|
a/src/internfile/mh_html.h |
|
b/src/internfile/mh_html.h |
|
... |
|
... |
| 14 |
* Free Software Foundation, Inc.,
|
14 |
* Free Software Foundation, Inc.,
|
| 15 |
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
15 |
* 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
|
| 16 |
*/
|
16 |
*/
|
| 17 |
#ifndef _HTML_H_INCLUDED_
|
17 |
#ifndef _HTML_H_INCLUDED_
|
| 18 |
#define _HTML_H_INCLUDED_
|
18 |
#define _HTML_H_INCLUDED_
|
| 19 |
/* @(#$Id: mh_html.h,v 1.7 2006-01-30 11:15:27 dockes Exp $ (C) 2004 J.F.Dockes */
|
19 |
/* @(#$Id: mh_html.h,v 1.8 2006-12-15 12:40:02 dockes Exp $ (C) 2004 J.F.Dockes */
|
| 20 |
|
20 |
|
| 21 |
#include <string>
|
21 |
#include <string>
|
| 22 |
|
22 |
|
| 23 |
#include "mimehandler.h"
|
23 |
#include "mimehandler.h"
|
| 24 |
|
24 |
|
| 25 |
/**
|
25 |
/**
|
| 26 |
Translate html document to internal one.
|
26 |
Translate html document to internal one.
|
| 27 |
|
|
|
| 28 |
There are 2 interfaces, depending if we're working on a file, or
|
|
|
| 29 |
on a string. The string form is applied to the output of external
|
|
|
| 30 |
handlers for foreign formats: they return a result in html, which
|
|
|
| 31 |
has the advantage to be text (easy to use in shell-scripts), and
|
|
|
| 32 |
semi-structured (can carry titles, abstracts, whatever)
|
|
|
| 33 |
*/
|
27 |
*/
|
| 34 |
class MimeHandlerHtml : public MimeHandler {
|
28 |
class MimeHandlerHtml : public RecollFilter {
|
| 35 |
public:
|
29 |
public:
|
| 36 |
std::string charsethint;
|
30 |
MimeHandlerHtml(const string& mt) : RecollFilter(mt) {}
|
| 37 |
|
31 |
virtual ~MimeHandlerHtml() {}
|
| 38 |
/** Create internal document from html file (standard interface) */
|
32 |
virtual bool set_document_file(const string &file_path);
|
| 39 |
virtual MimeHandler::Status
|
33 |
virtual bool set_document_string(const string &data);
|
| 40 |
mkDoc(RclConfig *conf, const std::string &fn,
|
34 |
virtual bool next_document();
|
| 41 |
const std::string &mtype, Rcl::Doc &docout, std::string&);
|
35 |
private:
|
| 42 |
|
36 |
string m_html;
|
| 43 |
/** Create internal doc from html string (postfilter for external ones) */
|
|
|
| 44 |
virtual MimeHandler::Status
|
|
|
| 45 |
mkDoc(RclConfig *conf, const std::string &fn, const std::string& htext,
|
|
|
| 46 |
const std::string &mtype, Rcl::Doc &docout);
|
|
|
| 47 |
};
|
37 |
};
|
| 48 |
|
38 |
|
| 49 |
#endif /* _HTML_H_INCLUDED_ */
|
39 |
#endif /* _HTML_H_INCLUDED_ */
|