From b6f166c4ed98f94bdd2cc82885d61173a101abfd Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Tue, 26 Jul 2016 15:12:54 +0200
Subject: Redesign path to store trailing slash for directories

---
 butl/path.ixx | 352 +++++++++++++++++++++++++++++++++++++++++++++-------------
 1 file changed, 273 insertions(+), 79 deletions(-)

(limited to 'butl/path.ixx')

diff --git a/butl/path.ixx b/butl/path.ixx
index 48d6576..3d1f20c 100644
--- a/butl/path.ixx
+++ b/butl/path.ixx
@@ -25,41 +25,57 @@ namespace butl
   }
 #endif
 
-  // @@ Should only enable_if P is basic_path<C, K1>.
-  //
+  template <class C, class K1, class K2>
+  inline basic_path<C, K1>
+  path_cast_impl (const basic_path<C, K2>& p, basic_path<C, K1>*)
+  {
+    typename basic_path<C, K1>::data_type d (
+      typename basic_path<C, K1>::string_type (p.path_), p.diff_);
+    K1::cast (d);
+    return basic_path<C, K1> (std::move (d));
+  }
+
+  template <class C, class K1, class K2>
+  inline basic_path<C, K1>
+  path_cast_impl (basic_path<C, K2>&& p, basic_path<C, K1>*)
+  {
+    typename basic_path<C, K1>::data_type d (std::move (p.path_), p.diff_);
+    K1::cast (d);
+    return basic_path<C, K1> (std::move (d));
+  }
+
   template <class P, class C, class K>
   inline P
   path_cast (const basic_path<C, K>& p)
   {
-    return P (p.path_, false);
+    return path_cast_impl (p, static_cast<P*> (nullptr));
   }
 
   template <class P, class C, class K>
   inline P
   path_cast (basic_path<C, K>&& p)
   {
-    return P (std::move (p.path_), false);
+    return path_cast_impl (std::move (p), static_cast<P*> (nullptr));
   }
 
   template <typename C, typename K>
   inline bool basic_path<C, K>::
   simple () const
   {
-    return
-#ifndef _WIN32
-      root () ||
-#endif
-      traits::find_separator (this->path_) == string_type::npos;
+    return empty () ||
+      traits::rfind_separator (this->path_, _size () - 1) == string_type::npos;
   }
 
   template <typename C, typename K>
   inline bool basic_path<C, K>::
   absolute () const
   {
+    const string_type& s (this->path_);
+
 #ifdef _WIN32
-    return this->path_.size () > 1 && this->path_[1] == ':';
+    return s.size () >  1 && s[1] == ':';
 #else
-    return !this->path_.empty () && traits::is_separator (this->path_[0]);
+    return s.size () != 0 && traits::is_separator (s[0]);
 #endif
   }
 
@@ -67,10 +83,12 @@ namespace butl
   inline bool basic_path<C, K>::
   root () const
   {
+    const string_type& s (this->path_);
+
 #ifdef _WIN32
-    return this->path_.size () == 2 && this->path_[1] == ':';
+    return s.size () == 2 && s[1] == ':';
 #else
-    return this->path_.size () == 1 && traits::is_separator (this->path_[0]);
+    return s.size () == 1 && traits::is_separator (s[0]);
 #endif
   }
 
@@ -78,95 +96,116 @@ namespace butl
   inline bool basic_path<C, K>::
   sub (const basic_path& p) const
   {
-    size_type n (p.path_.size ());
+    // The thinking here is that we can use the full string representations
+    // (including the trailing slash in "/").
+    //
+    const string_type& ps (p.path_);
+    size_type pn (ps.size ());
 
-    if (n == 0)
+    if (pn == 0)
       return true;
 
-    size_type m (this->path_.size ());
+    const string_type& s (this->path_);
+    size_type n (s.size ());
 
     // The second condition guards against the /foo-bar vs /foo case.
     //
-    return m >= n &&
-      traits::compare (this->path_.c_str (), n, p.path_.c_str (), n) == 0 &&
-      (traits::is_separator (p.path_.back ()) || // p ends with a separator
-       m == n                                 || // *this == p
-       traits::is_separator (this->path_[n]));   // next char is a separator
+    return n >= pn &&
+      traits::compare (s.c_str (), pn, ps.c_str (), pn) == 0 &&
+      (traits::is_separator (ps.back ())      || // p ends with a separator
+       n == pn                                || // *this == p
+       traits::is_separator (s[pn]));            // next char is a separator
   }
 
   template <typename C, typename K>
   inline bool basic_path<C, K>::
   sup (const basic_path& p) const
   {
-    size_type n (p.path_.size ());
+    // The thinking here is that we can use the full string representations
+    // (including the trailing slash in "/").
+    //
+    const string_type& ps (p.path_);
+    size_type pn (ps.size ());
 
-    if (n == 0)
+    if (pn == 0)
       return true;
 
-    size_type m (this->path_.size ());
+    const string_type& s (this->path_);
+    size_type n (s.size ());
 
     // The second condition guards against the /foo-bar vs bar case.
     //
-    return m >= n &&
-      traits::compare (
-        this->path_.c_str () +  m - n, n, p.path_.c_str (), n) == 0 &&
-      (m == n ||                                       // *this == p
-       traits::is_separator (this->path_[m - n - 1])); // prev char separator
+    return n >= pn &&
+      traits::compare (s.c_str () +  n - pn, pn, ps.c_str (), pn) == 0 &&
+      (n == pn ||                             // *this == p
+       traits::is_separator (s[n - pn - 1])); // previous char is a separator
+  }
+
+  template <typename C, typename K>
+  inline basic_path<C, K> basic_path<C, K>::
+  leaf () const
+  {
+    const string_type& s (this->path_);
+    size_type n (_size ());
+
+    size_type p (n != 0
+                 ? traits::rfind_separator (s, n - 1)
+                 : string_type::npos);
+
+    return p != string_type::npos
+      ? basic_path (data_type (string_type (s, p + 1), this->diff_))
+      : *this;
+  }
+
+  template <typename C, typename K>
+  inline typename basic_path<C, K>::dir_type basic_path<C, K>::
+  directory () const
+  {
+    const string_type& s (this->path_);
+    size_type n (_size ());
+
+    size_type p (n != 0
+                 ? traits::rfind_separator (s, n - 1)
+                 : string_type::npos);
+
+    return p != string_type::npos
+      ? dir_type (data_type (string_type (s, 0, p + 1))) // Include slash.
+      : dir_type ();
   }
 
   template <typename C, typename K>
   inline auto basic_path<C, K>::
   begin () const -> iterator
   {
-    size_type b, e;
+    const string_type& s (this->path_);
 
-    if (this->path_.empty ())
-      b = e = string_type::npos;
+    size_type b (s.empty () ? string_type::npos : 0);
+    size_type e (b == 0 ? traits::find_separator (s) : b);
 
-#ifndef _WIN32
-    else if (root ())
-    {
-      // We want to return a single empty component. Here we return
-      // the begin position one past the end. Not sure if this legal.
-      //
-      b = 1;
-      e = string_type::npos;
-    }
-#endif
-    else
-    {
-      b = 0;
-      e = traits::find_separator (this->path_);
-    }
-
-    return iterator (this->path_, b, e);
+    return iterator (this, b, e);
   }
 
   template <typename C, typename K>
   inline auto basic_path<C, K>::
   end () const -> iterator
   {
-    return iterator (this->path_, string_type::npos, string_type::npos);
+    return iterator (this, string_type::npos, string_type::npos);
   }
 
   template <typename C, typename K>
   inline basic_path<C, K>::
   basic_path (const iterator& b, const iterator& e)
+      : base_type (
+        b == e
+        ? data_type ()
+        // We need to include the trailing separator but it is implied if
+        // e == end().
+        //
+        : (e.b_ != string_type::npos
+           ? data_type (string_type (b.p_->path_, b.b_, e.b_ - b.b_))
+           : data_type (string_type (b.p_->path_, b.b_), b.p_->diff_)))
   {
     //assert (b.p_ == e.p_);
-
-    if (b != e)
-    {
-      this->path_.assign (
-        *b.p_, b.b_, (e.b_ != string_type::npos ? e.b_ - b.b_ - 1 : e.b_));
-
-#ifndef _WIN32
-      if (this->path_.empty ())
-        this->path_ = '/';
-#endif
-
-      // No init() should be necessary.
-    }
   }
 
   template <typename C, typename K>
@@ -187,7 +226,7 @@ namespace butl
     complete ();
     normalize ();
 #else
-    traits::realize (this->path_);
+    traits::realize (this->path_); // Note: we retail trailing slash.
 #endif
     return *this;
   }
@@ -196,24 +235,34 @@ namespace butl
   inline typename basic_path<C, K>::dir_type basic_path<C, K>::
   root_directory () const
   {
-    return absolute ()
 #ifdef _WIN32
-      // Disambiguate with dir_type(string_type, bool).
-      //
-      ? dir_type (this->path_, static_cast<size_type> (2))
+    // Note: on Windows we may have "c:" but still need to return "c:\".
+    //
+    const string_type& s (this->path_);
+
+    return absolute ()
+      ? dir_type (
+        s.size () > 2
+        ? data_type (string_type (s, 0, 3))
+        : data_type (string_type (s), this->diff_ != 0 ? this->diff_ : 1))
+      : dir_type ();
 #else
-      ? dir_type ("/")
-#endif
+    return absolute ()
+      ? dir_type (data_type ("/", -1))
       : dir_type ();
+#endif
+
   }
 
   template <typename C, typename K>
   inline basic_path<C, K> basic_path<C, K>::
   base () const
   {
-    size_type p (traits::find_extension (this->path_));
+    const string_type& s (this->path_);
+    size_type p (traits::find_extension (s));
+
     return p != string_type::npos
-      ? basic_path (this->path_.c_str (), p)
+      ? basic_path (data_type (string_type (s, 0, p), this->diff_))
       : *this;
   }
 
@@ -221,8 +270,9 @@ namespace butl
   inline const C* basic_path<C, K>::
   extension () const
   {
-    size_type p (traits::find_extension (this->path_));
-    return p != string_type::npos ? this->path_.c_str () + p + 1 : nullptr;
+    const string_type& s (this->path_);
+    size_type p (traits::find_extension (s));
+    return p != string_type::npos ? s.c_str () + p + 1 : nullptr;
   }
 
 #ifndef _WIN32
@@ -236,16 +286,160 @@ namespace butl
 
   template <typename C, typename K>
   inline void basic_path<C, K>::
-  combine (const C* r, size_type rn)
+  combine (const C* r, size_type rn, difference_type rd)
   {
-    size_type ln (this->path_.size ());
+    //assert (rn != 0);
 
-    if (ln != 0 && rn != 0)
+    string_type& l (this->path_);
+    difference_type& d (this->diff_);
+
+    // Handle the separator. LHS should be empty or already have one.
+    //
+    switch (d)
     {
-      if (!traits::is_separator (this->path_[ln - 1]))
-        this->path_ += traits::directory_separator;
+    case  0: if (!l.empty ()) throw invalid_basic_path<C> (l); break;
+    case -1: break; // Already in the string.
+    default: l += path_traits<C>::directory_separators[d - 1];
     }
 
+    l.append (r, rn);
+    d = rd; // New trailing separator from RHS.
+  }
+
+  template <typename C, typename K>
+  inline void basic_path<C, K>::
+  combine (const C* r, size_type rn)
+  {
+    // If we do (dir_path / path) then we will end up with path. What should
+    // we end up if we do (dir_path / "foo") vs (dir_path / "foo/")? We cannot
+    // choose at runtime what kind of path to return. One (elaborate) option
+    // would be to handle the trailing slash but also call K::cast() so that
+    // dir_path gets the canonical trailing slash if one wasn't there.
+    //
+    // For now we won't allow the slash and will always add the canonical one
+    // for dir_path (via cast()).
+    //
+    if (traits::find_separator (r, rn) != nullptr)
+      throw invalid_basic_path<C> (r);
+
+    combine (r, rn, 0);
+    K::cast (*this);
+  }
+
+  template <typename C, typename K>
+  inline basic_path<C, K>& basic_path<C, K>::
+  operator/= (basic_path<C, K> const& r)
+  {
+    if (r.absolute () && !empty ()) // Allow ('' / '/foo').
+      throw invalid_basic_path<C> (r.path_);
+
+    if (!r.empty ())
+      combine (r.path_.c_str (), r.path_.size (), r.diff_);
+
+    return *this;
+  }
+
+  template <typename C, typename K>
+  inline basic_path<C, K>& basic_path<C, K>::
+  operator/= (string_type const& r)
+  {
+    if (size_type rn = r.size ())
+      combine (r.c_str (), rn);
+
+    return *this;
+  }
+
+  template <typename C, typename K>
+  inline basic_path<C, K>& basic_path<C, K>::
+  operator/= (const C* r)
+  {
+    if (size_type rn = string_type::traits_type::length (r))
+      combine (r, rn);
+
+    return *this;
+  }
+
+  template <typename C, typename K>
+  inline void basic_path<C, K>::
+  append (const C* r, size_type rn)
+  {
+    //assert (this->diff_ != -1); // Append to root?
     this->path_.append (r, rn);
   }
+
+  template <typename C, typename K>
+  inline basic_path<C, K>& basic_path<C, K>::
+  operator+= (string_type const& s)
+  {
+    append (s.c_str (), s.size ());
+    return *this;
+  }
+
+  template <typename C, typename K>
+  inline basic_path<C, K>& basic_path<C, K>::
+  operator+= (const C* s)
+  {
+    append (s, string_type::traits_type::length (s));
+    return *this;
+  }
+
+  template <typename C, typename K>
+  inline basic_path<C, K>& basic_path<C, K>::
+  operator+= (C c)
+  {
+    append (&c, 1);
+    return *this;
+  }
+
+  template <typename C, typename K>
+  inline auto basic_path<C, K>::
+  representation () const& -> string_type
+  {
+    string_type r (this->path_);
+
+    if (this->diff_ > 0)
+      r += path_traits<C>::directory_separators[this->diff_ - 1];
+
+    return r;
+  }
+
+  template <typename C, typename K>
+  inline auto basic_path<C, K>::
+  representation () && -> string_type
+  {
+    string_type r;
+    r.swap (this->path_);
+
+    if (this->diff_ > 0)
+      r += path_traits<C>::directory_separators[this->diff_ - 1];
+
+    return r;
+  }
+
+  template <typename C, typename K>
+  inline C basic_path<C, K>::
+  separator () const
+  {
+    return (this->diff_ ==  0 ? 0 :
+            this->diff_ == -1 ? this->path_[0] :
+            path_traits<C>::directory_separators[this->diff_ - 1]);
+  }
+
+  template <typename C, typename K>
+  inline auto basic_path<C, K>::
+  separator_string () const -> string_type
+  {
+    C c (separator ());
+    return c == 0 ? string_type () : string_type (1, c);
+  }
+
+  template <typename C>
+  inline void dir_path_kind<C>::
+  cast (data_type& d)
+  {
+    // Add trailing slash if one isn't already there.
+    //
+    if (!d.path_.empty () && d.diff_ == 0)
+      d.diff_ = 1; // Canonical separator is always first.
+  }
 }
-- 
cgit v1.1