Move trim(), next_word() basic string utilities from build2

4 files changed, 133 insertions, 28 deletions

diff --git a/libbutl/path.mxx b/libbutl/path.mxx
index ab85a0f..5b9bec1 100644
--- a/libbutl/path.mxx
+++ b/libbutl/path.mxx
@@ -343,6 +343,9 @@ LIBBUTL_MODEXPORT namespace butl
     static int
     compare (const C* l, size_type ln, const C* r, size_type rn)
     {
+      //@@ TODO: would be nice to ignore difference in trailing slashes
+      //   (except for POSIX root).
+
       for (size_type i (0), n (ln < rn ? ln : rn); i != n; ++i)
       {
 #ifdef _WIN32
diff --git a/libbutl/utility.cxx b/libbutl/utility.cxx
index c8e4e85..4645b68 100644
--- a/libbutl/utility.cxx
+++ b/libbutl/utility.cxx
@@ -35,7 +35,6 @@ import std.io;
 
 #endif
 
-
 namespace butl
 {
   using namespace std;
@@ -88,6 +87,43 @@ namespace butl
                         win32::error_msg (system_code));
 #endif
   }
+
+  string&
+  trim (string& l)
+  {
+    /*
+    assert (trim (r = "") == "");
+    assert (trim (r = " ") == "");
+    assert (trim (r = " \t\r") == "");
+    assert (trim (r = "a") == "a");
+    assert (trim (r = " a") == "a");
+    assert (trim (r = "a ") == "a");
+    assert (trim (r = " \ta") == "a");
+    assert (trim (r = "a \r") == "a");
+    assert (trim (r = " a ") == "a");
+    assert (trim (r = " \ta \r") == "a");
+    */
+
+    size_t i (0), n (l.size ());
+
+    for (char c;
+         i != n && ((c = l[i]) == ' ' || c == '\t' || c == '\r');
+         ++i) ;
+
+    for (char c;
+         n != i && ((c = l[n - 1]) == ' ' || c == '\t' || c == '\r');
+         --n) ;
+
+    if (i != 0)
+    {
+      string s (l, i, n - i);
+      l.swap (s);
+    }
+    else if (n != l.size ())
+      l.resize (n);
+
+    return l;
+  }
 }
 
 namespace std
diff --git a/libbutl/utility.ixx b/libbutl/utility.ixx
index fcb8789..24271e4 100644
--- a/libbutl/utility.ixx
+++ b/libbutl/utility.ixx
@@ -4,18 +4,6 @@
 
 namespace butl
 {
-  inline bool
-  eof (std::istream& is)
-  {
-    if (!is.fail ())
-      return false;
-
-    if (is.eof ())
-      return true;
-
-    throw std::istream::failure ("");
-  }
-
   inline char
   ucase (char c)
   {
@@ -167,4 +155,48 @@ namespace butl
   {
     return std::iswxdigit (c);
   }
+
+  inline std::size_t
+  next_word (const std::string& s, std::size_t& b, std::size_t& e,
+             char d1, char d2)
+  {
+    return next_word (s, s.size (), b, e, d1, d2);
+  }
+
+  inline size_t
+  next_word (const std::string& s,
+             std::size_t n, std::size_t& b, std::size_t& e,
+             char d1, char d2)
+  {
+    if (b != e)
+      b = e;
+
+    // Skip leading delimiters.
+    //
+    for (; b != n && (s[b] == d1 || s[b] == d2); ++b) ;
+
+    if (b == n)
+    {
+      e = n;
+      return 0;
+    }
+
+    // Find first trailing delimiter.
+    //
+    for (e = b + 1; e != n && s[e] != d1 && s[e] != d2; ++e) ;
+
+    return e - b;
+  }
+
+  inline bool
+  eof (std::istream& is)
+  {
+    if (!is.fail ())
+      return false;
+
+    if (is.eof ())
+      return true;
+
+    throw std::istream::failure ("");
+  }
 }
diff --git a/libbutl/utility.mxx b/libbutl/utility.mxx
index 4c83827..b957282 100644
--- a/libbutl/utility.mxx
+++ b/libbutl/utility.mxx
@@ -57,21 +57,6 @@ LIBBUTL_MODEXPORT namespace butl
   [[noreturn]] LIBBUTL_SYMEXPORT void
   throw_system_error (int system_code, int fallback_errno_code = 0);
 
-  // If an input stream is in a failed state, then return true if this is
-  // because of the eof and throw istream::failure otherwise. If the stream
-  // is not in a failed state, return false. This helper function is normally
-  // used like this:
-  //
-  // is.exceptions (istream::badbit);
-  //
-  // for (string l; !eof (getline (is, l)); )
-  // {
-  //   ...
-  // }
-  //
-  bool
-  eof (std::istream&);
-
   // Convert ASCII character/string case. If there is no upper/lower case
   // counterpart, leave the character unchanged. The POSIX locale (also known
   // as C locale) must be the current application locale. Otherwise the
@@ -149,6 +134,55 @@ LIBBUTL_MODEXPORT namespace butl
   bool alnum  (wchar_t);
   bool xdigit (wchar_t);
 
+  // Basic string utilities.
+  //
+
+  // Trim leading/trailing whitespacec, including '\r'.
+  //
+  std::string&
+  trim (std::string&);
+
+  // Find the beginning and end poistions of the next word. Return the size
+  // of the word or 0 and set b = e = n if there are no more words. For
+  // example:
+  //
+  // for (size_t b (0), e (0); next_word (s, b, e); )
+  // {
+  //   string w (s, b, e - b);
+  // }
+  //
+  // Or:
+  //
+  // for (size_t b (0), e (0), n; n = next_word (s, b, e, ' ', ','); )
+  // {
+  //   string w (s, b, n);
+  // }
+  //
+  // The second version examines up to the n'th character in the string.
+  //
+  std::size_t
+  next_word (const std::string&, std::size_t& b, std::size_t& e,
+             char d1 = ' ', char d2 = '\0');
+
+  std::size_t
+  next_word (const std::string&, std::size_t n, std::size_t& b, std::size_t& e,
+             char d1 = ' ', char d2 = '\0');
+
+  // If an input stream is in a failed state, then return true if this is
+  // because of the eof and throw istream::failure otherwise. If the stream
+  // is not in a failed state, return false. This helper function is normally
+  // used like this:
+  //
+  // is.exceptions (istream::badbit);
+  //
+  // for (string l; !eof (getline (is, l)); )
+  // {
+  //   ...
+  // }
+  //
+  bool
+  eof (std::istream&);
+
   // Key comparators (i.e., to be used in sets, maps, etc).
   //
   struct compare_c_string