From 2caebcfecffad6d472db7b11c2082b43d6d5f270 Mon Sep 17 00:00:00 2001
From: Matt Baer <matt@write.as>
Date: Mon, 12 Mar 2018 11:20:56 -0400
Subject: [PATCH] Support blacklist for terms that prevent a public post

This adds a BLACKLIST_TERMS configuration value that contains a
comma-separated list of terms that indicate spam posts and will prevent
a link from being showcased on the Browse page, tweeted out, etc.
---
 README.md       |  1 +
 config.go       | 11 +++++++++++
 construction.go |  2 +-
 filter.go       |  6 ++++++
 4 files changed, 19 insertions(+), 1 deletion(-)
 create mode 100644 filter.go

diff --git a/README.md b/README.md
index 96f16ea..286e994 100644
--- a/README.md
+++ b/README.md
@@ -55,6 +55,7 @@ DB_USER=dbuser DB_PASSWORD=pass DB_DB=htmlhouse PRIVATE_KEY=keys/dev PUBLIC_KEY=
 | `PREVIEWS_HOST` | Fully-qualified URL (without trailing slash) of screenshot server | None. |
 | `ADMIN_PASS` | Password to perform admin functions via API | `uhoh` |
 | `BROWSE_ITEMS` | Number of items to show on Browse page | 10 |
+| `BLACKLIST_TERMS` | Comma-separated list of terms to prevent a post from being made public | None. |
 | `TWITTER_KEY` | Twitter consumer key | `notreal` |
 | `TWITTER_SECRET` | Twitter consumer secret | `notreal` |
 | `TWITTER_TOKEN` | Twitter access token of the posting Twitter account | `notreal` |
diff --git a/config.go b/config.go
index 41e64b5..cbd6943 100644
--- a/config.go
+++ b/config.go
@@ -2,6 +2,8 @@ package htmlhouse
 
 import (
 	"github.com/danryan/env"
+	"regexp"
+	"strings"
 )
 
 type config struct {
@@ -23,6 +25,9 @@ type config struct {
 	AdminPass    string `env:"key=ADMIN_PASS default=uhoh"`
 	BrowseItems  int    `env:"key=BROWSE_ITEMS default=10"`
 
+	BlacklistTerms string `env:"key=BLACKLIST_TERMS"`
+	BlacklistReg   *regexp.Regexp
+
 	// Twitter configuration
 	TwitterConsumerKey    string `env:"key=TWITTER_KEY default=notreal"`
 	TwitterConsumerSecret string `env:"key=TWITTER_SECRET default=notreal"`
@@ -36,5 +41,11 @@ func newConfig() (*config, error) {
 		return cfg, err
 	}
 
+	// Process anything
+	termsReg := `(?i)\b` + cfg.BlacklistTerms + `\b`
+	termsReg = strings.Replace(termsReg, ",", `\b|\b`, -1)
+	cfg.BlacklistReg = regexp.MustCompile(termsReg)
+
+	// Return result
 	return cfg, nil
 }
diff --git a/construction.go b/construction.go
index 3f8165c..018b449 100644
--- a/construction.go
+++ b/construction.go
@@ -39,7 +39,7 @@ func createHouse(app *app, w http.ResponseWriter, r *http.Request) error {
 
 	resUser := newSessionInfo(houseID)
 
-	if public {
+	if public && passesPublicFilter(app, html) {
 		go addPublicAccess(app, houseID, html)
 	}
 
diff --git a/filter.go b/filter.go
new file mode 100644
index 0000000..703f7e4
--- /dev/null
+++ b/filter.go
@@ -0,0 +1,6 @@
+package htmlhouse
+
+func passesPublicFilter(app *app, html string) bool {
+	spam := app.cfg.BlacklistReg.MatchString(html)
+	return !spam
+}