w3m

commit 5af6784ee134004c99c1670373bda6c271e52769
parent 8ba13c66798d5acf1da756da3837dc6edc2fe729
Author: ukai <ukai>
Date:   Thu, 23 Jan 2003 18:01:01 +0000

* etc.c (next_status): after = is R_ST_VALUE
		R_ST_VALUE > => R_ST_NORMAL
		R_ST_VALUE SP => R_ST_TAG
	(read_token): R_ST_VALUE
	(correct_irrtag): R_ST_VALUE
* rm.h (R_ST_VALUE): added
	(ST_IS_REAL_TAG): rewrite
	(ST_IS_COMMENT): deleted
	(ST_IS_TAG): deleted
* parsetagx.c (parse_tag): skip too long tagname
			skip too long attrname
			if attrvalue has quote char, need reconstruct
			if unknown attr, need reconstruct
* table.c (visible_length): R_ST_VALUE
From: Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>

Diffstat:
M
ChangeLog
 | 
17
+++++++++++++++++
M
etc.c
 | 
10
+++++++++-
M
fm.h
 | 
5
++---
M
parsetagx.c
 | 
20
++++++++++++++------
M
table.c
 | 
3
++-

5 files changed, 44 insertions(+), 11 deletions(-)
diff --git a/ChangeLog b/ChangeLog
@@ -1,5 +1,22 @@
 2003-01-24  Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>
 
+	* etc.c (next_status): after = is R_ST_VALUE
+			R_ST_VALUE > => R_ST_NORMAL
+			R_ST_VALUE SP => R_ST_TAG
+		(read_token): R_ST_VALUE
+		(correct_irrtag): R_ST_VALUE
+	* rm.h (R_ST_VALUE): added
+		(ST_IS_REAL_TAG): rewrite
+		(ST_IS_COMMENT): deleted
+		(ST_IS_TAG): deleted
+	* parsetagx.c (parse_tag): skip too long tagname
+				skip too long attrname
+				if attrvalue has quote char, need reconstruct
+				if unknown attr, need reconstruct
+	* table.c (visible_length): R_ST_VALUE
+
+2003-01-24  Hironori SAKAMOTO <hsaka@mth.biglobe.ne.jp>
+
 	* [w3m-dev 03680] showProgress() on fast system
 	* file.c (showProgress): check line len, time
 
diff --git a/etc.c b/etc.c
@@ -636,7 +636,7 @@ next_status(char c, int *status)
 	else if (c == '>')
 	    *status = R_ST_NORMAL;
 	else
-	    *status = R_ST_TAG;
+	    *status = R_ST_VALUE;
 	return 0;
     case R_ST_QUOTE:
 	if (c == '\'')
@@ -646,6 +646,12 @@ next_status(char c, int *status)
 	if (c == '"')
 	    *status = R_ST_TAG;
 	return 0;
+    case R_ST_VALUE:
+	if (c == '>')
+	    *status = R_ST_NORMAL;
+	else if (IS_SPACE(c))
+	    *status = R_ST_TAG;
+	return 0;
     case R_ST_AMP:
 	if (c == ';') {
 	    *status = R_ST_NORMAL;
@@ -792,6 +798,7 @@ read_token(Str buf, char **instr, int *status, int pre, int append)
 	case R_ST_EQL:
 	case R_ST_QUOTE:
 	case R_ST_DQUOTE:
+	case R_ST_VALUE:
 	case R_ST_AMP:
 	    Strcat_char(buf, *p);
 	    break;
@@ -838,6 +845,7 @@ correct_irrtag(int status)
 	case R_ST_TAG:
 	case R_ST_TAG0:
 	case R_ST_EQL:		/* required ">" */
+	case R_ST_VALUE:
 	    c = '>';
 	    break;
 	case R_ST_QUOTE:
diff --git a/fm.h b/fm.h
@@ -623,10 +623,9 @@ struct readbuffer {
 #define R_ST_NCMNT2 12		/* comment -- */
 #define R_ST_NCMNT3 13		/* comment -- space */
 #define R_ST_IRRTAG 14		/* within irregular tag */
+#define R_ST_VALUE  15		/* within tag attribule value */
 
-#define ST_IS_REAL_TAG(s)   ((s)==R_ST_TAG||(s)==R_ST_TAG0||(s)==R_ST_EQL)
-#define ST_IS_COMMENT(s)    ((s)>=R_ST_CMNT1)
-#define ST_IS_TAG(s)        ((s)!=R_ST_NORMAL&&(s)!=R_ST_AMP&&!ST_IS_COMMENT(s)&&(s)!=R_ST_EOL)
+#define ST_IS_REAL_TAG(s)   ((s)==R_ST_TAG||(s)==R_ST_TAG0||(s)==R_ST_EQL||(s)==R_ST_VALUE)
 
 /* is this '<' really means the beginning of a tag? */
 #define REALLY_THE_BEGINNING_OF_A_TAG(p) \
diff --git a/parsetagx.c b/parsetagx.c
@@ -132,6 +132,9 @@ parse_tag(char **s, int internal)
 	q++;
     }
     *p = '\0';
+    while (*q && !IS_SPACE(*q) && !(tagname[0] != '/' && *q == '/') &&
+	   *q != '>')
+	q++;
 
     tag_id = getHash_si(&tagtable, tagname, HTML_UNKNOWN);
 
@@ -165,11 +168,9 @@ parse_tag(char **s, int internal)
 	    *(p++) = TOLOWER(*q);
 	    q++;
 	}
-	if (q == p) {
-	    q++;
-	    continue;
-	}
 	*p = '\0';
+	while (*q && *q != '=' && !IS_SPACE(*q) && *q != '>')
+	    q++;
 	SKIP_BLANKS(q);
 	if (*q == '=') {
 	    /* get value */
@@ -201,8 +202,12 @@ parse_tag(char **s, int internal)
 		    q++;
 	    }
 	    else if (*q) {
-		while (*q && !IS_SPACE(*q) && *q != '>')
-		    Strcat_char(value, *q++);
+		while (*q && !IS_SPACE(*q) && *q != '>') {
+		    Strcat_char(value, *q);
+		    if (!tag->need_reconstruct && html_quote_char(*q))
+			tag->need_reconstruct = TRUE;
+		    q++;
+		}
 	    }
 	}
 	for (i = 0; i < nattr; i++) {
@@ -227,6 +232,9 @@ parse_tag(char **s, int internal)
 	    else
 		tag->value[i] = NULL;
 	}
+	else {
+	    tag->need_reconstruct = TRUE;
+	}
     }
 
   skip_parse_tagarg:
diff --git a/table.c b/table.c
@@ -484,7 +484,8 @@ visible_length(char *str)
 	    Strcat_char(tagbuf, *str);
 	}
 	else if (status == R_ST_TAG || status == R_ST_DQUOTE
-		 || status == R_ST_QUOTE || status == R_ST_EQL) {
+		 || status == R_ST_QUOTE || status == R_ST_EQL
+		 || status == R_ST_VALUE) {
 	    Strcat_char(tagbuf, *str);
 	}
 	else if (status == R_ST_AMP) {