vadimkantorov

namespace CSharpFromPythonNamespace
{
    public class CSharpFromPythonClass
    {
        public string Hi(string x)
        {
            return "Hi " + x;
        }

        public static string Hello(string x)

vadimkantorov

// gcc -shared -fPIC log_file_access_dynamic.c -o log_file_access_dynamic.so -ldl; LD_PRELOAD=$PWD/log_file_access_dynamic.so /usr/bin/cat log_file_access_dynamic.c

#define _GNU_SOURCE
#include <stdio.h>
#include <string.h>

#include <unistd.h>
#include <errno.h>
#include <dlfcn.h>
#include <sys/stat.h>

vadimkantorov

# Usage:
#  gdbpython script.py my_script_arguments
#  gdbbacktrace ./my_program arg1 arg2

alias gdbpython="gdb -ex run --args python"
alias gdbbacktrace="gdb -batch -ex run -ex bt --args"

alias gdbbacktracequit="gdb -ex 'set pagination off' -ex 'set confirm off' -iex 'set debuginfod enabled on' -ex run -ex 'thread apply all bt' -ex quit --args"

vadimkantorov

alias cmdline='ps --no-headers -o comm,args -p'
cmdline 1

# can produce JSON, but without process args
alias notbrokenjson='ps --no-headers -o '\''{"pcpu": %C, "group": "%G", "PPID": %P, "user": "%U", "comm": "%c", "rgroup": "%g", "nice": "%n", "pid": %p, "pgid": %r, "etime": "%t", "ruser": "%u", "time": "%x", "tty": "%y", "vsz": %z}'\'' -p'
notbrokenjson 1 > notbroken.json

# does not remove escape sequences or quotes in args - this will break JSON format for some processes
alias brokenjson='ps --no-headers -o '\''{"pcpu": %C, "group": "%G", "PPID": %P, "user": "%U", "args": "%a", "comm": "%c", "rgroup": "%g", "nice": "%n", "pid": %p, "pgid": %r, "etime": "%t", "ruser": "%u", "time": "%x", "tty": "%y", "vsz": %z}'\'' -p'
brokenjson 1 > brokenjson.json

vadimkantorov

# git clone https://gist.github.com/vadimkantorov/fe63f8628ff6cad460e934e1d7ed650b
# cd fe63f8628ff6cad460e934e1d7ed650b
# uv venv
# uv sync

# https://github.com/vllm-project/vllm/pull/20358#issuecomment-3247178818
# https://github.com/vllm-project/vllm/issues/9244
# https://github.com/astral-sh/uv/issues/8082
# https://github.com/vllm-project/vllm/issues/24126

vadimkantorov

# https://stackoverflow.com/questions/71944328/how-to-implement-json-format-logs-in-python

import json
import logging

# https://docs.python.org/3/library/logging.html#logrecord-attributes
formatter = type('JsonFormatter', (logging.Formatter, ), dict(format = lambda self, record: json.dumps(dict(time = self.formatTime(record), level = record.levelname, message = record.getMessage(), module = record.module, lineno = record.lineno)) ))()

# simpler version below does not escape quotes in message and does not delete newlines in message
# formatter = logging.Formatter('{\"time\": \"%(asctime)-s\", \"level\": \"%(levelname)-s\", \"message\": \"%(message)s\", \"module\": \"%(module)s\", \"lineno\": %(lineno)d}')

vadimkantorov

# Usage: 
# 1. Put in your email/SendGrid credentials or phone/Twilio credentials
# 2. Run as: "bash rdv_naturalisation.sh nanterre" or "bash rdv_naturalisation.sh bobigny" or "bash rdv_naturalisation.sh cre"
# 3. Get your RDV! The script will dump the HTML of the page and will continue execution even after the first RDV is found.

SENDGRID_BEARER_TOKEN='' # should start with "SG."
SENDGRID_EMAIL='' # put your e-mail here
TWILIO_SID='' # should start with "AC"
TWILIO_AUTH_TOKEN=''
TWILIO_FROM_NUMBER='' # put your Twilio Trial Phone Number here, should start with +

vadimkantorov

echo '{"site": {"hello": "world"}}' > foo.json
echo 'Hello {{ site.hello }}' > foo.txt.j2

bash j2.sh -t foo.txt.j2 -i foo.json -o foo.txt
cat foo.txt

source function.sh
j2 -t foo.txt.j2 -i foo.json -o foo.txt
cat foo.txt
unset -f j2

vadimkantorov

# supports only strings, dicts and lists
# does not support multiline strings as the first list-item key `- run: |`
# does not preserve whitespaces in " |" literal string blocks as described in : https://docs.ansible.com/ansible/latest/reference_appendices/YAMLSyntax.html

def yaml_loads(content, convert_bool = True, convert_int = True, convert_dict = True, convert_none = True):
    def procval(val):
        read_until = lambda tail, chars: ([(tail[:i], tail[i+1:]) for i, c in enumerate(tail) if c in chars] or [(tail, '')])[0]

        val = val.strip()
        is_quoted_string = len(val) >= 2 and ((val[0] == val[-1] == '"') or (val[0] == val[-1] == "'"))

vadimkantorov

# https://en.wikipedia.org/wiki/Base64
# 00123456  00ABCDEF  00abcdef  00uvwxyz
# 123456AB  CDEFabcd  efuvwxyz
# this code does not support batches. adapting for e.g. concatenated varlen format is possible, but need to handle/preserve varlen information and paddings in some way

import torch

def base64_encode_padded(input_as_uint8_tensor):
    base64_alphabet, base64_pad = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/', '='
    device = input_as_uint8_tensor.device