On file insertion, skip decoding if all bytes are ASCII.

[bpt/emacs.git] / src / coding.c

diff --git a/src/coding.c b/src/coding.c
index 868fb7d..f33b5e7 100644 (file)
--- a/src/coding.c
+++ b/src/coding.c
@@ -322,8 +322,7 @@ Lisp_Object Qcall_process, Qcall_process_region;
 Lisp_Object Qstart_process, Qopen_network_stream;
 static Lisp_Object Qtarget_idx;
 
-static Lisp_Object Qinsufficient_source, Qinconsistent_eol, Qinvalid_source;
-static Lisp_Object Qinterrupted, Qinsufficient_memory;
+static Lisp_Object Qinsufficient_source, Qinvalid_source, Qinterrupted;
 
 /* If a symbol has this property, evaluate the value to define the
    symbol as a coding system.  */
@@ -820,18 +819,12 @@ record_conversion_result (struct coding_system *coding,
     case CODING_RESULT_INSUFFICIENT_SRC:
       Vlast_code_conversion_error = Qinsufficient_source;
       break;
-    case CODING_RESULT_INCONSISTENT_EOL:
-      Vlast_code_conversion_error = Qinconsistent_eol;
-      break;
     case CODING_RESULT_INVALID_SRC:
       Vlast_code_conversion_error = Qinvalid_source;
       break;
     case CODING_RESULT_INTERRUPT:
       Vlast_code_conversion_error = Qinterrupted;
       break;
-    case CODING_RESULT_INSUFFICIENT_MEM:
-      Vlast_code_conversion_error = Qinsufficient_memory;
-      break;
     case CODING_RESULT_INSUFFICIENT_DST:
       /* Don't record this error in Vlast_code_conversion_error
         because it happens just temporarily and is resolved when the
@@ -6356,7 +6349,12 @@ 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;
-      coding->head_ascii = 0;
+      for (src = coding->source; src < src_end; src++)
+       {
+         if (*src & 0x80)
+           break;
+       }
+      coding->head_ascii = src - coding->source;
       if (CONSP (coding_systems)
          && detect_coding_utf_8 (coding, &detect_info))
        {
@@ -6884,22 +6882,8 @@ produce_charset (struct coding_system *coding, int *charbuf, ptrdiff_t pos)
 
 #define ALLOC_CONVERSION_WORK_AREA(coding)                             \
   do {                                                                 \
-    int size = CHARBUF_SIZE;                                           \
-                                                                       \
-    coding->charbuf = NULL;                                            \
-    while (size > 1024)                                                        \
-      {                                                                        \
-       coding->charbuf = alloca (sizeof (int) * size);                 \
-       if (coding->charbuf)                                            \
-         break;                                                        \
-       size >>= 1;                                                     \
-      }                                                                        \
-    if (! coding->charbuf)                                             \
-      {                                                                        \
-       record_conversion_result (coding, CODING_RESULT_INSUFFICIENT_MEM); \
-       return;                                                         \
-      }                                                                        \
-    coding->charbuf_size = size;                                       \
+    coding->charbuf = SAFE_ALLOCA (CHARBUF_SIZE * sizeof (int));       \
+    coding->charbuf_size = CHARBUF_SIZE;                               \
   } while (0)
 
 
@@ -6968,6 +6952,8 @@ decode_coding (struct coding_system *coding)
   int carryover;
   int i;
 
+  USE_SAFE_ALLOCA;
+
   if (BUFFERP (coding->src_object)
       && coding->src_pos > 0
       && coding->src_pos < GPT
@@ -7090,6 +7076,8 @@ decode_coding (struct coding_system *coding)
       bset_undo_list (current_buffer, undo_list);
       record_insert (coding->dst_pos, coding->produced_char);
     }
+
+  SAFE_FREE ();
 }
 
 
@@ -7373,6 +7361,8 @@ encode_coding (struct coding_system *coding)
   int max_lookup;
   struct ccl_spec cclspec;
 
+  USE_SAFE_ALLOCA;
+
   attrs = CODING_ID_ATTRS (coding->id);
   if (coding->encoder == encode_coding_raw_text)
     translation_table = Qnil, max_lookup = 0;
@@ -7407,6 +7397,8 @@ encode_coding (struct coding_system *coding)
 
   if (BUFFERP (coding->dst_object) && coding->produced_char > 0)
     insert_from_gap (coding->produced_char, coding->produced);
+
+  SAFE_FREE ();
 }
 
 
@@ -7500,8 +7492,6 @@ decode_coding_gap (struct coding_system *coding,
   ptrdiff_t count = SPECPDL_INDEX ();
   Lisp_Object attrs;
 
-  code_conversion_save (0, 0);
-
   coding->src_object = Fcurrent_buffer ();
   coding->src_chars = chars;
   coding->src_bytes = bytes;
@@ -7515,13 +7505,45 @@ decode_coding_gap (struct coding_system *coding,
 
   if (CODING_REQUIRE_DETECTION (coding))
     detect_coding (coding);
+  attrs = CODING_ID_ATTRS (coding->id);
+#ifndef CODING_DISABLE_ASCII_OPTIMIZATION
+  if (! NILP (CODING_ATTR_ASCII_COMPAT (attrs))
+      && NILP (CODING_ATTR_POST_READ (attrs))
+      && NILP (get_translation_table (attrs, 0, NULL)))
+    {
+      /* We can skip the conversion if all source bytes are ASCII.  */
+      if (coding->head_ascii < 0)
+       {
+         /* We have not yet counted the number of ASCII bytes at the
+            head of the source.  Do it now.  */
+         const unsigned char *src, *src_end;
+
+         coding_set_source (coding);
+         src_end = coding->source + coding->src_bytes;
+         for (src = coding->source; src < src_end; src++)
+           {
+             if (*src & 0x80)
+               break;
+           }
+         coding->head_ascii = src - coding->source;
+       }
+      if (coding->src_bytes == coding->head_ascii)
+       {
+         /* No need of conversion.  Use the data in the gap as is.  */
+         coding->produced_char = chars;
+         coding->produced = bytes;
+         adjust_after_replace (PT, PT_BYTE, Qnil, chars, bytes, 1);
+         return;
+       }
+    }
+#endif /* not CODING_DISABLE_ASCII_OPTIMIZATION */
+  code_conversion_save (0, 0);
 
   coding->mode |= CODING_MODE_LAST_BLOCK;
   current_buffer->text->inhibit_shrinking = 1;
   decode_coding (coding);
   current_buffer->text->inhibit_shrinking = 0;
 
-  attrs = CODING_ID_ATTRS (coding->id);
   if (! NILP (CODING_ATTR_POST_READ (attrs)))
     {
       ptrdiff_t prev_Z = Z, prev_Z_BYTE = Z_BYTE;
@@ -7695,14 +7717,8 @@ decode_coding_object (struct coding_system *coding,
       set_buffer_internal (XBUFFER (coding->dst_object));
       if (dst_bytes < coding->produced)
        {
+         eassert (coding->produced > 0);
          destination = xrealloc (destination, coding->produced);
-         if (! destination)
-           {
-             record_conversion_result (coding,
-                                       CODING_RESULT_INSUFFICIENT_MEM);
-             unbind_to (count, Qnil);
-             return;
-           }
          if (BEGV < GPT && GPT < BEGV + coding->produced_char)
            move_gap_both (BEGV, BEGV_BYTE);
          memcpy (destination, BEGV_ADDR, coding->produced);
@@ -10408,10 +10424,8 @@ syms_of_coding (void)
        intern_c_string ("coding-category-undecided"));
 
   DEFSYM (Qinsufficient_source, "insufficient-source");
-  DEFSYM (Qinconsistent_eol, "inconsistent-eol");
   DEFSYM (Qinvalid_source, "invalid-source");
   DEFSYM (Qinterrupted, "interrupted");
-  DEFSYM (Qinsufficient_memory, "insufficient-memory");
   DEFSYM (Qcoding_system_define_form, "coding-system-define-form");
 
   defsubr (&Scoding_system_p);