|
a/src/internfile/mimehandler.h |
|
b/src/internfile/mimehandler.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 _MIMEHANDLER_H_INCLUDED_
|
17 |
#ifndef _MIMEHANDLER_H_INCLUDED_
|
| 18 |
#define _MIMEHANDLER_H_INCLUDED_
|
18 |
#define _MIMEHANDLER_H_INCLUDED_
|
| 19 |
/* @(#$Id: mimehandler.h,v 1.12 2006-03-29 13:08:08 dockes Exp $ (C) 2004 J.F.Dockes */
|
19 |
/* @(#$Id: mimehandler.h,v 1.13 2006-12-15 12:40:02 dockes Exp $ (C) 2004 J.F.Dockes */
|
| 20 |
|
20 |
|
| 21 |
#include <string>
|
21 |
#include <string>
|
| 22 |
#include <list>
|
22 |
#include <list>
|
|
|
23 |
using std::string;
|
|
|
24 |
using std::list;
|
| 23 |
|
25 |
|
| 24 |
#include "rclconfig.h"
|
26 |
#include <Filter.h>
|
| 25 |
#include "rcldb.h"
|
|
|
| 26 |
|
27 |
|
|
|
28 |
class RclConfig;
|
| 27 |
|
29 |
|
| 28 |
/**
|
30 |
class RecollFilter : public Dijon::Filter {
|
| 29 |
* Document interner class.
|
|
|
| 30 |
*/
|
|
|
| 31 |
class MimeHandler {
|
|
|
| 32 |
public:
|
31 |
public:
|
| 33 |
MimeHandler() : m_forPreview(false) {}
|
32 |
RecollFilter(const string& mtype)
|
|
|
33 |
: Dijon::Filter(mtype), m_forPreview(false), m_havedoc(false)
|
|
|
34 |
{}
|
| 34 |
virtual ~MimeHandler() {}
|
35 |
virtual ~RecollFilter() {}
|
|
|
36 |
virtual bool set_property(Properties p, const string &v) {
|
|
|
37 |
switch (p) {
|
|
|
38 |
case DEFAULT_CHARSET:
|
|
|
39 |
m_defcharset = v;
|
|
|
40 |
break;
|
|
|
41 |
case OPERATING_MODE:
|
|
|
42 |
if (!v.empty() && v[0] == 'v')
|
|
|
43 |
m_forPreview = true;
|
|
|
44 |
else
|
|
|
45 |
m_forPreview = false;
|
|
|
46 |
break;
|
|
|
47 |
}
|
|
|
48 |
return true;
|
|
|
49 |
}
|
| 35 |
|
50 |
|
| 36 |
/// Status from mkDoc method.
|
51 |
// We don't use this for now
|
| 37 |
enum Status {MHError, MHDone, MHAgain};
|
52 |
virtual bool set_document_uri(const std::string &) {return false;}
|
| 38 |
/**
|
|
|
| 39 |
* Transform external data into internal utf8 document
|
|
|
| 40 |
*
|
|
|
| 41 |
* @param conf the global configuration
|
|
|
| 42 |
* @param filename File from which the data comes from
|
|
|
| 43 |
* @param mimetype its mime type (from the mimemap configuration file)
|
|
|
| 44 |
* @param outdoc The output document
|
|
|
| 45 |
* @param ipath the access path for the document inside the file.
|
|
|
| 46 |
* For mono-document file types, this will always be empty.
|
|
|
| 47 |
* It is used, for example for mbox files which may contain
|
|
|
| 48 |
* multiple emails. If this is not empty in input, then the
|
|
|
| 49 |
* caller is requesting a single document (ie: for display).
|
|
|
| 50 |
* If this is empty (during indexation), it will be filled-up
|
|
|
| 51 |
* by the function, and all the file's documents will be
|
|
|
| 52 |
* returned by successive calls.
|
|
|
| 53 |
* @return The return value indicates if there are more documents to be
|
|
|
| 54 |
* fetched from the same file.
|
|
|
| 55 |
*/
|
|
|
| 56 |
virtual MimeHandler::Status mkDoc(RclConfig * conf,
|
|
|
| 57 |
const std::string &filename,
|
|
|
| 58 |
const std::string &mimetype,
|
|
|
| 59 |
Rcl::Doc& outdoc,
|
|
|
| 60 |
string& ipath) = 0;
|
|
|
| 61 |
|
53 |
|
| 62 |
virtual void setForPreview(bool onoff) {m_forPreview = onoff;};
|
54 |
// Default implementations
|
|
|
55 |
virtual bool set_document_string(const std::string &) {return false;}
|
|
|
56 |
virtual bool set_document_data(const char *cp, unsigned int sz) {
|
|
|
57 |
return set_document_string(string(cp, sz));
|
|
|
58 |
}
|
| 63 |
|
59 |
|
|
|
60 |
virtual bool has_documents() const {return m_havedoc;}
|
|
|
61 |
|
|
|
62 |
// Most doc types are single-doc
|
|
|
63 |
virtual bool skip_to_document(const string& s) {
|
|
|
64 |
if (s.empty())
|
|
|
65 |
return true;
|
|
|
66 |
return false;
|
|
|
67 |
}
|
|
|
68 |
|
|
|
69 |
virtual DataInput get_required_data_input() const
|
|
|
70 |
{return DOCUMENT_FILE_NAME;}
|
|
|
71 |
virtual string get_error() const {
|
|
|
72 |
return m_reason;
|
|
|
73 |
}
|
|
|
74 |
|
| 64 |
protected:
|
75 |
protected:
|
| 65 |
bool m_forPreview;
|
76 |
bool m_forPreview;
|
|
|
77 |
string m_defcharset;
|
|
|
78 |
string m_reason;
|
|
|
79 |
bool m_havedoc;
|
| 66 |
};
|
80 |
};
|
| 67 |
|
81 |
|
| 68 |
/**
|
82 |
/**
|
| 69 |
* Return indexing handler object for the given mime type
|
83 |
* Return indexing handler object for the given mime type
|
| 70 |
* returned pointer should be deleted by caller
|
84 |
* returned pointer should be deleted by caller
|
| 71 |
*/
|
85 |
*/
|
| 72 |
extern MimeHandler *getMimeHandler(const std::string &mtyp, RclConfig *cfg);
|
86 |
extern Dijon::Filter *getMimeHandler(const std::string &mtyp, RclConfig *cfg);
|
| 73 |
|
87 |
|
| 74 |
/// Can this mime type be interned ?
|
88 |
/// Can this mime type be interned ?
|
| 75 |
extern bool canIntern(const std::string mimetype, RclConfig *cfg);
|
89 |
extern bool canIntern(const std::string mimetype, RclConfig *cfg);
|
| 76 |
|
90 |
|
| 77 |
#endif /* _MIMEHANDLER_H_INCLUDED_ */
|
91 |
#endif /* _MIMEHANDLER_H_INCLUDED_ */
|