deps: Upgrade Nu to 0.112.2 & hustcer/setup-nu to v3.23 (#202)

* deps: Upgrade Nu to 0.112.2

* deps: Upgrade hustcer/setup-nu@v3.23

.github/workflows/cr.yml

@@ -37,7 +37,7 @@ jobs:
             As a senior Nushell engineer, perform comprehensive script review with focus on:
 
             ### 1. Core Requirements:
-            - Validate Nu 0.90+ compatibility
+            - Validate Nu 0.100+ compatibility
             - Check structured data handling
             - Verify pipeline efficiency
             - Assess module organization
@@ -55,7 +55,7 @@ jobs:
             - Parallel execution opportunities
 
             **Rules:**
-            - Target Nu 0.90+ features
+            - Target Nu 0.100+ features
             - Highlight data flow vulnerabilities
             - Suggest structured data optimizations
             - Keep feedback Nu-specific

.github/workflows/tests.yml

@@ -15,6 +15,7 @@ on:
     branches:
       - main
       - develop
+      - feature/test
     paths-ignore:
       - '**.md'
       - 'docs/**'
@@ -41,12 +42,12 @@ jobs:
     runs-on: ${{ matrix.platform }}
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v6
 
       - name: Checkout Nutest Repo
-        uses: actions/checkout@v4
+        uses: actions/checkout@v6
         with:
-          ref: v1.1.0
+          ref: main
           path: nutest
           repository: vyadh/nutest
           sparse-checkout: nutest/

CHANGELOG.md

@@ -1,6 +1,60 @@
 # Changelog
 All notable changes to this project will be documented in this file.
 
+## [1.20.0] - 2026-01-23
+
+### Bug Fixes
+
+- Make config-check use default value for --config flag
+- Fix `from env` for .env file parsing (#193)
+
+### Miscellaneous Tasks
+
+- Update minimum required Nushell version to 0.110.0
+- Update nutest to main ref (#191)
+- Update README.md
+
+### Refactor
+
+- A better `from env` parser (#194)
+
+### Deps
+
+- Upgrade to actions/checkout@v5
+- Upgrade `hustcer/setup-nu` to v3.20
+- Upgrade Nushell version to 0.108.0 (#190)
+- Upgrade `hustcer/setup-nu` to v3.21 (#192)
+- Upgrade actions/checkout@v6
+- Upgrade Nu to 0.109.1 (#195)
+- Update Nushell to 0.110.0 (#196)
+- Upgrade `hustcer/setup-nu` to v3.22
+
+## [1.19.0] - 2025-07-23
+
+### Bug Fixes
+
+- Fix "variable not found" error (#185)
+- Fix getting Nu binary path for Nushell 0.106
+
+### Deps
+
+- Upgrade Nu to v0.106 (#186)
+
+## [1.18.0] - 2025-06-11
+
+### Features
+
+- Set default `temperature` to **0.3** for code review (#181)
+
+### Miscellaneous Tasks
+
+- Refine diff flag descriptions in docs and scripts (#177)
+- Upgrade `Nu` to 0.105 and pin [`hustcer/setup-nu`](https://github.com/hustcer/setup-nu) to v3.19 (#183)
+
+### Deps
+
+- Upgrade `nutest` to v1.1.0 (#179)
+
 ## [1.17.0] - 2025-04-11
 
 ### Bug Fixes

Justfile

@@ -28,7 +28,7 @@ alias cr := code-review
 
 # Used to handle the path separator issue
 DEEPSEEK_REVIEW_PATH := parent_directory(justfile())
-NU_DIR := parent_directory(`(which nu).path.0`)
+NU_DIR := parent_directory(`$nu.current-exe`)
 _query_plugin := if os_family() == 'windows' { 'nu_plugin_query.exe' } else { 'nu_plugin_query' }
 
 # To pass arguments to a dependency, put the dependency

README.md

@@ -7,31 +7,29 @@
 
 [中文说明](README.zh-CN.md)
 
-`deepseek-review` also offers seamless integration with DeepSeek models on SiliconCloud. [Sign Up Now](https://cloud.siliconflow.cn/i/rqCdIxzS) to **Claim Your Free 20 Million Tokens** and start exploring its capabilities!
-
 ## Features
 
 ### GitHub Action
 
 - Automate PR Reviews with DeepSeek via GitHub Action
-- Add `skip cr` or `skip review` to PR Title or Body to Disable Code Review in GitHub Actions
+- Add `skip cr` or `skip review` to the PR title or body to disable code review in GitHub Actions
 - Cross-platform Support: Compatible with GitHub Runners across `macOS`, `Ubuntu`, and `Windows`.
 
 ### Local Code Review
 
-- Streaming Output Support for Local Code Review
-- Review Remote GitHub PRs Directly from Your Local CLI
-- Review Commit Changes with DeepSeek for Any Local Repository by CLI
-- Support On-demand Changes Generation via `git show`/`git diff` Command for Further Code Review
-- Output Code Review Result to Specified File in Markdown Format
-- Cross-platform Compatibility: Designed to function seamlessly across all platforms capable of running [Nushell](https://github.com/nushell/nushell)
+- Streaming output support for local code review
+- Review remote GitHub PRs directly from your local CLI
+- Review commit changes with DeepSeek for any local repository via CLI
+- Support on-demand change generation via `git show`/`git diff` commands for further code review
+- Output code review results to a specified file in Markdown format
+- Cross-platform compatibility: Designed to function seamlessly across all platforms capable of running [Nushell](https://github.com/nushell/nushell)
 
 ### Both GH Action & Local
 
-- Support Both DeepSeek's `V3` & `R1` Models
-- Fully Customizable: Choose Models, Base URLs, and Prompts
-- Supports Self-Hosted DeepSeek Models for Enhanced Flexibility
-- Perform Code Reviews for Changes That either Include or Exclude Specific Files
+- Support both DeepSeek's `V3` and `R1` models
+- Fully customizable: Choose models, base URLs, and prompts
+- Support self-hosted DeepSeek models for enhanced flexibility
+- Perform code reviews for changes that either include or exclude specific files
 
 ## Planned Features
 
@@ -40,7 +38,7 @@
 
 ## Code Review with GitHub Action
 
-### Initiate Code Review When PR was Created
+### Initiate Code Review When a PR is Created
 
 Add a GitHub workflow with the following contents:
 
@@ -82,13 +80,13 @@ jobs:
 
 </details>
 
-When a PR is created, DeepSeek code review will be automatically triggered, and the review results(depend on your prompt) will be posted as comments on the corresponding PR. For example:
+When a PR is created, DeepSeek code review will be automatically triggered, and the review results (depending on your prompt) will be posted as comments on the corresponding PR. For example:
 - [Example 1](https://github.com/hustcer/deepseek-review/pull/30) with [default prompts](https://github.com/hustcer/deepseek-review/blob/main/action.yaml#L35) & [Run Log](https://github.com/hustcer/deepseek-review/actions/runs/13043609677/job/36390331791#step:2:53).
 - [Example 2](https://github.com/hustcer/deepseek-review/pull/68) with [this prompt](https://github.com/hustcer/deepseek-review/blob/eba892d969049caff00b51a31e5c093aeeb536e3/.github/workflows/cr.yml#L32)
 
-### Trigger CR When a Specific Label was Added
+### Trigger Code Review When a Specific Label is Added
 
-If you don't want automatic review on PR creation, you can choose to trigger code review by adding a label. For example, create the following workflow:
+If you don't want automatic code review on PR creation, you can choose to trigger code review by adding a label. For example, create the following workflow:
 
 ```yaml
 name: Code Review
@@ -121,21 +119,21 @@ With this setup, DeepSeek code review will not run automatically upon PR creatio
 | Name           | Type   | Description                                                             |
 | -------------- | ------ | ----------------------------------------------------------------------- |
 | chat-token     | String | Required, DeepSeek API Token                                            |
-| model          | String | Optional, The model used for code review, defaults to `deepseek-chat`   |
+| model          | String | Optional, The model used for code review, defaults to `deepseek-v4-flash`   |
 | base-url       | String | Optional, DeepSeek API Base URL, defaults to `https://api.deepseek.com` |
-| max-length     | Int    | Optional, Maximum length(Unicode width) of the content for review, if the content length exceeds this value, the review will be skipped. Default `0` means no limit. |
+| max-length     | Int    | Optional, Maximum length (Unicode width) of the content for review. If the content length exceeds this value, the review will be skipped. Default `0` means no limit. |
 | sys-prompt     | String | Optional, System prompt corresponding to `$sys_prompt` in the payload, default value see note below |
 | user-prompt    | String | Optional, User prompt corresponding to `$user_prompt` in the payload, default value see note below |
-| temperature    | Number | Optional, The temperature for the model to generate the response, between `0` and `2`, default value `0.3` |
-| include-patterns | String | Optional, The comma separated file patterns to include in the code review. No default |
-| exclude-patterns | String | Optional, The comma separated file patterns to exclude in the code review. Default to `pnpm-lock.yaml,package-lock.json,*.lock` |
+| temperature    | Number | Optional, The temperature for the model to generate the response, between `0` and `2`. Default value is `0.3` |
+| include-patterns | String | Optional, Comma-separated file patterns to include in the code review. No default |
+| exclude-patterns | String | Optional, Comma-separated file patterns to exclude from the code review. Defaults to `pnpm-lock.yaml,package-lock.json,*.lock` |
 | github-token   | String | Optional, The `GITHUB_TOKEN` secret or personal access token to authenticate. Defaults to `${{ github.token }}`. |
 
 **DeepSeek API Call Payload**:
 
 ```js
 {
-  // `$model` default value: deepseek-chat
+  // `$model` default value: deepseek-v4-flash
   model: $model,
   stream: false,
   temperature: $temperature,
@@ -154,18 +152,18 @@ With this setup, DeepSeek code review will not run automatically upon PR creatio
 
 > [!NOTE]
 >
-> You can control the language of the code review results by the language of the
-> Prompt. The default Prompt language is currently English. When you use a Chinese
-> Prompt, the generated code review results will be in Chinese.
+> You can control the language of the code review results through the language of the
+> prompt. The default prompt language is currently English. When you use a Chinese
+> prompt, the generated code review results will be in Chinese.
 
 ## Local Code Review
 
 ### Required Tools
 
-To perform code reviews locally(should works for `macOS`, `Ubuntu`, and `Windows`), you need to install the following tools:
+To perform code reviews locally (works on `macOS`, `Ubuntu`, and `Windows`), you need to install the following tools:
 
-- [`Nushell`](https://www.nushell.sh/book/installation.html). It is recommended to install the latest versions(min version required: `0.103`).
-- The latest version of [`awk`](https://github.com/onetrueawk/awk) or [`gawk`](https://www.gnu.org/software/gawk/) is required, with `gawk` being the preferred choice.
+- [`Nushell`](https://www.nushell.sh/book/installation.html). It is recommended to install the latest version (minimum version required: `0.112.2`).
+- The latest version of [`awk`](https://github.com/onetrueawk/awk) or [`gawk`](https://www.gnu.org/software/gawk/) is required, with `gawk` being preferred.
 - Clone this repository to your local machine, navigate to the repository directory, and run `nu cr -h`. You should see an output similar to the following:
 
 ```console
@@ -183,7 +181,7 @@ Flags:
   -t, --diff-to <string>: Git diff ending commit SHA
   -c, --patch-cmd <string>: The `git show` or `git diff` command to get the diff content, for local CR only
   -l, --max-length <int>: Maximum length of the content for review, 0 means no limit.
-  -m, --model <string>: Model name, or read from CHAT_MODEL env var, `deepseek-chat` by default
+  -m, --model <string>: Model name, or read from CHAT_MODEL env var, `deepseek-v4-flash` by default
   -b, --base-url <string>: DeepSeek API base URL, fallback to BASE_URL env var
   -U, --chat-url <string>: DeepSeek Model chat full API URL, e.g. http://localhost:11535/api/chat
   -s, --sys-prompt <string>: Default to $DEFAULT_OPTIONS.SYS_PROMPT,
@@ -202,27 +200,27 @@ Parameters:
 
 ### Environment Configuration
 
-To perform code review locally, you need to modify the configuration file. The repository already provides a configuration example [`config.example.yml`](https://github.com/hustcer/deepseek-review/blob/main/config.example.yml). Copy it to `config.yml` and modify it according to your actual needs. ​**Read the comments in the configuration file carefully**, as they explain the purpose of each configuration item.
+To perform code reviews locally, you need to modify the configuration file. The repository provides a configuration example [`config.example.yml`](https://github.com/hustcer/deepseek-review/blob/main/config.example.yml). Copy it to `config.yml` and modify it according to your needs. **Read the comments in the configuration file carefully**, as they explain the purpose of each configuration item.
 
 > [!WARNING]
 >
-> The `config.yml` configuration file is ​**only used locally** and will not be utilized in GitHub Workflow. ​**Sensitive information** in this file should be properly secured and ​**never committed** to the code repository.
+> The `config.yml` configuration file is **only used locally** and will not be utilized in GitHub Workflows. **Sensitive information** in this file should be properly secured and **never committed** to the code repository.
 >
 
-**Create Command Alias**
+**Create a Command Alias**
 
-For convenience in performing code review across any local repository, create a command alias. For example:
+For convenience when performing code reviews across any local repository, create a command alias. For example:
 
 ```sh
 # For Nushell: Modify config.nu and add:
 alias cr = nu /absolute/path/to/deepseek-review/cr --config /absolute/path/to/deepseek-review/config.yml
 
-# Modify ~/.zshrc for zsh or ~/.bashrc for bash or ~/.config/fish/config.fish for fish and add:
+# Modify ~/.zshrc for zsh, ~/.bashrc for bash, or ~/.config/fish/config.fish for fish and add:
 alias cr="nu /absolute/path/to/deepseek-review/cr --config /absolute/path/to/deepseek-review/config.yml"
 
-# After sourcing the modified profile, use `cr` for code review
+# After sourcing the modified profile, use `cr` for code reviews
 
-# For Windows powershell users please set cr alias by editing $PROFILE and add:
+# For Windows PowerShell users, set the cr alias by editing $PROFILE and add:
 function cr {
   nu D:\absolute\path\to\deepseek-review\cr --config D:\absolute\path\to\deepseek-review\config.yml @args
 }
@@ -235,20 +233,20 @@ function cr {
 To review a local repository:
 
 - Navigate to the Git repository directory.
-- Use the `cr` command to review current modifications, ​provided that `config.yml` is correctly configured.
+- Use the `cr` command to review current modifications, provided that `config.yml` is correctly configured.
 
 **Usage Examples**
 
 ```sh
-# Perform code review on the `git diff` changes in current directory
+# Perform code review on the `git diff` changes in the current directory
 cr
-# Perform code review on the `git diff f536acc` changes in current directory
+# Perform code review on the `git diff f536acc` changes in the current directory
 cr --diff-from f536acc
-# Perform code review on the `git diff f536acc` changes and output result to review.md
+# Perform code review on the `git diff f536acc` changes and output the result to review.md
 cr --diff-from f536acc --output review.md
-# Perform code review on the `git diff f536acc 0dd0eb5` changes in current directory
+# Perform code review on the `git diff f536acc 0dd0eb5` changes in the current directory
 cr --diff-from f536acc --diff-to 0dd0eb5
-# Review the changes in current directory using the `--patch-cmd` flag
+# Review the changes in the current directory using the `--patch-cmd` flag
 cr --patch-cmd 'git diff head~3'
 cr -c 'git show head~3'
 cr -c 'git diff 2393375 71f5a31'
@@ -261,17 +259,17 @@ cr -c 'git diff 2393375 71f5a31 :!nu/*'
 
 When reviewing a remote GitHub PR locally:
 
-- Always specify the PR number via `--pr-number`
+- Always specify the PR number via `--pr-number`.
 - Use `--repo` to indicate the target repository (e.g., `hustcer/deepseek-review`). If `--repo` is omitted, the tool reads `settings.default-github-repo` from `config.yml`.
 
 **Usage Examples**
 
 ```sh
-# Perform code review on PR #31 in the remote DEFAULT_GITHUB_REPO repo
+# Perform code review on PR #31 in the remote DEFAULT_GITHUB_REPO repository
 cr --pr-number 31
-# Perform code review on PR #31 in the remote hustcer/deepseek-review repo
+# Perform code review on PR #31 in the remote hustcer/deepseek-review repository
 cr --pr-number 31 --repo hustcer/deepseek-review
-# Perform code review on PR #31 and exclude changes of pnpm-lock.yaml
+# Perform code review on PR #31 and exclude changes in pnpm-lock.yaml
 cr --pr-number 31 --exclude pnpm-lock.yaml
 ```
 

README.zh-CN.md

@@ -5,8 +5,6 @@
 ![Failed](https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fgist.githubusercontent.com%2Fhustcer%2Fb99391ee59016b17d0befe3331387e89%2Fraw%2Ftest-summary.json&query=%24.failed&label=Failed&color=red)
 ![Skipped](https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fgist.githubusercontent.com%2Fhustcer%2Fb99391ee59016b17d0befe3331387e89%2Fraw%2Ftest-summary.json&query=%24.skipped&label=Skipped&color=yellow)
 
-本工具也支持使用 SiliconCloud 上的 DeepSeek 模型，[注册](https://cloud.siliconflow.cn/i/rqCdIxzS) 就**免费赠送 2000 万 Token**，赶紧试试吧！
-
 ## 特性
 
 ### GitHub Action
@@ -119,7 +117,7 @@ jobs:
 | 名称           | 类型   | 描述                                                           |
 | -------------- | ------ | -------------------------------------------------------------- |
 | chat-token     | String | 必填，DeepSeek API Token                                       |
-| model          | String | 可选，配置代码审查选用的模型，默认为 `deepseek-chat`           |
+| model          | String | 可选，配置代码审查选用的模型，默认为 `deepseek-v4-flash`           |
 | base-url       | String | 可选，DeepSeek API Base URL, 默认为 `https://api.deepseek.com` |
 | max-length     | Int    | 可选，待审查内容的最大 Unicode 长度, 默认 `0` 表示没有限制，超过非零值则跳过审查 |
 | sys-prompt     | String | 可选，系统提示词对应入参中的 `$sys_prompt`, 默认值见后文注释      |
@@ -133,7 +131,7 @@ DeepSeek 接口调用入参:
 
 ```js
 {
-  // `$model` default value: deepseek-chat
+  // `$model` default value: deepseek-v4-flash
   model: $model,
   stream: false,
   temperature: $temperature,
@@ -161,7 +159,7 @@ DeepSeek 接口调用入参:
 
 在本地进行代码审查，支持 `macOS`, `Ubuntu` & `Windows` 不过需要安装以下工具：
 
-- [`Nushell`](https://www.nushell.sh/book/installation.html), 建议安装最新版本(最低版本 `0.103`)
+- [`Nushell`](https://www.nushell.sh/book/installation.html), 建议安装最新版本(最低版本 `0.112.2`)
 - [`awk`](https://github.com/onetrueawk/awk) 或者 [`gawk`](https://www.gnu.org/software/gawk/) 的最新版版本，优先推荐 `gawk`
 - 接下来只需要把本仓库代码克隆到本地，然后进入仓库目录执行 `nu cr -h` 即可看到类似如下输出:
 
@@ -180,7 +178,7 @@ Flags:
   -t, --diff-to <string>: Git diff ending commit SHA
   -c, --patch-cmd <string>: The `git show` or `git diff` command to get the diff content, for local CR only
   -l, --max-length <int>: Maximum length of the content for review, 0 means no limit.
-  -m, --model <string>: Model name, or read from CHAT_MODEL env var, `deepseek-chat` by default
+  -m, --model <string>: Model name, or read from CHAT_MODEL env var, `deepseek-v4-flash` by default
   -b, --base-url <string>: DeepSeek API base URL, fallback to BASE_URL env var
   -U, --chat-url <string>: DeepSeek Model chat full API URL, e.g. http://localhost:11535/api/chat
   -s, --sys-prompt <string>: Default to $DEFAULT_OPTIONS.SYS_PROMPT,

action.yaml

@@ -24,7 +24,7 @@ inputs:
     description: 'The maximum length of the content for review, 0 means no limit.'
   model:
     required: false
-    default: 'deepseek-chat'
+    default: 'deepseek-v4-flash'
     description: 'The DeepSeek model to choose for code review.'
   temperature:
     required: false
@@ -58,9 +58,9 @@ runs:
   using: 'composite'
   steps:
     - name: Setup Nu
-      uses: hustcer/setup-nu@v3
+      uses: hustcer/setup-nu@v3.23
       with:
-        version: 0.103.0
+        version: 0.112.2
 
     - name: DeepSeek Code Review
       shell: nu {0}

config.example.yml

@@ -50,7 +50,7 @@ providers:
     token: 'YOUR_DEEPSEEK_TOKEN'    # Required, The API token for the provider
     base-url: 'https://api.deepseek.com'
     models:
-      - name: 'deepseek-chat'       # Required, Pass the model name to --model flag to use it
+      - name: 'deepseek-v4-flash'       # Required, Pass the model name to --model flag to use it
         alias: v3                   # Optional, Alias name could also be passed to --model flag
         enabled: true               # One and Only one model could be enabled in one model group
         description: 'DeepSeek V3'  # Optional, Description of the model, won't be used actually

cr

@@ -18,11 +18,11 @@ def main [
   --diff-to(-t): string,    # Git diff ending commit SHA
   --patch-cmd(-c): string,  # The `git show` or `git diff` command to get the diff content, for local CR only
   --max-length(-l): int,    # Maximum length of the content for review, 0 means no limit.
-  --model(-m): string,      # Model name, or read from CHAT_MODEL env var, `deepseek-chat` by default
+  --model(-m): string,      # Model name, or read from CHAT_MODEL env var, `deepseek-v4-flash` by default
   --base-url(-b): string,   # DeepSeek API base URL, fallback to BASE_URL env var
   --chat-url(-U): string,   # DeepSeek Model chat full API URL, e.g. http://localhost:11535/api/chat
-  --sys-prompt(-s): string  # Optional, System prompt message, fallback to SYSTEM_PROMPT env var
-  --user-prompt(-u): string # Default to $DEFAULT_OPTIONS.USER_PROMPT
+  --sys-prompt(-s): string  # Default to $DEFAULT_OPTIONS.SYS_PROMPT,
+  --user-prompt(-u): string # Default to $DEFAULT_OPTIONS.USER_PROMPT,
   --include(-i): string,    # Comma separated file patterns to include in the code review
   --exclude(-x): string,    # Comma separated file patterns to exclude in the code review
   --temperature(-T): float, # Temperature for the model, between `0` and `2`, default value `0.3`

cspell.yml

@@ -20,6 +20,7 @@ words:
 - hustcer
 - Nushell
 - creatio
+- pipefail
 - justfile
 - lefthook
 - deepseek

meta.json

@@ -1,7 +1,7 @@
 {
     "name": "deepseek-review",
-    "version": "1.17.0",
-    "actionVer": "v1.17",
+    "version": "1.20.0",
+    "actionVer": "v1.20",
     "author": "hustcer",
     "license": "MIT",
     "github": "https://github.com/hustcer/deepseek-review",

nu/common.nu

@@ -51,8 +51,8 @@ export def compare-ver [v1: string, v2: string] {
   # If you want to compare more parts use the following code:
   # for i in 0..([2 ($a | length) ($b | length)] | math max)
   for i in 0..2 {
-    let x = $a | get -i $i | default 0
-    let y = $b | get -i $i | default 0
+    let x = $a | get -o $i | default 0
+    let y = $b | get -o $i | default 0
     if $x > $y { return 1    }
     if $x < $y { return (-1) }
   }
@@ -81,28 +81,69 @@ export def check-nushell [--debug] {
 }
 
 # Converts a .env file into a record
-# may be used like this: open .env | load-env
-# works with quoted and unquoted .env files
-export def 'from env' []: string -> record {
-  lines
-    | split column '#' # remove comments
-    | get column1
-    | parse '{key}={value}'
-    | update value {
-        str trim                        # Trim whitespace between value and inline comments
-          | str trim -c '"'             # unquote double-quoted values
-          | str trim -c "'"             # unquote single-quoted values
-          | str replace -a "\\n" "\n"   # replace `\n` with newline char
-          | str replace -a "\\r" "\r"   # replace `\r` with carriage return
-          | str replace -a "\\t" "\t"   # replace `\t` with tab
-      }
-    | transpose -r -d
+# May be used like this: open .env | load-env
+# Works with quoted and unquoted .env files
+export def "from env" []: string -> record {
+  let input = $in
+
+  # Process escape sequences in double-quoted values using str replace chain
+  # Use NUL char as placeholder to avoid replacement conflicts
+  let process_escapes = {|content: string|
+    $content
+      | str replace -a '\\' (char nul)   # Placeholder for \\ to avoid conflicts
+      | str replace -a '\n' (char nl)
+      | str replace -a '\r' (char cr)
+      | str replace -a '\t' (char tab)
+      | str replace -a '\"' '"'
+      | str replace -a (char nul) '\'    # Restore \\ to single \
+  }
+
+  # Parse double-quoted value with escape sequence support
+  let parse_double_quoted = {|val: string|
+    let matched = ($val | parse -r '^"(?P<content>(?:[^"\\]|\\.)*)"')
+    if ($matched | is-empty) { $val | str trim -c '"' } else { do $process_escapes $matched.0.content }
+  }
+
+  # Parse single-quoted value (no escape processing)
+  let parse_single_quoted = {|val: string|
+    let matched = ($val | parse -r "^'(?P<content>[^']*)'")
+    if ($matched | is-empty) { $val | str trim -c "'" } else { $matched.0.content }
+  }
+
+  # Parse unquoted value: handle escaped hash (\#) and strip inline comments
+  let parse_unquoted = {|val: string|
+    $val
+      | str replace -a '\#' (char nul)    # Placeholder for \#
+      | split row '#'                     # Split by comment delimiter
+      | first                             # Take content before first #
+      | str replace -a (char nul) '#'     # Restore \# to #
+      | str trim
+  }
+
+  # Parse value based on its format
+  let parse_value = {|val: string|
+    match $val {
+      $v if ($v | str starts-with '"') => { do $parse_double_quoted $v }
+      $v if ($v | str starts-with "'") => { do $parse_single_quoted $v }
+      _ => { do $parse_unquoted $val }
+    }
+  }
+
+  let parsed = $input | lines
+    | str trim
+    | compact -e
+    | where {|line| not ($line | str starts-with '#') }
+    | parse "{key}={value}"
+    | update key {|row| $row.key | str trim | str replace -r '^export\s+' '' }
+    | update value {|row| do $parse_value ($row.value | str trim) }
+
+  if ($parsed | is-empty) { {} } else { $parsed | transpose -r -d -l }
 }
 
 # Compact the record by removing empty columns
 export def compact-record []: record -> record {
   let record = $in
-  let empties = $record | columns | filter {|it| $record | get $it | is-empty }
+  let empties = $record | columns | where {|it| $record | get $it | is-empty }
   $record | reject ...$empties
 }
 
@@ -152,7 +193,8 @@ export def git-check [
 # Check if current directory is a git repo
 export def is-repo [] {
   let checkRepo = try {
-      do -i { git rev-parse --is-inside-work-tree } | complete
+      # Put `complete` inside `do` block to avoid pipefail error in Nushell 0.110+
+      do { git rev-parse --is-inside-work-tree | complete }
     } catch {
       ({ stdout: 'false' })
     }
@@ -164,8 +206,8 @@ export def has-ref [
   ref: string   # The git ref to check
 ] {
   if not (is-repo) { return false }
-  # Brackets were required here, or error will occur
-  let parse = (do -i { git rev-parse --verify -q $ref } | complete)
+  # Put `complete` inside `do` block to avoid pipefail error in Nushell 0.110+
+  let parse = (do { git rev-parse --verify -q $ref | complete })
   if ($parse.stdout | is-empty) { false } else { true }
 }
 

nu/config.nu

@@ -23,21 +23,20 @@ def file-exists [file: string] {
 # Check if the prompt keys exist in the config.yml file
 def check-prompts [options: record] {
   check-prompt $options user
-  # System prompt is optional, so we don't exit if it's missing
-  # check-prompt $options system
+  check-prompt $options system
 }
 
 # Check if the specified type of prompt key exists in the config.yml file
 def check-prompt [options: record, type: string] {
-  let prompt_key = $options.settings | get -i $'($type)-prompt' | default ''
+  let prompt_key = $options.settings | get -o $'($type)-prompt' | default ''
   if ($prompt_key | is-empty) {
     print $'(ansi r)The ($type) prompt key is missing in `settings.($type)-prompt` config.yml file.(ansi reset)'
     exit $ECODE.INVALID_PARAMETER
   }
-  let prompt = $options.prompts | get -i $type
+  let prompt = $options.prompts | get -o $type
     | default []
     | where name == $prompt_key
-    | get -i 0.prompt
+    | get -o 0.prompt
   if ($prompt | is-empty) {
     print $'The ($type) prompt (ansi r)($prompt_key)(ansi reset) is missing in `prompts.($type)` of config.yml file.'
     exit $ECODE.INVALID_PARAMETER
@@ -60,11 +59,11 @@ def check-providers [options: record] {
     exit $ECODE.INVALID_PARAMETER
   }
   # Each provider should have name, token and models field
-  $options.providers | each {|it|
-    let empties = [name token models] | filter { |field| $it | get -i $field | is-empty }
+  $options.providers | each {|p|
+    let empties = [name token models] | where { |field| $p | get -o $field | is-empty }
     if ($empties | is-not-empty) {
       print $'Field (ansi r)`($empties | str join ,)`(ansi reset) should not be empty for provider:'
-      $it | table -e -t psql | print
+      $p | table -e -t psql | print
       exit $ECODE.INVALID_PARAMETER
     }
   }
@@ -92,7 +91,8 @@ def check-models [options: record] {
 }
 
 # Check if the config.yml file exists and if it's valid
-export def config-check [--config: string = $SETTING_FILE] {
+export def config-check [--config: string] {
+  let config = $config | default $SETTING_FILE
   file-exists $config
   let options = open $config
   check-prompts $options
@@ -106,7 +106,7 @@ def get-model-envs [settings: record, model?: string = ''] {
   let provider = $settings.providers
     | default []
     | where name == $name
-    | get -i 0
+    | get -o 0
     | default {}
   let model_name = $provider.models
     | default []
@@ -115,7 +115,7 @@ def get-model-envs [settings: record, model?: string = ''] {
       } else {
         $it.name == $model or $it.alias? == $model }
       }
-    | get -i 0.name
+    | get -o 0.name
     | default $model
 
   { CHAT_TOKEN: $provider.token?, BASE_URL: $provider.base-url?, CHAT_URL: $provider.chat-url?, CHAT_MODEL: $model_name }
@@ -133,12 +133,12 @@ export def --env config-load [
   let user_prompt = $all_settings.prompts?.user?
     | default []
     | where name == ($settings.user-prompt? | default '')
-    | get -i 0.prompt
+    | get -o 0.prompt
 
   let system_prompt = $all_settings.prompts?.system?
     | default []
     | where name == ($settings.system-prompt? | default '')
-    | get -i 0.prompt
+    | get -o 0.prompt
 
   let model_envs = get-model-envs $all_settings $model
 

nu/diff.nu

@@ -80,6 +80,13 @@ def get-pr-diff [
     exit $ECODE.SUCCESS
   }
 
+  let commit_msg = http get -H $BASE_HEADER $'($GITHUB_API_BASE)/repos/($repo)/pulls/($pr_number)/commits'
+                   | last | get commit.message
+  if ($IGNORE_REVIEW_KEYWORDS | any {|it| $commit_msg =~ $it }) {
+    print $'(ansi r)The latest PR commit message contains keywords to skip the review, bye...(ansi reset)'
+    exit $ECODE.SUCCESS
+  }
+
   # Get the diff content of the PR
   http get -H $DIFF_HEADER $'($GITHUB_API_BASE)/repos/($repo)/pulls/($pr_number)' | str trim
 }

nu/release.nu

@@ -46,9 +46,9 @@ export def 'make-release' [
 export def has-ref [
   ref: string   # The git ref to check
 ] {
-  let checkRepo = (do -i { git rev-parse --is-inside-work-tree } | complete)
+  # Put `complete` inside `do` block to avoid pipefail error in Nushell 0.110+
+  let checkRepo = (do { git rev-parse --is-inside-work-tree | complete })
   if not ($checkRepo.stdout =~ 'true') { return false }
-  # Brackets were required here, or error will occur
-  let parse = (do -i { git rev-parse --verify -q $ref } | complete)
+  let parse = (do { git rev-parse --verify -q $ref | complete })
   if ($parse.stdout | is-empty) { false } else { true }
 }

nu/review.nu

@@ -43,10 +43,11 @@ const IGNORED_MESSAGES = {
 const HTTP_HEADERS = [User-Agent curl/8.9]
 
 const DEFAULT_OPTIONS = {
-  MODEL: 'deepseek-chat',
+  MODEL: 'deepseek-v4-flash',
   TEMPERATURE: 0.3,
   BASE_URL: 'https://api.deepseek.com',
-  USER_PROMPT: 'You are a professional code review assistant responsible for analyzing code changes. Identify potential issues such as code style violations, logical errors, security vulnerabilities, and provide improvement suggestions. Clearly list the problems and recommendations in a concise manner. Please review the following code changes:',
+  USER_PROMPT: 'Please review the following code changes:',
+  SYS_PROMPT: 'You are a professional code review assistant responsible for analyzing code changes in GitHub Pull Requests. Identify potential issues such as code style violations, logical errors, security vulnerabilities, and provide improvement suggestions. Clearly list the problems and recommendations in a concise manner.',
 }
 
 # Use DeepSeek AI to review code changes locally or in GitHub Actions
@@ -61,11 +62,11 @@ export def --env deepseek-review [
   --diff-from(-f): string,  # Git diff starting commit SHA
   --patch-cmd(-c): string,  # The `git show` or `git diff` command to get the diff content, for local CR only
   --max-length(-l): int,    # Maximum length of the content for review, 0 means no limit.
-  --model(-m): string,      # Model name, or read from CHAT_MODEL env var, `deepseek-chat` by default
+  --model(-m): string,      # Model name, or read from CHAT_MODEL env var, `deepseek-v4-flash` by default
   --base-url(-b): string,   # DeepSeek API base URL, fallback to BASE_URL env var
   --chat-url(-U): string,   # DeepSeek Model chat full API URL, e.g. http://localhost:11535/api/chat
-  --sys-prompt(-s): string  # Optional, System prompt message, fallback to SYSTEM_PROMPT env var
-  --user-prompt(-u): string # Default to $DEFAULT_OPTIONS.USER_PROMPT
+  --sys-prompt(-s): string  # Default to $DEFAULT_OPTIONS.SYS_PROMPT,
+  --user-prompt(-u): string # Default to $DEFAULT_OPTIONS.USER_PROMPT,
   --include(-i): string,    # Comma separated file patterns to include in the code review
   --exclude(-x): string,    # Comma separated file patterns to exclude in the code review
   --temperature(-T): float, # Temperature for the model, between `0` and `2`, default value `0.3`
@@ -113,7 +114,7 @@ export def --env deepseek-review [
   print $hint; print -n (char nl)
   if ($pr_number | is-empty) {
     print 'Current Settings:'; hr-line
-    $setting | compact-record | reject -i repo | print; print -n (char nl)
+    $setting | compact-record | reject -o repo | print; print -n (char nl)
   }
 
   let content = (
@@ -125,15 +126,17 @@ export def --env deepseek-review [
     exit $ECODE.SUCCESS
   }
   print $'Review content length: (ansi g)($length)(ansi reset), current max length: (ansi g)($max_length)(ansi reset)'
-  let sys_prompt = $sys_prompt | default $env.SYSTEM_PROMPT?
+  let sys_prompt = $sys_prompt | default $env.SYSTEM_PROMPT? | default $DEFAULT_OPTIONS.SYS_PROMPT
   let user_prompt = $user_prompt | default $env.USER_PROMPT? | default $DEFAULT_OPTIONS.USER_PROMPT
-  mut messages = if ($sys_prompt | is-empty) { [] } else { [{ role: 'system', content: $sys_prompt }] }
-  $messages = $messages | append { role: 'user', content: $"($user_prompt)\n($content)" }
   let payload = {
     model: $model,
     stream: $stream,
-    messages: $messages
     temperature: $temperature,
+    messages: [
+      { role: 'system', content: $sys_prompt },
+      { role: 'user', content: $"($user_prompt):\n($content)" }
+    ],
+    thinking: { type: 'disabled' }
   }
   if $debug { print $'(char nl)Code Changes:'; hr-line; print $content }
   print $'(char nl)Waiting for response from (ansi g)($url)(ansi reset) ...'
@@ -149,9 +152,9 @@ export def --env deepseek-review [
     print $'✖️ Code review failed！Error: '; hr-line; print $response
     exit $ECODE.SERVER_ERROR
   }
-  let message = $response | get -i choices.0.message
+  let message = $response | get -o choices.0.message
   let reason = $message | coalesce-reasoning
-  let review = $message.content? | default ($response | get -i message.content)
+  let review = $message.content? | default ($response | get -o message.content)
   let result = ['<details>' '<summary> Reasoning Details</summary>' $reason "</details>\n" $review] | str join "\n"
   if ($review | is-empty) {
     print $'✖️ Code review failed！No review result returned from ($base_url) ...'
@@ -190,7 +193,7 @@ def write-review-to-file [
     '# DeepSeek Code Review Result', ''
     $"Generated at: (date now | format date '%Y/%m/%d %H:%M:%S')", ''
     '## Code Review Settings', ''
-    ($setting | compact-record | reject -i repo | transpose key val | to md --pretty)
+    ($setting | compact-record | reject -o repo | transpose key val | to md --pretty)
     '', '## Review Detail', '', $result, '', ...$token_usage
   ]
   try {
@@ -262,10 +265,10 @@ def streaming-output [
     | try { lines } catch { print $'(ansi r)Error Happened ...(ansi reset)'; exit $ECODE.SERVER_ERROR }
     | each {|line|
         if ($line | is-empty) { return }
-        if ($IGNORED_MESSAGES | get -i $line | default false) { return }
+        if ($IGNORED_MESSAGES | get -o $line | default false) { return }
         let $last = $line | parse-line
         if $debug { $last | to json | kv set last-reply }
-        $last | get -i choices.0.delta | default ($last | get -i message) | if ($in | is-not-empty) {
+        $last | get -o choices.0.delta | default ($last | get -o message) | if ($in | is-not-empty) {
           let delta = $in
           if ($delta | coalesce-reasoning | is-not-empty) { kv set reasoning ((kv get reasoning) + 1) }
           if (kv get reasoning) == 1 { print $'(char nl)Reasoning Details:'; hr-line }
@@ -277,7 +280,7 @@ def streaming-output [
 
   if $debug and (kv get last-reply | is-not-empty) {
     print $'(char nl)(char nl)Model & Token Usage:'; hr-line
-    kv get last-reply | from json | select -i model usage | table -e | print
+    kv get last-reply | from json | select -o model usage | table -e | print
   }
 }
 

nu/util.nu

@@ -132,7 +132,7 @@ def install-gawk-for-actions [] {
     Invoke-Expression (New-Object System.Net.WebClient).DownloadString("https://get.scoop.sh")
     $env:Path = "$env:USERPROFILE\scoop\shims;" + $env:Path; scoop update; scoop install gawk
     '# | complete | get stdout | print
-  let awk_bin = $'($nu.home-path)/scoop/shims/gawk.exe'
+  let awk_bin = $'($nu.home-dir)/scoop/shims/gawk.exe'
   let version = get-awk-ver $awk_bin
   { awk_bin: $awk_bin, version: $version }
 }