diff options
Diffstat (limited to 'src/doctokenizer.l')
| -rw-r--r-- | src/doctokenizer.l | 49 |
1 files changed, 28 insertions, 21 deletions
diff --git a/src/doctokenizer.l b/src/doctokenizer.l index 2a17819..1e4b2d1 100644 --- a/src/doctokenizer.l +++ b/src/doctokenizer.l @@ -229,6 +229,29 @@ static void processSection() } } +static void handleHtmlTag() +{ + g_token->name = yytext; + g_token->attribs.clear(); + int startNamePos=1; + if (g_token->name.at(1)=='/') startNamePos++; + int attSep=0; + while (attSep<yyleng && !isspace(yytext[attSep])) + { + attSep++; + } + if (attSep!=yyleng) // tag has one or more options + { + parseHtmlAttribs(g_token->name.mid(attSep+1,g_token->name.length()-attSep-2)); + g_token->name=g_token->name.mid(startNamePos,attSep-1).lower(); + } + else // tag without options, strip brackets + { + g_token->name=g_token->name.mid(startNamePos,g_token->name.length()-startNamePos-1).lower(); + } + g_token->endTag = startNamePos==2; +} + //-------------------------------------------------------------------------- #undef YY_INPUT @@ -255,7 +278,7 @@ ID [a-z_A-Z][a-z_A-Z0-9]* OPTSTARS ("//"{BLANK}*)?"*"*{BLANK}* LISTITEM {BLANK}*{OPTSTARS}"-"("#")?{WS} ENDLIST {BLANK}*{OPTSTARS}"."{BLANK}*\n -ATTRIB {ID}("="(("\""[^\"]*"\"")|("'"[^\']*"'")|[^ \t\r\n'"><]+))? +ATTRIB {ID}{WS}*("="{WS}*(("\""[^\"]*"\"")|("'"[^\']*"'")|[^ \t\r\n'"><]+))? URLCHAR [a-z_A-Z0-9\!\~\:\;\'\$\?\@\&\%\#\.\-\+\/\=] URLMASK (([a-z_A-Z][^\>\"\n]*{URLCHAR})|({URLCHAR}+))([({]{URLCHAR}*[)}])? FILESCHAR [a-z_A-Z0-9\\:\\\/\-\+] @@ -283,7 +306,7 @@ WORD1 [^ \t\n\r\\@<>()\[\]:;\?{}&$#,.]+|"{"|"}"|("\""[^"\n]*"\"") WORD2 "."|","|"("|")"|"["|"]"|":"|";"|"\?" WORD1NQ [^ \t\n\r\\@<>()\[\]:;\?{}&$#,."]+ WORD2NQ "."|","|"("|")"|"["|"]"|":"|";"|"\?" -HTMLTAG "<"(("/")?){ID}({WS}+{ATTRIB})*">" +HTMLTAG "<"(("/")?){ID}({WS}+{ATTRIB})*{WS}*">" HTMLKEYL "strong"|"center"|"table"|"caption"|"small"|"code"|"dfn"|"var"|"img"|"pre"|"sub"|"sup"|"tr"|"td"|"th"|"ol"|"ul"|"li"|"tt"|"kbd"|"em"|"hr"|"dl"|"dt"|"dd"|"br"|"i"|"a"|"b"|"p" HTMLKEYU "STRONG"|"CENTER"|"TABLE"|"CAPTION"|"SMALL"|"CODE"|"DFN"|"VAR"|"IMG"|"PRE"|"SUB"|"SUP"|"TR"|"TD"|"TH"|"OL"|"UL"|"LI"|"TT"|"KBD"|"EM"|"HR"|"DL"|"DT"|"DD"|"BR"|"I"|"A"|"B"|"P" HTMLKEYW {HTMLKEYL}|{HTMLKEYU} @@ -391,25 +414,7 @@ LABELID [a-z_A-Z][a-z_A-Z0-9\-]* for (int i=value.length()-1;i>=0;i--) unput(value.at(i)); } <St_Para>{HTMLTAG} { /* html tag */ - g_token->name = yytext; - g_token->attribs.clear(); - int startNamePos=1; - if (g_token->name.at(1)=='/') startNamePos++; - int attSep=0; - while (attSep<yyleng && !isspace(yytext[attSep])) - { - attSep++; - } - if (attSep!=yyleng) // tag has one or more options - { - parseHtmlAttribs(g_token->name.mid(attSep+1,g_token->name.length()-attSep-2)); - g_token->name=g_token->name.mid(startNamePos,attSep-1).lower(); - } - else // tag without options, strip brackets - { - g_token->name=g_token->name.mid(startNamePos,g_token->name.length()-startNamePos-1).lower(); - } - g_token->endTag = startNamePos==2; + handleHtmlTag(); return TK_HTMLTAG; } <St_Para,St_Text>"&"{ID}";" { /* special symbol */ @@ -538,6 +543,8 @@ LABELID [a-z_A-Z][a-z_A-Z0-9\-]* g_token->name = yytext; return TK_SYMBOL; } +<St_TitleN>{HTMLTAG} { + } <St_TitleN>{SPCMD1} | <St_TitleN>{SPCMD2} { /* special command */ g_token->name = yytext+1; |