recoll / Code / Diff of /src/query/xadump.cpp

Diff of /src/query/xadump.cpp [c0689d] .. [f581d3]

Switch to unified view


#ifndef lint
static char rcsid[] = "@(#$Id: xadump.cpp,v 1.20 2008-12-18 14:11:01 dockes Exp $ (C) 2004 J.F.Dockes";
#endif
/*
 *   This program is free software; you can redistribute it and/or modify
 *   it under the terms of the GNU General Public License as published by
 *   the Free Software Foundation; either version 2 of the License, or
...

#ifndef NO_NAMESPACES
using namespace std;
#endif /* NO_NAMESPACES */

#include "utf8iter.h"

#include "xapian.h"

static string thisprog;

static string usage =
...
    " -i docid -b : 'rebuild' document from term positions\n"
    " -t term -E  : term existence test\n"
    " -t term -F  : retrieve term frequency data for given term\n"
    " -t term -P  : retrieve postings for term\n"
    " -i docid -T : term list for doc docid\n"
    " -x          : separate each output char with a space\n"
    " -T          : list all terms\n"
    "    -f       : precede each term in the list with its occurrence count\n"
    "    -n : raw data (no [])\n"
    " -s          : special mode to dump recoll stem db\n"
    " -q term [term ...] : perform AND query\n"
...
#define OPT_s     0x400
#define OPT_f     0x800
#define OPT_q     0x1000
#define OPT_n     0x2000
#define OPT_X     0x4000
#define OPT_x     0x8000

// Compute an exploded version of string, inserting a space between each char.
// (no character combining possible)
static string detailstring(const string& in)
{
    if (!(op_flags & OPT_x))
  return in;
    string out;
    Utf8Iter  it(in);
    for (; !it.eof(); it++) {
  it.appendchartostring(out);
  out += ' ';
    }
    // Strip last space
    if (!out.empty())
  out.resize(out.size()-1);
    return out;
}

Xapian::Database *db;

static void cleanup()
{
...
        case 't':   op_flags |= OPT_t; if (argc < 2)  Usage();
        aterm = *(++argv);
        argc--; 
        goto b1;
        case 'X':   op_flags |= OPT_X; break;
      case 'x':   op_flags |= OPT_x; break;
        default: Usage();   break;
        }
    b1: argc--; argv++;
    }

...
        string op = (op_flags & OPT_n) ? string(): "[";
        string cl = (op_flags & OPT_n) ? string(): "]";
        if (op_flags & OPT_i) {
        for (term = db->termlist_begin(docid); 
             term != db->termlist_end(docid);term++) {
            cout << op << detailstring(*term) << cl << endl;
        }
        } else {
        for (term = db->allterms_begin(); 
             term != db->allterms_end();term++) {
            if (op_flags & OPT_f)
            cout << term.get_termfreq() << " ";
            cout << op << detailstring(*term) << cl << endl;
        }
        }
    } else if (op_flags & OPT_s) {
        for (unsigned int docid = 1;
         docid < db->get_lastdocid(); docid++) {
        // cout << docid << ": ";
        Xapian::TermIterator term;
        for (term = db->termlist_begin(docid); 
             term != db->termlist_end(docid);term++) {
            cout << detailstring(*term) << " ";
            Xapian::Document doc = db->get_document(docid);
            string data = doc.get_data();
            cout << data;
        }
        }
...
        Xapian::PositionIterator pos;
        for (pos = db->positionlist_begin(docid, *term); 
             pos != db->positionlist_end(docid, *term); pos++) {
            if (buf.size() <= *pos)
            buf.resize((*pos)+100);
            buf[(*pos)] = detailstring(*term);
        }
        }
        for (vector<string>::iterator it = buf.begin(); it != buf.end();
         it++) {
        cout << *it << " ";

	a/src/query/xadump.cpp		b/src/query/xadump.cpp
1	#ifndef lint	1	#ifndef lint
2	static char rcsid[] = "@(#$Id: xadump.cpp,v 1.19 2008-12-15 14:39:52 dockes Exp $ (C) 2004 J.F.Dockes";	2	static char rcsid[] = "@(#$Id: xadump.cpp,v 1.20 2008-12-18 14:11:01 dockes Exp $ (C) 2004 J.F.Dockes";
3	#endif	3	#endif
4	/*	4	/*
5	* This program is free software; you can redistribute it and/or modify	5	* This program is free software; you can redistribute it and/or modify
6	* it under the terms of the GNU General Public License as published by	6	* it under the terms of the GNU General Public License as published by
7	* the Free Software Foundation; either version 2 of the License, or	7	* the Free Software Foundation; either version 2 of the License, or
	...		...
27		27
28	#ifndef NO_NAMESPACES	28	#ifndef NO_NAMESPACES
29	using namespace std;	29	using namespace std;
30	#endif /* NO_NAMESPACES */	30	#endif /* NO_NAMESPACES */
31		31
		32	#include "utf8iter.h"
		33
32	#include "xapian.h"	34	#include "xapian.h"
33		35
34	static string thisprog;	36	static string thisprog;
35		37
36	static string usage =	38	static string usage =
	...		...
40	" -i docid -b : 'rebuild' document from term positions\n"	42	" -i docid -b : 'rebuild' document from term positions\n"
41	" -t term -E : term existence test\n"	43	" -t term -E : term existence test\n"
42	" -t term -F : retrieve term frequency data for given term\n"	44	" -t term -F : retrieve term frequency data for given term\n"
43	" -t term -P : retrieve postings for term\n"	45	" -t term -P : retrieve postings for term\n"
44	" -i docid -T : term list for doc docid\n"	46	" -i docid -T : term list for doc docid\n"
		47	" -x : separate each output char with a space\n"
45	" -T : list all terms\n"	48	" -T : list all terms\n"
46	" -f : precede each term in the list with its occurrence count\n"	49	" -f : precede each term in the list with its occurrence count\n"
47	" -n : raw data (no [])\n"	50	" -n : raw data (no [])\n"
48	" -s : special mode to dump recoll stem db\n"	51	" -s : special mode to dump recoll stem db\n"
49	" -q term [term ...] : perform AND query\n"	52	" -q term [term ...] : perform AND query\n"
	...		...
71	#define OPT_s 0x400	74	#define OPT_s 0x400
72	#define OPT_f 0x800	75	#define OPT_f 0x800
73	#define OPT_q 0x1000	76	#define OPT_q 0x1000
74	#define OPT_n 0x2000	77	#define OPT_n 0x2000
75	#define OPT_X 0x4000	78	#define OPT_X 0x4000
		79	#define OPT_x 0x8000
		80
		81	// Compute an exploded version of string, inserting a space between each char.
		82	// (no character combining possible)
		83	static string detailstring(const string& in)
		84	{
		85	if (!(op_flags & OPT_x))
		86	return in;
		87	string out;
		88	Utf8Iter it(in);
		89	for (; !it.eof(); it++) {
		90	it.appendchartostring(out);
		91	out += ' ';
		92	}
		93	// Strip last space
		94	if (!out.empty())
		95	out.resize(out.size()-1);
		96	return out;
		97	}
76		98
77	Xapian::Database *db;	99	Xapian::Database *db;
78		100
79	static void cleanup()	101	static void cleanup()
80	{	102	{
	...		...
130	case 't': op_flags \|= OPT_t; if (argc < 2) Usage();	152	case 't': op_flags \|= OPT_t; if (argc < 2) Usage();
131	aterm = *(++argv);	153	aterm = *(++argv);
132	argc--;	154	argc--;
133	goto b1;	155	goto b1;
134	case 'X': op_flags \|= OPT_X; break;	156	case 'X': op_flags \|= OPT_X; break;
		157	case 'x': op_flags \|= OPT_x; break;
135	default: Usage(); break;	158	default: Usage(); break;
136	}	159	}
137	b1: argc--; argv++;	160	b1: argc--; argv++;
138	}	161	}
139		162
	...		...
172	string op = (op_flags & OPT_n) ? string(): "[";	195	string op = (op_flags & OPT_n) ? string(): "[";
173	string cl = (op_flags & OPT_n) ? string(): "]";	196	string cl = (op_flags & OPT_n) ? string(): "]";
174	if (op_flags & OPT_i) {	197	if (op_flags & OPT_i) {
175	for (term = db->termlist_begin(docid);	198	for (term = db->termlist_begin(docid);
176	term != db->termlist_end(docid);term++) {	199	term != db->termlist_end(docid);term++) {
177	cout << op << *term << cl << endl;	200	cout << op << detailstring(*term) << cl << endl;
178	}	201	}
179	} else {	202	} else {
180	for (term = db->allterms_begin();	203	for (term = db->allterms_begin();
181	term != db->allterms_end();term++) {	204	term != db->allterms_end();term++) {
182	if (op_flags & OPT_f)	205	if (op_flags & OPT_f)
183	cout << term.get_termfreq() << " ";	206	cout << term.get_termfreq() << " ";
184	cout << op << *term << cl << endl;	207	cout << op << detailstring(*term) << cl << endl;
185	}	208	}
186	}	209	}
187	} else if (op_flags & OPT_s) {	210	} else if (op_flags & OPT_s) {
188	for (unsigned int docid = 1;	211	for (unsigned int docid = 1;
189	docid < db->get_lastdocid(); docid++) {	212	docid < db->get_lastdocid(); docid++) {
190	// cout << docid << ": ";	213	// cout << docid << ": ";
191	Xapian::TermIterator term;	214	Xapian::TermIterator term;
192	for (term = db->termlist_begin(docid);	215	for (term = db->termlist_begin(docid);
193	term != db->termlist_end(docid);term++) {	216	term != db->termlist_end(docid);term++) {
194	cout << *term << " ";	217	cout << detailstring(*term) << " ";
195	Xapian::Document doc = db->get_document(docid);	218	Xapian::Document doc = db->get_document(docid);
196	string data = doc.get_data();	219	string data = doc.get_data();
197	cout << data;	220	cout << data;
198	}	221	}
199	}	222	}
	...		...
223	Xapian::PositionIterator pos;	246	Xapian::PositionIterator pos;
224	for (pos = db->positionlist_begin(docid, *term);	247	for (pos = db->positionlist_begin(docid, *term);
225	pos != db->positionlist_end(docid, *term); pos++) {	248	pos != db->positionlist_end(docid, *term); pos++) {
226	if (buf.size() <= *pos)	249	if (buf.size() <= *pos)
227	buf.resize((*pos)+100);	250	buf.resize((*pos)+100);
228	buf[(pos)] = term;	251	buf[(pos)] = detailstring(term);
229	}	252	}
230	}	253	}
231	for (vector<string>::iterator it = buf.begin(); it != buf.end();	254	for (vector<string>::iterator it = buf.begin(); it != buf.end();
232	it++) {	255	it++) {
233	cout << *it << " ";	256	cout << *it << " ";