antville/helma
antville/helma
1
1
Fork 0
Code Issues 5 Pull requests 7 Projects Releases 5 Activity Actions

Implement basic HTML tag attribute recognition in order not to interpret

Browse source 
<a href=http://foo/> as empty tag.
This commit is contained in:
hns 2003-06-26 12:24:45 +00:00
parent 1edfed4583
commit df7d736c91
1 changed files with 41 additions and 3 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

42
src/helma/util/HtmlEncoder.java
View file

@ -342,6 +342,11 @@ public final class HtmlEncoder {
emptyTags.add("param"); emptyTags.add("param");
} }
final static byte TAG_NAME = 0;
final static byte TAG_SPACE = 1;
final static byte TAG_ATT_NAME = 2;
final static byte TAG_ATT_VAL = 3;
/** /**
* Do "smart" encodging on a string. This means that valid HTML entities and tags, * Do "smart" encodging on a string. This means that valid HTML entities and tags,
@ -397,6 +402,8 @@ public final class HtmlEncoder {
// are we inside an HTML tag? // are we inside an HTML tag?
boolean insideHtmlTag = false; boolean insideHtmlTag = false;
boolean insideCloseTag = false;
byte htmlTagMode = TAG_NAME;
// if we are inside a <code> tag, we encode everything to make // if we are inside a <code> tag, we encode everything to make
// documentation work easier // documentation work easier
@ -445,7 +452,7 @@ public final class HtmlEncoder {
} }
} else if (!insideTag) { } else if (!insideTag) {
// check if this is a HTML tag. // check if this is a HTML tag.
boolean insideCloseTag = ('/' == str.charAt(i + 1)); insideCloseTag = ('/' == str.charAt(i + 1));
int tagStart = insideCloseTag ? (i + 2) : (i + 1); int tagStart = insideCloseTag ? (i + 2) : (i + 1);
int j = tagStart; int j = tagStart;
@ -464,6 +471,7 @@ public final class HtmlEncoder {
allTags.contains(tagName) && !insideCodeTag) { allTags.contains(tagName) && !insideCodeTag) {
insideHtmlTag = insideTag = true; insideHtmlTag = insideTag = true;
htmlQuoteChar = '\u0000'; htmlQuoteChar = '\u0000';
htmlTagMode = TAG_NAME;
// set ignoreNewline on some tags, depending on wheather they're // set ignoreNewline on some tags, depending on wheather they're
// being opened or closed. // being opened or closed.
@ -610,6 +618,7 @@ public final class HtmlEncoder {
escape = false; escape = false;
} else if (htmlQuoteChar == c) { } else if (htmlQuoteChar == c) {
htmlQuoteChar = '\u0000'; htmlQuoteChar = '\u0000';
htmlTagMode = TAG_SPACE;
} else if (htmlQuoteChar == '\u0000') { } else if (htmlQuoteChar == '\u0000') {
htmlQuoteChar = c; htmlQuoteChar = c;
} }
@ -650,8 +659,12 @@ public final class HtmlEncoder {
// Check if this is an empty tag so we don't generate an // Check if this is an empty tag so we don't generate an
// additional </close> tag. // additional </close> tag.
if (str.charAt(i - 1) == '/') { if (str.charAt(i - 1) == '/') {
// this is to avoid misinterpreting tags like
// <a href=http://foo/> as empty
if (htmlTagMode != TAG_ATT_VAL && htmlTagMode != TAG_ATT_NAME) {
openTags.pop(); openTags.pop();
} }
}
} else { } else {
ret.append("&gt;"); ret.append("&gt;");
} }
@ -663,7 +676,32 @@ public final class HtmlEncoder {
default: default:
// ret.append (c); if (insideHtmlTag && !insideCloseTag) {
switch(htmlTagMode) {
case TAG_NAME:
if (!Character.isLetterOrDigit(c)) {
htmlTagMode = TAG_SPACE;
}
break;
case TAG_SPACE:
if (Character.isLetterOrDigit(c)) {
htmlTagMode = TAG_ATT_NAME;
}
break;
case TAG_ATT_NAME:
if (c == '=') {
htmlTagMode = TAG_ATT_VAL;
} else if (c == ' ') {
htmlTagMode = TAG_SPACE;
}
break;
case TAG_ATT_VAL:
if (Character.isWhitespace(c) && htmlQuoteChar == '\u0000') {
htmlTagMode = TAG_SPACE;
}
break;
}
}
if (c < 128) { if (c < 128) {
ret.append(c); ret.append(c);
} else if ((c >= 128) && (c < 256)) { } else if ((c >= 128) && (c < 256)) {