Merge pull request #222 from AuraVisionLabs/main

Add python bindings

.gitignore

@@ -8,3 +8,5 @@ nethogs.project
 libnethogs.so.*
 libnethogs.a
 results/
+build/
+nethogs.egg-info

contrib/README.md

@@ -1,4 +0,0 @@
-This folder contains user-contributed scripts.
-
-The Nethogs project does not make claims about the quality of the scripts, maintains
-them in any way, or necessarily think they're a good idea in the first place :).

pyproject.toml

@@ -0,0 +1,8 @@
+[build-system]
+requires = [
+    "setuptools>=42",
+    "wheel",
+    "pybind11>=2.8.0",
+]
+
+build-backend = "setuptools.build_meta"

python/README.md

@@ -0,0 +1,49 @@
+## Python stuff
+This directory contains the code that wraps `libnethogs` into a python module.
+
+### To build + install
+Install dependencies:
+``` bash
+apt-get install build-essential libncurses5-dev libpcap-dev pybind11-dev
+```
+All the code is compiled through the [pybind11 setuptools building system](https://pybind11.readthedocs.io/en/stable/compiling.html). So to build and install you just need to do:
+``` bash
+### if you have the code cloned locally
+pip install .
+```
+Or:
+``` bash
+### if you can't bother to clone the repo yourself
+pip install git+https://github.com/raboof/nethogs.git
+```
+
+### To use
+``` python
+import nethogs
+import threading
+
+def callback(action: int, record: nethogs.NethogsMonitorRecord) -> None:
+    do_whatever_with_record(record)
+    return
+    
+th = threading.Thread(target=nethogs.nethogsmonitor_loop, args(callback, filter, to_ms))
+th.start()
+do_whatever_you_need_to_do()
+nethogs.nethogsmonitor_breakloop()
+th.join()
+```
+
+### Caveats
+- **Ideally this should not be run in the main thread**, stopping the nethogsmonitor_loop with a SIGINT or SIGTERM is a bit hacky and could break things.
+- Only one instance of the loop can be run at a time, callbacks would get completely messed up if more than one is run. I believe nethogs only allows one at a time anyways.
+- There is no way of setting pidsToWatch, but it is not difficult to implement if anyone wants to do it :)
+- The package version is set in setup.py using the script determineVersion.sh, but it is very hacky as a PEP compatible string is needed. Something stronger should be implemented, ideally using the regex module.
+
+## Extra stuff
+The script `python-wrapper.py` used to be in a contrib directory. This script loads the libnethogs library directly, which you shouldn't really do anyways. The way of using nethogs in python by the `python-wrapper.py` is now deprecated. I'm just leaving it here to not destroy the previous contribution. This is the original comment that **raboof** made about the `python-wrapper.py` and the `contrib` directory:
+```
+This folder contains user-contributed scripts.
+
+The Nethogs project does not make claims about the quality of the scripts, maintains
+them in any way, or necessarily think they're a good idea in the first place :).
+```

python/bindings.cpp

@@ -0,0 +1,98 @@
+#include <pybind11/functional.h>
+#include <pybind11/pybind11.h>
+#include <pybind11/stl.h>
+#include <set>
+#include <iostream>
+
+#include "libnethogs.h"
+
+namespace py = pybind11;
+
+//--- for some reason this is a global defined in main.cpp
+std::set<pid_t> pidsToWatch;
+
+//--- hacky way to get callbacks working and handle signals
+std::function<void(int, NethogsMonitorRecord const *)> empty_callback;
+std::function<void(int, NethogsMonitorRecord const *)> loop_callback;
+void loop_callback_wrapper(
+  int action,
+  NethogsMonitorRecord const *record)
+{
+  py::gil_scoped_acquire acquire;
+  if (PyErr_CheckSignals() != 0) {
+    nethogsmonitor_breakloop();
+    PyErr_Clear();
+  }
+  else if (loop_callback) {
+    loop_callback(action, record);
+  }
+}
+
+int nethogsmonitor_loop_py(
+  std::function<void(int, NethogsMonitorRecord const *)> &cb,
+  char *filter,
+  int to_ms)
+{
+    int retval;
+    loop_callback = cb;
+    {
+      py::gil_scoped_release release;
+      retval = nethogsmonitor_loop(loop_callback_wrapper, filter, to_ms);
+    }
+    loop_callback = empty_callback;
+    return retval;
+}
+
+int nethogsmonitor_loop_devices_py(
+  std::function<void(int, NethogsMonitorRecord const *)> &cb,
+  char *filter,
+  std::vector<std::string> __devicenames,
+  bool all,
+  int to_ms)
+{
+    // this is ok because we only use the vector here
+    std::vector<char*> _devicenames;
+    for (auto _dn : __devicenames) _devicenames.push_back(const_cast<char*>(_dn.c_str()));
+    int devc = _devicenames.size();
+    char **devicenames = (_devicenames.empty()) ? NULL : _devicenames.data();
+
+    int retval;
+    loop_callback = cb;
+    {
+      py::gil_scoped_release release;
+      retval = nethogsmonitor_loop_devices(loop_callback_wrapper, filter, devc, devicenames, all, to_ms);
+    }
+    loop_callback = empty_callback;
+    return retval;
+}
+
+//--- python module binding
+PYBIND11_MODULE(nethogs, m) {
+    py::class_<NethogsMonitorRecord>(m, "NethogsMonitorRecord")
+        .def_readwrite("record_id", &NethogsMonitorRecord::record_id)
+        .def_readwrite("name", &NethogsMonitorRecord::name)
+        .def_readwrite("pid", &NethogsMonitorRecord::pid)
+        .def_readwrite("uid", &NethogsMonitorRecord::uid)
+        .def_readwrite("device_name", &NethogsMonitorRecord::device_name)
+        .def_readwrite("sent_bytes", &NethogsMonitorRecord::sent_bytes)
+        .def_readwrite("recv_bytes", &NethogsMonitorRecord::recv_bytes)
+        .def_readwrite("sent_kbs", &NethogsMonitorRecord::sent_kbs)
+        .def_readwrite("recv_kbs", &NethogsMonitorRecord::recv_kbs);
+
+    m.def("nethogsmonitor_loop", &nethogsmonitor_loop_py, R"pbdoc(
+        Nethogs monitor loop
+    )pbdoc");
+    m.def("nethogsmonitor_loop_devices", &nethogsmonitor_loop_devices_py, R"pbdoc(
+        Nethogs monitor loop
+    )pbdoc");
+    m.def("nethogsmonitor_breakloop", &nethogsmonitor_breakloop, R"pbdoc(
+        Nethogs monitor loop break
+    )pbdoc");
+
+#ifdef VERSION
+    m.attr("__version__") = VERSION;
+#else
+    m.attr("__version__") = "unknown";
+#endif
+
+}

setup.py

@@ -0,0 +1,63 @@
+import glob
+import subprocess
+import sys
+
+from pybind11 import get_cmake_dir
+# Available at setup time due to pyproject.toml
+from pybind11.setup_helpers import Pybind11Extension, build_ext
+from setuptools import setup
+
+_version_info = subprocess.run(['bash', "./determineVersion.sh"], stdout=subprocess.PIPE)
+__version__ = _version_info.stdout.decode("utf-8").rstrip("\n").split("-")[0] if _version_info else "0.0.0"
+
+OBJS = [
+    "python/bindings.cpp",
+    "src/libnethogs.cpp",
+    "src/packet.cpp",
+    "src/connection.cpp",
+    "src/process.cpp",
+    "src/decpcap.c",
+    "src/inode2prog.cpp",
+    "src/conninode.cpp",
+    "src/devices.cpp"
+]
+
+FLAGS = [
+    "-Wall",
+    "-Wextra",
+    "-Wno-missing-field-initializers",
+    "--std=c++0x",
+    "-O3",
+    "-fPIC",
+    '-DVERSION="{}"'.format(__version__)
+]
+# The main interface is through Pybind11Extension.
+# * You can add cxx_std=11/14/17, and then build_ext can be removed.
+# * You can set include_pybind11=false to add the include directory yourself,
+#   say from a submodule.
+#
+# Note:
+#   Sort input source files if you glob sources to ensure bit-for-bit
+#   reproducible builds (https://github.com/pybind/python_example/pull/53)
+
+ext_modules = [
+    Pybind11Extension(
+        "nethogs",
+        sources = OBJS,
+        include_dirs = ["src/"],
+        extra_compile_args = FLAGS,
+        libraries = ["pcap"]
+    ),
+]
+
+setup(
+    name="nethogs",
+    version=__version__,
+    author="raboof",
+    url="https://github.com/raboof/nethogs",
+    description="Nethogs python bindings",
+    ext_modules=ext_modules,
+    cmdclass={"build_ext": build_ext},
+    zip_safe=False,
+    python_requires=">=3.6",
+)

src/process.cpp

@@ -268,7 +268,7 @@ Process *getProcess(unsigned long inode, const char *devicename) {
   if (proc != NULL)
     return proc;
 
-  if ( !(pidsToWatch.empty()) && pidsToWatch.find(node->pid) == pidsToWatch.end() ) {    
+  if ( !(pidsToWatch.empty()) && pidsToWatch.find(node->pid) == pidsToWatch.end() ) {
     return NULL;
   }
 
@@ -440,4 +440,3 @@ void remove_timed_out_processes() {
 }
 
 void garbage_collect_processes() { garbage_collect_inodeproc(); }
-