Optimize the code for reading UTF-8 files.

2 files changed, 178 insertions, 44 deletions

diff --git a/src/ChangeLog b/src/ChangeLog
index 0634ec7cc1d..6b3ca9d3ff3 100644
--- a/src/ChangeLog
+++ b/src/ChangeLog
@@ -1,3 +1,17 @@
+2013-04-03  Kenichi Handa  <handa@gnu.org>
+
+        The following changes is to optimize the code for reading UTF-8
+        files.
+
+        * coding.c (check_ascii): Renamed from detect_ascii.  Return value
+        changed.  Check EOL format.  Do not call adjust_coding_eol_type
+        here.
+        (check_utf_8): New function.
+        (adjust_coding_eol_type): Do nothing if already adjusted.
+        (detect_coding): Compare the return value of check_ascii with
+        coding->src_bytes.  Call adjust_coding_eol_type if necessary.
+        (decode_coding_gap): Optimize for valid UTF-8.
+
 2013-03-21  Kenichi Handa  <handa@gnu.org>
 
         * coding.c (syms_of_coding): Cancel previous change.
@@ -89,7 +103,7 @@
 
         * coding.c (decode_coding_gap): Fix typo caught by static checking.
 
-2013-03-15  handa  <handa@gnu.org>
+2013-03-15  Kenichi Handa  <handa@gnu.org>
 
         * insdel.c (insert_from_gap): New arg text_at_gap_tail.
         (adjust_after_replace): Make it back to static.  Delete the third
diff --git a/src/coding.c b/src/coding.c
index 8a09cd67859..735af25502d 100644
--- a/src/coding.c
+++ b/src/coding.c
@@ -6072,17 +6072,18 @@ complement_process_encoding_system (Lisp_Object coding_system)
 #define EOL_SEEN_CRLF   4
 
 
-static Lisp_Object adjust_coding_eol_type (struct coding_system *coding, int eol_seen);
+static Lisp_Object adjust_coding_eol_type (struct coding_system *coding,
+                                           int eol_seen);
 
 
-/* Return true iff all the source bytes are ASCII.
+/* Return the number of ASCII characters at the head of the source.
    By side effects, set coding->head_ascii and coding->eol_seen.  The
    value of coding->eol_seen is "logical or" of EOL_SEEN_LF,
    EOL_SEEN_CR, and EOL_SEEN_CRLF, but the value is reliable only when
    all the source bytes are ASCII.  */
 
-static bool
-detect_ascii (struct coding_system *coding)
+static int
+check_ascii (struct coding_system *coding)
 {
   const unsigned char *src, *end;
   Lisp_Object eol_type = CODING_ID_EOL_TYPE (coding->id);
@@ -6096,21 +6097,20 @@ detect_ascii (struct coding_system *coding)
   src = coding->source;
   end = src + coding->src_bytes;
 
-  if (inhibit_eol_conversion)
+  if (inhibit_eol_conversion
+      || eol_seen != EOL_SEEN_NONE)
     {
       /* We don't have to check EOL format.  */
       while (src < end && !( *src & 0x80)) src++;
-      eol_seen = EOL_SEEN_LF;
-      adjust_coding_eol_type (coding, eol_seen);
-    }
-  else if (eol_seen != EOL_SEEN_NONE)
-    {
-      /* We don't have to check EOL format either.  */
-      while (src < end && !(*src & 0x80)) src++;
+      if (inhibit_eol_conversion)
+        {
+          eol_seen = EOL_SEEN_LF;
+          adjust_coding_eol_type (coding, eol_seen);
+        }
     }
   else
     {
-      end--;                    /* We look ahead one byte.  */
+      end--;                /* We look ahead one byte for "CR LF".  */
       while (src < end)
         {
           int c = *src;
@@ -6118,6 +6118,69 @@ detect_ascii (struct coding_system *coding)
           if (c & 0x80)
             break;
           src++;
+          if (c == '\r')
+            {
+              if (*src == '\n')
+                {
+                  eol_seen |= EOL_SEEN_CRLF;
+                  src++;
+                }
+              else
+                eol_seen |= EOL_SEEN_CR;
+            }
+          else if (c == '\n')
+            eol_seen |= EOL_SEEN_LF;
+        }
+      if (src == end)
+        {
+          int c = *src;
+
+          /* All bytes but the last one C are ASCII.  */
+          if (! (c & 0x80))
+            {
+              if (c == '\r')
+                eol_seen |= EOL_SEEN_CR;
+              else if (c  == '\n')
+                eol_seen |= EOL_SEEN_LF;
+              src++;
+            }
+        }
+    }
+  coding->head_ascii = src - coding->source;
+  coding->eol_seen = eol_seen;
+  return (coding->head_ascii);
+}
+
+
+/* Return the number of charcters at the source if all the bytes are
+   valid UTF-8 (of Unicode range).  Otherwise, return -1.  By side
+   effects, update coding->eol_seen.  The value of coding->eol_seen is
+   "logical or" of EOL_SEEN_LF, EOL_SEEN_CR, and EOL_SEEN_CRLF, but
+   the value is reliable only when all the source bytes are valid
+   UTF-8.  */
+
+static int
+check_utf_8 (struct coding_system *coding)
+{
+  const unsigned char *src, *end;
+  int eol_seen = coding->eol_seen;
+  int nchars = coding->head_ascii;
+
+  if (coding->head_ascii < 0)
+    check_ascii (coding);
+  else
+    coding_set_source (coding);
+  src = coding->source + coding->head_ascii;
+  /* We look ahead one byte for CR LF.  */
+  end = coding->source + coding->src_bytes - 1;
+
+  while (src < end)
+    {
+      int c = *src;
+
+      if (UTF_8_1_OCTET_P (*src))
+        {
+          src++;
           if (c < 0x20)
             {
               if (c == '\r')
@@ -6126,6 +6189,7 @@ detect_ascii (struct coding_system *coding)
                     {
                       eol_seen |= EOL_SEEN_CRLF;
                       src++;
+                      nchars++;
                     }
                   else
                     eol_seen |= EOL_SEEN_CR;
@@ -6134,27 +6198,58 @@ detect_ascii (struct coding_system *coding)
                 eol_seen |= EOL_SEEN_LF;
             }
         }
-      if (src > end)
-        /* The last two bytes are CR LF, which means that we have
-           scanned all bytes. */
-        end++;
-      else if (src == end)
+      else if (UTF_8_2_OCTET_LEADING_P (c))
         {
-          end++;
-          if (! (*src & 0x80))
-            {
-              if (*src == '\r')
-                eol_seen |= EOL_SEEN_CR;
-              else if (*src  == '\n')
-                eol_seen |= EOL_SEEN_LF;
-              src++;
-            }
+          if (c < 0xC2          /* overlong sequence */
+              || src + 1 >= end
+              || ! UTF_8_EXTRA_OCTET_P (src[1]))
+            return -1;
+          src += 2;
         }
-      adjust_coding_eol_type (coding, eol_seen);
+      else if (UTF_8_3_OCTET_LEADING_P (c))
+        {
+          if (src + 2 >= end
+              || ! (UTF_8_EXTRA_OCTET_P (src[1])
+                    && UTF_8_EXTRA_OCTET_P (src[2])))
+            return -1;
+          c = (((c & 0xF) << 12)
+               | ((src[1] & 0x3F) << 6) | (src[2] & 0x3F));
+          if (c < 0x800                       /* overlong sequence */
+              || (c >= 0xd800 && c < 0xe000)) /* surrogates (invalid) */
+            return -1;
+          src += 3;
+        }
+      else if (UTF_8_4_OCTET_LEADING_P (c))
+        {
+          if (src + 3 >= end
+              || ! (UTF_8_EXTRA_OCTET_P (src[1])
+                    && UTF_8_EXTRA_OCTET_P (src[2])
+                    && UTF_8_EXTRA_OCTET_P (src[3])))
+            return -1;
+          c = (((c & 0x7) << 18) | ((src[1] & 0x3F) << 12)
+               | ((src[2] & 0x3F) << 6) | (src[3] & 0x3F));
+          if (c < 0x10000       /* overlong sequence */
+              || c >= 0x110000) /* non-Unicode character  */
+            return -1;
+          src += 4;
+        }
+      else
+        return -1;
+      nchars++;
+    }
+
+  if (src == end)
+    {
+      if (! UTF_8_1_OCTET_P (*src))
+        return -1;
+      nchars++;
+      if (*src == '\r')
+        eol_seen |= EOL_SEEN_CR;
+      else if (*src  == '\n')
+        eol_seen |= EOL_SEEN_LF;
     }
-  coding->head_ascii = src - coding->source;
   coding->eol_seen = eol_seen;
-  return (src == end);
+  return nchars;
 }
 
 
@@ -6269,6 +6364,9 @@ adjust_coding_eol_type (struct coding_system *coding, int eol_seen)
   Lisp_Object eol_type;
 
   eol_type = CODING_ID_EOL_TYPE (coding->id);
+  if (! VECTORP (eol_type))
+    /* Already adjusted.  */
+    return eol_type;
   if (eol_seen & EOL_SEEN_LF)
     {
       coding->id = CODING_SYSTEM_ID (AREF (eol_type, 0));
@@ -6360,7 +6458,8 @@ detect_coding (struct coding_system *coding)
                         {
                           coding->eol_seen |= EOL_SEEN_CRLF;
                           src++;
-                          coding->head_ascii++;
+                          if (! eight_bit_found)
+                            coding->head_ascii++;
                         }
                       else
                         coding->eol_seen |= EOL_SEEN_CR;
@@ -6461,9 +6560,14 @@ detect_coding (struct coding_system *coding)
       coding_systems
         = AREF (CODING_ID_ATTRS (coding->id), coding_attr_utf_bom);
       detect_info.found = detect_info.rejected = 0;
-      if (detect_ascii (coding))
+      if (check_ascii (coding) == coding->src_bytes)
         {
+          int head_ascii = coding->head_ascii;
+
+          if (coding->eol_seen != EOL_SEEN_NONE)
+            adjust_coding_eol_type (coding, coding->eol_seen);
           setup_coding_system (XCDR (coding_systems), coding);
+          coding->head_ascii = head_ascii;
         }
       else
         {
@@ -7620,15 +7724,27 @@ decode_coding_gap (struct coding_system *coding,
   if (CODING_REQUIRE_DETECTION (coding))
     detect_coding (coding);
   attrs = CODING_ID_ATTRS (coding->id);
-  if (! disable_ascii_optimization)
-    {
-      if (! NILP (CODING_ATTR_ASCII_COMPAT (attrs))
-          && NILP (CODING_ATTR_POST_READ (attrs))
-          && NILP (get_translation_table (attrs, 0, NULL))
-          && (coding->head_ascii >= 0 /* We've already called detect_coding */
-              ? coding->head_ascii == bytes
-              : detect_ascii (coding)))
+  if (! disable_ascii_optimization
+      && ! coding->src_multibyte
+      && ! NILP (CODING_ATTR_ASCII_COMPAT (attrs))
+      && NILP (CODING_ATTR_POST_READ (attrs))
+      && NILP (get_translation_table (attrs, 0, NULL)))
+    {
+      chars = coding->head_ascii;
+      if (chars < 0)
+        chars = check_ascii (coding);
+      if (chars != bytes)
+        {
+          if (EQ (CODING_ATTR_TYPE (attrs), Qutf_8))
+            chars = check_utf_8 (coding);
+          else
+            chars = -1;
+        }
+      if (chars >= 0)
         {
+          if (coding->eol_seen != EOL_SEEN_NONE)
+            adjust_coding_eol_type (coding, coding->eol_seen);
+
           if (coding->eol_seen == EOL_SEEN_CR)
             {
               unsigned char *src_end = GAP_END_ADDR;
@@ -7645,6 +7761,7 @@ decode_coding_gap (struct coding_system *coding,
               unsigned char *src = GAP_END_ADDR;
               unsigned char *src_beg = src - coding->src_bytes;
               unsigned char *dst = src;
+              ptrdiff_t diff;
 
               while (src_beg < src)
                 {
@@ -7652,10 +7769,13 @@ decode_coding_gap (struct coding_system *coding,
                   if (*src == '\n')
                     src--;
                 }
-              bytes -= dst - src;
+              diff = dst - src;
+              bytes -= diff;
+              chars -= diff;
             }
-          coding->produced_char = coding->produced = bytes;
-          insert_from_gap (bytes, bytes, 1);
+          coding->produced = bytes;
+          coding->produced_char = chars;
+          insert_from_gap (chars, bytes, 1);
           return;
         }
     }