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

Diff of /src/query/xadump.cpp [2a3075] .. [7ad5f2]

Switch to unified view


#ifndef lint
static char rcsid[] = "@(#$Id: xadump.cpp,v 1.9 2006-01-25 08:09:41 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
...
static string thisprog;

static string usage =
    " -d <dbdir> -e <output encoding>\n"
    " -i docid -D : get document data for docid\n"
    " -i docid -b : 'rebuild' document from term positions\n"
    " -t term -E  : term existence test\n"
    " -t term -F  : retrieve term frequency data\n"
    " -t term -P  : retrieve postings for term\n"
    " -i docid -T : term list for doc docid\n"
    " -T          : list all terms\n"
...
#define OPT_D     0x10
#define OPT_t     0x20
#define OPT_P     0x40
#define OPT_F     0x80
#define OPT_E     0x100
#define OPT_b     0x200

Xapian::Database *db;

static void cleanup()
{
...
        case 'D':   op_flags |= OPT_D; break;
        case 'E':   op_flags |= OPT_E; break;
        case 'F':   op_flags |= OPT_F; break;
        case 'P':   op_flags |= OPT_P; break;
        case 'T':   op_flags |= OPT_T; break;
      case 'b':   op_flags |= OPT_b; break;
        case 'd':   op_flags |= OPT_d; if (argc < 2)  Usage();
        dbdir = *(++argv);
        argc--; 
        goto b1;
        case 'e':   op_flags |= OPT_d; if (argc < 2)  Usage();
...
        }
    } 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_b) {
      if (!(op_flags & OPT_i))
      Usage();
      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((*pos)+1);
          buf[(*pos)] = *term;
      }
      }
      for (vector<string>::iterator it = buf.begin(); it != buf.end();
       it++) {
      cout << *it << " ";
      }
    } else if (op_flags & OPT_P) {
        Xapian::PostingIterator doc;
        for (doc = db->postlist_begin(aterm);
         doc != db->postlist_end(aterm); doc++) {
        cout << *doc << " : " ;
      Xapian::PositionIterator pos;
      for (pos = doc.positionlist_begin(); 
           pos != doc.positionlist_end(); pos++) {
          cout << *pos << " " ;
      }
      cout << endl;
        }
        
    } else if (op_flags & OPT_F) {
        cout << "FreqFor " << aterm << " : " <<
        db->get_termfreq(aterm) << endl;

	a/src/query/xadump.cpp		b/src/query/xadump.cpp
1	#ifndef lint	1	#ifndef lint
2	static char rcsid[] = "@(#$Id: xadump.cpp,v 1.8 2006-01-23 13:32:28 dockes Exp $ (C) 2004 J.F.Dockes";	2	static char rcsid[] = "@(#$Id: xadump.cpp,v 1.9 2006-01-25 08:09:41 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
	...		...
35	static string thisprog;	35	static string thisprog;
36		36
37	static string usage =	37	static string usage =
38	" -d <dbdir> -e <output encoding>\n"	38	" -d <dbdir> -e <output encoding>\n"
39	" -i docid -D : get document data for docid\n"	39	" -i docid -D : get document data for docid\n"
		40	" -i docid -b : 'rebuild' document from term positions\n"
40	" -t term -E : term existence test\n"	41	" -t term -E : term existence test\n"
41	" -t term -F : retrieve term frequency data\n"	42	" -t term -F : retrieve term frequency data\n"
42	" -t term -P : retrieve postings for term\n"	43	" -t term -P : retrieve postings for term\n"
43	" -i docid -T : term list for doc docid\n"	44	" -i docid -T : term list for doc docid\n"
44	" -T : list all terms\n"	45	" -T : list all terms\n"
	...		...
60	#define OPT_D 0x10	61	#define OPT_D 0x10
61	#define OPT_t 0x20	62	#define OPT_t 0x20
62	#define OPT_P 0x40	63	#define OPT_P 0x40
63	#define OPT_F 0x80	64	#define OPT_F 0x80
64	#define OPT_E 0x100	65	#define OPT_E 0x100
		66	#define OPT_b 0x200
65		67
66	Xapian::Database *db;	68	Xapian::Database *db;
67		69
68	static void cleanup()	70	static void cleanup()
69	{	71	{
	...		...
97	case 'D': op_flags \|= OPT_D; break;	99	case 'D': op_flags \|= OPT_D; break;
98	case 'E': op_flags \|= OPT_E; break;	100	case 'E': op_flags \|= OPT_E; break;
99	case 'F': op_flags \|= OPT_F; break;	101	case 'F': op_flags \|= OPT_F; break;
100	case 'P': op_flags \|= OPT_P; break;	102	case 'P': op_flags \|= OPT_P; break;
101	case 'T': op_flags \|= OPT_T; break;	103	case 'T': op_flags \|= OPT_T; break;
		104	case 'b': op_flags \|= OPT_b; break;
102	case 'd': op_flags \|= OPT_d; if (argc < 2) Usage();	105	case 'd': op_flags \|= OPT_d; if (argc < 2) Usage();
103	dbdir = *(++argv);	106	dbdir = *(++argv);
104	argc--;	107	argc--;
105	goto b1;	108	goto b1;
106	case 'e': op_flags \|= OPT_d; if (argc < 2) Usage();	109	case 'e': op_flags \|= OPT_d; if (argc < 2) Usage();
	...		...
158	}	161	}
159	} else if (op_flags & OPT_D) {	162	} else if (op_flags & OPT_D) {
160	Xapian::Document doc = db->get_document(docid);	163	Xapian::Document doc = db->get_document(docid);
161	string data = doc.get_data();	164	string data = doc.get_data();
162	cout << data << endl;	165	cout << data << endl;
		166	} else if (op_flags & OPT_b) {
		167	if (!(op_flags & OPT_i))
		168	Usage();
		169	vector<string> buf;
		170	Xapian::TermIterator term;
		171	for (term = db->termlist_begin(docid);
		172	term != db->termlist_end(docid); term++) {
		173	Xapian::PositionIterator pos;
		174	for (pos = db->positionlist_begin(docid, *term);
		175	pos != db->positionlist_end(docid, *term); pos++) {
		176	if (buf.size() < *pos)
		177	buf.resize((*pos)+1);
		178	buf[(pos)] = term;
		179	}
		180	}
		181	for (vector<string>::iterator it = buf.begin(); it != buf.end();
		182	it++) {
		183	cout << *it << " ";
		184	}
163	} else if (op_flags & OPT_P) {	185	} else if (op_flags & OPT_P) {
164	Xapian::PostingIterator doc;	186	Xapian::PostingIterator doc;
165	for (doc = db->postlist_begin(aterm);	187	for (doc = db->postlist_begin(aterm);
166	doc != db->postlist_end(aterm);doc++) {	188	doc != db->postlist_end(aterm); doc++) {
167	cout << *doc << endl;	189	cout << *doc << " : " ;
		190	Xapian::PositionIterator pos;
		191	for (pos = doc.positionlist_begin();
		192	pos != doc.positionlist_end(); pos++) {
		193	cout << *pos << " " ;
		194	}
		195	cout << endl;
168	}	196	}
169		197
170	} else if (op_flags & OPT_F) {	198	} else if (op_flags & OPT_F) {
171	cout << "FreqFor " << aterm << " : " <<	199	cout << "FreqFor " << aterm << " : " <<
172	db->get_termfreq(aterm) << endl;	200	db->get_termfreq(aterm) << endl;