I'm using Python 2.6 and PyGTK 2.22.6 from the all-in-one installer on Windows XP, trying to build a single-file executable (via py2exe) for my app.
My problem is that when I run my app as a script (ie. not built into an .exe file, just as a loose collection of .py files), it uses the native-looking Windows theme, but when I run the built exe I see the default GTK theme.
I know that this problem can be fixed by copying a bunch of files into the dist directory created by py2exe, but everything I've read involves manually copying the data, whereas I want this to be an automatic part of the build process. Furthermore, everything on the topic (including the FAQ) is out of date - PyGTK now keeps its files in C:\Python2x\Lib\site-packages\gtk-2.0\runtime\..., and just copying the lib and etc directories doesn't fix the problem.
My questions are:
I'd like to be able to programmatically find the GTK runtime data in setup.py rather than hard coding paths. How do I do this?
What are the minimal resources I need to include?
Update: I may have almost answered #2 by trial-and-error. For the "wimp" (ie. MS Windows) theme to work, I need the files from:
runtime\lib\gtk-2.0\2.10.0\engines\libwimp.dll
runtime\etc\gtk-2.0\gtkrc
runtime\share\icons\*
runtime\share\themes\MS-Windows
...without the runtime prefix, but otherwise with the same directory structure, sitting directly in the dist directory produced by py2exe. But where does the 2.10.0 come from, given that gtk.gtk_version is (2,22,0)?
Answering my own question here, but if anyone knows better feel free to answer too. Some of it seems quite fragile (eg. version numbers in paths), so comment or edit if you know a better way.
1. Finding the files
Firstly, I use this code to actually find the root of the GTK runtime. This is very specific to how you install the runtime, though, and could probably be improved with a number of checks for common locations:
#gtk file inclusion
import gtk
# The runtime dir is in the same directory as the module:
GTK_RUNTIME_DIR = os.path.join(
os.path.split(os.path.dirname(gtk.__file__))[0], "runtime")
assert os.path.exists(GTK_RUNTIME_DIR), "Cannot find GTK runtime data"
2. What files to include
This depends on (a) how much of a concern size is, and (b) the context of your application's deployment. By that I mean, are you deploying it to the whole wide world where anyone can have an arbitrary locale setting, or is it just for internal corporate use where you don't need translated stock strings?
If you want Windows theming, you'll need to include:
GTK_THEME_DEFAULT = os.path.join("share", "themes", "Default")
GTK_THEME_WINDOWS = os.path.join("share", "themes", "MS-Windows")
GTK_GTKRC_DIR = os.path.join("etc", "gtk-2.0")
GTK_GTKRC = "gtkrc"
GTK_WIMP_DIR = os.path.join("lib", "gtk-2.0", "2.10.0", "engines")
GTK_WIMP_DLL = "libwimp.dll"
If you want the Tango icons:
GTK_ICONS = os.path.join("share", "icons")
There is also localisation data (which I omit, but you might not want to):
GTK_LOCALE_DATA = os.path.join("share", "locale")
3. Piecing it together
Firstly, here's a function that walks the filesystem tree at a given point and produces output suitable for the data_files option.
def generate_data_files(prefix, tree, file_filter=None):
"""
Walk the filesystem starting at "prefix" + "tree", producing a list of files
suitable for the data_files option to setup(). The prefix will be omitted
from the path given to setup(). For example, if you have
C:\Python26\Lib\site-packages\gtk-2.0\runtime\etc\...
...and you want your "dist\" dir to contain "etc\..." as a subdirectory,
invoke the function as
generate_data_files(
r"C:\Python26\Lib\site-packages\gtk-2.0\runtime",
r"etc")
If, instead, you want it to contain "runtime\etc\..." use:
generate_data_files(
r"C:\Python26\Lib\site-packages\gtk-2.0",
r"runtime\etc")
Empty directories are omitted.
file_filter(root, fl) is an optional function called with a containing
directory and filename of each file. If it returns False, the file is
omitted from the results.
"""
data_files = []
for root, dirs, files in os.walk(os.path.join(prefix, tree)):
to_dir = os.path.relpath(root, prefix)
if file_filter is not None:
file_iter = (fl for fl in files if file_filter(root, fl))
else:
file_iter = files
data_files.append((to_dir, [os.path.join(root, fl) for fl in file_iter]))
non_empties = [(to, fro) for (to, fro) in data_files if fro]
return non_empties
So now you can call setup() like so:
setup(
# Other setup args here...
data_files = (
# Use the function above...
generate_data_files(GTK_RUNTIME_DIR, GTK_THEME_DEFAULT) +
generate_data_files(GTK_RUNTIME_DIR, GTK_THEME_WINDOWS) +
generate_data_files(GTK_RUNTIME_DIR, GTK_ICONS) +
# ...or include single files manually
[
(GTK_GTKRC_DIR, [
os.path.join(GTK_RUNTIME_DIR,
GTK_GTKRC_DIR,
GTK_GTKRC)
]),
(GTK_WIMP_DIR, [
os.path.join(
GTK_RUNTIME_DIR,
GTK_WIMP_DIR,
GTK_WIMP_DLL)
])
]
)
)
I have a trained a classic CNN(pre-trained mobile net) for image classification. I want to now use this model from c++. From my understanding, I need to create a library of the model, that can accept the input and return its outputs. I have the model saved in format .pb (SavedModel).
I have already tried, CppFlow, where the error shows that it can't read my model. I assume it's due to incompatibility with TF 2.0.
I have also got the command line interface of SavedModel working, but I don't know how to use it in my cpp application.
I want to know how I can build a library of my model and use this library such that it can make predictions on the fly. Any guidance will be helpful. Please let me know if any additional information is required.
One way of using keras model in C++ is to convert it to TensorFlow .pb format. I've just composed a script for doing this, down below.
Usage: python script.py keras_model.hdf5
It outputs tensorflow model as input file name appended by .pb.
Then you can use TF C++ api for reading model and doing inference. Nice detailed example of using image recognition model to label images in C++ TF is located here.
Another option - you may use Keras directly by calling Python API from C++, it is not that difficult, there is standalone python which is compiled statically meaning having no dll/shared libs dependencies at all hence python interpreter can be fully compiled into C++ single binary. There are also many libraries in Internet that help you to easily run Python from C++.
import sys, os
from keras import backend as K
from keras.models import load_model
import tensorflow as tf
def freeze_session(session, keep_var_names=None, output_names=None, clear_devices=True):
"""
Freezes the state of a session into a pruned computation graph.
Creates a new computation graph where variable nodes are replaced by
constants taking their current value in the session. The new graph will be
pruned so subgraphs that are not necessary to compute the requested
outputs are removed.
#param session The TensorFlow session to be frozen.
#param keep_var_names A list of variable names that should not be frozen,
or None to freeze all the variables in the graph.
#param output_names Names of the relevant graph outputs.
#param clear_devices Remove the device directives from the graph for better portability.
#return The frozen graph definition.
"""
from tensorflow.python.framework.graph_util import convert_variables_to_constants
graph = session.graph
with graph.as_default():
freeze_var_names = list(set(v.op.name for v in tf.global_variables()).difference(keep_var_names or []))
output_names = output_names or []
output_names += [v.op.name for v in tf.global_variables()]
# Graph -> GraphDef ProtoBuf
input_graph_def = graph.as_graph_def()
if clear_devices:
for node in input_graph_def.node:
node.device = ""
frozen_graph = convert_variables_to_constants(session, input_graph_def,
output_names, freeze_var_names)
return frozen_graph
if len(sys.argv) <= 1:
print('Usage: python script.py keras_model.hdf5')
sys.exit(0)
else:
ifname = sys.argv[1]
model = load_model(ifname)
frozen_graph = freeze_session(
K.get_session(),
output_names = [out.op.name for out in model.outputs],
)
tf.io.write_graph(frozen_graph, os.path.dirname(ifname), ifname + '.pb', as_text = False)
There are standalone third-party libraries that can import keras model into c++ for inference without doing much of work from our side.
Examples are
Multithreaded library for image segmentation models such as U-Net etc. - https://github.com/upashu1/keras2cpp_multithreading_image_segmentation
A library that supports the most of the layers
https://github.com/Dobiasd/frugally-deep
My application uses log4j but OkHttpClient uses java util logging. So apart from log4j.properties, I created a logging.properties file with the following contents:
handlers=java.util.logging.FileHandler
.level=FINE
okhttp3.internal.http2.level=FINE
java.util.logging.FileHandler.pattern = logs/%hjava%u.log
java.util.logging.FileHandler.limit = 50000
java.util.logging.FileHandler.count = 1
java.util.logging.FileHandler.formatter = java.util.logging.XMLFormatter
java.util.logging.ConsoleHandler.level = FINE
java.util.logging.ConsoleHandler.formatter = java.util.logging.SimpleFormatter
I then added this to jvm params used for starting the application -Djava.util.logging.config.file="file://${BASE_DIR}/logging.properties"
But I don't see any new folders being created as indicated by the Filehandler. Any one know why?
But I don't see any new folders being created as indicated by the Filehandler. Any one know why?
The FileHandler will not create any new folders. A directory must be created before the FileHandler will create a file.
The system property requires a path to file that is located on the filesystem It will not expand system properties or environment variables by using the dollar sign syntax.
You can use a relative path based off of the working directory or you have to use an absolute path to the logging.properties. The logging properties can not be packaged inside of an archive.
If you want to work around this limitation then you want to create a custom config class and use the java.util.logging.config.class property in conjunction with the java.util.logging.config.file property. You then write a class that reads the file://${BASE_DIR}/logging.properties and performs the needed transformation into a path to a file. Then update the configuration if you are using JDK9 or newer. On older versions you need to use readConfiguration and add code to work work around limitations of the LogManager
I'm updating a C++ class wrapping the boost logger to use a recent boost version with the logger v2 API.
Most of the work have been done, but with the logger v1 we used an API to get the current log file name and path like in the example below:
boost::filesystem::path p = pAsyncFileSink->locked_backend()->get_log_filename()
I searched for some hour, but I don't find a way to do that with the current API ! if we use a rotating text file backend, how to get the latest or in use file name ?
Any help or advice will be welcome.
Alex.
There is no API to obtain the current file name. You can see in the implementation that the file name is generated only when file rotation is performed, making it possible for the sink to not have a file name in certain points in time.
I am writing some code to create a file from a Windows 8 app in an standard way, the code looks like below:
using namespace Windows::Storage;
StorageFolder^ folder = KnownFolders::DocumentsLibrary;
String ^filename = ref new String(L"file.txt");
auto createFile = folder->CreateFileAsync(filename, CreationCollisionOption::ReplaceExisting);
concurrency::create_task(createFile).wait;
Now instead of using DocumentsLibrary, I want to write thid file to an customized file path, like:
C:\Users\<username>\AppData\Local\ExampleApp\ExampleFolder
How should I change the code to be able to do this? Thanks!
WinRT can only access a few folders. You have a few standard libraries like Pictures, Music, etc (Documents requires elevated rights) and you have the application data folders that you can find under \AppData\Local\Packages\yourpackage.
Inside of this package folder you have two main folders to store data: LocalState and RoamingState. As the names convey: the former is to store data locally while the latter will synchronize its contents whenever possible (according to the rules you define).
You can access these folders using the C++ equivalent of Windows.Storage.ApplicationData.Current.LocalFolder and Windows.Storage.ApplicationData.Current.RoamingFolder.
What you can do though is request explicit access through a FilePicker but this will prompt the user a window where he should target the directory himself.