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

Diff of /src/query/xadump.cpp [7876fb] .. [23140e]

Switch to unified view


...
    "-e <output encoding>\n"
    " -i docid -D : get document data for docid\n"
    " -i docid -X : delete document docid\n"
    " -i docid -b : 'rebuild' document from term positions\n"
    " -i docid -T : term list for doc docid\n"
    " -i docid -r : reconstructed text for docid\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"
    " -T          : list all terms\n"
    "    -f       : precede each term in the list with its occurrence counts\n"
...
#define OPT_q     0x1000
#define OPT_s     0x2000
#define OPT_t     0x4000
#define OPT_x     0x8000
#define OPT_l     0x10000
#define OPT_r     0x20000

// Compute an exploded version of string, inserting a space between each char.
// (no character combining possible)
static string detailstring(const string& in)
{
...
{
    if (o_index_stripchars) {
    return trm.size() && 'A' <= trm[0] && trm[0] <= 'Z';
    } else {
    return trm.size() > 0 && trm[0] == ':';
    }
}


void wholedoc(Xapian::Database* db, int docid)
{
    vector<string> buf;
    Xapian::TermIterator term;
    for (term = db->termlist_begin(docid);
         term != db->termlist_end(docid); term++) {
        Xapian::PositionIterator pos;
        for (pos = db->positionlist_begin(docid, *term);
             pos != db->positionlist_end(docid, *term); pos++) {
            if (buf.size() < *pos)
                buf.resize(2*((*pos)+1));
            buf[(*pos)] = *term;
        }
    }
    for (vector<string>::iterator it = buf.begin(); it != buf.end(); it++) {
        if (!it->empty())
            cout << *it << " ";
    }
}

int main(int argc, char **argv)
{
...
        goto b1;
        case 'l':   op_flags |= OPT_l; break;
        case 'n':   op_flags |= OPT_n; break;
        case 'P':   op_flags |= OPT_P; break;
        case 'q':   op_flags |= OPT_q; break;
      case 'r':   op_flags |= OPT_r; break;
        case 's':   op_flags |= OPT_s; break;
        case 'T':   op_flags |= OPT_T; break;
        case 't':   op_flags |= OPT_t; if (argc < 2)  Usage();
        aterm = *(++argv);
        argc--; 
...
        }
    } else if (op_flags & OPT_D) {
        Xapian::Document doc = db->get_document(docid);
        string data = doc.get_data();
        cout << data << endl;
  } else if (op_flags & OPT_r) {
      wholedoc(db, docid);
    } else if (op_flags & OPT_X) {
        Xapian::Document doc = db->get_document(docid);
        string data = doc.get_data();
        cout << data << endl;
        cout << "Really delete xapian document ?" << endl;

	a/src/query/xadump.cpp		b/src/query/xadump.cpp
	...		...
43	"-e <output encoding>\n"	43	"-e <output encoding>\n"
44	" -i docid -D : get document data for docid\n"	44	" -i docid -D : get document data for docid\n"
45	" -i docid -X : delete document docid\n"	45	" -i docid -X : delete document docid\n"
46	" -i docid -b : 'rebuild' document from term positions\n"	46	" -i docid -b : 'rebuild' document from term positions\n"
47	" -i docid -T : term list for doc docid\n"	47	" -i docid -T : term list for doc docid\n"
		48	" -i docid -r : reconstructed text for docid\n"
48	" -t term -E : term existence test\n"	49	" -t term -E : term existence test\n"
49	" -t term -F : retrieve term frequency data for given term\n"	50	" -t term -F : retrieve term frequency data for given term\n"
50	" -t term -P : retrieve postings for term\n"	51	" -t term -P : retrieve postings for term\n"
51	" -T : list all terms\n"	52	" -T : list all terms\n"
52	" -f : precede each term in the list with its occurrence counts\n"	53	" -f : precede each term in the list with its occurrence counts\n"
	...		...
81	#define OPT_q 0x1000	82	#define OPT_q 0x1000
82	#define OPT_s 0x2000	83	#define OPT_s 0x2000
83	#define OPT_t 0x4000	84	#define OPT_t 0x4000
84	#define OPT_x 0x8000	85	#define OPT_x 0x8000
85	#define OPT_l 0x10000	86	#define OPT_l 0x10000
		87	#define OPT_r 0x20000
86		88
87	// Compute an exploded version of string, inserting a space between each char.	89	// Compute an exploded version of string, inserting a space between each char.
88	// (no character combining possible)	90	// (no character combining possible)
89	static string detailstring(const string& in)	91	static string detailstring(const string& in)
90	{	92	{
	...		...
122	{	124	{
123	if (o_index_stripchars) {	125	if (o_index_stripchars) {
124	return trm.size() && 'A' <= trm[0] && trm[0] <= 'Z';	126	return trm.size() && 'A' <= trm[0] && trm[0] <= 'Z';
125	} else {	127	} else {
126	return trm.size() > 0 && trm[0] == ':';	128	return trm.size() > 0 && trm[0] == ':';
		129	}
		130	}
		131
		132
		133	void wholedoc(Xapian::Database* db, int docid)
		134	{
		135	vector<string> buf;
		136	Xapian::TermIterator term;
		137	for (term = db->termlist_begin(docid);
		138	term != db->termlist_end(docid); term++) {
		139	Xapian::PositionIterator pos;
		140	for (pos = db->positionlist_begin(docid, *term);
		141	pos != db->positionlist_end(docid, *term); pos++) {
		142	if (buf.size() < *pos)
		143	buf.resize(2((pos)+1));
		144	buf[(pos)] = term;
		145	}
		146	}
		147	for (vector<string>::iterator it = buf.begin(); it != buf.end(); it++) {
		148	if (!it->empty())
		149	cout << *it << " ";
127	}	150	}
128	}	151	}
129		152
130	int main(int argc, char **argv)	153	int main(int argc, char **argv)
131	{	154	{
	...		...
163	goto b1;	186	goto b1;
164	case 'l': op_flags \|= OPT_l; break;	187	case 'l': op_flags \|= OPT_l; break;
165	case 'n': op_flags \|= OPT_n; break;	188	case 'n': op_flags \|= OPT_n; break;
166	case 'P': op_flags \|= OPT_P; break;	189	case 'P': op_flags \|= OPT_P; break;
167	case 'q': op_flags \|= OPT_q; break;	190	case 'q': op_flags \|= OPT_q; break;
		191	case 'r': op_flags \|= OPT_r; break;
168	case 's': op_flags \|= OPT_s; break;	192	case 's': op_flags \|= OPT_s; break;
169	case 'T': op_flags \|= OPT_T; break;	193	case 'T': op_flags \|= OPT_T; break;
170	case 't': op_flags \|= OPT_t; if (argc < 2) Usage();	194	case 't': op_flags \|= OPT_t; if (argc < 2) Usage();
171	aterm = *(++argv);	195	aterm = *(++argv);
172	argc--;	196	argc--;
	...		...
257	}	281	}
258	} else if (op_flags & OPT_D) {	282	} else if (op_flags & OPT_D) {
259	Xapian::Document doc = db->get_document(docid);	283	Xapian::Document doc = db->get_document(docid);
260	string data = doc.get_data();	284	string data = doc.get_data();
261	cout << data << endl;	285	cout << data << endl;
		286	} else if (op_flags & OPT_r) {
		287	wholedoc(db, docid);
262	} else if (op_flags & OPT_X) {	288	} else if (op_flags & OPT_X) {
263	Xapian::Document doc = db->get_document(docid);	289	Xapian::Document doc = db->get_document(docid);
264	string data = doc.get_data();	290	string data = doc.get_data();
265	cout << data << endl;	291	cout << data << endl;
266	cout << "Really delete xapian document ?" << endl;	292	cout << "Really delete xapian document ?" << endl;