Merge pull request #3 from dense-analysis/codex/extend-sea-go-generator-nginx-map-for-backend-routing

Enhance nginx map routing

Author: w0rp

.gitignore

@@ -7,6 +7,7 @@
 *.dll
 *.so
 *.dylib
+/sea
 
 # Test binary, built with `go test -c`
 *.test

AGENTS.md

@@ -0,0 +1,25 @@
+# AGENTS.md
+
+The purpose of this project is to produce an nginx configuration for redirecting
+to search providers based on the phrases used.
+
+You can run the main executable like so:
+
+```
+go run ./cmd/sea/main.go
+```
+
+`./cmd/sea/nginx.conf.tmpl` is the nginx template, and you should keep
+`nginx.conf` committed to the repository so people don't have to run the program
+to get it. Commit `config.toml` as a default configuration people can override
+if they so choose.
+
+You may validate that the nginx configuration works with the docker compose file
+like so:
+
+```
+# Leave an nginx docker image running on port 57321
+docker compose up
+# Check a redirect. The hostname comes from `server_name` in `config.toml`
+curl -s -D - http://search.localhost:57321/?q=how+can+i -o /dev/null | grep -i '^Location:'
+```

README.md

@@ -15,6 +15,8 @@ applied to `nginx`.
 - **Simple Configuration**: Ships as a ready-to-use NGINX `.conf` file.
 - **Lightweight & Fast**: Minimal dependencies for high throughput. Very simple
   security profile, as there is no web app to exploit.
+- **Optimized Matching**: Regex rules are evaluated once per request using
+  NGINX's `map` directive. Enable `pcre_jit` for the best performance.
 
 ## Development Requirements
 

cmd/sea/main.go

@@ -19,6 +19,13 @@ type Config struct {
 	CustomKeywords []KeywordRule `toml:"custom_keywords"`
 }
 
+// TemplateData combines the parsed configuration with the destination
+// to URL mappings used to construct the final nginx file.
+type TemplateData struct {
+	Config
+	Targets map[string]string
+}
+
 // KeywordRule maps a phrase to a destination.
 type KeywordRule struct {
 	Phrase string `toml:"phrase"`
@@ -73,8 +80,19 @@ func generateNginx(cfg Config) (string, error) {
 		return "", err
 	}
 
+	data := TemplateData{
+		Config: cfg,
+		Targets: map[string]string{
+			"google":        "https://www.google.com/search?q=$arg_q",
+			"chatgpt":       "https://chatgpt.com/?q=$arg_q",
+			"wikipedia":     "https://en.wikipedia.org/wiki/$arg_q",
+			"google_images": "https://www.google.com/search?tbm=isch&q=$arg_q",
+			"google_maps":   "https://www.google.com/maps/search/?q=$arg_q",
+		},
+	}
+
 	var b bytes.Buffer
-	if err := tmpl.Execute(&b, cfg); err != nil {
+	if err := tmpl.Execute(&b, data); err != nil {
 		return "", err
 	}
 	return b.String(), nil

cmd/sea/nginx.conf.tmpl

@@ -1,23 +1,29 @@
+# Map routes search queries to the appropriate backend.
+# Regex is evaluated once per request; enable `pcre_jit` for best speed.
 map $arg_q $dest {
-    default google;
-    ~*(?i)^\s*how\s+to\b                chatgpt;
-    ~*(?i)^\s*what\s+is\b               chatgpt;
-    ~*(?i)^\s*(when|where|why)\b         chatgpt;
-    ~*(?i)^\s*who\s+(?:is|was)\b       wikipedia;
-    ~*(?i)\bpictures?\s+of\b            google_images;
-    ~*(?i)\bvs\b                        google;
-    ~*(?i)\bdownload\b                   google;
-    ~*(?i)\bwikipedia\b|\bwiki\b        wikipedia;
+    # direct image search
+    ~*(?i)\bpictures?\s+of\b                         google_images;
+    # map and direction queries
+    ~*(?i)\b(near\s+me|directions?\s+to|map\s+of)\b  google_maps;
+    # Wikipedia lookups
+    ~*(?i)\b(?:biography|history|life\s+of)\b        wikipedia;
+    # question detection
+    ~*(?i)\b(who|what|when|where|why|how|can|could|would|should|do|did|is|are|was|were|am|will|whom|whose|which)\b chatgpt;
+    # instructional queries
+    ~*(?i)\b(explain|describe|compare|define)\b      chatgpt;
+    # explicit wiki keywords
+    ~*(?i)\bwikipedia\b|\bwiki\b                     wikipedia;
 {{- range .CustomKeywords }}
     ~*(?i)^{{ escape .Phrase }}$                {{ .Dest }};
 {{- end }}
+    default                                          google;
 }
 
+# Map engine names to full URL targets. Update this list in Go.
 map $dest $target {
-    google        "https://www.google.com/search?q=$arg_q";
-    google_images "https://www.google.com/search?tbm=isch&q=$arg_q";
-    chatgpt       "https://chat.openai.com/?q=$arg_q";
-    wikipedia     "https://en.wikipedia.org/wiki/$arg_q";
+{{- range $name, $url := .Targets }}
+    {{ $name }} {{ $url }};
+{{- end }}
 }
 
 server {
@@ -28,3 +34,4 @@ server {
         return 302 $target;
     }
 }
+

compose.yml

@@ -0,0 +1,8 @@
+---
+services:
+  nginx:
+    image: nginx:latest
+    ports:
+      - '57321:80'
+    volumes:
+      - ./nginx.conf:/etc/nginx/conf.d/default.conf:ro

config.toml

@@ -1,5 +1,5 @@
 listen = 80
-server_name = "search.local"
+server_name = "search.localhost"
 
 [[custom_keywords]]
 phrase = "nginx"

nginx.conf

@@ -0,0 +1,40 @@
+# Map routes search queries to the appropriate backend.
+# Regex is evaluated once per request; enable `pcre_jit` for best speed.
+map $arg_q $dest {
+    # direct image search
+    ~*(?i)\bpictures?\s+of\b                         google_images;
+    # map and direction queries
+    ~*(?i)\b(near\s+me|directions?\s+to|map\s+of)\b  google_maps;
+    # Wikipedia lookups
+    ~*(?i)\b(?:biography|history|life\s+of)\b        wikipedia;
+    # question detection
+    ~*(?i)\b(who|what|when|where|why|how|can|could|would|should|do|did|is|are|was|were|am|will|whom|whose|which)\b chatgpt;
+    # instructional queries
+    ~*(?i)\b(explain|describe|compare|define)\b      chatgpt;
+    # explicit wiki keywords
+    ~*(?i)\bwikipedia\b|\bwiki\b                     wikipedia;
+    ~*(?i)^nginx$                wikipedia;
+    ~*(?i)^New\ York$                wikipedia;
+    ~*(?i)^openai$                wikipedia;
+    ~*(?i)^cat\ pictures$                google_images;
+    default                                          google;
+}
+
+# Map engine names to full URL targets. Update this list in Go.
+map $dest $target {
+    chatgpt https://chatgpt.com/?q=$arg_q;
+    google https://www.google.com/search?q=$arg_q;
+    google_images https://www.google.com/search?tbm=isch&q=$arg_q;
+    google_maps https://www.google.com/maps/search/?q=$arg_q;
+    wikipedia https://en.wikipedia.org/wiki/$arg_q;
+}
+
+server {
+    listen 80;
+    server_name search.localhost;
+
+    location / {
+        return 302 $target;
+    }
+}
+