Skip to content

Commit

Permalink
Rewrote SysLink tag parsing for control auto-sizing code.
Browse files Browse the repository at this point in the history
  • Loading branch information
@Lexikos
Lexikos committed Feb 1, 2012
1 parent 48e6b3f commit b6ce337
Showing 1 changed file with 59 additions and 89 deletions.
148 changes: 59 additions & 89 deletions source/script_gui.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2656,111 +2656,81 @@ ResultType GuiType::AddControl(GuiControls aControlType, LPTSTR aOptions, LPTSTR

int draw_height;

//Link controls don't strip the HTML tags out for size calculation, so it needs to be done manually here.
// Since a Link control's text contains markup which isn't actually rendered, we need
// to strip it out before calling DrawText or it will put out the size calculation:
if (aControlType == GUI_CONTROL_LINK)
{
if (style & LWS_NOPREFIX) // This is necessary to auto-width the control properly if its contents include any ampersands.
// Unless the control has the LWS_NOPREFIX style, the first ampersand causes the character
// following it to be rendered with an underline, and the ampersand itself is not rendered.
// Since DrawText has this same behaviour by default, we can simply set DT_NOPREFIX when
// appropriate and it should handle the ampersands correctly.
if (style & LWS_NOPREFIX)
draw_format |= DT_NOPREFIX;

TCHAR aChar = 0, tagChar = 0;
LPTSTR aTextCopy = new TCHAR[_tcslen(aText) + 1];
ZeroMemory(aTextCopy, sizeof(TCHAR) * (_tcslen(aText) +1));
int pos = 0; //position in the new string
bool foundAmp = style & SS_NOPREFIX; //remove first & if this style is not set
bool inLink = false; //true if the current character belongs to a link. Needed for treating closing tags.
for(int i = 0, aChar = aText[0]; aChar; aChar = aText[++i])

TCHAR *src, *dst;
for (src = aText, dst = aTextCopy; *src; ++src)
{
if(!foundAmp && aChar == '&')
foundAmp = true;
else if(!inLink && aChar == '<' && ctoupper(aText[i+1]) == 'A') //possible opening of tag
if (*src == '<' && ctoupper(src[1]) == 'A')
{
int tagstate = 0;
bool isTag = false;
for(int j = i + 2, tagChar = aText[j]; tagChar; tagChar = aText[++j])
TCHAR *linkText = NULL;
TCHAR *cp = omit_leading_whitespace(src + 2);

// It is important to note that while the SysLink control's syntax is similar to HTML,
// it is not HTML. Some valid HTML won't work, and some invalid HTML will. Testing
// indicates that whitespace is not required between "<A" and the attribute, so there
// is no check to see if the above actually omitted whitespace.
for (;;)
{
if(tagstate == 0) //0 = looking for spaces or end of empty tags
{
if(tagChar == ' ') //spaces between tag type and attribute are allowed
continue;
else if(tagChar == '>') //no attribute, valid link!
{
inLink = true;
i = j;
break;
}
else if(tagChar == '=' && aText[j+1] && aText[j+1] == '"') //end of empty tag and start of URL
{
j++;
tagstate = 3;
continue;
}
else if(tagChar != '"') //no unexpected character here: start of attribute
tagstate = 1;
else //unexpected character, treat as part of the text
break;
}
if(tagstate == 1) //1 = require character for attribute
{
if(tagChar == ' ' || tagChar == '=' || tagChar == '"' || tagChar == '<') //invalid characters in attribute
break;
else //atleast one character found for attribute
{
tagstate = 2;
continue;
}
}
else if(tagstate == 2) //2 = further characters in attribute
{
if(tagChar == ' ' || tagChar == '"') //invalid characters in attribute
break;
else if(tagChar == '=' && aText[j+1] && aText[j+1] == '"') //end of tag and start of URL
{
j++;
tagstate = 3;
continue;
}
else //more attribute characters
continue;
}
else if(tagstate == 3) //3 = in URL
if (*cp == '>')
{
if(tagChar == '"') //end of URL
tagstate = 4;
else if(tagChar == '>') //invalid URL, treat as text
break;
else //part of the URL
continue;
linkText = cp + 1;
break;
}
else if(tagstate == 4) //expecting spaces or closing tag

// Testing indicates that attribute names can only be alphanumeric chars. This
// method of testing for the attribute name does not treat <a=""> as an error,
// which is perfect since the control tolerates it (the attribute is ignored).
while (cisalnum(*cp))
cp++;

// What follows next must be the attribute value. Spaces are not tolerated.
if (*cp != '=' || '"' != cp[1])
break; // Not valid.

// Testing indicates that the attribute value can contain virtually anything,
// including </a> but not quotation marks.
cp = _tcschr(cp + 2, '"');
if (!cp)
break; // Not valid.

// Allow whitespace after the attribute (before '>' or the next attribute).
cp = omit_leading_whitespace(cp + 1);
} // for (attribute-parsing loop)

if (linkText)
{
// Testing indicates that spaces in the end tag are not tolerated:
if (LPTSTR endTag = tcscasestr(linkText, _T("</a>")))
{
if(tagChar == ' ') //spaces after link are allowed
continue;
else if(tagChar == '>') //end of tag, valid!
{
inLink = true;
i = j;
break;
}
else //unexpected characters, treat as text
break;
// Copy the link text and skip over the end tag.
for (cp = linkText; cp < endTag; *dst++ = *cp++);
src = endTag + 3; // The outer loop's increment makes it + 4.
continue;
}
// Otherwise, there's no end tag so the begin tag is treated as text.
}
if(inLink)
continue;
}

if(inLink && aText[i+1] == '<' && aText[i+2] == '/' && ctoupper(aText[i+3]) == 'A' && aText[i+4] == '>') //closing tag of link needs to be skipped
{
i += 3;
inLink = false;
continue;
}
else //normal text, unrecognized tags, nested <A> tags and redundant closing </a> tags are used for calculation
aTextCopy[pos++] = aChar;
}
// Otherwise, it wasn't a valid tag, so it is treated as text.
} // if (found a tag)
*dst++ = *src;
} // for (searching for tag, copying text)

*dst = '\0';

// If no text, "H" is used in case the function requires a non-empty string to give consistent results:
draw_height = DrawText(hdc, *aTextCopy ? aTextCopy : _T("H"), -1, &draw_rect, draw_format);

delete[] aTextCopy;
}
else
Expand Down

0 comments on commit b6ce337

Please sign in to comment.