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

Diff of /src/query/wasastringtoquery.cpp [b2515b] .. [4a84b6]

Switch to unified view


...
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <regex.h>

#include "smallut.h"
#include "wasastringtoquery.h"

#undef DEB_WASASTRINGTOQ
#ifdef DEB_WASASTRINGTOQ
#define DPRINT(X) fprintf X
#define DUMPQ(Q) {string D;Q->describe(D);fprintf(stderr, "%s\n", D.c_str());}
#else
#define DPRINT(X)
...
    if (m_modifiers & WQM_BOOST)     desc += "BOOST|";
    if (m_modifiers & WQM_CASESENS)  desc += "CASESENS|";
    if (m_modifiers & WQM_DIACSENS)  desc += "DIACSENS|";
    if (m_modifiers & WQM_FUZZY)     desc += "FUZZY|";
    if (m_modifiers & WQM_NOSTEM)    desc += "NOSTEM|";
    if (m_modifiers & WQM_PHRASESLACK) {
      char buf[100];
      sprintf(buf, "%d", m_slack);
      desc += "PHRASESLACK(" + string(buf) + string(")|");
  }
    if (m_modifiers & WQM_PROX)      desc += "PROX|";
    if (m_modifiers & WQM_REGEX)     desc += "REGEX|";
    if (m_modifiers & WQM_SLOPPY)    desc += "SLOPPY|";
    if (m_modifiers & WQM_WORDS)     desc += "WORDS|";

    if (desc.length() > 0 && desc[desc.length()-1] == '|')
        desc.erase(desc.length()-1);
    }
    desc += " ";
}

// The string query parser code:
...
}

WasaQuery *
StringToWasaQuery::stringToQuery(const string& str, string& reason)
{
    if (internal == 0)
  return 0;
    WasaQuery *wq = internal->stringToQuery(str, reason);
    DUMPQ(wq);
    return wq;
}

WasaQuery *
StringToWasaQuery::Internal::stringToQuery(const string& str, string& reason)
{
...
        reason = "Out of memory";
        return 0;
        }

        // Check for quoted or unquoted value
      unsigned int mods = 0;
        if (checkSubMatch(SMI_QUOTED, match, reason)) {
        nclause->m_value = match;
        } else if (checkSubMatch(SMI_TERM, match, reason)) {
        nclause->m_value = match;
        }
...
        goto nextfield;
        }
        
        if (checkSubMatch(SMI_MODIF, match, reason)) {
        DPRINT((stderr, "Got modifiers: [%s]\n", match));

        for (unsigned int i = 0; i < strlen(match); i++) {
            switch (match[i]) {
            case 'b': 
            mods |= WasaQuery::WQM_BOOST; 
            nclause->m_weight = 10.0;
...
                WasaQuery::WQM_NOSTEM; 
            break;
            case 'f': mods |= WasaQuery::WQM_FUZZY; break;
            case 'l': mods |= WasaQuery::WQM_NOSTEM; break;
            case 'L': break;
          case 'o': 
          mods |= WasaQuery::WQM_PHRASESLACK; 
          // Default slack if specified only by 'o' is 10.
          nclause->m_slack = 10;
          if (i < strlen(match) - 1) {
              char *endptr;
              int slack = strtol(match+i+1, &endptr, 10);
              if (endptr != match+i+1) {
              i += endptr - (match+i+1);
              nclause->m_slack = slack;
              }
          }
          break;
            case 'p': mods |= WasaQuery::WQM_PROX; break;
            case 'r': mods |= WasaQuery::WQM_REGEX; break;
            case 's': mods |= WasaQuery::WQM_SLOPPY; break;
            case 'w': mods |= WasaQuery::WQM_WORDS; break;
            case '.':case '0':case '1':case '2':case '3':case '4':
...
            if (n)
                i += n-1;
            }
            }
        }
      }
        nclause->m_modifiers = WasaQuery::Modifier(mods);


        // Field indicator ?
        if (checkSubMatch(SMI_FIELD, match, reason)) {
        // We used Check for special fields indicating sorting
        // etc. here but this went away from the spec. See 1.4

	a/src/query/wasastringtoquery.cpp		b/src/query/wasastringtoquery.cpp
	...		...
18	#include <stdio.h>	18	#include <stdio.h>
19	#include <stdlib.h>	19	#include <stdlib.h>
20	#include <string.h>	20	#include <string.h>
21	#include <regex.h>	21	#include <regex.h>
22		22
		23	#include "smallut.h"
23	#include "wasastringtoquery.h"	24	#include "wasastringtoquery.h"
24		25
25	// #define DEB_WASASTRINGTOQ 1	26	#undef DEB_WASASTRINGTOQ
26	#ifdef DEB_WASASTRINGTOQ	27	#ifdef DEB_WASASTRINGTOQ
27	#define DPRINT(X) fprintf X	28	#define DPRINT(X) fprintf X
28	#define DUMPQ(Q) {string D;Q->describe(D);fprintf(stderr, "%s\n", D.c_str());}	29	#define DUMPQ(Q) {string D;Q->describe(D);fprintf(stderr, "%s\n", D.c_str());}
29	#else	30	#else
30	#define DPRINT(X)	31	#define DPRINT(X)
	...		...
87	if (m_modifiers & WQM_BOOST) desc += "BOOST\|";	88	if (m_modifiers & WQM_BOOST) desc += "BOOST\|";
88	if (m_modifiers & WQM_CASESENS) desc += "CASESENS\|";	89	if (m_modifiers & WQM_CASESENS) desc += "CASESENS\|";
89	if (m_modifiers & WQM_DIACSENS) desc += "DIACSENS\|";	90	if (m_modifiers & WQM_DIACSENS) desc += "DIACSENS\|";
90	if (m_modifiers & WQM_FUZZY) desc += "FUZZY\|";	91	if (m_modifiers & WQM_FUZZY) desc += "FUZZY\|";
91	if (m_modifiers & WQM_NOSTEM) desc += "NOSTEM\|";	92	if (m_modifiers & WQM_NOSTEM) desc += "NOSTEM\|";
92	if (m_modifiers & WQM_PHRASESLACK) desc += "PHRASESLACK\|";	93	if (m_modifiers & WQM_PHRASESLACK) {
		94	char buf[100];
		95	sprintf(buf, "%d", m_slack);
		96	desc += "PHRASESLACK(" + string(buf) + string(")\|");
		97	}
93	if (m_modifiers & WQM_PROX) desc += "PROX\|";	98	if (m_modifiers & WQM_PROX) desc += "PROX\|";
94	if (m_modifiers & WQM_REGEX) desc += "REGEX\|";	99	if (m_modifiers & WQM_REGEX) desc += "REGEX\|";
95	if (m_modifiers & WQM_SLOPPY) desc += "SLOPPY\|";	100	if (m_modifiers & WQM_SLOPPY) desc += "SLOPPY\|";
96	if (m_modifiers & WQM_WORDS) desc += "WORDS\|";	101	if (m_modifiers & WQM_WORDS) desc += "WORDS\|";
		102
97	if (desc.length() > 0 && desc[desc.length()-1] == '\|')	103	if (desc.length() > 0 && desc[desc.length()-1] == '\|')
98	desc = desc.substr(0, desc.length()-1);	104	desc.erase(desc.length()-1);
99	}	105	}
100	desc += " ";	106	desc += " ";
101	}	107	}
102		108
103	// The string query parser code:	109	// The string query parser code:
	...		...
222	}	228	}
223		229
224	WasaQuery *	230	WasaQuery *
225	StringToWasaQuery::stringToQuery(const string& str, string& reason)	231	StringToWasaQuery::stringToQuery(const string& str, string& reason)
226	{	232	{
		233	if (internal == 0)
		234	return 0;
227	return internal ? internal->stringToQuery(str, reason) : 0;	235	WasaQuery *wq = internal->stringToQuery(str, reason);
		236	DUMPQ(wq);
		237	return wq;
228	}	238	}
229		239
230	WasaQuery *	240	WasaQuery *
231	StringToWasaQuery::Internal::stringToQuery(const string& str, string& reason)	241	StringToWasaQuery::Internal::stringToQuery(const string& str, string& reason)
232	{	242	{
	...		...
314	reason = "Out of memory";	324	reason = "Out of memory";
315	return 0;	325	return 0;
316	}	326	}
317		327
318	// Check for quoted or unquoted value	328	// Check for quoted or unquoted value
		329	unsigned int mods = 0;
319	if (checkSubMatch(SMI_QUOTED, match, reason)) {	330	if (checkSubMatch(SMI_QUOTED, match, reason)) {
320	nclause->m_value = match;	331	nclause->m_value = match;
321	} else if (checkSubMatch(SMI_TERM, match, reason)) {	332	} else if (checkSubMatch(SMI_TERM, match, reason)) {
322	nclause->m_value = match;	333	nclause->m_value = match;
323	}	334	}
	...		...
330	goto nextfield;	341	goto nextfield;
331	}	342	}
332		343
333	if (checkSubMatch(SMI_MODIF, match, reason)) {	344	if (checkSubMatch(SMI_MODIF, match, reason)) {
334	DPRINT((stderr, "Got modifiers: [%s]\n", match));	345	DPRINT((stderr, "Got modifiers: [%s]\n", match));
335	unsigned int mods = 0;
336	for (unsigned int i = 0; i < strlen(match); i++) {	346	for (unsigned int i = 0; i < strlen(match); i++) {
337	switch (match[i]) {	347	switch (match[i]) {
338	case 'b':	348	case 'b':
339	mods \|= WasaQuery::WQM_BOOST;	349	mods \|= WasaQuery::WQM_BOOST;
340	nclause->m_weight = 10.0;	350	nclause->m_weight = 10.0;
	...		...
348	WasaQuery::WQM_NOSTEM;	358	WasaQuery::WQM_NOSTEM;
349	break;	359	break;
350	case 'f': mods \|= WasaQuery::WQM_FUZZY; break;	360	case 'f': mods \|= WasaQuery::WQM_FUZZY; break;
351	case 'l': mods \|= WasaQuery::WQM_NOSTEM; break;	361	case 'l': mods \|= WasaQuery::WQM_NOSTEM; break;
352	case 'L': break;	362	case 'L': break;
		363	case 'o':
353	case 'o': mods \|= WasaQuery::WQM_PHRASESLACK; break;	364	mods \|= WasaQuery::WQM_PHRASESLACK;
		365	// Default slack if specified only by 'o' is 10.
		366	nclause->m_slack = 10;
		367	if (i < strlen(match) - 1) {
		368	char *endptr;
		369	int slack = strtol(match+i+1, &endptr, 10);
		370	if (endptr != match+i+1) {
		371	i += endptr - (match+i+1);
		372	nclause->m_slack = slack;
		373	}
		374	}
		375	break;
354	case 'p': mods \|= WasaQuery::WQM_PROX; break;	376	case 'p': mods \|= WasaQuery::WQM_PROX; break;
355	case 'r': mods \|= WasaQuery::WQM_REGEX; break;	377	case 'r': mods \|= WasaQuery::WQM_REGEX; break;
356	case 's': mods \|= WasaQuery::WQM_SLOPPY; break;	378	case 's': mods \|= WasaQuery::WQM_SLOPPY; break;
357	case 'w': mods \|= WasaQuery::WQM_WORDS; break;	379	case 'w': mods \|= WasaQuery::WQM_WORDS; break;
358	case '.':case '0':case '1':case '2':case '3':case '4':	380	case '.':case '0':case '1':case '2':case '3':case '4':
	...		...
368	if (n)	390	if (n)
369	i += n-1;	391	i += n-1;
370	}	392	}
371	}	393	}
372	}	394	}
		395	}
373	nclause->m_modifiers = WasaQuery::Modifier(mods);	396	nclause->m_modifiers = WasaQuery::Modifier(mods);
374	}
375		397
376	// Field indicator ?	398	// Field indicator ?
377	if (checkSubMatch(SMI_FIELD, match, reason)) {	399	if (checkSubMatch(SMI_FIELD, match, reason)) {
378	// We used Check for special fields indicating sorting	400	// We used Check for special fields indicating sorting
379	// etc. here but this went away from the spec. See 1.4	401	// etc. here but this went away from the spec. See 1.4