Add url::traits::find(), checks whether string looks like URL

2 files changed, 46 insertions, 1 deletions

diff --git a/libbutl/url.mxx b/libbutl/url.mxx
index fe091f1..577434f 100644
--- a/libbutl/url.mxx
+++ b/libbutl/url.mxx
@@ -216,6 +216,17 @@ LIBBUTL_MODEXPORT namespace butl
     //
     static string_type
     translate_path (const path_type& path) {return string_type (path);}
+
+    // Check whether a string looks like a URL by searching for the first ':'
+    // (unless its position is specified with the second argument) and then
+    // making sure it's followed by '/' (e.g., http:// or file:/) and preceded
+    // by the scheme at least 2 characters long (so we don't confuse it with
+    // an absolute Windows path, e.g., c:/).
+    //
+    // Return the start of the URL substring or string_type::npos.
+    //
+    static std::size_t
+    find (const string_type&, std::size_t pos = string_type::npos);
   };
 
   template <typename H, // scheme
diff --git a/libbutl/url.txx b/libbutl/url.txx
index e511bed..3a2c387 100644
--- a/libbutl/url.txx
+++ b/libbutl/url.txx
@@ -182,6 +182,40 @@ LIBBUTL_MODEXPORT namespace butl //@@ MOD Clang needs this for some reason.
     return r;
   }
 
+  // url_traits
+  //
+  template <typename H, typename S, typename P>
+  std::size_t url_traits<H, S, P>::
+  find (const string_type& s, std::size_t p)
+  {
+    if (p == string_type::npos)
+      p = s.find (':');
+
+    if (p == string_type::npos ||
+        p < 2 ||
+        p + 1 == s.size () || s[p + 1] != '/')
+      return string_type::npos;
+
+    // Scan backwards for as long as it is a valid scheme.
+    //
+    std::size_t i (p);
+
+    for (; i != 0; --i)
+    {
+      auto c (s[i - 1]);
+      if (!(alnum (c) || c == '+' || c == '-' || c == '.'))
+        break;
+    }
+
+    if (i != p && !alpha (s[i])) // First must be alpha.
+      ++i;
+
+    if (p - i < 2)
+      return string_type::npos;
+
+    return i;
+  }
+
   // basic_url
   //
   template <typename S, typename T>
@@ -209,7 +243,7 @@ LIBBUTL_MODEXPORT namespace butl //@@ MOD Clang needs this for some reason.
 
       // Extract scheme.
       //
-      for(char_type c; i != e && (c = *i) != ':'; ++i)
+      for (char_type c; i != e && (c = *i) != ':'; ++i)
       {
         if (!(i == b
               ? alpha (c)