recoll / Code / Diff of /src/internfile/mimehandler.h

Diff of /src/internfile/mimehandler.h [ef8338] .. [a7728c]

Switch to unified view


...
#include "autoconfig.h"

#include <stdio.h>

#include <string>




#include "Filter.h"
#include "cstr.h"

class RclConfig;

class RecollFilter : public Dijon::Filter {
public:
    RecollFilter(RclConfig *config, const std::string& id)
  : m_config(config), m_forPreview(false), m_havedoc(false), m_id(id)

    {}
    virtual ~RecollFilter() {}
    virtual void setConfig(RclConfig *config)
    {
    m_config = config;
    }
    virtual bool set_property(Properties p, const std::string &v) {
    switch (p) {
    case DJF_UDI: 
        m_udi = v;
        break;
    case DEFAULT_CHARSET: 
...
    }
    return true;
    }

    // We don't use this for now
    virtual bool set_document_uri(const std::string& mtype, 
                const std::string &) 
    {
  m_mimeType = mtype;
  return false;
    }

    // This does nothing right now but should be called from the
    // subclass method in case we need some common processing one day
    // (was used for xattrs at some point).  Yes this is the "call
    // super" anti-pattern, bad, but we have several layers of derived
    // classes, so that implementing the template method approach (by
    // having a pure virtual called from here and implemented in the
    // subclass) would have to be repeated in each derived class. It's
    // just simpler this way.
    virtual bool set_document_file(const std::string& mtype, 
                 const std::string & /*file_path*/) 
    {
  m_mimeType = mtype;
  return true;
    }

    // Default implementations
    virtual bool set_document_string(const std::string& mtype, 
                   const std::string &) 
    {
  m_mimeType = mtype;
  return false;
    }
    virtual bool set_document_data(const std::string& mtype, 
                 const char *cp, unsigned int sz) 
    {
    return set_document_string(mtype, std::string(cp, sz));
    }

    virtual void set_docsize(size_t size)
    {
    char csize[30];
...
    }

    virtual bool has_documents() const {return m_havedoc;}

    // Most doc types are single-doc
    virtual bool skip_to_document(const std::string& s) {
    if (s.empty())
        return true;
    return false;
    }

...
    if (input == DOCUMENT_FILE_NAME)
        return true;
    return false;
    }

    virtual std::string get_error() const {
    return m_reason;
    }

    virtual const std::string& get_id() const
    {
  return m_id;
    }

    // "Call super" anti-pattern again. Must be called from derived
    // classes which reimplement clear()
    virtual void clear() {
...
    m_reason.clear();
    }

    // This only makes sense if the contents are currently txt/plain
    // It converts from keyorigcharset to UTF-8 and sets keycharset.
    bool txtdcode(const std::string& who);

protected:
    bool preview() {return m_forPreview;}

    RclConfig *m_config;
    bool   m_forPreview;
    std::string m_dfltInputCharset;
    std::string m_reason;
    bool   m_havedoc;
    std::string m_udi; // May be set by creator as a hint
    // m_id is and md5 of the filter definition line (from mimeconf) and
    // is used when fetching/returning filters to / from the cache.
    std::string m_id;
};

/**
 * Return indexing handler object for the given mime type. The returned 
 * pointer should be passed to returnMimeHandler() for recycling, after use.
 * @param mtyp input mime type, ie text/plain
 * @param cfg  the recoll config object to be used
 * @param filtertypes decide if we should restrict to types in 
 *     indexedmimetypes (if this is set at all).
 */
extern RecollFilter *getMimeHandler(const std::string &mtyp, RclConfig *cfg,
                     bool filtertypes=false);

/// Free up filter for reuse (you can also delete it)
extern void returnMimeHandler(RecollFilter *);

/// Clean up cache at the end of an indexing pass. For people who use
/// the GUI to index: avoid all those filter processes forever hanging
/// off recoll.
extern void clearMimeHandlerCache();

	a/src/internfile/mimehandler.h		b/src/internfile/mimehandler.h
	...		...
19	#include "autoconfig.h"	19	#include "autoconfig.h"
20		20
21	#include <stdio.h>	21	#include <stdio.h>
22		22
23	#include <string>	23	#include <string>
24	#include <list>
25	using std::string;
26	using std::list;
27		24
28	#include <Filter.h>	25	#include "Filter.h"
		26	#include "cstr.h"
29		27
30	class RclConfig;	28	class RclConfig;
31		29
32	class RecollFilter : public Dijon::Filter {	30	class RecollFilter : public Dijon::Filter {
33	public:	31	public:
34	RecollFilter(RclConfig *config, const string& mtype)	32	RecollFilter(RclConfig *config, const std::string& id)
35	: Dijon::Filter(mtype), m_config(config),	33	: m_config(config), m_forPreview(false), m_havedoc(false), m_id(id)
36	m_forPreview(false), m_havedoc(false)
37	{}	34	{}
38	virtual ~RecollFilter() {}	35	virtual ~RecollFilter() {}
39	virtual void setConfig(RclConfig *config)	36	virtual void setConfig(RclConfig *config)
40	{	37	{
41	m_config = config;	38	m_config = config;
42	}	39	}
43	virtual bool set_property(Properties p, const string &v) {	40	virtual bool set_property(Properties p, const std::string &v) {
44	switch (p) {	41	switch (p) {
45	case DJF_UDI:	42	case DJF_UDI:
46	m_udi = v;	43	m_udi = v;
47	break;	44	break;
48	case DEFAULT_CHARSET:	45	case DEFAULT_CHARSET:
	...		...
57	}	54	}
58	return true;	55	return true;
59	}	56	}
60		57
61	// We don't use this for now	58	// We don't use this for now
62	virtual bool set_document_uri(const std::string &) {return false;}	59	virtual bool set_document_uri(const std::string& mtype,
		60	const std::string &)
		61	{
		62	m_mimeType = mtype;
		63	return false;
		64	}
63		65
64	// This does nothing right now but should be called from the	66	// This does nothing right now but should be called from the
65	// subclass method in case we need some common processing one day	67	// subclass method in case we need some common processing one day
66	// (was used for xattrs at some point). Yes this is the "call	68	// (was used for xattrs at some point). Yes this is the "call
67	// super" anti-pattern, bad, but we have several layers of derived	69	// super" anti-pattern, bad, but we have several layers of derived
68	// classes, so that implementing the template method approach (by	70	// classes, so that implementing the template method approach (by
69	// having a pure virtual called from here and implemented in the	71	// having a pure virtual called from here and implemented in the
70	// subclass) would have to be repeated in each derived class. It's	72	// subclass) would have to be repeated in each derived class. It's
71	// just simpler this way.	73	// just simpler this way.
72	virtual bool set_document_file(const string & /file_path/) {return true;}	74	virtual bool set_document_file(const std::string& mtype,
		75	const std::string & /file_path/)
		76	{
		77	m_mimeType = mtype;
		78	return true;
		79	}
73		80
74	// Default implementations	81	// Default implementations
75	virtual bool set_document_string(const std::string &) {return false;}	82	virtual bool set_document_string(const std::string& mtype,
76	virtual bool set_document_data(const char *cp, unsigned int sz) {	83	const std::string &)
		84	{
		85	m_mimeType = mtype;
		86	return false;
		87	}
		88	virtual bool set_document_data(const std::string& mtype,
		89	const char *cp, unsigned int sz)
		90	{
77	return set_document_string(string(cp, sz));	91	return set_document_string(mtype, std::string(cp, sz));
78	}	92	}
79		93
80	virtual void set_docsize(size_t size)	94	virtual void set_docsize(size_t size)
81	{	95	{
82	char csize[30];	96	char csize[30];
	...		...
85	}	99	}
86		100
87	virtual bool has_documents() const {return m_havedoc;}	101	virtual bool has_documents() const {return m_havedoc;}
88		102
89	// Most doc types are single-doc	103	// Most doc types are single-doc
90	virtual bool skip_to_document(const string& s) {	104	virtual bool skip_to_document(const std::string& s) {
91	if (s.empty())	105	if (s.empty())
92	return true;	106	return true;
93	return false;	107	return false;
94	}	108	}
95		109
	...		...
97	if (input == DOCUMENT_FILE_NAME)	111	if (input == DOCUMENT_FILE_NAME)
98	return true;	112	return true;
99	return false;	113	return false;
100	}	114	}
101		115
102	virtual string get_error() const {	116	virtual std::string get_error() const {
103	return m_reason;	117	return m_reason;
		118	}
		119
		120	virtual const std::string& get_id() const
		121	{
		122	return m_id;
104	}	123	}
105		124
106	// "Call super" anti-pattern again. Must be called from derived	125	// "Call super" anti-pattern again. Must be called from derived
107	// classes which reimplement clear()	126	// classes which reimplement clear()
108	virtual void clear() {	127	virtual void clear() {
	...		...
112	m_reason.clear();	131	m_reason.clear();
113	}	132	}
114		133
115	// This only makes sense if the contents are currently txt/plain	134	// This only makes sense if the contents are currently txt/plain
116	// It converts from keyorigcharset to UTF-8 and sets keycharset.	135	// It converts from keyorigcharset to UTF-8 and sets keycharset.
117	bool txtdcode(const string& who);	136	bool txtdcode(const std::string& who);
118		137
119	protected:	138	protected:
120	bool preview() {return m_forPreview;}	139	bool preview() {return m_forPreview;}
121		140
122	RclConfig *m_config;	141	RclConfig *m_config;
123	bool m_forPreview;	142	bool m_forPreview;
124	string m_dfltInputCharset;	143	std::string m_dfltInputCharset;
125	string m_reason;	144	std::string m_reason;
126	bool m_havedoc;	145	bool m_havedoc;
127	string m_udi; // May be set by creator as a hint	146	std::string m_udi; // May be set by creator as a hint
		147	// m_id is and md5 of the filter definition line (from mimeconf) and
		148	// is used when fetching/returning filters to / from the cache.
		149	std::string m_id;
128	};	150	};
129		151
130	/**	152	/**
131	* Return indexing handler object for the given mime type. The returned	153	* Return indexing handler object for the given mime type. The returned
132	* pointer should be passed to returnMimeHandler() for recycling, after use.	154	* pointer should be passed to returnMimeHandler() for recycling, after use.
133	* @param mtyp input mime type, ie text/plain	155	* @param mtyp input mime type, ie text/plain
134	* @param cfg the recoll config object to be used	156	* @param cfg the recoll config object to be used
135	* @param filtertypes decide if we should restrict to types in	157	* @param filtertypes decide if we should restrict to types in
136	* indexedmimetypes (if this is set at all).	158	* indexedmimetypes (if this is set at all).
137	*/	159	*/
138	extern Dijon::Filter getMimeHandler(const std::string &mtyp, RclConfig cfg,	160	extern RecollFilter getMimeHandler(const std::string &mtyp, RclConfig cfg,
139	bool filtertypes=false);	161	bool filtertypes=false);
140		162
141	/// Free up filter for reuse (you can also delete it)	163	/// Free up filter for reuse (you can also delete it)
142	extern void returnMimeHandler(Dijon::Filter *);	164	extern void returnMimeHandler(RecollFilter *);
143		165
144	/// Clean up cache at the end of an indexing pass. For people who use	166	/// Clean up cache at the end of an indexing pass. For people who use
145	/// the GUI to index: avoid all those filter processes forever hanging	167	/// the GUI to index: avoid all those filter processes forever hanging
146	/// off recoll.	168	/// off recoll.
147	extern void clearMimeHandlerCache();	169	extern void clearMimeHandlerCache();