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

Diff of /src/query/wasatorcl.cpp [bf120c] .. [e883c4]

Switch to unified view


...
#include "smallut.h"
#include "rclconfig.h"
#include "refcntr.h"
#include "textsplit.h"

static Rcl::SearchData *wasaQueryToRcl(RclConfig *config, WasaQuery *wasa, 
                       const string& autosuffs, string& reason)

{


    if (wasa == 0) {
  reason = "NULL query";



    return 0;
    }








    if (wasa->m_op != WasaQuery::OP_AND && wasa->m_op != WasaQuery::OP_OR) {
  reason = "Top query neither AND nor OR ?";
    LOGERR(("wasaQueryToRcl: top query neither AND nor OR!\n"));
    return 0;
    }

    Rcl::SearchData *sdata = new 
...
         "AND" : "OR"));

    WasaQuery::subqlist_t::iterator it;
    Rcl::SearchDataClause *nclause;

    // Walk the list of clauses. Some pseudo-field types need special
    // processing, which results in setting data in the top struct
    // instead of adding a clause. We check for these first
    for (it = wasa->m_subs.begin(); it != wasa->m_subs.end(); it++) {

  if (!stringicmp("mime", (*it)->m_fieldspec) ||
      !stringicmp("format", (*it)->m_fieldspec)) {
      if ((*it)->m_op != WasaQuery::OP_LEAF) {
      reason = "Negative mime/format clauses not supported yet";
      return 0;
      }
      sdata->addFiletype((*it)->m_value);
      continue;
  } 

  // Xesam uses "type", we also support "rclcat", for broad
  // categories like "audio", "presentation", etc.
  if (!stringicmp("rclcat", (*it)->m_fieldspec) ||
      !stringicmp("type", (*it)->m_fieldspec)) {
      if ((*it)->m_op != WasaQuery::OP_LEAF) {
      reason = "Negative rclcat/type clauses not supported yet";
      return 0;
      }
      list<string> mtypes;
      if (config && config->getMimeCatTypes((*it)->m_value, mtypes)
      && !mtypes.empty()) {
      for (list<string>::iterator mit = mtypes.begin();
           mit != mtypes.end(); mit++) {
          sdata->addFiletype(*mit);
      }
      } else {
      reason = "Unknown rclcat/type value: no mime types found";
      return 0;
      }
      continue;
  }

  // Filtering on location
  if (!stringicmp("dir", (*it)->m_fieldspec)) {
      sdata->setTopdir((*it)->m_value, (*it)->m_op == WasaQuery::OP_EXCL);
      continue;
  } 

  // Handle "date" spec
  if (!stringicmp("date", (*it)->m_fieldspec)) {
      if ((*it)->m_op != WasaQuery::OP_LEAF) {
      reason = "Negative date filtering not supported";
      return 0;
      }
      DateInterval di;
      if (!parsedateinterval((*it)->m_value, &di)) {
      LOGERR(("wasaQueryToRcl: bad date interval format\n"));
      reason = "Bad date interval format";
      return 0;
      }
      LOGDEB(("wasaQueryToRcl:: date span:  %d-%d-%d/%d-%d-%d\n",
          di.y1,di.m1,di.d1, di.y2,di.m2,di.d2));
      sdata->setDateSpan(&di);
      continue;
  } 

  // "Regular" processing follows:
    switch ((*it)->m_op) {
    case WasaQuery::OP_NULL:
    case WasaQuery::OP_AND:
    default:
      reason = "Found bad NULL or AND query type in list";
        LOGERR(("wasaQueryToRcl: found bad NULL or AND q type in list\n"));
        continue;

    case WasaQuery::OP_LEAF: {
        LOGDEB2(("wasaQueryToRcl: leaf clause [%s]:[%s]\n", 
             (*it)->m_fieldspec.c_str(), (*it)->m_value.c_str()));




























            // Change terms found in the "autosuffs" list into "ext"
            // field queries
            if ((*it)->m_fieldspec.empty() && !autosuffs.empty()) {
                vector<string> asfv;
...
                        (*it)->m_modifiers |= WasaQuery::WQM_NOSTEM;
                    }
                }
            }
















        unsigned int mods = (unsigned int)(*it)->m_modifiers;

        if (TextSplit::hasVisibleWhite((*it)->m_value)) {
        int slack = (mods & WasaQuery::WQM_PHRASESLACK) ? 10 : 0;
        Rcl::SClType tp = Rcl::SCLT_PHRASE;
...
        nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_AND, 
                              (*it)->m_value, 
                              (*it)->m_fieldspec);
        }
        if (nclause == 0) {
      reason = "Out of memory";
        LOGERR(("wasaQueryToRcl: out of memory\n"));
        return 0;
        }
        if (mods & WasaQuery::WQM_NOSTEM) {
        nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
...
        // Note: have to add dquotes which will be translated to
        // phrase if there are several words in there. Not pretty
        // but should work. If there is actually a single
        // word, it will not be taken as a phrase, and
        // stem-expansion will work normally
            // Have to do this because searchdata has nothing like and_not
        nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_EXCL, 
                              string("\"") + 
                              (*it)->m_value + "\"",
                              (*it)->m_fieldspec);
        
        if (nclause == 0) {
      reason = "Out of memory";
        LOGERR(("wasaQueryToRcl: out of memory\n"));
        return 0;
        }
        if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
        nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
...

    case WasaQuery::OP_OR:
        LOGDEB2(("wasaQueryToRcl: OR clause [%s]:[%s]\n", 
             (*it)->m_fieldspec.c_str(), (*it)->m_value.c_str()));
        // Create a subquery.
      Rcl::SearchData *sub = 
      wasaQueryToRcl(config, *it, autosuffs, reason);
        if (sub == 0) {
        continue;
        }
        nclause = 
        new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB, 
                         RefCntr<Rcl::SearchData>(sub));
        if (nclause == 0) {
        LOGERR(("wasaQueryToRcl: out of memory\n"));
      reason = "Out of memory";
        return 0;
        }
        if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
        nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
        sdata->addClause(nclause);
    }
    }

    return sdata;
}

Rcl::SearchData *wasaStringToRcl(RclConfig *config, 
               const string &qs, string &reason, 
                                 const string& autosuffs)
{
    StringToWasaQuery parser;
    WasaQuery *wq = parser.stringToQuery(qs, reason);
    if (wq == 0) 
  return 0;
    return wasaQueryToRcl(config, wq, autosuffs, reason);
}

	a/src/query/wasatorcl.cpp		b/src/query/wasatorcl.cpp
	...		...
33	#include "smallut.h"	33	#include "smallut.h"
34	#include "rclconfig.h"	34	#include "rclconfig.h"
35	#include "refcntr.h"	35	#include "refcntr.h"
36	#include "textsplit.h"	36	#include "textsplit.h"
37		37
38	Rcl::SearchData wasaStringToRcl(RclConfig config,	38	static Rcl::SearchData wasaQueryToRcl(RclConfig config, WasaQuery *wasa,
39	const string &qs, string &reason,	39	const string& autosuffs, string& reason)
40	const string& autosuffs)
41	{	40	{
42	StringToWasaQuery parser;
43	WasaQuery *wq = parser.stringToQuery(qs, reason);
44	if (wq == 0)	41	if (wasa == 0) {
45	return 0;	42	reason = "NULL query";
46	Rcl::SearchData *rq = wasaQueryToRcl(config, wq, autosuffs);
47	if (rq == 0) {
48	reason = "Failed translating xesam query structure to recoll";
49	return 0;	43	return 0;
50	}	44	}
51	return rq;
52	}
53
54	Rcl::SearchData wasaQueryToRcl(RclConfig config,
55	WasaQuery *wasa, const string& autosuffs)
56	{
57	if (wasa == 0)
58	return 0;
59	if (wasa->m_op != WasaQuery::OP_AND && wasa->m_op != WasaQuery::OP_OR) {	45	if (wasa->m_op != WasaQuery::OP_AND && wasa->m_op != WasaQuery::OP_OR) {
		46	reason = "Top query neither AND nor OR ?";
60	LOGERR(("wasaQueryToRcl: top query neither AND nor OR!\n"));	47	LOGERR(("wasaQueryToRcl: top query neither AND nor OR!\n"));
61	return 0;	48	return 0;
62	}	49	}
63		50
64	Rcl::SearchData *sdata = new	51	Rcl::SearchData *sdata = new
	...		...
68	"AND" : "OR"));	55	"AND" : "OR"));
69		56
70	WasaQuery::subqlist_t::iterator it;	57	WasaQuery::subqlist_t::iterator it;
71	Rcl::SearchDataClause *nclause;	58	Rcl::SearchDataClause *nclause;
72		59
		60	// Walk the list of clauses. Some pseudo-field types need special
		61	// processing, which results in setting data in the top struct
		62	// instead of adding a clause. We check for these first
73	for (it = wasa->m_subs.begin(); it != wasa->m_subs.end(); it++) {	63	for (it = wasa->m_subs.begin(); it != wasa->m_subs.end(); it++) {
		64
		65	if (!stringicmp("mime", (*it)->m_fieldspec) \|\|
		66	!stringicmp("format", (*it)->m_fieldspec)) {
		67	if ((*it)->m_op != WasaQuery::OP_LEAF) {
		68	reason = "Negative mime/format clauses not supported yet";
		69	return 0;
		70	}
		71	sdata->addFiletype((*it)->m_value);
		72	continue;
		73	}
		74
		75	// Xesam uses "type", we also support "rclcat", for broad
		76	// categories like "audio", "presentation", etc.
		77	if (!stringicmp("rclcat", (*it)->m_fieldspec) \|\|
		78	!stringicmp("type", (*it)->m_fieldspec)) {
		79	if ((*it)->m_op != WasaQuery::OP_LEAF) {
		80	reason = "Negative rclcat/type clauses not supported yet";
		81	return 0;
		82	}
		83	list<string> mtypes;
		84	if (config && config->getMimeCatTypes((*it)->m_value, mtypes)
		85	&& !mtypes.empty()) {
		86	for (list<string>::iterator mit = mtypes.begin();
		87	mit != mtypes.end(); mit++) {
		88	sdata->addFiletype(*mit);
		89	}
		90	} else {
		91	reason = "Unknown rclcat/type value: no mime types found";
		92	return 0;
		93	}
		94	continue;
		95	}
		96
		97	// Filtering on location
		98	if (!stringicmp("dir", (*it)->m_fieldspec)) {
		99	sdata->setTopdir((it)->m_value, (it)->m_op == WasaQuery::OP_EXCL);
		100	continue;
		101	}
		102
		103	// Handle "date" spec
		104	if (!stringicmp("date", (*it)->m_fieldspec)) {
		105	if ((*it)->m_op != WasaQuery::OP_LEAF) {
		106	reason = "Negative date filtering not supported";
		107	return 0;
		108	}
		109	DateInterval di;
		110	if (!parsedateinterval((*it)->m_value, &di)) {
		111	LOGERR(("wasaQueryToRcl: bad date interval format\n"));
		112	reason = "Bad date interval format";
		113	return 0;
		114	}
		115	LOGDEB(("wasaQueryToRcl:: date span: %d-%d-%d/%d-%d-%d\n",
		116	di.y1,di.m1,di.d1, di.y2,di.m2,di.d2));
		117	sdata->setDateSpan(&di);
		118	continue;
		119	}
		120
		121	// "Regular" processing follows:
74	switch ((*it)->m_op) {	122	switch ((*it)->m_op) {
75	case WasaQuery::OP_NULL:	123	case WasaQuery::OP_NULL:
76	case WasaQuery::OP_AND:	124	case WasaQuery::OP_AND:
77	default:	125	default:
		126	reason = "Found bad NULL or AND query type in list";
78	LOGINFO(("wasaQueryToRcl: found bad NULL or AND q type in list\n"));	127	LOGERR(("wasaQueryToRcl: found bad NULL or AND q type in list\n"));
79	continue;	128	continue;
		129
80	case WasaQuery::OP_LEAF: {	130	case WasaQuery::OP_LEAF: {
81	LOGDEB2(("wasaQueryToRcl: leaf clause [%s]:[%s]\n",	131	LOGDEB2(("wasaQueryToRcl: leaf clause [%s]:[%s]\n",
82	(it)->m_fieldspec.c_str(), (it)->m_value.c_str()));	132	(it)->m_fieldspec.c_str(), (it)->m_value.c_str()));
83
84	// Special cases (mime, category, dir filter ...). Not pretty.
85
86	if (!stringicmp("mime", (*it)->m_fieldspec) \|\|
87	!stringicmp("format", (*it)->m_fieldspec)
88	) {
89	sdata->addFiletype((*it)->m_value);
90	break;
91	}
92
93	// Xesam uses "type", we also support "rclcat", for broad
94	// categories like "audio", "presentation", etc.
95	if (!stringicmp("rclcat", (*it)->m_fieldspec) \|\|
96	!stringicmp("type", (*it)->m_fieldspec)) {
97	list<string> mtypes;
98	if (config && config->getMimeCatTypes((*it)->m_value, mtypes)) {
99	for (list<string>::iterator mit = mtypes.begin();
100	mit != mtypes.end(); mit++) {
101	sdata->addFiletype(*mit);
102	}
103	}
104	break;
105	}
106	if (!stringicmp("dir", (*it)->m_fieldspec)) {
107	sdata->setTopdir((*it)->m_value);
108	break;
109	}
110		133
111	// Change terms found in the "autosuffs" list into "ext"	134	// Change terms found in the "autosuffs" list into "ext"
112	// field queries	135	// field queries
113	if ((*it)->m_fieldspec.empty() && !autosuffs.empty()) {	136	if ((*it)->m_fieldspec.empty() && !autosuffs.empty()) {
114	vector<string> asfv;	137	vector<string> asfv;
	...		...
119	(*it)->m_modifiers \|= WasaQuery::WQM_NOSTEM;	142	(*it)->m_modifiers \|= WasaQuery::WQM_NOSTEM;
120	}	143	}
121	}	144	}
122	}	145	}
123		146
124	// Handle "date" spec
125	if (!stringicmp("date", (*it)->m_fieldspec)) {
126	DateInterval di;
127	if (!parsedateinterval((*it)->m_value, &di)) {
128	LOGERR(("wasaQueryToRcl: bad date interval format\n"));
129	// Process rest of query anyway ?
130	break;
131	}
132	LOGDEB(("wasaQueryToRcl:: date span: %d-%d-%d/%d-%d-%d\n",
133	di.y1,di.m1,di.d1, di.y2,di.m2,di.d2))
134	sdata->setDateSpan(&di);
135	break;
136	}
137
138	// "Regular" processing follows:
139	unsigned int mods = (unsigned int)(*it)->m_modifiers;	147	unsigned int mods = (unsigned int)(*it)->m_modifiers;
140		148
141	if (TextSplit::hasVisibleWhite((*it)->m_value)) {	149	if (TextSplit::hasVisibleWhite((*it)->m_value)) {
142	int slack = (mods & WasaQuery::WQM_PHRASESLACK) ? 10 : 0;	150	int slack = (mods & WasaQuery::WQM_PHRASESLACK) ? 10 : 0;
143	Rcl::SClType tp = Rcl::SCLT_PHRASE;	151	Rcl::SClType tp = Rcl::SCLT_PHRASE;
	...		...
152	nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_AND,	160	nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_AND,
153	(*it)->m_value,	161	(*it)->m_value,
154	(*it)->m_fieldspec);	162	(*it)->m_fieldspec);
155	}	163	}
156	if (nclause == 0) {	164	if (nclause == 0) {
		165	reason = "Out of memory";
157	LOGERR(("wasaQueryToRcl: out of memory\n"));	166	LOGERR(("wasaQueryToRcl: out of memory\n"));
158	return 0;	167	return 0;
159	}	168	}
160	if (mods & WasaQuery::WQM_NOSTEM) {	169	if (mods & WasaQuery::WQM_NOSTEM) {
161	nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);	170	nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
	...		...
174	// Note: have to add dquotes which will be translated to	183	// Note: have to add dquotes which will be translated to
175	// phrase if there are several words in there. Not pretty	184	// phrase if there are several words in there. Not pretty
176	// but should work. If there is actually a single	185	// but should work. If there is actually a single
177	// word, it will not be taken as a phrase, and	186	// word, it will not be taken as a phrase, and
178	// stem-expansion will work normally	187	// stem-expansion will work normally
		188	// Have to do this because searchdata has nothing like and_not
179	nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_EXCL,	189	nclause = new Rcl::SearchDataClauseSimple(Rcl::SCLT_EXCL,
180	string("\"") +	190	string("\"") +
181	(*it)->m_value + "\"",	191	(*it)->m_value + "\"",
182	(*it)->m_fieldspec);	192	(*it)->m_fieldspec);
183		193
184	if (nclause == 0) {	194	if (nclause == 0) {
		195	reason = "Out of memory";
185	LOGERR(("wasaQueryToRcl: out of memory\n"));	196	LOGERR(("wasaQueryToRcl: out of memory\n"));
186	return 0;	197	return 0;
187	}	198	}
188	if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)	199	if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
189	nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);	200	nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
	...		...
192		203
193	case WasaQuery::OP_OR:	204	case WasaQuery::OP_OR:
194	LOGDEB2(("wasaQueryToRcl: OR clause [%s]:[%s]\n",	205	LOGDEB2(("wasaQueryToRcl: OR clause [%s]:[%s]\n",
195	(it)->m_fieldspec.c_str(), (it)->m_value.c_str()));	206	(it)->m_fieldspec.c_str(), (it)->m_value.c_str()));
196	// Create a subquery.	207	// Create a subquery.
197	Rcl::SearchData sub = wasaQueryToRcl(config, it);	208	Rcl::SearchData *sub =
		209	wasaQueryToRcl(config, *it, autosuffs, reason);
198	if (sub == 0) {	210	if (sub == 0) {
199	continue;	211	continue;
200	}	212	}
201	nclause =	213	nclause =
202	new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB,	214	new Rcl::SearchDataClauseSub(Rcl::SCLT_SUB,
203	RefCntr<Rcl::SearchData>(sub));	215	RefCntr<Rcl::SearchData>(sub));
204	if (nclause == 0) {	216	if (nclause == 0) {
205	LOGERR(("wasaQueryToRcl: out of memory\n"));	217	LOGERR(("wasaQueryToRcl: out of memory\n"));
		218	reason = "Out of memory";
206	return 0;	219	return 0;
207	}	220	}
208	if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)	221	if ((*it)->m_modifiers & WasaQuery::WQM_NOSTEM)
209	nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);	222	nclause->setModifiers(Rcl::SearchDataClause::SDCM_NOSTEMMING);
210	sdata->addClause(nclause);	223	sdata->addClause(nclause);
211	}	224	}
212	}	225	}
213		226
214	return sdata;	227	return sdata;
215	}	228	}
		229
		230	Rcl::SearchData wasaStringToRcl(RclConfig config,
		231	const string &qs, string &reason,
		232	const string& autosuffs)
		233	{
		234	StringToWasaQuery parser;
		235	WasaQuery *wq = parser.stringToQuery(qs, reason);
		236	if (wq == 0)
		237	return 0;
		238	return wasaQueryToRcl(config, wq, autosuffs, reason);
		239	}