From 8d15119f34ab1742437251c9683d18291ab26d5b Mon Sep 17 00:00:00 2001
From: thezero <io@thezero.org>
Date: Fri, 24 Apr 2020 12:54:38 +0200
Subject: [PATCH] add parse version in html regexes, fix RDoc detection

---
 webtech/__version__.py |  2 +-
 webtech/target.py      | 13 ++++++++++++-
 webtech/webtech.json   | 10 ++++++++++
 3 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/webtech/__version__.py b/webtech/__version__.py
index 5214924..72dad44 100644
--- a/webtech/__version__.py
+++ b/webtech/__version__.py
@@ -1,2 +1,2 @@
 # DON'T EDIT THIS FILE
-__version__ = "1.2.7"
+__version__ = "1.2.8"
diff --git a/webtech/target.py b/webtech/target.py
index ac9a46a..53b2ede 100644
--- a/webtech/target.py
+++ b/webtech/target.py
@@ -3,6 +3,7 @@
 import sys
 import json
 import re
+import sre_constants
 from io import open
 from .__burp__ import BURP
 
@@ -214,9 +215,19 @@ def check_html(self, tech, html):
             html = [html]
 
         for source in html:
-            matches = re.search(source, self.data['html'], re.IGNORECASE)
+            # Parse the matching regex
+            attr, extra = parse_regex_string(source)
+            matches = None
+            try:
+                matches = re.search(attr, self.data['html'], re.IGNORECASE)
+            except sre_constants.error:
+                pass
             if matches is not None:
                 matched_tech = Tech(name=tech, version=None)
+                # The version extra data is present
+                if extra and 'version' in extra:
+                    if matches.group(1):
+                        matched_tech = matched_tech._replace(version=matches.group(1))
                 self.report['tech'].add(matched_tech)
                 # this tech is matched, GOTO next
                 return
diff --git a/webtech/webtech.json b/webtech/webtech.json
index 0c624f9..32d09d8 100644
--- a/webtech/webtech.json
+++ b/webtech/webtech.json
@@ -31,6 +31,16 @@
         ".*/([\\d.]+)/scripts/microsoft.owa\\S*.js\\;version:\\1",
         ".*/([\\d.]+)/scripts/owa.mail.js\\;version:\\1"
       ]
+    },
+    "RDoc": {
+      "html": [
+        "<link[^>]+href=\"[^\"]*rdoc-style\\.css",
+        "Generated by <a[^>]+href=\"(?:[^>])+>RDoc<\\/a> ([\\d.]*\\d)\\;version:\\1",
+        "<footer(?:(?!<\\/footer>).){1,500}<p>\\s*Generated by <a href=\"https:\\/\\/ruby\\.github\\.io\\/rdoc\\/\">RDoc<\\/a> ([\\d.]*\\d)\\;version:\\1"
+      ],
+      "js": {
+        "rdoc_rel_prefix": ""
+      }
     }
   }
 }
\ No newline at end of file