as necessary, you build the application by passing the spec file to the pyinstaller command:

pyinstaller options name.spec

When you create a spec file, most command options are encoded in the spec file. When you build from a spec file, those options cannot be changed. If they are given on the command line they are ignored and replaced by the options in the spec file.

Only the following command-line options have an effect when building from a spec file:

Spec File Operation

After PyInstaller creates a spec file, or opens a spec file when one is given instead of a script, the pyinstaller command executes the spec file as code. Your bundled application is created by the execution of the spec file. The following is a shortened example of a spec file for a minimal, one-folder app:

a = Analysis([''],
pyz = PYZ(a.pure)
exe = EXE(pyz,... )
coll = COLLECT(...)

The statements in a spec file create instances of four classes, Analysis, PYZ, EXE and COLLECT.

  • A new instance of class Analysis takes a list of script names as input. It analyzes all imports and other dependencies. The resulting object (assigned to a) contains lists of dependencies in class members named:

    • scripts: the python scripts named on the command line;

    • pure: pure python modules needed by the scripts;

    • pathex: a list of paths to search for imports (like using PYTHONPATH), including paths given by the --paths option.

    • binaries: non-python modules needed by the scripts, including names given by the --add-binary option;

    • datas: non-binary files included in the app, including names given by the --add-data option.

  • An instance of class PYZ is a .pyz archive (described under Inspecting Archives below), which contains all the Python modules from a.pure.

  • An instance of EXE is built from the analyzed scripts and the PYZ archive. This object creates the executable file.

  • An instance of COLLECT creates the output folder from all the other parts.

In one-file mode, there is no call to COLLECT, and the EXE instance receives all of the scripts, modules and binaries.

You modify the spec file to pass additional values to Analysis and to EXE.

Adding Files to the Bundle

To add files to the bundle, you create a list that describes the files and supply it to the Analysis call. When you bundle to a single folder (see Bundling to One Folder), the added data files are copied into the folder with the executable. When you bundle to a single executable (see Bundling to One File), copies of added files are compressed into the executable, and expanded to the _MEIxxxxxx temporary folder before execution. This means that any changes a one-file executable makes to an added file will be lost when the application ends.

In either case, to find the data files at run-time, see Run-time Information.

Adding Data Files

You can add data files to the bundle by using the --add-data command option, or by adding them as a list to the spec file.

When using the spec file, provide a list that describes the files as the value of the datas= argument to Analysis. The list of data files is a list of tuples. Each tuple has two values, both of which must be strings:

  • The first string specifies the file or files as they are in this system now.

  • The second specifies the name of the folder to contain the files at run-time.

For example, to add a single README file to the top level of a one-folder app, you could modify the spec file as follows:

a = Analysis(...
         datas=[ ('src/README.txt', '.') ],

And the command line equivalent:

pyinstaller --add-data "src/README.txt:."

You have made the datas= argument a one-item list. The item is a tuple in which the first string says the existing file is src/README.txt. That file will be looked up (relative to the location of the spec file) and copied into the top level of the bundled app.

The strings may use either / or \ as the path separator character. You can specify input files using “glob” abbreviations. For example to include all the .mp3 files from a certain folder:

a = Analysis(...
         datas= [ ('/mygame/sfx/*.mp3', 'sfx' ) ],

All the .mp3 files in the folder /mygame/sfx will be copied into a folder named sfx in the bundled app.

The spec file is more readable if you create the list of added files in a separate statement:

added_files = [
         ( 'src/README.txt', '.' ),
         ( '/mygame/sfx/*.mp3', 'sfx' )
a = Analysis(...
         datas = added_files,

You can also include the entire contents of a folder:

added_files = [
         ( 'src/README.txt', '.' ),
         ( '/mygame/data', 'data' ),
         ( '/mygame/sfx/*.mp3', 'sfx' )

The folder /mygame/data will be reproduced under the name data in the bundle.

Using Data Files from a Module

If the data files you are adding are contained within a Python module, you can retrieve them using pkgutil.get_data().

For example, suppose that part of your application is a module named helpmod. In the same folder as your script and its spec file you have this folder arrangement:


Because your script includes the statement import helpmod, PyInstaller will create this folder arrangement in your bundled app. However, it will only include the .py files. The data file help_data.txt will not be automatically included. To cause it to be included also, you would add a datas tuple to the spec file:

a = Analysis(...
         datas= [ ('helpmod/help_data.txt', 'helpmod' ) ],

When your script executes, you could find help_data.txt by using its base folder path, as described in the previous section. However, this data file is part of a module, so you can also retrieve its contents using the standard library function pkgutil.get_data():

import pkgutil
help_bin = pkgutil.get_data( 'helpmod', 'help_data.txt' )

This returns the contents of the help_data.txt file as a binary string. If it is actually characters, you must decode it:

help_utf = help_bin.decode('UTF-8', 'ignore')

Adding Binary Files


Binary files refers to DLLs, dynamic libraries, shared object-files, and such, which PyInstaller is going to search for further binary dependencies. Files like images and PDFs should go into the datas.

You can add binary files to the bundle by using the --add-binary command option, or by adding them as a list to the spec file. In the spec file, make a list of tuples that describe the files needed. Assign the list of tuples to the binaries= argument of Analysis.

Adding binary files works in a similar way as adding data files. As described in Adding Binary Files, each tuple should have two values:

  • The first string specifies the file or files as they are in this system now.

  • The second specifies the name of the folder to contain the files at run-time.

Normally PyInstaller learns about .so and .dll libraries by analyzing the imported modules. Sometimes it is not clear that a module is imported; in that case you use a --hidden-import command option. But even that might not find all dependencies.

Suppose you have a module that is written in C and uses the Python C-API. Your program imports special_ops, and PyInstaller finds and includes But perhaps links to libiodbc.2.dylib. PyInstaller does not find this dependency. You could add it to the bundle this way:

a = Analysis(...
         binaries=[ ( '/usr/lib/libiodbc.2.dylib', '.' ) ],

Or via the command line:

pyinstaller --add-binary "/usr/lib/libiodbc.2.dylib:."

If you wish to store libiodbc.2.dylib on a specific folder inside the bundle, for example vendor, then you could specify it, using the second element of the tuple:

a = Analysis(...
         binaries=[ ( '/usr/lib/libiodbc.2.dylib', 'vendor' ) ],

As with data files, if you have multiple binary files to add, to improve readability, create the list in a separate statement and pass the list by name.

Advanced Methods of Adding Files

PyInstaller supports a more advanced (and complex) way of adding files to the bundle that may be useful for special cases. See The Table of Contents (TOC) lists and the Tree Class below.

Specifying Python Interpreter Options

PyInstaller-frozen application runs the application code in isolated, embedded Python interpreter. Therefore, the typical means of passing options to Python interpreter do not apply, including:

  • environment variables (such as PYTHONUTF8 and PYTHONHASHSEED) - because the frozen application is supposed to be isolated from python environment that might be present on the target system

  • command-line arguments (such as -v and -O) - because command-line arguments are reserved for application.

Instead, PyInstaller offers an option to specify permanent run-time options for the application’s Python interpreter via its own OPTIONS mechanism. To pass run-time options, create a list of three-element tuples: (‘option string’, None, ‘OPTION’), and pass it as an additional argument to EXE before the keyword arguments. The first element of the option tuple is the option string (see below for valid options), the second is always None, and the third is always ‘OPTION’.

An example spec file, modified to specify two run-time options:

options = [
    ('v', None, 'OPTION'),
    ('W ignore', None, 'OPTION'),

a = Analysis(
exe = EXE(
    options,  # <-- the options list, passed to EXE

The following options are supported by this mechanism:

  • 'v' or 'verbose': increment the value of sys.flags.verbose, which causes messages to be written to stdout each time a module is initialized. This option is equivalent to Python’s -v command-line option. It is automatically enabled when verbose imports are enabled via PyInstaller’s own ˙˙–debug imports`` option.

  • 'u' or 'unbuffered': enable unbuffered stdout and stderr. Equivalent to Python’s -u command-line option.

  • 'O' or 'optimize': increment the value of sys.flags.optimize. Equivalent to Python’s -O command-line option.


The optimization level reflected by sys.flags.optimize affects only bytecode that python interpreter would end up compiling at the run time. In a PyInstaller-frozen application, however, most of python modules are available as pre-compiled bytecode, hence the run-time bytecode optimization level does not affect them at all.

For details on how to enforce the bytecode optimization level for collected modules, see Bytecode Optimization Level.

  • 'W <arg>': a pass-through for Python’s W-options that control warning messages.

  • 'X <arg>': a pass-through for Python’s X-options. The utf8 and dev X-options, which control UTF-8 mode and developer mode, are explicitly parsed by PyInstaller’s bootloader and used during interpreter pre-initialization; the rest of X-options are just passed on to the interpreter configuration.

  • 'hash_seed=<value>': an option to set Python’s hash seed within the frozen application to a fixed value. Equivalent to PYTHONHASHSEED environment variable. At the time of writing, this does not exist as an X-option, so it is implemented as a custom option.

Further examples to illustrate the syntax:

options = [
    # Warning control
    ('W ignore', None, 'OPTION'),  # disable all warnings
    ('W ignore::DeprecationWarning', None, 'OPTION')  # disable deprecation warnings

    # UTF-8 mode; unless explicitly enabled/disabled, it is auto enabled based on locale
    ('X utf8', None, 'OPTION),  # force UTF-8 mode on
    ('X utf8=1', None, 'OPTION),  # force UTF-8 mode on
    ('X utf8=0', None, 'OPTION),  # force UTF-8 mode off

    # Developer mode; disabled by default
    ('X dev', None, 'OPTION),  # enable dev mode
    ('X dev=1', None, 'OPTION),  # enable dev mode

    # Hash seed
    ('hash_seed=0', None, 'OPTION'),  # disable hash randomization; sys.flags.hash_randomization=0
    ('hash_seed=123', None, 'OPTION'),  # hash randomization with fixed seed value

Spec File Options for a macOS Bundle

When you build a windowed macOS app (that is, running under macOS, you specify the --windowed option), the spec file contains an additional statement to create the macOS application bundle, or app folder:

app = BUNDLE(exe,

The icon= argument to BUNDLE will have the path to an icon file that you specify using the --icon option. The bundle_identifier will have the value you specify with the --osx-bundle-identifier option.

An Info.plist file is an important part of a macOS app bundle. (See the Apple bundle overview for a discussion of the contents of Info.plist.)

PyInstaller creates a minimal Info.plist. The version option can be used to set the application version using the CFBundleShortVersionString Core Foundation Key.

You can add or overwrite entries in the plist by passing an info_plist= parameter to the BUNDLE call. Its argument should be a Python dict with keys and values to be included in the Info.plist file. PyInstaller creates Info.plist from the info_plist dict using the Python Standard Library module plistlib. plistlib can handle nested Python objects (which are translated to nested XML), and translates Python data types to the proper Info.plist XML types. Here’s an example:

app = BUNDLE(exe,
            'NSPrincipalClass': 'NSApplication',
            'NSAppleScriptEnabled': False,
            'CFBundleDocumentTypes': [
                    'CFBundleTypeName': 'My File Format',
                    'CFBundleTypeIconFile': 'MyFileIcon.icns',
                    'LSItemContentTypes': ['com.example.myformat'],
                    'LSHandlerRank': 'Owner'

In the above example, the key/value 'NSPrincipalClass': 'NSApplication' is necessary to allow macOS to render applications using retina resolution. The key 'NSAppleScriptEnabled' is assigned the Python boolean False, which will be output to Info.plist properly as <false/>. Finally the key CFBundleDocumentTypes tells macOS what filetypes your application supports (see Apple document types).

POSIX Specific Options

By default all required system libraries are bundled. To exclude all or most non-Python shared system libraries from the bundle, you can add a call to the function exclude_system_libraries from the Analysis class. System libraries are defined as files that come from under /lib* or /usr/lib* as is the case on POSIX and related operating systems. The function accepts an optional parameter that is a list of file wildcards exceptions, to not exclude library files that match those wildcards in the bundle. For example to exclude all non-Python system libraries except “libexpat” and anything containing “krb” use this:

a = Analysis(...)

a.exclude_system_libraries(list_of_exceptions=['libexpat*', '*krb*'])

The Splash Target

For a splash screen to be displayed by the bootloader, the Splash target must be called at build time. This class can be added when the spec file is created with the command-line option --splash IMAGE_FILE. By default, the option to display the optional text is disabled (text_pos=None). For more information about the splash screen, see Splash Screen (Experimental) section. The Splash Target looks like this:

a = Analysis(...)

splash = Splash('image.png',
                text_pos=(10, 50),

Splash bundles the required resources for the splash screen into a file, which will be included in the CArchive.

A Splash has two outputs, one is itself and one is stored in splash.binaries. Both need to be passed on to other build targets in order to enable the splash screen. To use the splash screen in a onefile application, please follow this example:

a = Analysis(...)

splash = Splash(...)

# onefile
exe = EXE(pyz,
          splash,                   # <-- both, splash target
          splash.binaries,          # <-- and splash binaries

In order to use the splash screen in a onedir application, only a small change needs to be made. The splash.binaries attribute has to be moved into the COLLECT target, since the splash binaries do not need to be included into the executable:

a = Analysis(...)

splash = Splash(...)

# onedir
exe = EXE(pyz,
          splash,                   # <-- splash target
coll = COLLECT(exe,
               splash.binaries,     # <-- splash binaries

On Windows/macOS images with per-pixel transparency are supported. This allows non-rectangular splash screen images. On Windows the transparent borders of the image are hard-cuted, meaning that fading transparent values are not supported. There is no common implementation for non-rectangular windows on Linux, so images with per- pixel transparency is not supported.

The splash target can be configured in various ways. The constructor of the Splash target is as follows:

Splash.__init__(image_file, binaries, datas, **kwargs)
  • image_file (str) –

    A path-like object to the image to be used. Only the PNG file format is supported.


    If a different file format is supplied and PIL (Pillow) is installed, the file will be converted automatically.


    Windows: The color 'magenta' / '#ff00ff' must not be used in the image or text, as it is used by splash screen to indicate transparent areas. Use a similar color (e.g., '#ff00fe') instead.


    If PIL (Pillow) is installed and the image is bigger than max_img_size, the image will be resized to fit into the specified area.

  • binaries (list) – The TOC list of binaries the Analysis build target found. This TOC includes all extension modules and their binary dependencies. This is required to determine whether the user’s program uses tkinter.

  • datas (list) – The TOC list of data the Analysis build target found. This TOC includes all data-file dependencies of the modules. This is required to check if all splash screen requirements can be bundled.

Keyword Arguments
  • text_pos – An optional two-integer tuple that represents the origin of the text on the splash screen image. The origin of the text is its lower left corner. A unit in the respective coordinate system is a pixel of the image, its origin lies in the top left corner of the image. This parameter also acts like a switch for the text feature. If omitted, no text will be displayed on the splash screen. This text will be used to show textual progress in onefile mode.

  • text_size – The desired size of the font. If the size argument is a positive number, it is interpreted as a size in points. If size is a negative number, its absolute value is interpreted as a size in pixels. Default: 12

  • text_font – An optional name of a font for the text. This font must be installed on the user system, otherwise the system default font is used. If this parameter is omitted, the default font is also used.

  • text_color – An optional color for the text. HTML color codes ('#40e0d0') and color names ('turquoise') are supported. Default: 'black' (Windows: the color 'magenta' / '#ff00ff' is used to indicate transparency, and should not be used)

  • text_default – The default text which will be displayed before the extraction starts. Default: "Initializing"

  • full_tk – By default Splash bundles only the necessary files for the splash screen (some tk components). This options enables adding full tk and making it a requirement, meaning all tk files will be unpacked before the splash screen can be started. This is useful during development of the splash screen script. Default: False

  • minify_script – The splash screen is created by executing an Tcl/Tk script. This option enables minimizing the script, meaning removing all non essential parts from the script. Default: True

  • name – An optional alternative filename for the .res file. If not specified, a name is generated.

  • script_name – An optional alternative filename for the Tcl script, that will be generated. If not specified, a name is generated.

  • max_img_size – Maximum size of the splash screen image as a tuple. If the supplied image exceeds this limit, it will be resized to fit the maximum width (to keep the original aspect ratio). This option can be disabled by setting it to None. Default: (760, 480)

  • always_on_top – Force the splashscreen to be always on top of other windows. If disabled, other windows (e.g., from other applications) can cover the splash screen by user bringing them to front. This might be useful for frozen applications with long startup times. Default: True

Multipackage Bundles

Some products are made of several different apps, each of which might depend on a common set of third-party libraries, or share code in other ways. When packaging such a product it would be a pity to treat each app in isolation, bundling it with all its dependencies, because that means storing duplicate copies of code and libraries.

You can use the multipackage feature to bundle a set of executable apps so that they share single copies of libraries. You can do this with either one-file or one-folder apps.

Multipackaging with One-Folder Apps

For combining multiple one-folder applications, use a shared COLLECT statement. This will collect the external resources for all of the one-folder apps into one directory.

Multipackaging with One-File Apps

Each dependency (a DLL, for example) is packaged only once, in one of the apps. Any other apps in the set that depend on that DLL have an “external reference” to it, telling them to extract that dependency from the executable file of the app that contains it.

This saves disk space because each dependency is stored only once. However, to follow an external reference takes extra time when an app is starting up. All but one of the apps in the set will have slightly slower launch times.

The external references between binaries include hard-coded paths to the output directory, and cannot be rearranged. You must place all the related applications in the same directory when you install the application.

To build such a set of apps you must code a custom spec file that contains a call to the MERGE function. This function takes a list of analyzed scripts, finds their common dependencies, and modifies the analyses to minimize the storage cost.

The order of the analysis objects in the argument list matters. The MERGE function packages each dependency into the first script from left to right that needs that dependency. A script that comes later in the list and needs the same file will have an external reference to the prior script in the list. You might sequence the scripts to place the most-used scripts first in the list.

A custom spec file for a multipackage bundle contains one call to the MERGE function:


MERGE is used after the analysis phase and before EXE. Its variable-length list of arguments consists of a list of tuples, each tuple having three elements:

  • The first element is an Analysis object, an instance of class Analysis, as applied to one of the apps.

  • The second element is the script name of the analyzed app (without the .py extension).

  • The third element is the name for the executable (usually the same as the script).

MERGE examines the Analysis objects to learn the dependencies of each script. It modifies these objects to avoid duplication of libraries and modules. As a result the packages generated will be connected.

Example MERGE spec file

One way to construct a spec file for a multipackage bundle is to first build a spec file for each app in the package. Suppose you have a product that comprises three apps named (because we have no imagination) foo, bar and zap:

pyi-makespec options as appropriate…

pyi-makespec options as appropriate…

pyi-makespec options as appropriate…

Check for warnings and test each of the apps individually. Deal with any hidden imports and other problems. When all three work correctly, combine the statements from the three files foo.spec, bar.spec and zap.spec as follows.

First copy the Analysis statements from each, changing them to give each Analysis object a unique name:

foo_a = Analysis([''],

bar_a = Analysis([''], etc., etc...

zap_a = Analysis([''], etc., etc...

Now call the MERGE method to process the three Analysis objects:

MERGE( (foo_a, 'foo', 'foo'), (bar_a, 'bar', 'bar'), (zap_a, 'zap', 'zap') )

The Analysis objects foo_a, bar_a, and zap_a are modified so that the latter two refer to the first for common dependencies.

Following this you can copy the PYZ, EXE and COLLECT statements from the original three spec files, substituting the unique names of the Analysis objects where the original spec files have a. Modify the EXE statements to pass in Analysis.dependencies, in addition to all other arguments that are passed in the original EXE statements. For example:

foo_pyz = PYZ(foo_a.pure)
foo_exe = EXE(foo_pyz, foo_a.dependencies, foo_a.scripts, ... etc.

bar_pyz = PYZ(bar_a.pure)
bar_exe = EXE(bar_pyz, bar_a.dependencies, bar_a.scripts, ... etc.

Save the combined spec file as foobarzap.spec and then build it:

pyinstaller foobarzap.spec

The output in the dist folder will be all three apps, but the apps dist/bar and dist/zap will refer to the contents of dist/foo for shared dependencies.

Remember that a spec file is executable Python. You can use all the Python facilities (for and with and the members of sys and io) in creating the Analysis objects and performing the PYZ, EXE and COLLECT statements. You may also need to know and use The Table of Contents (TOC) lists and the Tree Class described below.

Globals Available to the Spec File

While a spec file is executing it has access to a limited set of global names. These names include the classes defined by PyInstaller: Analysis, BUNDLE, COLLECT, EXE, MERGE, PYZ, TOC, Tree and Splash, which are discussed in the preceding sections.

Other globals contain information about the build environment:


The relative path to the dist folder where the application will be stored. The default path is relative to the current directory. If the --distpath option is used, DISTPATH contains that value.


The absolute path to the PyInstaller distribution, typically in the current Python site-packages folder.


The complete spec file argument given to the pyinstaller command, for example myscript.spec or source/myscript.spec.


The path prefix to the SPEC value as returned by os.path.split().


The name of the spec file, for example myscript.


The path to the build directory. The default is relative to the current directory. If the workpath= option is used, workpath contains that value.


The full path to the warnings file in the build directory, for example build/warn-myscript.txt.

Adding parameters to spec files

Sometimes, you may wish to have different build modes (e.g. a debug build and a production build) from the same spec file. Any command line arguments to pyinstaller given after a -- separator will not be parsed by PyInstaller and will instead be forwarded to the spec file where you can implement your own argument parsing and handle the options accordingly. For example, the following spec file will create a onedir application with console enabled if invoked via pyinstaller example.spec -- --debug or a onefile console-less application if invoked with just pyinstaller example.spec. If you use an argparse based parser rather than rolling your own using sys.argv then pyinstaller example.spec -- --help will display your spec options.

# example.spec

import argparse

parser = argparse.ArgumentParser()
parser.add_argument("--debug", action="store_true")
options = parser.parse_args()

a = Analysis(
pyz = PYZ(a.pure)

if options.debug:
    exe = EXE(
    coll = COLLECT(
    exe = EXE(