From: Andy Wingo <wingo@pobox.com>
Date: Fri, 24 Feb 2012 22:05:02 +0000 (+0100)
Subject: port i/o optimizations for iso-8859-1
X-Git-Url: http://git.hcoop.net/bpt/guile.git/commitdiff_plain/79eb47ea47650ef42c545931726277a7118a0210

port i/o optimizations for iso-8859-1

* libguile/ports.h (scm_t_port_encoding_mode):
* libguile/ports.c (scm_c_make_port_with_encoding):
  (scm_i_set_port_encoding_x): Add special treatment for latin1
  encoding.
  (get_latin1_codepoint, get_codepoint): Add latin1 fast-path.

* libguile/print.c (display_string_as_latin1): Add latin1 fastpath.
---

diff --git a/libguile/ports.c b/libguile/ports.c
index 5b98bf9f7..e17ea069d 100644
--- a/libguile/ports.c
+++ b/libguile/ports.c
@@ -605,6 +605,8 @@ scm_c_make_port_with_encoding (scm_t_bits tag, unsigned long mode_bits,
   entry->encoding = encoding ? scm_gc_strdup (encoding, "port") : NULL;
   if (encoding && strcmp (encoding, "UTF-8") == 0)
     entry->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8;
+  else if (!encoding || strcmp (encoding, "ISO-8859-1") == 0)
+    entry->encoding_mode = SCM_PORT_ENCODING_MODE_LATIN1;
   else
     entry->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV;
   entry->ilseq_handler = handler;
@@ -941,15 +943,18 @@ scm_i_set_port_encoding_x (SCM port, const char *encoding)
   pt = SCM_PTAB_ENTRY (port);
   prev = pt->iconv_descriptors;
 
-  if (encoding == NULL)
-    encoding = "ISO-8859-1";
-
-  if (strcmp (encoding, "UTF-8") == 0)
+  if (encoding && strcmp (encoding, "UTF-8") == 0)
     {
       pt->encoding = "UTF-8";
       pt->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8;
       pt->iconv_descriptors = NULL;
     }
+  else if (!encoding || strcmp (encoding, "ISO-8859-1") == 0)
+    {
+      pt->encoding = "ISO-8859-1";
+      pt->encoding_mode = SCM_PORT_ENCODING_MODE_LATIN1;
+      pt->iconv_descriptors = NULL;
+    }
   else
     {
       /* Open descriptors before mutating the port. */
@@ -1582,6 +1587,26 @@ get_utf8_codepoint (SCM port, scm_t_wchar *codepoint,
 #undef ASSERT_NOT_EOF
 }
 
+/* Read an ISO-8859-1 codepoint (a byte) from PORT.  On success, return
+   *0 and set CODEPOINT to the codepoint that was read, fill BUF with
+   *its UTF-8 representation, and set *LEN to the length in bytes.
+   *Return `EILSEQ' on error.  */
+static int
+get_latin1_codepoint (SCM port, scm_t_wchar *codepoint,
+                      char buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
+{
+  *codepoint = scm_get_byte_or_eof_unlocked (port);
+
+  if (*codepoint == EOF)
+    *len = 0;
+  else
+    {
+      *len = 1;
+      buf[0] = *codepoint;
+    }
+  return 0;
+}
+
 /* Likewise, read a byte sequence from PORT, passing it through its
    input conversion descriptor.  */
 static int
@@ -1662,6 +1687,8 @@ get_codepoint (SCM port, scm_t_wchar *codepoint,
 
   if (pt->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8)
     err = get_utf8_codepoint (port, codepoint, (scm_t_uint8 *) buf, len);
+  else if (pt->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1)
+    err = get_latin1_codepoint (port, codepoint, buf, len);
   else
     err = get_iconv_codepoint (port, codepoint, buf, len);
 
diff --git a/libguile/ports.h b/libguile/ports.h
index c42b501fd..2d277e031 100644
--- a/libguile/ports.h
+++ b/libguile/ports.h
@@ -50,6 +50,7 @@ typedef enum scm_t_port_rw_active {
 
 typedef enum scm_t_port_encoding_mode {
   SCM_PORT_ENCODING_MODE_UTF8,
+  SCM_PORT_ENCODING_MODE_LATIN1,
   SCM_PORT_ENCODING_MODE_ICONV
 } scm_t_port_encoding_mode;
 
diff --git a/libguile/print.c b/libguile/print.c
index a1bf5eded..eb601322e 100644
--- a/libguile/print.c
+++ b/libguile/print.c
@@ -853,6 +853,54 @@ display_string_as_utf8 (const void *str, int narrow_p, size_t len,
   return len;
 }
 
+/* Write STR to PORT as ISO-8859-1.  STR is a LEN-codepoint string; it
+   is narrow if NARROW_P is true, wide otherwise.  Return LEN.  */
+static size_t
+display_string_as_latin1 (const void *str, int narrow_p, size_t len,
+                          SCM port,
+                          scm_t_string_failed_conversion_handler strategy)
+{
+  size_t printed = 0;
+
+  if (narrow_p)
+    {
+      scm_lfwrite_unlocked (str, len, port);
+      return len;
+    }
+
+  while (printed < len)
+    {
+      char buf[256];
+      size_t i;
+
+      for (i = 0; i < sizeof(buf) && printed < len; i++, printed++)
+        {
+          scm_t_wchar c = STR_REF (str, printed);
+
+          if (c < 256)
+            buf[i] = c;
+          else
+            break;
+        }
+
+      scm_lfwrite_unlocked (buf, i, port);
+
+      if (i < sizeof(buf) && printed < len)
+        {
+          if (strategy == SCM_FAILED_CONVERSION_ERROR)
+            break;
+          else if (strategy == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
+            write_character_escaped (STR_REF (str, printed), 1, port);
+          else
+            /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'.  */
+            display_string ("?", 1, 1, port, strategy);
+          printed++;
+        }
+    }
+
+  return printed;
+}
+
 /* Convert STR through PORT's output conversion descriptor and write the
    output to PORT.  Return the number of codepoints written.  */
 static size_t
@@ -968,9 +1016,10 @@ display_string (const void *str, int narrow_p,
 
   if (pt->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8)
     return display_string_as_utf8 (str, narrow_p, len, port);
+  else if (pt->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1)
+    return display_string_as_latin1 (str, narrow_p, len, port, strategy);
   else
-    return display_string_using_iconv (str, narrow_p, len,
-				       port, strategy);
+    return display_string_using_iconv (str, narrow_p, len, port, strategy);
 }
 
 /* Attempt to display CH to PORT according to STRATEGY.  Return non-zero