check-safety/check_htmlsanitize.go
Alex Dunmow cd88c808b0 initial: standalone check-safety module hoisted from CMS
Static safety/lint runner for the BlockNinja codebase. ~25 invariant
checks across Go and frontend sources. Was at git.dev.alexdunmow.com:block/ninja
in backend/cmd/check-safety/ until the 2026-06-06 consolidation moved
the BlockNinja repos under a shared ~/src/blockninja/ parent.

This repo is the standalone extraction:
- Own go.mod (git.dev.alexdunmow.com/block/check-safety, go 1.26.4)
- Vendored internal/{helpers,theme} from CMS (Go's internal/ rule
  blocks cross-module imports; vendoring is the workaround)
- CLI contract unchanged: `check-safety <target-dir> [--flags]`
- CMS Makefile shells into ../check-safety for safety-check /
  install-safety-checker targets

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-06-06 13:04:02 +08:00

51 lines
2.7 KiB
Go

package main
import "fmt"
func init() {
register(Check{
Seq: 220,
ID: "22",
Title: "No hand-rolled HTML sanitization (use bluemonday)",
Run: func(ctx *ScanContext, rep *Reporter) {
// Check 22: No hand-rolled HTML sanitization (use bluemonday)
fmt.Println("=== Check 22: No hand-rolled HTML sanitization (use bluemonday) ===")
var htmlSanViolations []htmlSanitizeViolation
for _, target := range ctx.backendTargets {
for _, v := range checkHTMLSanitize(target.root) {
v.file = prefixDisplayPath(target.display, v.file)
htmlSanViolations = append(htmlSanViolations, v)
}
}
for _, target := range ctx.pluginTargets {
for _, v := range checkHTMLSanitize(target.root) {
v.file = prefixDisplayPath(target.display, v.file)
htmlSanViolations = append(htmlSanViolations, v)
}
}
if len(htmlSanViolations) > 0 {
fmt.Printf(" FAIL: %d hand-rolled HTML sanitization pattern(s):\n", len(htmlSanViolations))
for _, v := range htmlSanViolations {
fmt.Printf(" %s:%d [%s] %s\n", v.file, v.line, v.rule, v.snippet)
}
fmt.Println(`
Fix: Use github.com/microcosm-cc/bluemonday for ALL HTML sanitization.
┌────────────────────────────────────────┬────────────────────────────────────────┐
│ Instead of │ Use │
├────────────────────────────────────────┼────────────────────────────────────────┤
│ regexp strip <tags> │ bluemonday.StrictPolicy().Sanitize(s) │
│ func stripHTML() { char loop } │ bluemonday.StrictPolicy().Sanitize(s) │
│ strings.NewReplacer("<br>","") │ bluemonday.StrictPolicy().Sanitize(s) │
│ helpers.StripHTML(s) │ bluemonday.StrictPolicy().Sanitize(s) │
└────────────────────────────────────────┴────────────────────────────────────────┘
Regex-based HTML stripping is a security risk — it misses edge cases,
nested tags, and encoded entities. bluemonday is battle-tested.`)
rep.Fail()
} else {
fmt.Println(" OK: No hand-rolled HTML sanitization detected")
printPerTargetOKLines(pluginTargetLabels(ctx.pluginTargets))
}
},
})
}