* src/coding.c (detect_coding_iso_2022): Reorganize code to clarify

structure.
author: Andreas Schwab 2011-03-15 19:38:57 +0100
committer: Andreas Schwab 2011-03-15 19:38:57 +0100
commit: 0e48bb227a5b9cdabeb845422de33d62ccb1edc5 (patch)
tree: 1555e62e7146144b2c5dc344e55e7e1fb570ec7f /src/coding.c
parent: 0adf561883e07549e657a39d0a4a95cafa4d04fd (diff)
download: emacs-0e48bb227a5b9cdabeb845422de33d62ccb1edc5.tar.gz
emacs-0e48bb227a5b9cdabeb845422de33d62ccb1edc5.zip
1 files changed, 80 insertions, 79 deletions
diff --git a/src/coding.c b/src/coding.c
index 9a6a4484e50..0c2836c19f6 100644
--- a/src/coding.c
+++ b/src/coding.c
@@ -2954,12 +2954,7 @@ detect_coding_iso_2022 (struct coding_system *coding,
  const unsigned char *src_end = coding->source + coding->src_bytes;
  int multibytep = coding->src_multibyte;
  int single_shifting = 0;
+  int id;
-  /* FIXME: Does ID need to be initialized here?  The "End of composition"
-     code below does not initialize ID even though ID is used
-     afterwards, and perhaps that is a bug.  */
-  int id = 0;
  int c, c1;
  int consumed_chars = 0;
  int i;
@@ -2999,40 +2994,11 @@ detect_coding_iso_2022 (struct coding_system *coding,
            break;
          single_shifting = 0;
          ONE_MORE_BYTE (c);
-          if (c >= '(' && c <= '/')
+          if (c == 'N' || c == 'O')
-            {
-              /* Designation sequence for a charset of dimension 1.  */
-              ONE_MORE_BYTE (c1);
-              if (c1 < ' ' || c1 >= 0x80
-                  || (id = iso_charset_table[0][c >= ','][c1]) < 0)
-                /* Invalid designation sequence.  Just ignore.  */
-                break;
-            }
-          else if (c == '$')
-            {
-              /* Designation sequence for a charset of dimension 2.  */
-              ONE_MORE_BYTE (c);
-              if (c >= '@' && c <= 'B')
-                /* Designation for JISX0208.1978, GB2312, or JISX0208.  */
-                id = iso_charset_table[1][0][c];
-              else if (c >= '(' && c <= '/')
-                {
-                  ONE_MORE_BYTE (c1);
-                  if (c1 < ' ' || c1 >= 0x80
-                      || (id = iso_charset_table[1][c >= ','][c1]) < 0)
-                    /* Invalid designation sequence.  Just ignore.  */
-                    break;
-                }
-              else
-                /* Invalid designation sequence.  Just ignore it.  */
-                break;
-            }
-          else if (c == 'N' || c == 'O')
            {
              /* ESC <Fe> for SS2 or SS3.  */
              single_shifting = 1;
              rejected |= CATEGORY_MASK_ISO_7BIT | CATEGORY_MASK_ISO_8BIT;
-              break;
            }
          else if (c == '1')
            {
@@ -3048,36 +3014,66 @@ detect_coding_iso_2022 (struct coding_system *coding,
            {
              /* ESC <Fp> for start/end composition.  */
              composition_count = 0;
-              break;
            }
          else
            {
-              /* Invalid escape sequence.  Just ignore it.  */
+              if (c >= '(' && c <= '/')
-              break;
+                {
-            }
+                  /* Designation sequence for a charset of dimension 1.  */
+                  ONE_MORE_BYTE (c1);
+                  if (c1 < ' ' || c1 >= 0x80
+                      || (id = iso_charset_table[0][c >= ','][c1]) < 0)
+                    /* Invalid designation sequence.  Just ignore.  */
+                    break;
+                }
+              else if (c == '$')
+                {
+                  /* Designation sequence for a charset of dimension 2.  */
+                  ONE_MORE_BYTE (c);
+                  if (c >= '@' && c <= 'B')
+                    /* Designation for JISX0208.1978, GB2312, or JISX0208.  */
+                    id = iso_charset_table[1][0][c];
+                  else if (c >= '(' && c <= '/')
+                    {
+                      ONE_MORE_BYTE (c1);
+                      if (c1 < ' ' || c1 >= 0x80
+                          || (id = iso_charset_table[1][c >= ','][c1]) < 0)
+                        /* Invalid designation sequence.  Just ignore.  */
+                        break;
+                    }
+                  else
+                    /* Invalid designation sequence.  Just ignore it.  */
+                    break;
+                }
+              else
+                {
+                  /* Invalid escape sequence.  Just ignore it.  */
+                  break;
+                }
-          /* We found a valid designation sequence for CHARSET.  */
+              /* We found a valid designation sequence for CHARSET.  */
-          rejected |= CATEGORY_MASK_ISO_8BIT;
+              rejected |= CATEGORY_MASK_ISO_8BIT;
-          if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7],
+              if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7],
-                              id))
+                                  id))
-            found |= CATEGORY_MASK_ISO_7;
+                found |= CATEGORY_MASK_ISO_7;
-          else
+              else
-            rejected |= CATEGORY_MASK_ISO_7;
+                rejected |= CATEGORY_MASK_ISO_7;
-          if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_tight],
+              if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_tight],
-                              id))
+                                  id))
-            found |= CATEGORY_MASK_ISO_7_TIGHT;
+                found |= CATEGORY_MASK_ISO_7_TIGHT;
-          else
+              else
-            rejected |= CATEGORY_MASK_ISO_7_TIGHT;
+                rejected |= CATEGORY_MASK_ISO_7_TIGHT;
-          if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_else],
+              if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_else],
-                              id))
+                                  id))
-            found |= CATEGORY_MASK_ISO_7_ELSE;
+                found |= CATEGORY_MASK_ISO_7_ELSE;
-          else
+              else
-            rejected |= CATEGORY_MASK_ISO_7_ELSE;
+                rejected |= CATEGORY_MASK_ISO_7_ELSE;
-          if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_8_else],
+              if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_8_else],
-                              id))
+                                  id))
-            found |= CATEGORY_MASK_ISO_8_ELSE;
+                found |= CATEGORY_MASK_ISO_8_ELSE;
-          else
+              else
-            rejected |= CATEGORY_MASK_ISO_8_ELSE;
+                rejected |= CATEGORY_MASK_ISO_8_ELSE;
+            }
          break;
        case ISO_CODE_SO:
@@ -3105,13 +3101,32 @@ detect_coding_iso_2022 (struct coding_system *coding,
          rejected |= CATEGORY_MASK_ISO_7BIT;
          if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])
              & CODING_ISO_FLAG_SINGLE_SHIFT)
-            found |= CATEGORY_MASK_ISO_8_1, single_shifting = 1;
+            {
+              found |= CATEGORY_MASK_ISO_8_1;
+              single_shifting = 1;
+            }
          if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_2])
              & CODING_ISO_FLAG_SINGLE_SHIFT)
-            found |= CATEGORY_MASK_ISO_8_2, single_shifting = 1;
+            {
+              found |= CATEGORY_MASK_ISO_8_2;
+              single_shifting = 1;
+            }
          if (single_shifting)
            break;
-          goto check_extra_latin;
+        check_extra_latin:
+          if (! VECTORP (Vlatin_extra_code_table)
+              || NILP (XVECTOR (Vlatin_extra_code_table)->contents[c]))
+            {
+              rejected = CATEGORY_MASK_ISO;
+              break;
+            }
+          if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])
+              & CODING_ISO_FLAG_LATIN_EXTRA)
+            found |= CATEGORY_MASK_ISO_8_1;
+          else
+            rejected |= CATEGORY_MASK_ISO_8_1;
+          rejected |= CATEGORY_MASK_ISO_8_2;
+          break;
        default:
          if (c < 0)
@@ -3162,20 +3177,6 @@ detect_coding_iso_2022 (struct coding_system *coding,
                }
              break;
            }
-        check_extra_latin:
-          single_shifting = 0;
-          if (! VECTORP (Vlatin_extra_code_table)
-              || NILP (XVECTOR (Vlatin_extra_code_table)->contents[c]))
-            {
-              rejected = CATEGORY_MASK_ISO;
-              break;
-            }
-          if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])
-              & CODING_ISO_FLAG_LATIN_EXTRA)
-            found |= CATEGORY_MASK_ISO_8_1;
-          else
-            rejected |= CATEGORY_MASK_ISO_8_1;
-          rejected |= CATEGORY_MASK_ISO_8_2;
        }
    }
  detect_info->rejected |= CATEGORY_MASK_ISO;
author	Andreas Schwab	2011-03-15 19:38:57 +0100
committer	Andreas Schwab	2011-03-15 19:38:57 +0100
commit	0e48bb227a5b9cdabeb845422de33d62ccb1edc5 (patch)
tree	1555e62e7146144b2c5dc344e55e7e1fb570ec7f /src/coding.c
parent	0adf561883e07549e657a39d0a4a95cafa4d04fd (diff)
download	emacs-0e48bb227a5b9cdabeb845422de33d62ccb1edc5.tar.gz emacs-0e48bb227a5b9cdabeb845422de33d62ccb1edc5.zip

diff --git a/src/coding.c b/src/coding.c index 9a6a4484e50..0c2836c19f6 100644 --- a/src/coding.c +++ b/src/coding.c
@@ -2954,12 +2954,7 @@ detect_coding_iso_2022 (struct coding_system *coding,
2954	const unsigned char *src_end = coding->source + coding->src_bytes;	2954	const unsigned char *src_end = coding->source + coding->src_bytes;
2955	int multibytep = coding->src_multibyte;	2955	int multibytep = coding->src_multibyte;
2956	int single_shifting = 0;	2956	int single_shifting = 0;
2957		2957	int id;
2958	/* FIXME: Does ID need to be initialized here? The "End of composition"
2959	code below does not initialize ID even though ID is used
2960	afterwards, and perhaps that is a bug. */
2961	int id = 0;
2962
2963	int c, c1;	2958	int c, c1;
2964	int consumed_chars = 0;	2959	int consumed_chars = 0;
2965	int i;	2960	int i;
@@ -2999,40 +2994,11 @@ detect_coding_iso_2022 (struct coding_system *coding,
2999	break;	2994	break;
3000	single_shifting = 0;	2995	single_shifting = 0;
3001	ONE_MORE_BYTE (c);	2996	ONE_MORE_BYTE (c);
3002	if (c >= '(' && c <= '/')	2997	if (c == 'N' \|\| c == 'O')
3003	{
3004	/* Designation sequence for a charset of dimension 1. */
3005	ONE_MORE_BYTE (c1);
3006	if (c1 < ' ' \|\| c1 >= 0x80
3007	\|\| (id = iso_charset_table[0][c >= ','][c1]) < 0)
3008	/* Invalid designation sequence. Just ignore. */
3009	break;
3010	}
3011	else if (c == '$')
3012	{
3013	/* Designation sequence for a charset of dimension 2. */
3014	ONE_MORE_BYTE (c);
3015	if (c >= '@' && c <= 'B')
3016	/* Designation for JISX0208.1978, GB2312, or JISX0208. */
3017	id = iso_charset_table[1][0][c];
3018	else if (c >= '(' && c <= '/')
3019	{
3020	ONE_MORE_BYTE (c1);
3021	if (c1 < ' ' \|\| c1 >= 0x80
3022	\|\| (id = iso_charset_table[1][c >= ','][c1]) < 0)
3023	/* Invalid designation sequence. Just ignore. */
3024	break;
3025	}
3026	else
3027	/* Invalid designation sequence. Just ignore it. */
3028	break;
3029	}
3030	else if (c == 'N' \|\| c == 'O')
3031	{	2998	{
3032	/* ESC <Fe> for SS2 or SS3. */	2999	/* ESC <Fe> for SS2 or SS3. */
3033	single_shifting = 1;	3000	single_shifting = 1;
3034	rejected \|= CATEGORY_MASK_ISO_7BIT \| CATEGORY_MASK_ISO_8BIT;	3001	rejected \|= CATEGORY_MASK_ISO_7BIT \| CATEGORY_MASK_ISO_8BIT;
3035	break;
3036	}	3002	}
3037	else if (c == '1')	3003	else if (c == '1')
3038	{	3004	{
@@ -3048,36 +3014,66 @@ detect_coding_iso_2022 (struct coding_system *coding,
3048	{	3014	{
3049	/* ESC <Fp> for start/end composition. */	3015	/* ESC <Fp> for start/end composition. */
3050	composition_count = 0;	3016	composition_count = 0;
3051	break;
3052	}	3017	}
3053	else	3018	else
3054	{	3019	{
3055	/* Invalid escape sequence. Just ignore it. */	3020	if (c >= '(' && c <= '/')
3056	break;	3021	{
3057	}	3022	/* Designation sequence for a charset of dimension 1. */
		3023	ONE_MORE_BYTE (c1);
		3024	if (c1 < ' ' \|\| c1 >= 0x80
		3025	\|\| (id = iso_charset_table[0][c >= ','][c1]) < 0)
		3026	/* Invalid designation sequence. Just ignore. */
		3027	break;
		3028	}
		3029	else if (c == '$')
		3030	{
		3031	/* Designation sequence for a charset of dimension 2. */
		3032	ONE_MORE_BYTE (c);
		3033	if (c >= '@' && c <= 'B')
		3034	/* Designation for JISX0208.1978, GB2312, or JISX0208. */
		3035	id = iso_charset_table[1][0][c];
		3036	else if (c >= '(' && c <= '/')
		3037	{
		3038	ONE_MORE_BYTE (c1);
		3039	if (c1 < ' ' \|\| c1 >= 0x80
		3040	\|\| (id = iso_charset_table[1][c >= ','][c1]) < 0)
		3041	/* Invalid designation sequence. Just ignore. */
		3042	break;
		3043	}
		3044	else
		3045	/* Invalid designation sequence. Just ignore it. */
		3046	break;
		3047	}
		3048	else
		3049	{
		3050	/* Invalid escape sequence. Just ignore it. */
		3051	break;
		3052	}
3058		3053
3059	/* We found a valid designation sequence for CHARSET. */	3054	/* We found a valid designation sequence for CHARSET. */
3060	rejected \|= CATEGORY_MASK_ISO_8BIT;	3055	rejected \|= CATEGORY_MASK_ISO_8BIT;
3061	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7],	3056	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7],
3062	id))	3057	id))
3063	found \|= CATEGORY_MASK_ISO_7;	3058	found \|= CATEGORY_MASK_ISO_7;
3064	else	3059	else
3065	rejected \|= CATEGORY_MASK_ISO_7;	3060	rejected \|= CATEGORY_MASK_ISO_7;
3066	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_tight],	3061	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_tight],
3067	id))	3062	id))
3068	found \|= CATEGORY_MASK_ISO_7_TIGHT;	3063	found \|= CATEGORY_MASK_ISO_7_TIGHT;
3069	else	3064	else
3070	rejected \|= CATEGORY_MASK_ISO_7_TIGHT;	3065	rejected \|= CATEGORY_MASK_ISO_7_TIGHT;
3071	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_else],	3066	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_7_else],
3072	id))	3067	id))
3073	found \|= CATEGORY_MASK_ISO_7_ELSE;	3068	found \|= CATEGORY_MASK_ISO_7_ELSE;
3074	else	3069	else
3075	rejected \|= CATEGORY_MASK_ISO_7_ELSE;	3070	rejected \|= CATEGORY_MASK_ISO_7_ELSE;
3076	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_8_else],	3071	if (SAFE_CHARSET_P (&coding_categories[coding_category_iso_8_else],
3077	id))	3072	id))
3078	found \|= CATEGORY_MASK_ISO_8_ELSE;	3073	found \|= CATEGORY_MASK_ISO_8_ELSE;
3079	else	3074	else
3080	rejected \|= CATEGORY_MASK_ISO_8_ELSE;	3075	rejected \|= CATEGORY_MASK_ISO_8_ELSE;
		3076	}
3081	break;	3077	break;
3082		3078
3083	case ISO_CODE_SO:	3079	case ISO_CODE_SO:
@@ -3105,13 +3101,32 @@ detect_coding_iso_2022 (struct coding_system *coding,
3105	rejected \|= CATEGORY_MASK_ISO_7BIT;	3101	rejected \|= CATEGORY_MASK_ISO_7BIT;
3106	if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])	3102	if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])
3107	& CODING_ISO_FLAG_SINGLE_SHIFT)	3103	& CODING_ISO_FLAG_SINGLE_SHIFT)
3108	found \|= CATEGORY_MASK_ISO_8_1, single_shifting = 1;	3104	{
		3105	found \|= CATEGORY_MASK_ISO_8_1;
		3106	single_shifting = 1;
		3107	}
3109	if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_2])	3108	if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_2])
3110	& CODING_ISO_FLAG_SINGLE_SHIFT)	3109	& CODING_ISO_FLAG_SINGLE_SHIFT)
3111	found \|= CATEGORY_MASK_ISO_8_2, single_shifting = 1;	3110	{
		3111	found \|= CATEGORY_MASK_ISO_8_2;
		3112	single_shifting = 1;
		3113	}
3112	if (single_shifting)	3114	if (single_shifting)
3113	break;	3115	break;
3114	goto check_extra_latin;	3116	check_extra_latin:
		3117	if (! VECTORP (Vlatin_extra_code_table)
		3118	\|\| NILP (XVECTOR (Vlatin_extra_code_table)->contents[c]))
		3119	{
		3120	rejected = CATEGORY_MASK_ISO;
		3121	break;
		3122	}
		3123	if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])
		3124	& CODING_ISO_FLAG_LATIN_EXTRA)
		3125	found \|= CATEGORY_MASK_ISO_8_1;
		3126	else
		3127	rejected \|= CATEGORY_MASK_ISO_8_1;
		3128	rejected \|= CATEGORY_MASK_ISO_8_2;
		3129	break;
3115		3130
3116	default:	3131	default:
3117	if (c < 0)	3132	if (c < 0)
@@ -3162,20 +3177,6 @@ detect_coding_iso_2022 (struct coding_system *coding,
3162	}	3177	}
3163	break;	3178	break;
3164	}	3179	}
3165	check_extra_latin:
3166	single_shifting = 0;
3167	if (! VECTORP (Vlatin_extra_code_table)
3168	\|\| NILP (XVECTOR (Vlatin_extra_code_table)->contents[c]))
3169	{
3170	rejected = CATEGORY_MASK_ISO;
3171	break;
3172	}
3173	if (CODING_ISO_FLAGS (&coding_categories[coding_category_iso_8_1])
3174	& CODING_ISO_FLAG_LATIN_EXTRA)
3175	found \|= CATEGORY_MASK_ISO_8_1;
3176	else
3177	rejected \|= CATEGORY_MASK_ISO_8_1;
3178	rejected \|= CATEGORY_MASK_ISO_8_2;
3179	}	3180	}
3180	}	3181	}
3181	detect_info->rejected \|= CATEGORY_MASK_ISO;	3182	detect_info->rejected \|= CATEGORY_MASK_ISO;