From a50ba784b61e7e1c534b10ad23b62290ee707203 Mon Sep 17 00:00:00 2001
From: Jeff Bezanson <bezanson@post.harvard.edu>
Date: Wed, 11 Apr 2012 22:24:48 -0400
Subject: [PATCH] line reading performance tweaks (issue #661)

---
 base/io.jl  |  9 ++++-----
 src/julia.h |  2 +-
 src/sys.c   | 42 ++++++++++++++++++++++++++++++------------
 3 files changed, 35 insertions(+), 18 deletions(-)

diff --git a/base/io.jl b/base/io.jl
index 1de6475c70d94..0a4704349a169 100644
--- a/base/io.jl
+++ b/base/io.jl
@@ -234,10 +234,9 @@ function read{T<:Union(Int8,Uint8,Int16,Uint16,Int32,Uint32,Int64,Uint64,Float32
     end
 end
 
-function readuntil(s::IOStream, delim::Uint8)
-    a = ccall(:jl_readuntil, Any, (Ptr{Void}, Uint8), s.ios, delim)
-    # TODO: faster versions that avoid this encoding check
-    ccall(:jl_array_to_string, Any, (Any,), a)::ByteString
+function readuntil(s::IOStream, delim)
+    # TODO: faster versions that avoid the encoding check
+    ccall(:jl_readuntil, Any, (Ptr{Void}, Uint8), s.ios, delim)
 end
 
 function readall(s::IOStream)
@@ -246,7 +245,7 @@ function readall(s::IOStream)
     takebuf_string(dest)
 end
 
-readline(s::IOStream) = readuntil(s, uint8('\n'))
+readline(s::IOStream) = readuntil(s, '\n')
 
 flush(s::IOStream) = ccall(:ios_flush, Void, (Ptr{Void},), s.ios)
 
diff --git a/src/julia.h b/src/julia.h
index 4cbf4e9617a12..bbd35bf838f51 100644
--- a/src/julia.h
+++ b/src/julia.h
@@ -920,7 +920,7 @@ DLLEXPORT void jl_set_current_output_stream_obj(jl_value_t *v);
 
 DLLEXPORT jl_array_t *jl_takebuf_array(ios_t *s);
 DLLEXPORT jl_value_t *jl_takebuf_string(ios_t *s);
-DLLEXPORT jl_array_t *jl_readuntil(ios_t *s, uint8_t delim);
+DLLEXPORT jl_value_t *jl_readuntil(ios_t *s, uint8_t delim);
 
 static inline void jl_eh_restore_state(jl_savestate_t *ss)
 {
diff --git a/src/sys.c b/src/sys.c
index a2f731d218731..54b498b708f64 100644
--- a/src/sys.c
+++ b/src/sys.c
@@ -153,22 +153,40 @@ jl_value_t *jl_takebuf_string(ios_t *s)
     return str;
 }
 
-jl_array_t *jl_readuntil(ios_t *s, uint8_t delim)
+jl_value_t *jl_readuntil(ios_t *s, uint8_t delim)
 {
-    jl_array_t *a = jl_alloc_array_1d(jl_array_uint8_type, 80);
-    ios_t dest;
-    jl_ios_mem(&dest, 0);
-    ios_setbuf(&dest, a->data, 80, 0);
-    size_t n = ios_copyuntil(&dest, s, delim);
-    if (dest.buf != a->data) {
-        return jl_takebuf_array(&dest);
+    jl_array_t *a;
+    // manually inlined common case
+    char *pd = (char*)memchr(s->buf+s->bpos, delim, s->size - s->bpos);
+    if (pd) {
+        size_t n = pd-(s->buf+s->bpos)+1;
+        a = jl_alloc_array_1d(jl_array_uint8_type, n);
+        memcpy(jl_array_data(a), s->buf+s->bpos, n);
+        s->bpos += n;
     }
     else {
-        a->length = n;
-        a->nrows = n;
-        ((char*)a->data)[n] = '\0';
+        a = jl_alloc_array_1d(jl_array_uint8_type, 80);
+        ios_t dest;
+        jl_ios_mem(&dest, 0);
+        ios_setbuf(&dest, a->data, 80, 0);
+        size_t n = ios_copyuntil(&dest, s, delim);
+        if (dest.buf != a->data) {
+            a = jl_takebuf_array(&dest);
+        }
+        else {
+            a->length = n;
+            a->nrows = n;
+            ((char*)a->data)[n] = '\0';
+        }
     }
-    return a;
+    JL_GC_PUSH(&a);
+    jl_struct_type_t* string_type = u8_isvalid(a->data, a->length) == 1 ? // ASCII
+        jl_ascii_string_type : jl_utf8_string_type;
+    jl_value_t *str = alloc_2w();
+    str->type = (jl_type_t*)string_type;
+    jl_fieldref(str,0) = (jl_value_t*)a;
+    JL_GC_POP();
+    return str;
 }
 
 // -- syscall utilities --