LeetCodeTools for Sublime Text

LeetCodeTools is A Sublime Text plugin for practicing LeetCode without leaving your editor. It lets you search problems, fetch problems and official solutions, run your code offline against the official examples, submit directly to LeetCode, and pick problems from study plans like Hot 100 and Top Interview 150. Problems and solutions are saved locally as Markdown files; log in once and you're good to go.

LeetCodeTools 是一个 Sublime Text 插件，让你不用离开编辑器就能刷力扣（LeetCode）。它支持按题号或关键字搜索题目、拉取题目和官方题解、在本地离线运行代码并与官方示例对比、一键提交到力扣判题，还能从「热题 100」「面试经典 150 题」等题集里选题。题目和题解都会以 Markdown 文件保存到本地，登录一次即可长期使用。 一个让你不用离开编辑器、直接在 Sublime Text 里刷力扣（LeetCode.cn）的插件：搜索题目、拉取题目与官方题解、离线运行、直接提交。

How to login / 如何登录

1. 在浏览器里登录 leetcode.cn（页面能正常打开，说明已经过掉 Cloudflare）
2. 按 F12 → 选「网络 / Network」标签 → 刷新一次页面
3. 点左侧任意一个 leetcode.cn 的请求
4. 右侧找「请求标头 / Request Headers」→ 找到 Cookie 这一行
5. 复制整条值，随便用哪种办法： · 右键那个值 → Copy value（最省事） · 在值上连点三下选中整段 → Ctrl+C（别按 Ctrl+A，那会选中整个面板） · 或者右键左侧请求 → Copy → Copy as cURL，整段粘过来也行（插件会自己抠） （一定要从「请求标头」复制，不要从响应头的 Set-Cookie 复制）
6. 回到 Sublime，粘进输入框，回车

（leetcode.cn 只贴 LEETCODE_SESSION 的值也能用；整条粘更好。）

1. Log in to leetcode.cn in your browser.
2. F12 → Network tab → reload the page.
3. Click any request to leetcode.cn.
4. Request Headers → find the Cookie line.
5. Copy the whole value: right-click it → Copy value, or triple-click it → Ctrl+C, or right-click the request → Copy → Copy as cURL and paste that whole block (the plugin extracts the cookie from it). Copy from Request Headers, never from a Set-Cookie response header.
6. Paste it into the input box and press Enter. 打开浏览器 / Open browser

Features / 功能

Command 命令	Description 说明
LeetCode Tools: Login	Open the browser to log in and paste the LEETCODE_SESSION cookie. 打开浏览器登录并粘贴 LEETCODE_SESSION Cookie
LeetCode Tools: Search	Open a filterable list of every problem (type an id or a title keyword in the panel, press Enter to fetch). 打开可过滤的题目列表（在面板里输入题号或标题关键字，回车拉题）
LeetCode Tools: Update	Rebuild the local problem-list and study-plan caches. 重建本地题目列表和题集缓存
LeetCode Tools: Run	Run your code offline and compare with the official examples. 离线运行代码，与官方示例对比
LeetCode Tools: Run Online	Same output as Run, but executed by LeetCode (like Ctrl+' in the web editor); works for every language, needs the login cookie. 输出格式和 Run 一致，但代码发到力扣跑（等价网页端 Ctrl+'）；支持所有语言，需要登录 Cookie
LeetCode Tools: Reload Problem	Force-refetch the current problem: the statement is overwritten and _in.json / _out.json are regenerated (the "insert return stubs and Run Code" step included). Your own solution file is kept. 强制重拉当前题：题面覆盖、用例与期望输出重新抓（含"插空返回值骗 Run Code"那一步）；你自己写的解法不会被换掉
LeetCode Tools: Edit Testcases	Open the current problem's _in.json (inputs) and _out.json (expected outputs) for editing. 打开当前题的 _in.json（输入用例）和 _out.json（期望输出）直接编辑
LeetCode Tools: Submit	Submit the current code to LeetCode for judging. 把当前代码提交到力扣判题
LeetCode Tools: Open in Browser	Open the current problem's page in the browser. 在浏览器打开当前题目的网页
LeetCode Tools: Fetch Official Explanations	Fetch the official solution as xxx_explanation.md. 拉取官方题解为 xxx_explanation.md
LeetCode Tools: Select from Problem Set	Pick a problem from a study plan (Hot 100, Top Interview 150, …). 从题集（热题100 / 面试经典150题等学习计划）里选题
LeetCode Tools: Daily Question	Fetch today's daily question. 拉取今日的每日一题

Installation / 安装

1. Install Package Control first if you don't have it yet. 先安装 Package Control（如果还没有）。

2. Press Ctrl+Shift+P (Cmd+Shift+P on macOS), run Package Control: Install Package, search for LeetCodeTools, and press Enter to install. 按 Ctrl+Shift+P（macOS 为 Cmd+Shift+P），运行 Package Control: Install Package，搜索 LeetCodeTools 并回车安装。

3. Install a system Python 3 and add it to PATH. It is required by the offline Run (not by Login). 安装一个系统 Python 3 并加入 PATH。离线 Run 需要它（登录不需要）。

4. Press Ctrl+Shift+P, run LeetCode Tools: Login; it opens the site in your browser. Log in, then press F12 → Network → reload → click any request → Request Headers → Cookie and copy the whole value — either right-click it → Copy value, or triple-click it → Ctrl+C (a plain Ctrl+A there selects the whole panel, not this value), or right-click the request → Copy → Copy as cURL and paste that whole block (the plugin extracts the cookie from it). Copy from Request Headers, never from a Set-Cookie response header: only the request header carries the cf_clearance / __cf_bm cookies Cloudflare issues, and leetcode.com needs them for Submit / Run Online. 按 Ctrl+Shift+P，运行 LeetCode Tools: Login；它会在浏览器打开站点。登录后按 F12 → Network（网络） → 刷新 → 点任意一个请求 → Request Headers（请求标头）→ Cookie，把整条值复制下来：右键那个值 → Copy value 最省事；也可以在值上连点三下选中后 Ctrl+C（直接按 Ctrl+A 在这里选中的是整个面板，不是这条值）；或者右键请求 → Copy → Copy as cURL，整段粘过来也行（插件会自己把 Cookie 抠出来）。注意要从请求标头复制，不要从响应头的 Set-Cookie 复制：只有请求头那一份才带 Cloudflare 发的 cf_clearance / __cf_bm，leetcode.com 的 Submit / Run Online 需要它。

Note: Search / Update / Submit / Run Online need the login cookie; Open in Browser, Run, and Fetch Official Explanations do not (the official solution uses the public GraphQL API). 说明：Search / Update / Submit / Run Online 需要登录 Cookie；Open in Browser、Run、Fetch Official Explanations 不需要登录（其中官方题解走的是公开 GraphQL 接口）。

Configuration / 配置

Settings are edited via Preferences: LeetCodeTools Settings in the command palette, or Preferences → Package Settings → LeetCodeTools → Settings in the menu: 设置通过命令面板的 Preferences: LeetCodeTools Settings，或菜单 Preferences → Package Settings → LeetCodeTools → Settings 编辑：

Key 配置项	Default 默认值	Description 说明
working_dir	~/leetcode	Where problems are saved. 题目保存目录
default_lang	python3	Default language used when fetching a problem. 拉题时的默认语言
language	zh	Problem language (zh / en), independent of site. leetcode.com has no Chinese content, so com + zh falls back to English. The title cache is kept per language (problem_list_zh.json / problem_list_en.json). 题目语言（zh / en），和 site 无关。英文站没有中文内容，所以 com + zh 会回滚成英文。标题缓存按语言分成 problem_list_zh.json / problem_list_en.json 两份
site	cn	cn = leetcode.cn; com = leetcode.com. Cookies and per-problem caches are stored per site, so each site needs its own Login. cn = leetcode.cn；com = leetcode.com。Cookie 和每题缓存按站点分开存，两个站各要 Login 一次
browser_ua	""	User-Agent used for requests. Leave empty for a built-in Chrome UA; set it to your own browser's exact UA when you want a pasted cf_clearance to be accepted (see the leetcode.com note below). 请求使用的 User-Agent。留空用内置 Chrome UA；想让粘贴的 cf_clearance 生效就填自己浏览器的 UA（见下方 leetcode.com 说明）
cache_age_days	7	Cache age in days for the problem list and study plans; auto-refreshes when expired. 题目列表与题集的缓存天数，过期自动刷新
run_timeout	1	Time limit in seconds for the offline Run (0 = no limit). 离线 Run 的超时秒数（0 表示不限时）

Supported languages / 支持的语言

These languages are available for fetched code templates and for Submit; the offline Run is Python-only (see limitations below). 以下语言可用于拉取的代码模板和 Submit；离线 Run 目前仅支持 Python（见下方限制）。

Language slug 语言	Extension 扩展名	Run offline 离线运行
python3 / python	.py	✔️
java	.java	❌️
cpp	.cpp	❌️
c	.c	❌️
csharp	.cs	❌️
javascript	.js	❌️
typescript	.ts	❌️
golang	.go	❌️
rust	.rs	❌️
kotlin	.kt	❌️
swift	.swift	❌️
scala	.scala	❌️
ruby	.rb	❌️
php	.php	❌️

Usage / 使用流程

1. Log in (once): LeetCode Tools: Login → log in in the browser → copy LEETCODE_SESSION → paste it. 登录（只需一次）：LeetCode Tools: Login → 浏览器登录 → 复制 LEETCODE_SESSION → 粘贴。

2. Pick a problem: LeetCode Tools: Search → the panel lists every problem; type an id or a title keyword to filter, press Enter to fetch and open the .md and code file. 选题：LeetCode Tools: Search → 面板里列出全部题目，直接输入题号或标题关键字过滤，回车自动拉取并打开 .md 和代码文件。

3. Solve it: write your solution in the code file. 做题：在代码文件里写解法。

4. Run: LeetCode Tools: Run executes locally (Python only) and compares with the official examples; LeetCode Tools: Run Online has the same output layout but is executed by LeetCode (like Ctrl+' in the web editor), so it works for any language. 运行：LeetCode Tools: Run 在本地跑（仅 Python）并与官方示例对比；LeetCode Tools: Run Online 输出格式完全一致，但代码是发到力扣跑的（等价网页端 Ctrl+'），所以任何语言都能用。

5. Submit: LeetCode Tools: Submit; failed cases are appended to the local _in.json / _out.json for offline replay. 提交：LeetCode Tools: Submit，失败用例会自动追加到本地 _in.json / _out.json，方便离线复现。

6. Read the solution: LeetCode Tools: Fetch Official Explanations generates and opens xxx_explanation.md. 看题解：LeetCode Tools: Fetch Official Explanations，生成并打开 xxx_explanation.md。

Generated files / 生成的文件

User-facing files go under working_dir; the plugin's internal files go into the cache. 用户可见的文件存在 working_dir，插件内部文件存在缓存目录。

Under working_dir（在 working_dir 下）:

File 文件	Content 内容
{slug}.md	Problem description (HTML → Markdown). 题目描述（HTML 转成 Markdown）
{slug}.py	Official code template. 官方代码模板
{slug}_explanation.md	Official solution (from Fetch Official Explanations). 官方题解（由 Fetch Official Explanations 生成）

In the cache {working_dir}/.cache/（在缓存目录下）:

Everything cookie-side is per site; the problem/title list is per language. leetcode.cn keeps the plain cookie names and can produce both title lists; leetcode.com only ever writes the English one, so com + language: zh reuses the Chinese list downloaded from cn and falls back to English for problems it does not contain. Cookie 那份按站点分；题目列表（标题）按语言分。leetcode.cn 沿用原来的 Cookie 文件名，并且能同时产出中英两份标题列表；leetcode.com 只会写英文那份，所以 com + language: zh 会复用从 cn 下过的中文列表，列表里没有的题再回滚英文。

File 文件	Content 内容
cookie.json (cn) / cookie_com.json (com)	Login cookie for that site. 该站点的登录 Cookie
problem_list_zh.json / problem_list_en.json	Title list per language. 按语言分的题目列表（标题缓存）
last_update_zh.json / last_update_en.json	Refresh timestamp of each list. 各自列表的更新时间
problems[_com]/{slug}.json	Metadata (id / metaData / examples). 元数据（题号 / metaData / 示例用例）
problems[_com]/{slug}_in.json	Parsed input cases. 解析后的输入用例
problems[_com]/{slug}_out.json	Expected outputs. 预期输出
images/{slug}/	Problem-description images (shared). 题目描述的图片（共用）
images/{slug}_explanation/	Official-solution images (shared). 官方题解的图片（共用）

Current limitations / 当前限制

• Offline Run currently supports only Python: the local judge runs your code with Python exec and includes built-in parsing for ListNode / TreeNode / Node. Other languages (java / cpp / javascript / golang, …) can be run with LeetCode Tools: Run Online (LeetCode executes them) or submitted with Submit — they just cannot run locally.
• 离线 Run 目前仅支持 Python：本地判题用 Python exec 运行你的代码，并内置了 ListNode / TreeNode / Node 的用例解析。其它语言（java / cpp / javascript / golang 等）可以用 LeetCode Tools: Run Online（交给力扣执行）或 Submit 提交，只是不能本地跑。
• Offline Run compares values exactly, list order included, while some problems accept answers in any order (Two Sum indices, subsets, permutations, …) — a FAIL on those can be the same answer written in a different order. Run Online and Submit are judged by LeetCode's own checker.
• 离线 Run 是按值严格比较、顺序也算的，而有些题的答案顺序无关（两数之和的下标、子集、全排列等）—— 这类题上出现 FAIL 可能只是写法顺序不同。Run Online 和 Submit 走的是力扣自己的判定。
• On leetcode.com, everything except Submit and Run Online works with a normal login: search, fetch, study plans, daily question and official explanations all use the public API (the plugin's queries are site-aware). The two write endpoints — /problems/<slug>/submit/ and /problems/<slug>/interpret_solution/ — sit behind a Cloudflare managed challenge (cf-mitigated: challenge, HTTP 403) and only a browser's cf_clearance cookie gets through it. If Cloudflare challenges a request the error message says exactly that. To give it the best chance: pass the challenge in your browser, run Login and paste the whole Cookie request header (so cf_clearance / __cf_bm come along), and set browser_ua to that same browser's User-Agent — Cloudflare binds cf_clearance to the UA (and IP) that obtained it. Without it, Submit / Run Online will keep failing, and the offline Run can only show your own outputs (no EXPECT lines, and the panel says no expected outputs, nothing compared instead of faking failures).
• leetcode.com 上除了 Submit 和 Run Online，其余都能用正常登录跑通：搜题、拉题、题集、每日一题、官方题解走的都是公开接口（插件里的查询已按站点区分）。只有两个写接口 —— /problems/<slug>/submit/ 和 /problems/<slug>/interpret_solution/ —— 后面是 Cloudflare 托管挑战（cf-mitigated: challenge / HTTP 403），只有浏览器的 cf_clearance 能过。真被挑战时，报错里会直接写明。想尽量过掉：先在浏览器里过掉挑战，再运行 Login 并粘整条 Cookie 请求头（这样 cf_clearance / __cf_bm 才带上），同时把 browser_ua 设成同一个浏览器的 User-Agent —— Cloudflare 的 cf_clearance 和拿到它的 UA（以及 IP）是绑定的。搞不定的话 Submit / Run Online 会一直失败，离线 Run 也只能看到你自己的输出（没有 EXPECT 行，面板会写 no expected outputs, nothing compared，而不是假装全错）。
• Only the China site (leetcode.cn) is fully supported; the global site (leetcode.com) works for everything the public API allows, and for Submit / Run Online as far as Cloudflare lets it through.

Roadmap / 待办

• Separate the parser (test-case parsing + linked-list/tree/graph building + offline judge) out of leetcodetools.py into a standalone module/script, so it can run from the command line outside Sublime (e.g. python -m leetcodetools_offline solution.py). 把 parser（用例解析 + 链表/树/图构建 + 离线判题）从 leetcodetools.py 里分离成独立模块 / 脚本，使其可以脱离 Sublime、在命令行本地执行（例如 python -m leetcodetools_offline solution.py）。
• Class / design problems (e.g. Min Stack, where you implement a class with multiple methods) are not supported by the offline Run yet. 「最小栈」这类 class 设计题（需要实现一个类、多个方法）暂不支持本地 Run。

Directory structure / 目录结构

LeetCodeTools/
├── leetcodetools.py                # Main plugin (commands + GraphQL client + offline judge). 主插件（命令 + GraphQL 客户端 + 离线判题）
├── Default.sublime-commands        # Command-palette menu. 命令面板菜单
├── LeetCodeTools.sublime-settings  # Default settings. 默认配置
└── .python-version                 # Dev Python version. 开发环境 Python 版本

Notes / 其它

• The login cookie is cached at {working_dir}/.cache/cookie.json; the problem list at {working_dir}/.cache/problem_list.json; study plans at {working_dir}/.cache/study_plans.json and {working_dir}/.cache/study_plan_problems.json. 登录 Cookie 缓存在 {working_dir}/.cache/cookie.json，题目列表缓存在 {working_dir}/.cache/problem_list.json，题集缓存在 {working_dir}/.cache/study_plans.json 和 {working_dir}/.cache/study_plan_problems.json。
• When you pick a problem from a study plan (Select from Problem Set), submitting it also marks it as done in that study plan. 从题集（Select from Problem Set）选的题，提交时也会在对应题集里标记完成。