Merge PR #189: Add workflow duplicate issue detection

# Conflicts:
#	README.md
#	README.zh-CN.md
#	shortcuts/workflow/workflow.go
This commit is contained in:
wbtiger 2026-07-14 22:49:58 +08:00
commit 4ab4682887
6 changed files with 621 additions and 194 deletions

196
README.md
View File

@ -5,7 +5,7 @@
[![Go Version](https://img.shields.io/badge/Go-1.26%2B-blue.svg)](https://golang.org)
[![npm version](https://img.shields.io/npm/v/@gitlink-ai/cli.svg)](https://www.npmjs.com/package/@gitlink-ai/cli)
The official [GitLink](https://www.gitlink.org.cn) CLI tool — built for humans and AI Agents. Supports **macOS, Linux, and Windows**. Covers repository management, issue tracking, pull requests, webhooks, member collaboration, CI/CD, and AI-powered workflows, with 40+ commands and AI Agent [Skills](./skills/README.md).
The official [GitLink](https://www.gitlink.org.cn) CLI tool — built for humans and AI Agents. Supports **macOS, Linux, and Windows**. Covers repository management, issue tracking, pull requests, webhooks, member collaboration, CI/CD, and AI-powered workflows, with 40+ commands and AI Agent [Skills](./skills/).
**[中文文档](./README.zh-CN.md)**
@ -78,19 +78,11 @@ The official [GitLink](https://www.gitlink.org.cn) CLI tool — built for humans
<a href="https://www.gitlink.org.cn/jiangtx" title="jiangtx"><img src="https://www.gitlink.org.cn/system/lets/letter_avatars/2/J/67_157_94/120.png" width="40" height="40" alt="jiangtx" style="border-radius: 50%;"></a>
<br><sub><a href="https://www.gitlink.org.cn/jiangtx">jiangtx</a></sub>
</div>
<div align="center">
<a href="https://www.gitlink.org.cn/luwanzhou" title="luwanzhou"><img src="https://www.gitlink.org.cn/system/lets/letter_avatars/2/L/165_135_246/120.png" width="40" height="40" alt="luwanzhou" style="border-radius: 50%;"></a>
<br><sub><a href="https://www.gitlink.org.cn/luwanzhou">luwanzhou</a></sub>
</div>
<div align="center">
<a href="https://www.gitlink.org.cn/whale_hihihi" title="whale_hihihi"><img src="https://www.gitlink.org.cn/images/avatars/User/137722?t=1778575729" width="40" height="40" alt="whale_hihihi" style="border-radius: 50%;"></a>
<br><sub><a href="https://www.gitlink.org.cn/whale_hihihi">whale_hihihi</a></sub>
</div>
</div>
## Why gitlink-cli?
- **Agent-Native Design** — Structured [Skills](./skills/README.md) out of the box, compatible with Claude Code, OpenClaw, and other AI platforms — Agents can operate GitLink with zero extra setup
- **Agent-Native Design** — Structured [Skills](./skills/) out of the box, compatible with Claude Code, OpenClaw, and other AI platforms — Agents can operate GitLink with zero extra setup
- **Wide Coverage** — Repository, Issue, PR, Webhook, Member, Branch, Release, CI, Pipeline, Org, Search, and User workflows are covered by high-level commands
- **AI-Friendly & Optimized** — Every command is tested with real Agents, featuring concise parameters, smart defaults, and structured output
- **Cross-Platform** — Runs on macOS, Linux, and Windows (x64/arm64), install via `npm install -g @gitlink-ai/cli` in one command, binary auto-downloaded
@ -103,8 +95,8 @@ The official [GitLink](https://www.gitlink.org.cn) CLI tool — built for humans
| Category | Capabilities |
|----------|-------------|
| 📦 Repo | List, create, fork, delete repositories, read and search files, batch commit file changes, view repo info, insights, and interactions |
| 🐛 Issue | Create, update, close, batch close/update/delete, comment on issues |
| 📦 Repo | List, create, fork, delete repositories, view repo info, insights, and interactions |
| 🐛 Issue | Create, update, close, batch close, comment on issues |
| 🔖 Label | Create, list, update, delete issue labels |
| 🔀 PR | Create, merge, review pull requests, view changed files |
| 👥 Member | List, add, remove repository members, change roles, create and accept invite links |
@ -113,14 +105,9 @@ The official [GitLink](https://www.gitlink.org.cn) CLI tool — built for humans
| 🏢 Org | Manage organizations, members, teams |
| 🔧 CI | View builds, logs, CI/CD operations |
| ⚙️ Pipeline | Run, inspect, enable, disable, delete pipeline workflows and logs |
| 🔔 Message Settings | Inspect and update personal message delivery preferences |
| 🔔 Webhook | Manage repo webhooks and test deliveries |
| 📖 Wiki | List, view, create, update, and delete wiki pages |
| 🔔 Notification | List, read, and delete user messages |
| 🔍 Search | Search repositories, users |
| 📊 Dataset | Query research datasets by project |
| 👤 User | View user profiles and info |
| 📊 Profile | User ability, role, major, activity, and contribution statistics |
| 📋 PM | Sprint management, kanban boards, weekly reports |
| 🤖 Workflow | AI-powered issue triage, PR review, release notes |
@ -225,26 +212,10 @@ gitlink-cli repo +info --owner Gitlink --repo forgeplus
# Read repository README
gitlink-cli repo +readme --owner Gitlink --repo forgeplus --ref master
# Read one repository file with metadata
gitlink-cli repo +file --owner Gitlink --repo forgeplus --path README.md --ref main
# List repository files at root or a directory
gitlink-cli repo +tree --owner Gitlink --repo forgeplus --ref master
gitlink-cli repo +tree --owner Gitlink --repo forgeplus --path src --ref main
# Search repository files by name
gitlink-cli repo +files --owner Gitlink --repo forgeplus --search README --ref main
# Commit one file change
gitlink-cli repo +commit-files --owner Gitlink --repo forgeplus \
--branch main --message "docs: update README" \
--path README.md --content "# Project"
# Commit several file changes from a JSON operations file
gitlink-cli repo +commit-files --owner Gitlink --repo forgeplus \
--branch main --new-branch docs/batch-update \
--message "docs: batch update" --ops changes.json --dry-run
# Show language breakdown
gitlink-cli repo +languages --owner Gitlink --repo forgeplus
@ -292,47 +263,6 @@ gitlink-cli webhook +test --owner Gitlink --repo forgeplus --id 68
gitlink-cli webhook +tasks --owner Gitlink --repo forgeplus --id 68
```
### Wiki Management
```bash
# List wiki pages (table of contents)
gitlink-cli wiki +list --owner Gitlink --repo forgeplus --project-id 12345
# View a wiki page by page name
gitlink-cli wiki +view --owner Gitlink --repo forgeplus --project-id 12345 -n home
# Create a wiki page
gitlink-cli wiki +create --owner Gitlink --repo forgeplus --project-id 12345 \
-n getting-started -t "Getting Started" -c "# Getting Started Guide"
# Update a wiki page title and/or content
gitlink-cli wiki +update --owner Gitlink --repo forgeplus --project-id 12345 -n home -t "New Title"
gitlink-cli wiki +update --owner Gitlink --repo forgeplus --project-id 12345 -n home -c "# Updated content"
gitlink-cli wiki +update --owner Gitlink --repo forgeplus --project-id 12345 -n home -t "New Title" -c "New content"
# Delete a wiki page
gitlink-cli wiki +delete --owner Gitlink --repo forgeplus --project-id 12345 -n old-page
```
### Notifications
```bash
# List current user's unread notifications
gitlink-cli notification +list --type notification --status unread
# List @me messages for an explicit user
gitlink-cli notification +list --user Mengz --type atme
# Mark messages as read
gitlink-cli notification +read --type atme --ids 101,102
# Mark all unread notifications as read
gitlink-cli notification +read --type notification --ids -1
# Delete messages
gitlink-cli notification +delete --type notification --ids 101,102
```
### Member Management
```bash
@ -382,14 +312,6 @@ gitlink-cli issue +batch-close --owner Gitlink --repo forgeplus --numbers 123,12
# Batch close issues from a CSV file
gitlink-cli issue +batch-close --owner Gitlink --repo forgeplus --from issues.csv
# Preview batch metadata update by API issue IDs
# Note: --ids uses API issue IDs, not web URL issue numbers.
gitlink-cli issue +batch-update --owner Gitlink --repo forgeplus --ids 101,102 --status-id 3 --priority-id 2 --dry-run
# Destructive batch delete requires both dry-run first and --yes for real execution
gitlink-cli issue +batch-delete --owner Gitlink --repo forgeplus --ids 101,102 --dry-run
gitlink-cli issue +batch-delete --owner Gitlink --repo forgeplus --ids 101,102 --yes
# Add a comment
gitlink-cli issue +comment --owner Gitlink --repo forgeplus -i 123 -b "Fixed"
@ -476,10 +398,6 @@ gitlink-cli pr +review --owner Gitlink --repo forgeplus -i 42 --status approved
```bash
# List branches
gitlink-cli branch +list --owner Gitlink --repo forgeplus
gitlink-cli branch +list --owner Gitlink --repo forgeplus --keyword fix --state deleted
# List all branches without pagination
gitlink-cli branch +all --owner Gitlink --repo forgeplus
# Create a branch
gitlink-cli branch +create --name feature/new-feature
@ -492,10 +410,6 @@ gitlink-cli branch +protect --name main
# Remove branch protection
gitlink-cli branch +unprotect --name main
# Set default branch or restore a deleted branch (preview first)
gitlink-cli branch +set-default --owner Gitlink --repo forgeplus --name main --dry-run
gitlink-cli branch +restore --owner Gitlink --repo forgeplus --id 7 --name feature/old --dry-run
```
### Release Management
@ -553,16 +467,6 @@ gitlink-cli pipeline +disable --owner Gitlink --repo forgeplus --id 7 --workflow
gitlink-cli pipeline +delete --owner Gitlink --repo forgeplus --id 7 --dry-run
```
### Ignore File Templates
```bash
# List all available .gitignore templates
gitlink-cli ignore +list
# Filter templates by name
gitlink-cli ignore +list --name Go
```
### Search
```bash
@ -573,29 +477,6 @@ gitlink-cli search +repos -k "machine learning"
gitlink-cli search +users -k "zhangsan"
```
### User Profile
`profile` surfaces GitLink's native user statistics (ability, role, major, activity,
contribution). When `--user` is omitted it defaults to the authenticated user.
```bash
# Development ability scores + language breakdown
gitlink-cli profile +ability --user zhangsan
# Role positioning / major (discipline) categories
gitlink-cli profile +role --user zhangsan
gitlink-cli profile +major --user zhangsan
# Ability within a time window (Unix timestamps)
gitlink-cli profile +ability --user zhangsan --start-time 1704067200 --end-time 1735689600
# Recent activity (issues / PRs / commits per day) for the current user
gitlink-cli profile +activity
# Contribution heatmap for a given year
gitlink-cli profile +contribution --user zhangsan --year 2025
```
### Workflow Agent Commands
`workflow` provides rule-based repository analysis for maintainers and AI Agents. It currently supports:
@ -604,11 +485,11 @@ gitlink-cli profile +contribution --user zhangsan --year 2025
- `workflow +health`
- `workflow +pr-summary`
- `workflow +repo-report`
- `workflow +release-readiness`
- `workflow +issue-dedupe`
`workflow +pr-summary` defaults to `table` when `--format` is omitted.
`workflow +repo-report` defaults to `markdown` when `--format` is omitted.
`workflow +release-readiness` defaults to `table` when `--format` is omitted.
`workflow +issue-dedupe` defaults to `table` when `--format` is omitted.
Examples:
@ -682,17 +563,11 @@ gitlink-cli workflow +repo-report --owner Gitlink --repo gitlink-cli --format ma
# Repository workflow report from a local JSON file
gitlink-cli workflow +repo-report --from shortcuts/workflow/testdata/repo_report.json --format json
# Evaluate whether a release is ready to publish
gitlink-cli workflow +release-readiness \
--version v1.2.0 \
--changes "Add workflow gate,Fix release upload" \
--tests "go test ./...=passed,go build ./...=passed" \
--artifacts "windows zip=passed,linux tar=passed" \
--rollback-plan "Revert tag and restore previous release assets" \
--format table
# Find likely duplicate issues by read-only GitLink fetch
gitlink-cli workflow +issue-dedupe --owner Gitlink --repo gitlink-cli --limit 50 --threshold 55 --format table
# Evaluate release readiness from a JSON gate file
gitlink-cli workflow +release-readiness --from release_readiness.json --format markdown
# Find duplicate candidates from a local issue JSON file
gitlink-cli workflow +issue-dedupe --from issues.json --threshold 60 --format markdown
```
Output formats:
@ -708,33 +583,7 @@ Safety:
- They do not depend on LLM APIs.
- `workflow +pr-summary` does not comment, approve, reject, or merge pull requests.
- `workflow +repo-report` aggregates health, issue triage, and PR review summary signals without remote writes.
- `workflow +release-readiness` evaluates local gate signals only; it does not create tags, releases, or assets.
### Dataset
`dataset` manages and queries GitLink research datasets (title, description,
paper content, license, owning project).
```bash
# List datasets for one or more projects (by numeric project ID)
gitlink-cli dataset +list --ids 5988
# View a repository's dataset and attachments
gitlink-cli dataset +view --owner Gitlink --repo forgeplus
# Create / update a repository's dataset (preview first with --dry-run)
gitlink-cli dataset +create --owner me --repo proj -t "My dataset" -d "..." --license-id 359 --dry-run
gitlink-cli dataset +update --owner me --repo proj -t "My dataset" -d "updated"
# Delete a dataset attachment (destructive: preview, then confirm with --yes)
gitlink-cli dataset +delete-attachment --owner me --repo proj --uuid <uuid> --dry-run
gitlink-cli dataset +delete-attachment --owner me --repo proj --uuid <uuid> --yes
```
> Note: `dataset +list` (platform dataset query) is verified on production
> gitlink.org.cn. The per-repo `+view`/`+create`/`+update` routes follow the
> published OpenAPI contract but are not yet deployed on production (they return
> 404 there); they will work once the platform enables them.
- `workflow +issue-dedupe` only reports duplicate candidates; it does not close, comment on, or edit issues.
### Raw API
@ -755,20 +604,6 @@ Get-Content issue.json | gitlink-cli api POST /Gitlink/forgeplus/issues --body-s
# With query parameters
gitlink-cli api GET /Gitlink/forgeplus/commits --query 'page=1&limit=5'
# Render owner/repo from git context or explicit variables
gitlink-cli api POST /:owner/:repo/issues --body '{"subject":"Bug","description":"..."}'
# Reuse template variables in a single request
gitlink-cli api POST /v1/{{owner}}/{{repo}}/issues/{{number}}/journals \
--body '{"notes":"handled by {{actor}}"}' \
--var owner=Gitlink --var repo=gitlink-cli --var number=42 --var actor=bot
# Preview a rendered single request without sending it
gitlink-cli api POST /v1/{{owner}}/{{repo}}/issues \
--body-file issue.json \
--var owner=Gitlink --var repo=gitlink-cli \
--dry-run
```
## Global Parameters
@ -806,20 +641,19 @@ git push gitlink
The `skills/` directory contains Agent Skill files for AI-automated GitLink operations.
See [skills/README.md](./skills/README.md) for details.
See [skills/README.md](skills/README.md) for details.
| Skill | Description |
|-------|-------------|
| `gitlink-shared` | Authentication, global parameters, safety rules, API notes |
| `gitlink-repo` | Repository operations (create, view, delete, fork, insights, etc.) |
| `gitlink-issue` | Issue operations (create, update, close, batch update/delete, comment, etc.) |
| `gitlink-issue` | Issue operations (create, update, close, comment, etc.) |
| `gitlink-pr` | Pull request operations (create, merge, review, etc.) |
| `gitlink-member` | Repository member and invite link management |
| `gitlink-branch` | Branch management (create, delete, list, protect, unprotect) |
| `gitlink-release` | Release management (create, edit, update, view, delete, etc.) |
| `gitlink-ci` | CI/CD operations (builds, logs, etc.) |
| `gitlink-pipeline` | Pipeline workflow operations (runs, logs, enable, disable, delete, etc.) |
| `gitlink-notification` | User messages (list, mark read, delete) |
| `gitlink-search` | Search (repositories, users, etc.) |
| `gitlink-org` | Organization management (members, teams, etc.) |
| `gitlink-user` | User management (profile info, etc.) |
@ -877,7 +711,7 @@ gitlink-cli/
## Documentation
- [Skills Guide](./skills/README.md) — AI Agent Skills detailed documentation
- [Skills Guide](skills/README.md) — AI Agent Skills detailed documentation
- [Design Document](doc/design.md) — Architecture design and development plan
## FAQ
@ -942,7 +776,7 @@ gitlink-cli uses Windows Credential Manager for secure token storage. If Credent
### Q: Where can I find the full API reference?
See [skills/gitlink-shared/references/api-reference.md](./skills/gitlink-shared/references/api-reference.md).
See [skills/gitlink-shared/REFERENCE.md](skills/gitlink-shared/REFERENCE.md).
## License

View File

@ -457,7 +457,7 @@ gitlink-cli search +users -k "zhangsan"
### 工作流命令
`workflow` 提供面向维护者和 AI Agent 的只读分析能力,可用于 Issue 分流、仓库健康度评估、PR 审查摘要、仓库工作流报告和发布就绪度检查
`workflow` 提供面向维护者和 AI Agent 的只读分析能力,可用于 Issue 分流、仓库健康度评估、PR 审查摘要、仓库工作流报告和重复 Issue 候选检测
```bash
# 生成单个 PR 的审查摘要
@ -466,20 +466,14 @@ gitlink-cli workflow +pr-summary --owner Gitlink --repo gitlink-cli --number 1 -
# 生成仓库工作流报告
gitlink-cli workflow +repo-report --owner Gitlink --repo gitlink-cli --format markdown
# 根据本地门禁信号判断发布是否可以推进
gitlink-cli workflow +release-readiness \
--version v1.2.0 \
--changes "新增工作流门禁,修复 Release 上传" \
--tests "go test ./...=passed,go build ./...=passed" \
--artifacts "windows zip=passed,linux tar=passed" \
--rollback-plan "回滚 tag 并恢复上一版制品" \
--format table
# 从远端只读拉取 Issue 并检测重复候选
gitlink-cli workflow +issue-dedupe --owner Gitlink --repo gitlink-cli --limit 50 --threshold 55 --format table
# 从 JSON 门禁文件生成发布就绪度报告
gitlink-cli workflow +release-readiness --from release_readiness.json --format markdown
# 从本地 JSON 文件检测重复候选
gitlink-cli workflow +issue-dedupe --from issues.json --threshold 60 --format markdown
```
`workflow +release-readiness` 默认使用 `table` 输出,会综合版本号、变更摘要、已知阻塞项、测试结果、发布制品、破坏性变更回滚方案和依赖风险,给出 `pass/review/block` 门禁结论;命令只分析本地输入,不会创建 tag、Release 或附件
`workflow +issue-dedupe` 默认使用 `table` 输出,会根据标题、正文和标签的共享关键词计算相似度,帮助维护者优先确认高置信重复候选;命令只输出候选对,不会关闭、评论或修改 Issue。
### Raw API

View File

@ -0,0 +1,7 @@
# 新增重复 Issue 候选检测工作流
`gitlink-cli workflow +issue-dedupe` 新增了只读的重复 Issue 候选检测能力。命令可以读取本地 Issue JSON也可以通过 GitLink API 只读拉取指定仓库的 Issue 列表,再基于标题、正文和标签分词计算相似度,输出候选重复对、置信度、共享关键词和处理建议。
该功能面向 Issue 较多的仓库维护场景:维护者可以先处理高置信重复候选,把后创建的 Issue 链接到原始问题,减少重复排查成本;也可以通过 `--threshold` 调整相似度阈值,通过 `--max-pairs` 控制输出规模。输出支持 `table`、`markdown` 和 `json`适合终端查看、Issue 评论草稿或自动化脚本消费。
命令保持安全边界,不会自动关闭、评论或修改 Issue。测试覆盖了本地 JSON 输入、状态和数量过滤、远端只读查询参数、相似度排序、中文 markdown 渲染、table 输出和 shortcut 注册。

View File

@ -0,0 +1,444 @@
package workflow
import (
"bytes"
"fmt"
"math"
"os"
"sort"
"strings"
"text/tabwriter"
"unicode"
"github.com/gitlink-org/gitlink-cli/cmd/cmdutil"
"github.com/gitlink-org/gitlink-cli/shortcuts/common"
)
type IssueDedupeInput struct {
Repository string `json:"repository"`
Issues []IssueInput `json:"issues"`
Source string `json:"source"`
}
type IssueDedupeResult struct {
Repository string `json:"repository"`
TotalIssues int `json:"total_issues"`
CandidatePairs int `json:"candidate_pairs"`
HighConfidence int `json:"high_confidence"`
MediumConfidence int `json:"medium_confidence"`
LowConfidence int `json:"low_confidence"`
Threshold int `json:"threshold"`
Pairs []IssueDedupePair `json:"pairs"`
Recommendations []string `json:"recommendations"`
Source string `json:"source"`
}
type IssueDedupePair struct {
Score int `json:"score"`
Confidence string `json:"confidence"`
Primary IssueRef `json:"primary"`
Duplicate IssueRef `json:"duplicate"`
SharedTerms []string `json:"shared_terms"`
Reason string `json:"reason"`
}
func newIssueDedupeShortcut() *common.Shortcut {
return &common.Shortcut{
Name: "issue-dedupe",
Description: "Find likely duplicate issues with read-only local similarity rules",
Flags: []common.Flag{
{Name: "from", Usage: "Read issues from a JSON file. Supports an array or an object with an issues field"},
{Name: "state", Short: "s", Usage: "Remote issue state to fetch", Default: "open"},
{Name: "page", Short: "p", Usage: "Remote issue page", Default: "1"},
{Name: "limit", Short: "l", Usage: "Maximum issues to analyze", Default: "50"},
{Name: "threshold", Usage: "Minimum duplicate score from 0 to 100", Default: "55"},
{Name: "max-pairs", Usage: "Maximum duplicate pairs to output", Default: "20"},
{Name: "lang", Usage: "Output language: en or zh-CN", Default: langEN},
},
Run: runIssueDedupe,
}
}
func runIssueDedupe(ctx *common.RuntimeContext) error {
lang := normalizeLang(ctx.Arg("lang"))
input, threshold, maxPairs, err := collectIssueDedupeInput(ctx)
if err != nil {
return err
}
result := AnalyzeIssueDedupe(input, threshold, maxPairs, lang)
format := ctx.Format
if strings.TrimSpace(cmdutil.Format) == "" {
format = "table"
}
rendered, err := RenderIssueDedupe(result, format, lang)
if err != nil {
return err
}
_, err = fmt.Fprint(os.Stdout, rendered)
return err
}
func collectIssueDedupeInput(ctx *common.RuntimeContext) (IssueDedupeInput, int, int, error) {
threshold, err := parseIntArg(ctx.Arg("threshold"), 55, "threshold")
if err != nil {
return IssueDedupeInput{}, 0, 0, err
}
if threshold > 100 {
return IssueDedupeInput{}, 0, 0, fmt.Errorf("invalid --threshold %q: must be between 0 and 100", ctx.Arg("threshold"))
}
maxPairs, err := parseIntArg(ctx.Arg("max-pairs"), 20, "max-pairs")
if err != nil {
return IssueDedupeInput{}, 0, 0, err
}
if path := strings.TrimSpace(ctx.Arg("from")); path != "" {
issues, err := readIssueInputs(path)
if err != nil {
return IssueDedupeInput{}, 0, 0, err
}
limit, err := parseIntArg(ctx.Arg("limit"), 50, "limit")
if err != nil {
return IssueDedupeInput{}, 0, 0, err
}
return IssueDedupeInput{
Repository: repositoryFromContext(ctx, ""),
Issues: filterIssueInputs(issues, ctx.Arg("state"), limit),
Source: "local-json",
}, threshold, maxPairs, nil
}
limit, err := parseIntArg(ctx.Arg("limit"), 50, "limit")
if err != nil {
return IssueDedupeInput{}, 0, 0, err
}
page, err := parseIntArg(ctx.Arg("page"), 1, "page")
if err != nil {
return IssueDedupeInput{}, 0, 0, err
}
state := strings.TrimSpace(ctx.Arg("state"))
if state == "" {
state = "open"
}
issues, err := FetchIssuesForTriage(ctx, TriageFetchOptions{
State: state,
Limit: limit,
Page: page,
})
if err != nil {
return IssueDedupeInput{}, 0, 0, fmt.Errorf("workflow +issue-dedupe remote mode requires --owner and --repo or a Git remote: %w", err)
}
return IssueDedupeInput{
Repository: repositoryFromContext(ctx, ""),
Issues: issues,
Source: "remote-read-only-fetch",
}, threshold, maxPairs, nil
}
func AnalyzeIssueDedupe(input IssueDedupeInput, threshold int, maxPairs int, lang string) IssueDedupeResult {
lang = normalizeLang(lang)
if threshold < 0 {
threshold = 0
}
if threshold > 100 {
threshold = 100
}
repository := strings.TrimSpace(input.Repository)
if repository == "" {
repository = "local"
}
source := strings.TrimSpace(input.Source)
if source == "" {
source = "local"
}
fingerprints := make([]issueFingerprint, 0, len(input.Issues))
for _, issue := range input.Issues {
fp := buildIssueFingerprint(issue)
if len(fp.Terms) > 0 {
fingerprints = append(fingerprints, fp)
}
}
pairs := []IssueDedupePair{}
for i := 0; i < len(fingerprints); i++ {
for j := i + 1; j < len(fingerprints); j++ {
score, shared := scoreIssueSimilarity(fingerprints[i], fingerprints[j])
if score < threshold {
continue
}
pairs = append(pairs, buildIssueDedupePair(fingerprints[i], fingerprints[j], score, shared, lang))
}
}
sort.SliceStable(pairs, func(i, j int) bool {
if pairs[i].Score != pairs[j].Score {
return pairs[i].Score > pairs[j].Score
}
return pairs[i].Primary.Number < pairs[j].Primary.Number
})
if maxPairs > 0 && len(pairs) > maxPairs {
pairs = pairs[:maxPairs]
}
result := IssueDedupeResult{
Repository: repository,
TotalIssues: len(input.Issues),
CandidatePairs: len(pairs),
Threshold: threshold,
Pairs: pairs,
Source: source,
}
for _, pair := range pairs {
switch pair.Confidence {
case "high":
result.HighConfidence++
case "medium":
result.MediumConfidence++
default:
result.LowConfidence++
}
}
result.Recommendations = issueDedupeRecommendations(result, lang)
return result
}
type issueFingerprint struct {
Issue IssueInput
Terms map[string]int
}
func buildIssueFingerprint(issue IssueInput) issueFingerprint {
terms := map[string]int{}
for _, token := range tokenizeIssueText(issue.Title + " " + issue.Body) {
terms[token]++
}
for _, label := range issue.Labels {
for _, token := range tokenizeIssueText(label) {
terms[token] += 2
}
}
return issueFingerprint{Issue: issue, Terms: terms}
}
func tokenizeIssueText(value string) []string {
value = strings.ToLower(value)
var b strings.Builder
for _, r := range value {
if unicode.IsLetter(r) || unicode.IsDigit(r) {
b.WriteRune(r)
} else {
b.WriteByte(' ')
}
}
raw := strings.Fields(b.String())
tokens := make([]string, 0, len(raw))
for _, token := range raw {
if len([]rune(token)) < 3 || issueDedupeStopWords[token] {
continue
}
tokens = append(tokens, token)
}
return tokens
}
var issueDedupeStopWords = map[string]bool{
"the": true, "and": true, "for": true, "with": true, "when": true, "from": true, "this": true, "that": true,
"can": true, "not": true, "cannot": true, "should": true, "would": true, "could": true, "please": true,
"issue": true, "bug": true, "error": true, "failed": true, "failure": true, "using": true, "after": true,
}
func scoreIssueSimilarity(a, b issueFingerprint) (int, []string) {
intersectionWeight := 0
unionWeight := 0
shared := []string{}
seen := map[string]bool{}
for term, weightA := range a.Terms {
weightB := b.Terms[term]
if weightB > 0 {
intersectionWeight += minInt(weightA, weightB)
shared = append(shared, term)
}
unionWeight += maxIssueDedupeInt(weightA, weightB)
seen[term] = true
}
for term, weightB := range b.Terms {
if seen[term] {
continue
}
unionWeight += weightB
}
if unionWeight == 0 {
return 0, nil
}
sort.Strings(shared)
score := int(math.Round(float64(intersectionWeight) / float64(unionWeight) * 100))
if sameNonZeroIssueType(a.Issue, b.Issue) {
score += 8
}
if score > 100 {
score = 100
}
return score, limitIssueDedupeTerms(shared, 8)
}
func sameNonZeroIssueType(a, b IssueInput) bool {
typeA := AnalyzeIssue(a, langEN).DetectedType
typeB := AnalyzeIssue(b, langEN).DetectedType
return typeA != IssueTypeUnknown && typeA == typeB
}
func buildIssueDedupePair(a, b issueFingerprint, score int, shared []string, lang string) IssueDedupePair {
confidence := "low"
if score >= 80 {
confidence = "high"
} else if score >= 65 {
confidence = "medium"
}
primary := issueRefFromInput(a.Issue)
duplicate := issueRefFromInput(b.Issue)
if duplicate.Number > 0 && (primary.Number == 0 || duplicate.Number < primary.Number) {
primary, duplicate = duplicate, primary
}
reason := fmt.Sprintf("shared terms: %s", strings.Join(shared, ", "))
if normalizeLang(lang) == langZH {
reason = fmt.Sprintf("共享关键词:%s", strings.Join(shared, "、"))
}
return IssueDedupePair{
Score: score,
Confidence: confidence,
Primary: primary,
Duplicate: duplicate,
SharedTerms: shared,
Reason: reason,
}
}
func issueRefFromInput(issue IssueInput) IssueRef {
return IssueRef{
ID: issue.ID,
Number: issue.Number,
Title: issue.Title,
URL: issue.URL,
Author: issue.Author,
State: issue.State,
}
}
func issueDedupeRecommendations(result IssueDedupeResult, lang string) []string {
zh := normalizeLang(lang) == langZH
if result.CandidatePairs == 0 {
if zh {
return []string{"未发现达到阈值的重复候选,可以按常规 Issue 流程继续处理。"}
}
return []string{"No duplicate candidates reached the threshold; continue with normal issue triage."}
}
recs := []string{}
if result.HighConfidence > 0 {
if zh {
recs = append(recs, "优先人工确认高置信候选,必要时在后创建的 Issue 中链接原始 Issue。")
} else {
recs = append(recs, "Review high-confidence candidates first and link newer issues to the original when appropriate.")
}
}
if zh {
recs = append(recs, "命令只提供候选对,不会自动关闭、评论或修改 Issue。")
} else {
recs = append(recs, "The command only reports candidates; it never closes, comments on, or edits issues.")
}
return recs
}
func RenderIssueDedupe(result IssueDedupeResult, format string, lang string) (string, error) {
var buf bytes.Buffer
switch normalizeFormat(format) {
case "json":
if err := writeJSON(&buf, result); err != nil {
return "", err
}
case "markdown":
if err := writeIssueDedupeMarkdown(&buf, result, lang); err != nil {
return "", err
}
case "table":
if err := writeIssueDedupeTable(&buf, result); err != nil {
return "", err
}
default:
return "", fmt.Errorf("unsupported workflow output format %q", format)
}
return buf.String(), nil
}
func writeIssueDedupeMarkdown(buf *bytes.Buffer, result IssueDedupeResult, lang string) error {
title := "Issue Duplicate Candidates"
if normalizeLang(lang) == langZH {
title = "Issue 重复候选"
}
if _, err := fmt.Fprintf(buf, "# %s\n\n", title); err != nil {
return err
}
if _, err := fmt.Fprintf(buf, "- Repository: `%s`\n- Issues analyzed: `%d`\n- Candidate pairs: `%d`\n- Threshold: `%d`\n- Source: `%s`\n\n",
result.Repository, result.TotalIssues, result.CandidatePairs, result.Threshold, result.Source); err != nil {
return err
}
if _, err := fmt.Fprintln(buf, "## Candidates"); err != nil {
return err
}
if _, err := fmt.Fprintln(buf); err != nil {
return err
}
if len(result.Pairs) == 0 {
_, err := fmt.Fprintln(buf, "- No duplicate candidates.")
return err
}
for _, pair := range result.Pairs {
if _, err := fmt.Fprintf(buf, "- `%s` score `%d`: #%d %s -> #%d %s\n",
pair.Confidence, pair.Score, pair.Primary.Number, pair.Primary.Title, pair.Duplicate.Number, pair.Duplicate.Title); err != nil {
return err
}
if _, err := fmt.Fprintf(buf, " - %s\n", pair.Reason); err != nil {
return err
}
}
if _, err := fmt.Fprintln(buf); err != nil {
return err
}
if _, err := fmt.Fprintln(buf, "## Recommendations"); err != nil {
return err
}
for _, rec := range result.Recommendations {
if _, err := fmt.Fprintf(buf, "- %s\n", rec); err != nil {
return err
}
}
return nil
}
func writeIssueDedupeTable(buf *bytes.Buffer, result IssueDedupeResult) error {
tw := tabwriter.NewWriter(buf, 0, 0, 2, ' ', 0)
if _, err := fmt.Fprintln(tw, "SCORE\tCONFIDENCE\tPRIMARY\tDUPLICATE\tSHARED_TERMS"); err != nil {
return err
}
for _, pair := range result.Pairs {
if _, err := fmt.Fprintf(tw, "%d\t%s\t#%d %s\t#%d %s\t%s\n",
pair.Score,
pair.Confidence,
pair.Primary.Number,
truncateTableText(pair.Primary.Title, 48),
pair.Duplicate.Number,
truncateTableText(pair.Duplicate.Title, 48),
strings.Join(pair.SharedTerms, ","),
); err != nil {
return err
}
}
return tw.Flush()
}
func limitIssueDedupeTerms(values []string, limit int) []string {
if limit <= 0 || len(values) <= limit {
return values
}
return values[:limit]
}
func maxIssueDedupeInt(a, b int) int {
if a > b {
return a
}
return b
}

View File

@ -0,0 +1,148 @@
package workflow
import (
"net/http"
"net/http/httptest"
"path/filepath"
"strings"
"testing"
"github.com/gitlink-org/gitlink-cli/internal/client"
"github.com/gitlink-org/gitlink-cli/shortcuts/common"
)
func TestAnalyzeIssueDedupeFindsDuplicatePairs(t *testing.T) {
result := AnalyzeIssueDedupe(IssueDedupeInput{
Repository: "owner/repo",
Source: "local-json",
Issues: []IssueInput{
{Number: 1, Title: "Login fails on Windows with token auth", Body: "The token login command fails on Windows credential manager.", State: "open", Labels: []string{"bug", "auth"}},
{Number: 2, Title: "Token login failure in Windows credential manager", Body: "Auth token login cannot complete on Windows.", State: "open", Labels: []string{"bug", "auth"}},
{Number: 3, Title: "Add docs for release upload", Body: "Document release asset upload examples.", State: "open", Labels: []string{"docs"}},
},
}, 35, 10, "en")
if result.CandidatePairs != 1 {
t.Fatalf("CandidatePairs = %d, want 1: %+v", result.CandidatePairs, result.Pairs)
}
pair := result.Pairs[0]
if pair.Primary.Number != 1 || pair.Duplicate.Number != 2 {
t.Fatalf("pair = #%d/#%d, want #1/#2", pair.Primary.Number, pair.Duplicate.Number)
}
if pair.Score < 35 || len(pair.SharedTerms) == 0 {
t.Fatalf("score/shared = %d/%v, want duplicate signal", pair.Score, pair.SharedTerms)
}
if len(result.Recommendations) == 0 {
t.Fatal("expected recommendations")
}
}
func TestCollectIssueDedupeInputFromFileFiltersStateAndLimit(t *testing.T) {
path := filepath.Join(t.TempDir(), "issues.json")
writeJSONFixture(t, path, map[string]interface{}{
"issues": []IssueInput{
{Number: 1, Title: "open issue", State: "open"},
{Number: 2, Title: "closed issue", State: "closed"},
{Number: 3, Title: "second open issue", State: "open"},
},
})
ctx := &common.RuntimeContext{
Args: map[string]string{
"from": path,
"state": "open",
"limit": "1",
"threshold": "55",
"max-pairs": "20",
},
}
input, threshold, maxPairs, err := collectIssueDedupeInput(ctx)
if err != nil {
t.Fatalf("collectIssueDedupeInput returned error: %v", err)
}
if threshold != 55 || maxPairs != 20 {
t.Fatalf("threshold/maxPairs = %d/%d, want 55/20", threshold, maxPairs)
}
if len(input.Issues) != 1 || input.Issues[0].Number != 1 {
t.Fatalf("issues = %+v, want only first open issue", input.Issues)
}
}
func TestCollectIssueDedupeInputRemoteFetch(t *testing.T) {
server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.Method != "GET" || r.URL.Path != "/v1/owner/repo/issues.json" {
t.Fatalf("unexpected request: %s %s", r.Method, r.URL.Path)
}
if got := r.URL.Query().Get("state"); got != "open" {
t.Fatalf("state = %q, want open", got)
}
if got := r.URL.Query().Get("page"); got != "2" {
t.Fatalf("page = %q, want 2", got)
}
if got := r.URL.Query().Get("limit"); got != "5" {
t.Fatalf("limit = %q, want 5", got)
}
writeWorkflowJSON(t, w, map[string]interface{}{
"issues": []map[string]interface{}{
{"number": 7, "title": "Login fails", "description": "Token login fails", "state": "open"},
},
})
}))
defer server.Close()
ctx := &common.RuntimeContext{
Client: &client.Client{HTTP: server.Client(), BaseURL: server.URL},
Owner: "owner",
Repo: "repo",
Args: map[string]string{
"state": "open",
"page": "2",
"limit": "5",
"threshold": "55",
"max-pairs": "20",
},
}
input, _, _, err := collectIssueDedupeInput(ctx)
if err != nil {
t.Fatalf("collectIssueDedupeInput remote returned error: %v", err)
}
if input.Source != "remote-read-only-fetch" || len(input.Issues) != 1 || input.Issues[0].Number != 7 {
t.Fatalf("input = %+v, want remote issue #7", input)
}
}
func TestRenderIssueDedupeMarkdownAndTable(t *testing.T) {
result := AnalyzeIssueDedupe(IssueDedupeInput{
Repository: "owner/repo",
Issues: []IssueInput{
{Number: 1, Title: "Login token fails on Windows", Body: "credential manager auth failure", Labels: []string{"auth"}},
{Number: 2, Title: "Windows token login auth failure", Body: "credential manager login fails", Labels: []string{"auth"}},
},
}, 30, 10, "zh-CN")
markdown, err := RenderIssueDedupe(result, "markdown", "zh-CN")
if err != nil {
t.Fatalf("RenderIssueDedupe markdown returned error: %v", err)
}
if !strings.Contains(markdown, "# Issue 重复候选") || !strings.Contains(markdown, "共享关键词") {
t.Fatalf("markdown output missing expected content:\n%s", markdown)
}
table, err := RenderIssueDedupe(result, "table", "en")
if err != nil {
t.Fatalf("RenderIssueDedupe table returned error: %v", err)
}
if !strings.Contains(table, "SCORE") || !strings.Contains(table, "#1") {
t.Fatalf("table output missing expected content:\n%s", table)
}
}
func TestShortcutsExposeIssueDedupe(t *testing.T) {
names := map[string]bool{}
for _, shortcut := range Shortcuts() {
names[shortcut.Name] = true
}
if !names["issue-dedupe"] {
t.Fatal("Shortcuts missing issue-dedupe")
}
}

View File

@ -25,7 +25,7 @@ func Shortcuts() []*common.Shortcut {
newHealthShortcut(),
newPRSummaryShortcut(),
newRepoReportShortcut(),
newReleaseReadinessShortcut(),
newIssueDedupeShortcut(),
}
}