From 3e346a2a845320c19beab6c281fe4b7c14b88a4a Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Tue, 1 Nov 2022 15:45:41 +0200
Subject: Add getline_non_blocking(ifdstream&)

---
 libbutl/fdstream.cxx | 61 +++++++++++++++++++++++++++++++++++++++++++++++-----
 libbutl/fdstream.hxx | 50 +++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 105 insertions(+), 6 deletions(-)

diff --git a/libbutl/fdstream.cxx b/libbutl/fdstream.cxx
index 03c007c..b4fe4b6 100644
--- a/libbutl/fdstream.cxx
+++ b/libbutl/fdstream.cxx
@@ -41,7 +41,7 @@
 #include <new>          // bad_alloc
 #include <limits>       // numeric_limits
 #include <cassert>
-#include <cstring>      // memcpy(), memmove()
+#include <cstring>      // memcpy(), memmove(), memchr()
 #include <iostream>     // cin, cout
 #include <exception>    // uncaught_exception[s]()
 #include <stdexcept>    // invalid_argument
@@ -846,7 +846,7 @@ namespace butl
   }
 
   ifdstream&
-  getline (ifdstream& is, string& s, char delim)
+  getline (ifdstream& is, string& l, char delim)
   {
     ifdstream::iostate eb (is.exceptions ());
     assert (eb & ifdstream::badbit);
@@ -863,7 +863,7 @@ namespace butl
     if (eb != ifdstream::badbit)
       is.exceptions (ifdstream::badbit);
 
-    std::getline (is, s, delim);
+    std::getline (is, l, delim);
 
     // Throw if any of the newly set bits are present in the exception mask.
     //
@@ -876,6 +876,55 @@ namespace butl
     return is;
   }
 
+  bool
+  getline_non_blocking (ifdstream& is, std::string& l, char delim)
+  {
+    assert (!is.blocking () && (is.exceptions () &ifdstream::badbit) != 0);
+
+    fdstreambuf& sb (*static_cast<fdstreambuf*> (is.rdbuf ()));
+
+    // Read until blocked (0), EOF (-1) or encounter the delimiter.
+    //
+    // Note that here we reasonably assume that any failure in in_avail()
+    // will lead to badbit and thus an exception (see showmanyc()).
+    //
+    streamsize s;
+    while ((s = sb.in_avail ()) > 0)
+    {
+      const char* p (sb.gptr ());
+      size_t n (sb.egptr () - p);
+
+      const char* e (static_cast<const char*> (memchr (p, delim, n)));
+      if (e != nullptr)
+        n = e - p;
+
+      l.append (p, n);
+
+      // Note: consume the delimiter if found.
+      //
+      sb.gbump (static_cast<int> (n + (e != nullptr ? 1 : 0)));
+
+      if (e != nullptr)
+        break;
+    }
+
+    // Here s can be:
+    //
+    // -1 -- EOF.
+    //  0 -- blocked before encountering delimiter/EOF.
+    // >0 -- encountered the delimiter.
+    //
+    if (s == -1 && l.empty ())
+    {
+      // If we couldn't extract anything, not even the delimiter, then this is
+      // a failure per the getline() interface.
+      //
+      is.setstate (ifdstream::failbit);
+    }
+
+    return s != 0;
+  }
+
   // ofdstream
   //
   ofdstream::
@@ -1412,6 +1461,8 @@ namespace butl
 
       for (fdselect_state& s: from)
       {
+        s.ready = false;
+
         if (s.fd == nullfd)
           continue;
 
@@ -1419,7 +1470,6 @@ namespace butl
           throw invalid_argument ("invalid file descriptor");
 
         FD_SET (s.fd, &to);
-        s.ready = false;
 
         if (max_fd < s.fd)
           max_fd = s.fd;
@@ -1907,13 +1957,14 @@ namespace butl
 
     for (fdselect_state& s: read)
     {
+      s.ready = false;
+
       if (s.fd == nullfd)
         continue;
 
       if (s.fd < 0)
         throw invalid_argument ("invalid file descriptor");
 
-      s.ready = false;
       ++n;
     }
 
diff --git a/libbutl/fdstream.hxx b/libbutl/fdstream.hxx
index 730d4dd..1a8f9a6 100644
--- a/libbutl/fdstream.hxx
+++ b/libbutl/fdstream.hxx
@@ -652,6 +652,54 @@ namespace butl
   LIBBUTL_SYMEXPORT ifdstream&
   getline (ifdstream&, std::string&, char delim = '\n');
 
+  // The non-blocking getline() version that reads the line in potentially
+  // multiple calls. Key differences compared to getline():
+  //
+  // - Stream must be in the non-blocking mode and exception mask must have
+  //   at least badbit.
+  //
+  // - Return type is bool instead of stream. Return true if the line has been
+  //   read or false if it should be called again once the stream has more
+  //   data to read. Also return true on failure.
+  //
+  // - The string must be empty on the first call.
+  //
+  // - There could still be data to read in the stream's buffer (as opposed to
+  //   file descriptor) after this function returns true and you should be
+  //   careful not to block on fdselect() in this case. In fact, the
+  //   recommended pattern is to call this function first and only call
+  //   fdselect() if it returns false.
+  //
+  // The typical usage in combination with the eof() helper:
+  //
+  // fdselect_set fds {is.fd (), ...};
+  // fdselect_state& ist (fds[0]);
+  // fdselect_state& ...;
+  //
+  // for (string l; ist.fd != nullfd || ...; )
+  // {
+  //   if (ist.fd != nullfd && getline_non_blocking (is, l))
+  //   {
+  //     if (eof (is))
+  //       ist.fd = nullfd;
+  //     else
+  //     {
+  //       // Consume line.
+  //
+  //       l.clear ();
+  //     }
+  //
+  //     continue;
+  //   }
+  //
+  //   ifdselect (fds);
+  //
+  //   // Handle other ready fds.
+  // }
+  //
+  LIBBUTL_SYMEXPORT bool
+  getline_non_blocking (ifdstream&, std::string&, char delim = '\n');
+
   // Open a file returning an auto_fd that holds its file descriptor on
   // success and throwing ios::failure otherwise.
   //
@@ -858,7 +906,7 @@ namespace butl
   // underlying OS error.
   //
   // Note that the function clears all the previously-ready entries on each
-  // call. Entries with nullfd are ignored.
+  // call. Entries with nullfd are ignored (but cleared).
   //
   // On Windows only pipes and only their input (read) ends are supported.
   //
-- 
cgit v1.1