metaproxy_1::HTMLParser::Rep Class Reference

Collaboration diagram for metaproxy_1::HTMLParser::Rep:

Public Member Functions
void	parse_str (HTMLParserEvent &event, const char *cp)
void	tagText (HTMLParserEvent &event, const char *text_start, const char *text_end)
int	tagEnd (HTMLParserEvent &event, const char *tag, int tag_len, const char *cp)
int	tagAttrs (HTMLParserEvent &event, const char *name, int len, const char *cp)
int	skipAttribute (HTMLParserEvent &event, const char *cp, int *attr_len, const char **value, int *val_len, int *tr)
	Rep ()
	~Rep ()

Public Attributes
int	m_verbose
bool	nest

Friends
class	HTMLParser

Detailed Description

Definition at line 34 of file html_parser.cpp.

Constructor & Destructor Documentation

Rep()

metaproxy_1::HTMLParser::Rep::Rep

(

)

Definition at line 57 of file html_parser.cpp.

{
    m_verbose = 0;
    nest = true;
}

References m_verbose, and nest.

~Rep()

metaproxy_1::HTMLParser::Rep::~Rep

(

)

Definition at line 63 of file html_parser.cpp.

{
}

Member Function Documentation

parse_str()

void metaproxy_1::HTMLParser::Rep::parse_str	(	HTMLParserEvent &	event,
		const char *	cp )

Definition at line 221 of file html_parser.cpp.

{
    const char *text_start = cp;
    while (*cp)
    {
        if (*cp++ != '<')
            continue;
 
        if (nest && *cp == '!')
        {
            int i;
            tagText(event, text_start, cp - 1);
            if (cp[1] == '-' && cp[2] == '-')
            {
                for (i = 3; cp[i]; i++)
                    if (cp[i] == '-' && cp[i+1] == '-' && cp[i+2] == '>')
                    {
                        i+= 2;
                        event.openTagStart(cp, i);
                        break;
                    }
            }
            else
            {
                for (i = 1; cp[i] && cp[i] != '>'; i++)
                    ;
                event.openTagStart(cp, i);
            }
            if (m_verbose)
                printf("------ dtd %.*s\n", i, cp);
            i += tagEnd(event, cp, i, cp + i);
            cp += i;
            text_start = cp;
        }
        else if (nest && *cp == '?')
        {
            int i;
            tagText(event, text_start, cp - 1);
            for (i = 1; cp[i] && cp[i] != '>'; i++)
                ;
            event.openTagStart(cp, i);
            if (m_verbose)
                printf("------ pi %.*s\n", i, cp);
            i += tagEnd(event, cp, i, cp + i);
            cp += i;
            text_start = cp;
        }
        else if (*cp == '/' && isAlpha(cp[1]))
        {
            int i;
 
            i = skipName(++cp);
 
            if (!nest)
            {
                if (i == 6 && !yaz_strncasecmp(cp, "script", i))
                {
                    int ws = skipSpace(cp + 6);
                    if (cp[ws + 6] == '>')
                        nest = true; /* really terminated */
                }
                if (!nest)
                    continue;
            }
            tagText(event, text_start, cp - 2);
            event.closeTag(cp, i);
            if (m_verbose)
                printf("------ tag close %.*s\n", i, cp);
            i += tagEnd(event, cp, i, cp + i);
            cp += i;
            text_start = cp;
        }
        else if (nest && isAlpha(*cp))
        {
            int i, j;
            tagText(event, text_start, cp - 1);
            i = skipName(cp);
            event.openTagStart(cp, i);
            if (m_verbose)
                printf("------ tag open %.*s\n", i, cp);
            j = tagAttrs(event, cp, i, cp + i);
            j += tagEnd(event, cp, i, cp + i + j);
 
            if (i == 6 && !yaz_strncasecmp(cp, "script", i))
                nest = false;
 
            cp += i + j;
            text_start = cp;
        }
    }
    tagText(event, text_start, cp);
}

References isAlpha(), m_verbose, nest, skipName(), skipSpace(), tagAttrs(), tagEnd(), and tagText().

Here is the call graph for this function:

skipAttribute()

int metaproxy_1::HTMLParser::Rep::skipAttribute	(	HTMLParserEvent &	event,
		const char *	cp,
		int *	attr_len,
		const char **	value,
		int *	val_len,
		int *	tr )

Definition at line 107 of file html_parser.cpp.

{
    int v0, v1;
    int i = skipName(cp);
    *attr_len = i;
    *value = NULL;
    if (!i)
        return skipSpace(cp);
    i += skipSpace(cp + i);
    if (cp[i] == '=')
    {
        i++;
        i += skipSpace(cp + i);
        if (cp[i] == '\"' || cp[i] == '\'')
        {
            *tr = cp[i];
            v0 = ++i;
            while (cp[i] != *tr && cp[i])
                i++;
            v1 = i;
            if (cp[i])
                i++;
        }
        else
        {
            *tr = 0;
            v0 = i;
            while (cp[i] && !strchr(SPACECHR ">", cp[i]))
                i++;
            v1 = i;
        }
        *value = cp + v0;
        *val_len = v1 - v0;
        i += skipSpace(cp + i);
    }
    return i;
}

References skipName(), skipSpace(), and SPACECHR.

Referenced by tagAttrs().

Here is the call graph for this function:

tagAttrs()

int metaproxy_1::HTMLParser::Rep::tagAttrs	(	HTMLParserEvent &	event,
		const char *	name,
		int	len,
		const char *	cp )

Definition at line 148 of file html_parser.cpp.

{
    int i = skipSpace(cp);
    while (cp[i] && !strchr("/><", cp[i]))
    {
        const char *attr_name = cp + i;
        int attr_len;
        const char *value;
        int val_len;
        int tr;
        char x[2];
        int nor = skipAttribute(event, cp+i, &attr_len, &value, &val_len, &tr);
        if (!nor)
            break;
        i += nor;
 
        x[0] = tr;
        x[1] = 0;
        if (m_verbose)
        {
            printf("------ attr %.*s", attr_len, attr_name);
            if (value)
                printf("=%.*s", val_len, value);
            printf("\n");
        }
        event.attribute(name, len, attr_name, attr_len, value, val_len, x);
    }
    return i;
}

References m_verbose, skipAttribute(), and skipSpace().

Referenced by parse_str().

Here is the call graph for this function:

tagEnd()

int metaproxy_1::HTMLParser::Rep::tagEnd	(	HTMLParserEvent &	event,
		const char *	tag,
		int	tag_len,
		const char *	cp )

Definition at line 180 of file html_parser.cpp.

{
    int i = 0;
    int close_it = 0;
    for (; cp[i] && !strchr("/><", cp[i]); i++)
        ;
    if (i > 0)
    {
        if (m_verbose)
            printf("------ text %.*s\n", i, cp);
        event.text(cp, i);
    }
    if (cp[i] == '/')
    {
        close_it = 1;
        i++;
    }
    if (cp[i] == '>')
    {
        if (m_verbose)
            printf("------ any tag %s %.*s\n",
                   close_it ? "close" : "end", tag_len, tag);
        event.anyTagEnd(tag, tag_len, close_it);
        i++;
    }
    return i;
}

References m_verbose.

Referenced by parse_str().

tagText()

void metaproxy_1::HTMLParser::Rep::tagText	(	HTMLParserEvent &	event,
		const char *	text_start,
		const char *	text_end )

Definition at line 209 of file html_parser.cpp.

{
    if (text_end - text_start) //got text to flush
    {
        if (m_verbose)
            printf("------ text %.*s\n",
                   (int) (text_end - text_start), text_start);
        event.text(text_start, text_end-text_start);
    }
}

References m_verbose.

Referenced by parse_str().

Friends And Related Symbol Documentation

HTMLParser

friend class HTMLParser

friend

Definition at line 35 of file html_parser.cpp.

References HTMLParser.

Referenced by HTMLParser.

Member Data Documentation

m_verbose

int metaproxy_1::HTMLParser::Rep::m_verbose

Definition at line 50 of file html_parser.cpp.

Referenced by parse_str(), Rep(), tagAttrs(), tagEnd(), and tagText().

nest

bool metaproxy_1::HTMLParser::Rep::nest

Definition at line 51 of file html_parser.cpp.

Referenced by parse_str(), and Rep().

---

The documentation for this class was generated from the following file:

• html_parser.cpp