emacs: src/coding.c comparison

comparison src/coding.c @ 95533:831c8ee4d884

(detect_coding): Fix handling of coding->head_ascii. Be sure to call setup_coding_system when a proper coding system is found. (detect_coding_system): Fix handling of coding->head_ascii.

author	Kenichi Handa <handa@m17n.org>
date	Wed, 04 Jun 2008 07:52:46 +0000
parents	c99f0a16c077
children	862c7386145c

comparison

equal deleted inserted replaced

-:b657ba21e4d3
+:831c8ee4d884
 coding->consumed = coding->consumed_char = 0;
 coding->produced = coding->produced_char = 0;
 coding_set_source (coding);
 src_end = coding->source + coding->src_bytes;
+coding->head_ascii = 0;
 /* If we have not yet decided the text encoding type, detect it
 now.  */
 if (EQ (CODING_ATTR_TYPE (CODING_ID_ATTRS (coding->id)), Qundecided))
 {
 int c, i;
 struct coding_detection_info detect_info;
 int null_byte_found = 0, eight_bit_found = 0;
 detect_info.checked = detect_info.found = detect_info.rejected = 0;
-coding->head_ascii = -1;
 for (src = coding->source; src < src_end; src++)
 	{
 	  c = *src;
 	  if (c & 0x80)
 	    {
 	      eight_bit_found = 1;
-	      if (coding->head_ascii < 0)
-		coding->head_ascii = src - coding->source;
 	      if (null_byte_found)
 		break;
 	    }
 	  else if (c < 0x20)
 	    {
 	      if ((c == ISO_CODE_ESC || c == ISO_CODE_SI || c == ISO_CODE_SO)
 		  && ! inhibit_iso_escape_detection
 		  && ! detect_info.checked)
 		{
-		  if (coding->head_ascii < 0)
-		    coding->head_ascii = src - coding->source;
 		  if (detect_coding_iso_2022 (coding, &detect_info))
 		    {
 		      /* We have scanned the whole data.  */
 		      if (! (detect_info.rejected & CATEGORY_MASK_ISO_7_ELSE))
-			/* We didn't find an 8-bit code.  We may have
+			{
-			   found a null-byte, but it's very rare that
+			  /* We didn't find an 8-bit code.  We may
-			   a binary file confirm to ISO-2022.  */
+			     have found a null-byte, but it's very
-			src = src_end;
+			     rare that a binary file confirm to
+			     ISO-2022.  */
+			  src = src_end;
+			  coding->head_ascii = src - coding->source;
+			}
+		      detect_info.rejected |= ~CATEGORY_MASK_ISO_ESCAPE;
 		      break;
 		    }
 		}
 	      else if (! c)
 		{
 		  null_byte_found = 1;
 		  if (eight_bit_found)
 		    break;
 		}
+	      coding->head_ascii++;
 	    }
-	}
+	  else
-if (coding->head_ascii < 0)
+	    coding->head_ascii++;
-	coding->head_ascii = src - coding->source;
+	}
 if (null_byte_found || eight_bit_found
 	  || coding->head_ascii < coding->src_bytes
 	  || detect_info.found)
 	{
 			    category = coding_category_utf_16_be;
 			}
 		      break;
 		    }
 		}
-	      if (i < coding_category_raw_text)
-		setup_coding_system (CODING_ID_NAME (this->id), coding);
-	      else if (null_byte_found)
-		setup_coding_system (Qno_conversion, coding);
-	      else if ((detect_info.rejected & CATEGORY_MASK_ANY)
-		       == CATEGORY_MASK_ANY)
-		setup_coding_system (Qraw_text, coding);
-	      else if (detect_info.rejected)
-		for (i = 0; i < coding_category_raw_text; i++)
-		  if (! (detect_info.rejected & (1 << coding_priorities[i])))
-		    {
-		      this = coding_categories + coding_priorities[i];
-		      setup_coding_system (CODING_ID_NAME (this->id), coding);
-		      break;
-		    }
 	    }
+	  if (i < coding_category_raw_text)
+	    setup_coding_system (CODING_ID_NAME (this->id), coding);
+	  else if (null_byte_found)
+	    setup_coding_system (Qno_conversion, coding);
+	  else if ((detect_info.rejected & CATEGORY_MASK_ANY)
+		   == CATEGORY_MASK_ANY)
+	    setup_coding_system (Qraw_text, coding);
+	  else if (detect_info.rejected)
+	    for (i = 0; i < coding_category_raw_text; i++)
+	      if (! (detect_info.rejected & (1 << coding_priorities[i])))
+		{
+		  this = coding_categories + coding_priorities[i];
+		  setup_coding_system (CODING_ID_NAME (this->id), coding);
+		  break;
+		}
 	}
 }
 else if (XINT (CODING_ATTR_CATEGORY (CODING_ID_ATTRS (coding->id)))
 	   == coding_category_utf_8_auto)
 {
 coding.src_chars = src_chars;
 coding.src_bytes = src_bytes;
 coding.src_multibyte = multibytep;
 coding.consumed = 0;
 coding.mode |= CODING_MODE_LAST_BLOCK;
+coding.head_ascii = 0;
 detect_info.checked = detect_info.found = detect_info.rejected = 0;
 /* At first, detect text-format if necessary.  */
 base_category = XINT (CODING_ATTR_CATEGORY (attrs));
 {
 enum coding_category category;
 struct coding_system *this;
 int c, i;
-coding.head_ascii = -1;
 /* Skip all ASCII bytes except for a few ISO2022 controls.  */
 for (; src < src_end; src++)
 	{
 	  c = *src;
 	  if (c & 0x80)
 	    {
 	      eight_bit_found = 1;
-	      if (coding.head_ascii < 0)
-		coding.head_ascii = src - coding.source;
 	      if (null_byte_found)
 		break;
 	    }
-	  if (c < 0x20)
+	  else if (c < 0x20)
 	    {
 	      if ((c == ISO_CODE_ESC || c == ISO_CODE_SI || c == ISO_CODE_SO)
 		  && ! inhibit_iso_escape_detection
 		  && ! detect_info.checked)
 		{
-		  if (coding.head_ascii < 0)
-		    coding.head_ascii = src - coding.source;
 		  if (detect_coding_iso_2022 (&coding, &detect_info))
 		    {
 		      /* We have scanned the whole data.  */
 		      if (! (detect_info.rejected & CATEGORY_MASK_ISO_7_ELSE))
-			/* We didn't find an 8-bit code.  We may have
+			{
-			   found a null-byte, but it's very rare that
+			  /* We didn't find an 8-bit code.  We may
-			   a binary file confirm to ISO-2022.  */
+			     have found a null-byte, but it's very
-			src = src_end;
+			     rare that a binary file confirm to
+			     ISO-2022.  */
+			  src = src_end;
+			  coding.head_ascii = src - coding.source;
+			}
+		      detect_info.rejected |= ~CATEGORY_MASK_ISO_ESCAPE;
 		      break;
 		    }
 		}
 	      else if (! c)
 		{
 		  null_byte_found = 1;
 		  if (eight_bit_found)
 		    break;
 		}
+	      coding.head_ascii++;
 	    }
-	}
+	  else
-if (coding.head_ascii < 0)
+	    coding.head_ascii++;
-	coding.head_ascii = src - coding.source;
+	}
 if (null_byte_found || eight_bit_found
 	  || coding.head_ascii < coding.src_bytes
 	  || detect_info.found)
 	{

Mercurial > emacs

comparison src/coding.c @ 95533:831c8ee4d884