Check commands that you can add to tldr

to_tldr.py

#!/usr/bin/env python3
# This script will check for commands that you could most easily contribute to tldr (https://tldr.sh/)
# Run it in the root of the https://github.com/tldr-pages/tldr repository
import os
import sys

def get_used_commands():
    used_cmds = {}
    for filepath, complex in {"~/.zsh_history": True, "~/.bash_history": False}.items():
        with open(os.path.expanduser(filepath), "rb") as file:
            line_n = 0
            for line in file:
                line_n += 1
                try:
                    line = line.decode()
                except UnicodeDecodeError as e:
                    print(f"Failed to parse line {line_n} of {filepath}: {e}", file=sys.stderr)
                    continue

                if complex:
                    full_cmd = line[line.find(";")+1:]
                else:
                    full_cmd = line
                cmd = full_cmd[:full_cmd.find(" ")]
                cmd = os.path.basename(cmd)
                
                if cmd in used_cmds:
                    used_cmds[cmd] += 1
                else:
                    used_cmds[cmd] = 1
    return used_cmds


tldr_cmds = []
for subdir in os.listdir("./pages"):
    for file in os.listdir(os.path.join("./pages", subdir)):
        tldr_cmds.append(os.path.splitext(file)[0])

existing_cmds = []
for path in os.getenv("PATH").split(":"):
    if not os.path.exists(path):
        continue
    for cmd in os.listdir(path):
        existing_cmds.append(cmd)

used_cmds = get_used_commands()

todo_cmds = {}
for cmd, count in used_cmds.items():
    if cmd in existing_cmds and cmd not in tldr_cmds:
        todo_cmds[cmd] = count

for cmd, count in sorted(todo_cmds.items(), key=lambda item: item[1]):
    print(f"{cmd:<20} count={count}")

print("\n--- Stats:")
print(f"Commands in tldr db: {len(tldr_cmds)}")
print(f"Commands in PATH: {len(existing_cmds)}")
print(f"Unique commands used: {len(used_cmds)}")
print(f"Commands to put in tldr db that are used (shown above): {len(todo_cmds)}")