fix(robots2policy): handle multiple user agents under one block (#925)

cmd/robots2policy/testdata/blacklist.yaml

@@ -25,6 +25,6 @@
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("Googlebot")
-        - path.startsWith("/search")
-  name: robots-txt-policy-disallow-7
+    - userAgent.contains("Googlebot")
+    - path.startsWith("/search")
+  name: robots-txt-policy-disallow-7

cmd/robots2policy/testdata/complex.yaml

@@ -20,8 +20,8 @@
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("Googlebot")
-        - path.startsWith("/search/")
+    - userAgent.contains("Googlebot")
+    - path.startsWith("/search/")
   name: robots-txt-policy-disallow-6
 - action: WEIGH
   expression: userAgent.contains("Bingbot")
@@ -31,14 +31,14 @@
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("Bingbot")
-        - path.startsWith("/search/")
+    - userAgent.contains("Bingbot")
+    - path.startsWith("/search/")
   name: robots-txt-policy-disallow-8
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("Bingbot")
-        - path.startsWith("/admin/")
+    - userAgent.contains("Bingbot")
+    - path.startsWith("/admin/")
   name: robots-txt-policy-disallow-9
 - action: DENY
   expression: userAgent.contains("BadBot")
@@ -54,18 +54,18 @@
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("TestBot")
-        - path.matches("^/.*/admin")
+    - userAgent.contains("TestBot")
+    - path.matches("^/.*/admin")
   name: robots-txt-policy-disallow-13
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("TestBot")
-        - path.matches("^/temp.*\\.html")
+    - userAgent.contains("TestBot")
+    - path.matches("^/temp.*\\.html")
   name: robots-txt-policy-disallow-14
 - action: CHALLENGE
   expression:
     all:
-        - userAgent.contains("TestBot")
-        - path.matches("^/file.\\.log")
+    - userAgent.contains("TestBot")
+    - path.matches("^/file.\\.log")
   name: robots-txt-policy-disallow-15

cmd/robots2policy/testdata/consecutive.robots.txt

@@ -0,0 +1,25 @@
+# Test consecutive user agents that should be grouped into any: blocks
+User-agent: *
+Disallow: /admin
+Crawl-delay: 10
+
+# Multiple consecutive user agents - should be grouped
+User-agent: BadBot
+User-agent: SpamBot
+User-agent: EvilBot
+Disallow: /
+
+# Single user agent - should be separate
+User-agent: GoodBot
+Disallow: /private
+
+# Multiple consecutive user agents with crawl delay
+User-agent: SlowBot1
+User-agent: SlowBot2
+Crawl-delay: 5
+
+# Multiple consecutive user agents with specific path
+User-agent: SearchBot1
+User-agent: SearchBot2
+User-agent: SearchBot3
+Disallow: /search 

cmd/robots2policy/testdata/consecutive.yaml

@@ -0,0 +1,47 @@
+- action: WEIGH
+  expression: "true"
+  name: robots-txt-policy-crawl-delay-1
+  weight:
+    adjust: 3
+- action: CHALLENGE
+  expression: path.startsWith("/admin")
+  name: robots-txt-policy-disallow-2
+- action: DENY
+  expression:
+    any:
+    - userAgent.contains("BadBot")
+    - userAgent.contains("SpamBot")
+    - userAgent.contains("EvilBot")
+  name: robots-txt-policy-blacklist-3
+- action: CHALLENGE
+  expression:
+    all:
+    - userAgent.contains("GoodBot")
+    - path.startsWith("/private")
+  name: robots-txt-policy-disallow-4
+- action: WEIGH
+  expression:
+    any:
+    - userAgent.contains("SlowBot1")
+    - userAgent.contains("SlowBot2")
+  name: robots-txt-policy-crawl-delay-5
+  weight:
+    adjust: 3
+- action: CHALLENGE
+  expression:
+    all:
+    - userAgent.contains("SearchBot1")
+    - path.startsWith("/search")
+  name: robots-txt-policy-disallow-7
+- action: CHALLENGE
+  expression:
+    all:
+    - userAgent.contains("SearchBot2")
+    - path.startsWith("/search")
+  name: robots-txt-policy-disallow-8
+- action: CHALLENGE
+  expression:
+    all:
+    - userAgent.contains("SearchBot3")
+    - path.startsWith("/search")
+  name: robots-txt-policy-disallow-9

cmd/robots2policy/testdata/simple.json

@@ -1,12 +1,12 @@
 [
   {
-    "action": "CHALLENGE",
     "expression": "path.startsWith(\"/admin/\")",
-    "name": "robots-txt-policy-disallow-1"
+    "name": "robots-txt-policy-disallow-1",
+    "action": "CHALLENGE"
   },
   {
-    "action": "CHALLENGE",
     "expression": "path.startsWith(\"/private\")",
-    "name": "robots-txt-policy-disallow-2"
+    "name": "robots-txt-policy-disallow-2",
+    "action": "CHALLENGE"
   }
 ]