/01/ 剪贴板管理器
❓你是否發現自己忙於處理多個文本片段,而忘記了自己複製了什麼?你有沒有想過有一個工具可以記錄你一天中複製的所有內容?
👋該自動化腳本利用Pyperclip庫的強大功能完全捕獲複製數據,並集成了Tkinter以可視化方式跟蹤和管理複製的文本。它可以監控你複製的所有內容,將複製的每個文本無縫地存儲在一個時尚的圖形界面中,這樣就不必在無盡的標籤頁中搜索,也不會丟失一些有價值的讯息。
import tkinter as tk
from tkinter import ttk
import pyperclip
def update_listbox():
new_item = pyperclip.paste()
if new_item not in X:
X.append(new_item)
listbox.insert(tk.END, new_item)
listbox.insert(tk.END, "----------------------")
listbox.yview(tk.END)
root.after(1000, update_listbox)
def copy_to_clipboard(event):
selected_item = listbox.get(listbox.curselection())
if selected_item:
pyperclip.copy(selected_item)
X = []
root = tk.Tk()
root.title("Clipboard Manager")
root.geometry("500x500")
root.configure(bg="#f0f0f0")
frame = tk.Frame(root, bg="#f0f0f0")
frame.pack(padx=10, pady=10)
label = tk.Label(frame, text="Clipboard Contents:", bg="#f0f0f0")
label.grid(row=0, column=0)
scrollbar = tk.Scrollbar(root)
scrollbar.pack(side=tk.RIGHT, fill=tk.Y)
listbox = tk.Listbox(root, width=150, height=150, yscrollcommand=scrollbar.set)
listbox.pack(pady=10)
scrollbar.config(command=listbox.yview)
update_listbox()
listbox.bind("<Double-Button-1>", copy_to_clipboard)
root.mainloop()
應用:
·捕捉從各種來源複製的研究筆記並進行分類。
·擴展腳本可以捕捉重要的日曆事件、提醒事項、密碼等。
/02/ 代码质量检查器
👉每個開發人員都會遇到這樣的挫折:在 Python 程式碼中查找錯誤,卻發現自己迷失在錯誤的迷宮中。作爲開發人員,我們深知編寫簡潔高效代碼的重要性,但手動分析程式碼品質可能是一項壓力巨大的任務。
✅該自動化腳本利用 Pylint 和 Flake8 Python 軟體包對程式碼質量進行全面審查。將你的程式碼與編碼標準進行比較,並找出邏輯錯誤。它可確保的代碼符合行業最佳實踐並保持無錯。
import os
import subprocess
def analyze_code(directory):
# List Python files in the directory
python_files = [file for file in os.listdir(directory) if file.endswith('.py')]
if not python_files:
print("No Python files found in the specified directory.")
return
# Analyze each Python file using pylint and flake8
for file in python_files:
print(f"Analyzing file: {file}")
file_path = os.path.join(directory, file)
# Run pylint
print("\nRunning pylint...")
pylint_command = f"pylint {file_path}"
subprocess.run(pylint_command, shell=True)
# Run flake8
print("\nRunning flake8...")
flake8_command = f"flake8 {file_path}"
subprocess.run(flake8_command, shell=True)
if __name__ == "__main__":
directory = r"C:\Users\abhay\OneDrive\Desktop\Part7"
analyze_code(directory)
應用:
·自動程式碼增強器 — 對該腳本稍作擴展,可用於創建一個 Python 腳本,用於識別程式碼中的問題並作出相應修改。
·自動程式碼審查。
/03/ 不篡改文件
文件篡改又稱數據篡改,是指未經授權故意更改信息,包括破壞、篡改或編輯數據。
❗全球許多威脅行爲者濫用文件篡改技術,在關鍵系統文件中引入不同的漏洞或後門,從而危及安全性並使未經授權的訪問成爲可能。
🔎要降低這種風險,關鍵是要驗證文件的完整性,確保它們與原始狀態保持一致。這個自動化腳本可以幫助你測試任何文件,並識別其是否被篡改。
import hashlib
import os
def calculate_sha256(file_path):
sha256 = hashlib.sha256()
with open(file_path, 'rb') as file:
for chunk in iter(lambda: file.read(4096), b''):
sha256.update(chunk)
return sha256.hexdigest()
def check_integrity(file_path, expected_checksum):
actual_checksum = calculate_sha256(file_path)
return actual_checksum == expected_checksum
if __name__ == "__main__":
file_path = input("Enter the path to the file: ")
expected_checksum = input("Enter the expected SHA-256 checksum: ")
if os.path.isfile(file_path):
if check_integrity(file_path, expected_checksum):
print("File integrity verified: The file has not been tampered with.")
else:
print("File integrity check failed: The file may have been tampered with.")
else:
print("Error: File not found.")
/04/ 智能交易
👨💻交易是指買賣股票、債券、貨幣、商品或衍生品等金融工具,以獲取利潤。交易員整天都在查看不同的儀表盤,試圖找出買入或賣出的完美決策。
✅這個自動化腳本可以幫助交易者和投資者很好地瞭解你願意投資的任何股票。它根據從雅虎財經獲取的歷史股票數據預測近期股票價格。
import streamlit as st
from datetime import date
import yfinance as yf
from prophet import Prophet
from prophet.plot import plot_plotly
from plotly import graph_objs as go
START = "2015-01-01"
TODAY = date.today().strftime("%Y-%m-%d")
st.title('Stock Forecast App')
stocks = ('MSFT',"TSLA",'GOOG', 'AAPL', "NVDA")
selected_stock = st.selectbox('Select dataset for prediction', stocks)
n_years = st.slider('Years of prediction:', 1, 4)
period = n_years * 365
@st.cache
def load_data(ticker):
data = yf.download(ticker, START, TODAY)
data.reset_index(inplace=True)
return data
data_load_state = st.text('Loading data...')
data = load_data(selected_stock)
data_load_state.text('Loading data... done!')
st.subheader('Raw data')
st.write(data.tail())
# Plot raw data
def plot_raw_data():
fig = go.Figure()
fig.add_trace(go.Scatter(x=data['Date'], y=data['Open'], name="stock_open"))
fig.add_trace(go.Scatter(x=data['Date'], y=data['Close'], name="stock_close"))
fig.layout.update(title_text='Time Series data with Rangeslider', xaxis_rangeslider_visible=True)
st.plotly_chart(fig)
plot_raw_data()
# Predict forecast with Prophet.
df_train = data[['Date','Close']]
df_train = df_train.rename(columns={"Date": "ds", "Close": "y"})
m = Prophet()
m.fit(df_train)
future = m.make_future_dataframe(periods=period)
forecast = m.predict(future)
# Show and plot forecast
st.subheader('Forecast data')
st.write(forecast.tail())
st.write(f'Forecast plot for {n_years} years')
fig1 = plot_plotly(m, forecast)
st.plotly_chart(fig1)
st.write("Forecast components")
fig2 = m.plot_components(forecast)
st.write(fig2)
要運行此程序,首先需要使用 pip 安裝 Streamlit、yfinance、prophet 和 plotly python 庫。然後使用命令streamlit run smart_trade.py 運行它。
應用:
·算法交易
·股票價格比較儀表板
/05/ 自動圖像下載器
收集圖像數據是計算機視覺項目的關鍵挑戰。正如 Andrew Ng 所說,如果你能收集到大量數據集,那麼算法就不重要了。數據對提高模型的性能和準確性起著關鍵作用。
✅只需幾行程式碼,這個自動化腳本就能幫助你在幾分鐘內從網上下載數據。
# Importing the necessary module and function
from simple_image_download import simple_image_download as simp
# Creating a response object
response = simp.simple_image_download
## Keyword
keyword = "Dog"
# Downloading images
try:
response().download(keyword, 20)
print("Images downloaded successfully.")
except Exception as e:
print("An error occurred:", e)
/06/ 端口掃描程序
在計算機網絡中,端口是允許不同進程或服務通過網絡連接和交換數據的通信端點。端口由數值標識,並與特定協議相關聯。
🔧開放端口就像建築物中的門窗,每個端口都是網站與外部網絡通信的潛在入口。然而,在沒有採取適當安全措施的情況下開放端口,會使網站容易受到網路攻擊。
✅這個自動化腳本會將網站 URL 作爲輸入,並檢查該網站是否存在任何開放端口。無論你是作爲 “紅隊 “的一員在執行任務,還是在 “藍隊 “中堅守陣地,這個腳本都能爲你提供有用的便利工具。
import socket
from prettytable import PrettyTable
# Dictionary mapping common ports to vulnerabilities (Top 15)
vulnerabilities = {
80: "HTTP (Hypertext Transfer Protocol) - Used for unencrypted web traffic",
443: "HTTPS (HTTP Secure) - Used for encrypted web traffic",
22: "SSH (Secure Shell) - Used for secure remote access",
21: "FTP (File Transfer Protocol) - Used for file transfers",
25: "SMTP (Simple Mail Transfer Protocol) - Used for email transmission",
23: "Telnet - Used for remote terminal access",
53: "DNS (Domain Name System) - Used for domain name resolution",
110: "POP3 (Post Office Protocol version 3) - Used for email retrieval",
143: "IMAP (Internet Message Access Protocol) - Used for email retrieval",
3306: "MySQL - Used for MySQL database access",
3389: "RDP (Remote Desktop Protocol) - Used for remote desktop connections (Windows)",
8080: "HTTP Alternate - Commonly used as a secondary HTTP port",
8000: "HTTP Alternate - Commonly used as a secondary HTTP port",
8443: "HTTPS Alternate - Commonly used as a secondary HTTPS port",
5900: "VNC (Virtual Network Computing) - Used for remote desktop access",
# Add more ports and vulnerabilities as needed
}
def display_table(open_ports):
table = PrettyTable(["Open Port", "Vulnerability"])
for port in open_ports:
vulnerability = vulnerabilities.get(port, "No known vulnerabilities associated with common services")
table.add_row([port, vulnerability])
print(table)
def scan_top_ports(target):
open_ports = []
top_ports = [21, 22, 23, 25, 53, 80, 110, 143, 443, 3306, 3389, 5900, 8000, 8080, 8443] # Top 15 ports
for port in top_ports:
try:
sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
sock.settimeout(1) # Adjust timeout as needed
result = sock.connect_ex((target, port))
if result == 0:
open_ports.append(port)
sock.close()
except KeyboardInterrupt:
sys.exit()
except socket.error:
pass
return open_ports
def main():
target = input("Enter the website URL or IP address to scan for open ports: ")
open_ports = scan_top_ports(target)
if not open_ports:
print("No open ports found on the target.")
else:
print("Open ports and associated vulnerabilities:")
display_table(open_ports)
if __name__ == "__main__":
main()
/07/ 密碼管理器
👨💼作爲一名數位遊牧工作者,需要記住很多密碼,因爲每天都要註冊新網站。
這個自動化腳本可以幫你管理所有密碼,使用不同的加密技術確保密碼安全,只有你自己才能訪問。
這個腳本最棒的地方在於它是一個網絡應用程序,只需稍作改動,你就可以將它部署到雲端硬碟上,隨時隨地訪問你的所有密碼,而無需擔心安全問題。
import streamlit as st
import csv
from cryptography.fernet import Fernet
from cryptography.fernet import InvalidToken
# Custom encryption key (hardcoded)
CUSTOM_ENCRYPTION_KEY = b'u7wGgNdDFefqpr_kGxb8wJf6XRVsRwvb3QgITsD5Ft4=' ## 如果您打算在共享平台上使用此脚本,请确保将此密钥保存在一个单独的安全文件中。
# Function to encrypt password
def encrypt_password(password):
cipher_suite = Fernet(CUSTOM_ENCRYPTION_KEY)
encrypted_password = cipher_suite.encrypt(password.encode())
return encrypted_password
# Function to decrypt password
def decrypt_password(encrypted_password):
if isinstance(encrypted_password, bytes):
try:
cipher_suite = Fernet(CUSTOM_ENCRYPTION_KEY)
decrypted_password = cipher_suite.decrypt(encrypted_password)
return decrypted_password.decode()
except InvalidToken:
return "Invalid Token"
else:
return None
# Function to save website name and password to CSV file
def save_credentials(website_name, password):
encrypted_password = encrypt_password(password)
with open('credentials.csv', 'a', newline='') as csvfile:
writer = csv.writer(csvfile)
writer.writerow([website_name, encrypted_password.decode()]) # Ensure storing string representation
# Function to retrieve password from CSV file
def retrieve_password(website_name):
with open('credentials.csv', 'r') as csvfile:
reader = csv.reader(csvfile)
for row in reader:
if row[0] == website_name:
encrypted_password = row[1].encode()
return encrypted_password
return None
# Streamlit UI
st.title("Password Manager")
# Input fields for website name and password
website_name = st.text_input("Enter website name:")
password = st.text_input("Enter password:", type="password")
# Save button to save website name and password
if st.button("Save"):
if website_name and password:
save_credentials(website_name, password)
st.success("Website name and password saved successfully.")
else:
st.error("Please fill in all fields.")
# Retrieve button to retrieve password
if st.checkbox("Retrieve Password"):
website_name = st.selectbox("Select website name:", options=[""] + [row[0] for row in csv.reader(open('credentials.csv', 'r'))])
key = st.text_input("Enter Your Encryption Key:", type="password")
if st.button("Retrieve Password"):
if key == str(CUSTOM_ENCRYPTION_KEY.decode()):
if website_name:
encrypted_password = retrieve_password(website_name)
if encrypted_password:
decrypted_password = decrypt_password(encrypted_password)
st.success(f"Password for **{website_name}** -> **{decrypted_password}**")
else:
st.error("Password not found in database.")
elif key == "":
pass
else:
st.error("Invalid Encryption Key!!!")
❗它使用寫死的自定義密鑰進行加密和解密。如果你打算在共享平臺上使用此腳本,請確保將此密鑰保存在一個單獨的安全文件中。
/08/ 群發
電子郵件列表是線上的貨幣,每個訂閱者都是您數字部落格的重要成員。
🤝作爲一個有影響力的數位行銷者,與你的粉絲保持互動非常重要,但手動操作是一項艱鉅的工作,而使用相關工具則需要花費大量資金。
這款自動化腳本可利用 Gmail 自帶的 SMTP 服務器,在幾分鐘內批量發送電子郵件,讓你可以完全自定義並行使權力。
import smtplib
import ssl
# SMTP server details
smtp_server = 'data.STUDIO.com'
smtp_port = 465
# Sender and recipient details
from_address = 'Winzo Shop'
to_address = ['',''] ## Recepients List
# Authentication details
username = '' ## Sender Email
password = '' ## Sender Password
# Email message details
subject = '🎉 Exclusive Offer Inside! Get 10% Off Your Next Purchase'
body = '''
'''
# Create an SSL/TLS context
context = ssl.create_default_context()
# Connect to the SMTP server using SSL/TLS
with smtplib.SMTP_SSL(smtp_server, smtp_port, context=context) as server:
# Enable debugging to print the server's responses
server.set_debuglevel(1)
# Login to the SMTP server
server.login(username, password)
# Create the email message
message = f'From: {from_address}\r\nSubject: {subject}\r\nTo: {to_address}\r\n\r\n{body}'
message = message.encode() # Convert the message to bytes
# Send the email
server.sendmail(from_address, to_address, message)
/09/ Readme.md 生成器
README.md 文件是項目的入口,提供基本信息並吸引訪客進一步探索。它被認爲是版本庫中最重要的文件之一,但創建起來也相當耗時。
這個自動化腳本可以根據版本庫名稱、鏈接和描述等輸入信息輕鬆生成 README.md 文件,從而爲您節省大量時間。
def generate_markdown_file():
# Prompting user for inputs
repository_name = input("\n Enter the name of your GitHub repository: ")
project_description = input("Enter a short description of your project: ")
installation_instructions = input("Enter installation instructions for your project: ")
usage_instructions = input("Enter usage instructions for your project: ")
contributors = input("Enter the contributors to your project (separated by commas): ")
license = select_license()
# Generating badges
stars_badge = "[](https://github.com/{}/stargazers)".format(repository_name, repository_name)
forks_badge = "[](https://github.com/{}/network/members)".format(repository_name, repository_name)
issues_badge = "[](https://github.com/{}/issues)".format(repository_name, repository_name)
license_badge = "[](https://github.com/{}/blob/master/LICENSE)".format(repository_name, repository_name)
# Generating Markdown content
markdown_content = f"""
# {repository_name}
{project_description}
## Table of Contents
- [Installation](#installation)
- [Usage](#usage)
- [Contributors](#contributors)
- [License](#license)
- [Badges](#badges)
- [GitHub Repository](#github-repository)
## Installation
```
{installation_instructions}
```
## Usage
```
{usage_instructions}
```
## Contributors
{contributors}
## License
This project is licensed under the {license} License - see the [LICENSE](LICENSE) file for details.
## Badges
{stars_badge} {forks_badge} {issues_badge} {license_badge}
## GitHub Repository
[Link to GitHub repository](https://github.com/{repository_name})
"""
# Writing content to Markdown file
markdown_file_name = f"{repository_name}_README.md"
with open(markdown_file_name, "w") as markdown_file:
markdown_file.write(markdown_content)
print(f"Markdown file '{markdown_file_name}' generated successfully!")
def select_license():
licenses = {
"MIT": "MIT License",
"Apache": "Apache License 2.0",
"GPL": "GNU General Public License v3.0",
# Add more licenses as needed
}
print("Select a license for your project:")
for key, value in licenses.items():
print(f"{key}: {value}")
while True:
selected_license = input("Enter the number corresponding to your selected license: ")
if selected_license in licenses:
return licenses[selected_license]
else:
print("Invalid input. Please enter a valid license number.")
if __name__ == "__main__":
generate_markdown_file()
/10/ OrganizeIT 2.0
📣你的下載文件夾是否一團糟?你是否發現自己總是找不到重要文件,嘗試過無數次整理文件夾,但都以失敗告終?
✅這個自動化腳本可以幫你在幾分鐘內整理好文件夾。你只需指定需要清理的路徑,本腳本就會根據文件擴展名自動將所有文件劃分到不同的文件夾中。
import os
import hashlib
import shutil
def get_file_hash(file_path):
with open(file_path, 'rb') as f:
return hashlib.sha256(f.read()).hexdigest()
def organize_and_move_duplicates(folder_path):
# Create a dictionary to store destination folders based on file extensions
extension_folders = {}
# Create the "Duplicates" folder if it doesn't exist
duplicates_folder = os.path.join(folder_path, 'Duplicates')
os.makedirs(duplicates_folder, exist_ok=True)
# Create a dictionary to store file hashes
file_hashes = {}
# Iterate through files in the folder
for filename in os.listdir(folder_path):
file_path = os.path.join(folder_path, filename)
if os.path.isfile(file_path):
# Get the file extension
_, extension = os.path.splitext(filename)
extension = extension.lower() # Convert extension to lowercase
# Determine the destination folder
if extension in extension_folders:
destination_folder = extension_folders[extension]
else:
destination_folder = os.path.join(folder_path, extension[1:]) # Remove the leading dot from the extension
os.makedirs(destination_folder, exist_ok=True)
extension_folders[extension] = destination_folder
# Calculate the file hash
file_hash = get_file_hash(file_path)
# Check for duplicates
if file_hash in file_hashes:
# File is a duplicate, move it to the "Duplicates" folder
shutil.move(file_path, os.path.join(duplicates_folder, filename))
print(f"Moved duplicate file {filename} to Duplicates folder.")
else:
# Store the file hash
file_hashes[file_hash] = filename
# Move the file to the destination folder
shutil.move(file_path, destination_folder)
print(f"Moved {filename} to {destination_folder}")
if __name__ == "__main__":
folder_path = input("Enter the path to the folder to organize: ")
organize_and_move_duplicates(folder_path)
以上就是本期的內容分享~~,碼字不易,如果覺得對你有一點點幫助,歡迎「追蹤」,「按讚」,「分享」喔,我會持續為大家創作優質的內容~~
※※※※※※※※※※※※※※※※※※※※※※※※※※※※※※※※※
我是「數據分析那些事」。常年分享數據分析乾貨,不定期分享好用的職場技能工具。各位也可以關注我的Facebook,按讚我的臉書並私訊「10」,送你十週入門數據分析電子書唷!期待你與我互動起來~
文章推薦:
回顧十週入門數據分析系列文:
關注數據君的臉書,ins(全網同名)
我是「數據分析那些事」。常年在臉書,ins分享數據分析乾貨,不定期分享好用的職場技能工具。按贊我的臉書,並在臉書置頂帖子下回復SQL50,會有MySQL經典50題及答案贈送唷!