released 4.0.1

fix #277 permit legacy compilation; faster option -l for compressed files; enable an optimization that wasn't enabled

lib/matcher.cpp

@@ -2566,7 +2566,7 @@ bool Matcher::advance()
       }
     }
 #endif
-    if (min >= 2 && pat_->npy_ < 16)
+    if (pat_->npy_ < 16)
     {
       if (min >= 4)
       {
@@ -2695,6 +2695,40 @@ bool Matcher::advance()
           }
         }
       }
+      const Pattern::Pred *bit = pat_->bit_;
+      while (true)
+      {
+        const char *s = buf_ + loc;
+        const char *e = buf_ + end_ - 3;
+        bool f = true;
+        while (s < e &&
+            (f = ((bit[static_cast<uint8_t>(*s)] & 1) &&
+                  (bit[static_cast<uint8_t>(*++s)] & 1) &&
+                  (bit[static_cast<uint8_t>(*++s)] & 1) &&
+                  (bit[static_cast<uint8_t>(*++s)] & 1))))
+        {
+          ++s;
+        }
+        loc = s - buf_;
+        if (!f)
+        {
+          if (s < e && Pattern::predict_match(pma, s))
+          {
+            ++loc;
+            continue;
+          }
+          set_current(loc);
+          return true;
+        }
+        set_current_match(loc - 1);
+        (void)peek_more();
+        loc = cur_ + 1;
+        if (loc + 3 >= end_)
+        {
+          set_current(loc);
+          return false;
+        }
+      }
     }
     if (min >= 4)
     {

lib/pattern.cpp

@@ -347,6 +347,12 @@ void Pattern::init(const char *options, const uint8_t *pred)
         }
       }
     }
+    // only one position to pin
+    if (min_ == 1)
+    {
+      nlcs = nlcp;
+      lcs_ = lcp_;
+    }
     // number of needles required
     uint16_t n = nlcp > nlcs ? nlcp : nlcs;
     // determine if a needle-based search is worthwhile, below or meeting the thresholds

lib/simd_avx512bw.cpp

@@ -49,7 +49,7 @@ size_t simd_nlcount_avx512bw(const char*& b, const char *e)
     return 0;
   size_t n = 0;
   // align on 64 bytes
-  while ((reinterpret_cast<ptrdiff_t>(s) & 0x3f) != 0)
+  while ((reinterpret_cast<std::ptrdiff_t>(s) & 0x3f) != 0)
     n += (*s++ == '\n');
   __m512i vlcn = _mm512_set1_epi8('\n');
   while (s <= e)

src/ugrep.cpp

@@ -1066,7 +1066,13 @@ struct Zthread {
           {
             // write buffer data to the pipe, if the pipe is broken then the receiver is waiting for this thread to join so we drain the rest of the decompressed data
             if (is_selected && !drain && write(pipe_fd[1], buf, static_cast<size_t>(len)) < len)
+            {
+              // if no next decompression thread and decompressing a single file (not zip), then stop immediately
+              if (ztchain == NULL && zipinfo == NULL)
+                break;
+
               drain = true;
+            }
 
             // decompress the next block of data into the buffer
             len = zstream->decompress(buf, maxlen);

src/ugrep.hpp

@@ -38,7 +38,7 @@
 #define UGREP_HPP
 
 // ugrep version
-#define UGREP_VERSION "4.0.0"
+#define UGREP_VERSION "4.0.1"
 
 // disable mmap because mmap is almost always slower than the file reading speed improvements since 3.0.0
 #define WITH_NO_MMAP