SQLite

  $(TOP)/src/test_syscall.c \
  $(TOP)/src/test_stat.c \
  $(TOP)/src/test_tclvar.c \
  $(TOP)/src/test_thread.c \
  $(TOP)/src/test_vfs.c \
  $(TOP)/src/test_wholenumber.c \
  $(TOP)/src/test_wsd.c       \
  $(TOP)/ext/fts3/fts3_term.c \
  $(TOP)/ext/fts3/fts3_test.c 

# Source code to the library files needed by the test fixture
#
TESTSRC2 = \
  $(TOP)/src/attach.c \
  $(TOP)/src/backup.c \
  $(TOP)/src/bitvec.c \

/*
** When called, *ppPoslist must point to the byte immediately following the
** end of a position-list. i.e. ( (*ppPoslist)[-1]==POS_END ). This function
** moves *ppPoslist so that it instead points to the first byte of the
** same position list.
*/
static void fts3ReversePoslist(char *pStart, char **ppPoslist){
  char *p = &(*ppPoslist)[-2];
  char c;

  while( p>pStart && (c=*p--)==0 );
  while( p>pStart && (*p & 0x80) | c ){ 
    c = *p--; 
  }
  if( p>pStart ){ p = &p[2]; }
  while( *p++&0x80 );
  *ppPoslist = p;
}

    fts3PoslistCopy(0, &pDocid);
    while( pDocid<pEnd ){
      sqlite3_int64 iDelta;
      pDocid += sqlite3Fts3GetVarint(pDocid, &iDelta);
      iDocid += (iMul * iDelta);
      pNext = pDocid;
      fts3PoslistCopy(0, &pDocid);
      while( pDocid<pEnd && *pDocid==0 ) pDocid++;
    }

    *pnList = pEnd - pNext;
    *ppIter = pNext;
    *piDocid = iDocid;
  }else{
    sqlite3_int64 iDelta;

/*
** 2011 Jun 13
**
** The author disclaims copyright to this source code.  In place of
** a legal notice, here is a blessing:
**
**    May you do good and not evil.
**    May you find forgiveness for yourself and forgive others.
**    May you share freely, never taking more than you give.
**
******************************************************************************
**
** This file is not part of the production FTS code. It is only used for
** testing. It contains a Tcl command that can be used to test if a document
** matches an FTS NEAR expression.
*/

#include <tcl.h>
#include <string.h>
#include <assert.h>

#define NM_MAX_TOKEN 12

typedef struct NearPhrase NearPhrase;
typedef struct NearDocument NearDocument;
typedef struct NearToken NearToken;

struct NearDocument {
  int nToken;                     /* Length of token in bytes */
  NearToken *aToken;              /* Token array */
};

struct NearToken {
  int n;                          /* Length of token in bytes */
  const char *z;                  /* Pointer to token string */
};

struct NearPhrase {
  int nNear;                      /* Preceding NEAR value */
  int nToken;                     /* Number of tokens in this phrase */
  NearToken aToken[NM_MAX_TOKEN]; /* Array of tokens in this phrase */
};

static int nm_phrase_match(
  NearPhrase *p,
  NearToken *aToken
){
  int ii;

  for(ii=0; ii<p->nToken; ii++){
    NearToken *pToken = &p->aToken[ii];
    if( pToken->n>0 && pToken->z[pToken->n-1]=='*' ){
      if( aToken[ii].n<(pToken->n-1) ) return 0;
      if( memcmp(aToken[ii].z, pToken->z, pToken->n-1) ) return 0;
    }else{
      if( aToken[ii].n!=pToken->n ) return 0;
      if( memcmp(aToken[ii].z, pToken->z, pToken->n) ) return 0;
    }
  }

  return 1;
}

static int nm_near_chain(
  int iDir,                       /* Direction to iterate through aPhrase[] */
  NearDocument *pDoc,             /* Document to match against */
  int iPos,                       /* Position at which iPhrase was found */
  int nPhrase,                    /* Size of phrase array */
  NearPhrase *aPhrase,            /* Phrase array */
  int iPhrase                     /* Index of phrase found */
){
  int iStart;
  int iStop;
  int ii;
  int nNear;
  int iPhrase2;
  NearPhrase *p;
  NearPhrase *pPrev;

  assert( iDir==1 || iDir==-1 );

  if( iDir==1 ){
    if( (iPhrase+1)==nPhrase ) return 1;
    nNear = aPhrase[iPhrase+1].nNear;
  }else{
    if( iPhrase==0 ) return 1;
    nNear = aPhrase[iPhrase].nNear;
  }
  pPrev = &aPhrase[iPhrase];
  iPhrase2 = iPhrase+iDir;
  p = &aPhrase[iPhrase2];

  iStart = iPos - nNear - p->nToken;
  iStop = iPos + nNear + pPrev->nToken;

  if( iStart<0 ) iStart = 0;
  if( iStop > pDoc->nToken - p->nToken ) iStop = pDoc->nToken - p->nToken;

  for(ii=iStart; ii<=iStop; ii++){
    if( nm_phrase_match(p, &pDoc->aToken[ii]) ){
      if( nm_near_chain(iDir, pDoc, ii, nPhrase, aPhrase, iPhrase2) ) return 1;
    }
  }

  return 0;
}

static int nm_match_count(
  NearDocument *pDoc,             /* Document to match against */
  int nPhrase,                    /* Size of phrase array */
  NearPhrase *aPhrase,            /* Phrase array */
  int iPhrase                     /* Index of phrase to count matches for */
){
  int nOcc = 0;
  int ii;
  NearPhrase *p = &aPhrase[iPhrase];

  for(ii=0; ii<(pDoc->nToken + 1 - p->nToken); ii++){
    if( nm_phrase_match(p, &pDoc->aToken[ii]) ){
      /* Test forward NEAR chain (i>iPhrase) */
      if( 0==nm_near_chain(1, pDoc, ii, nPhrase, aPhrase, iPhrase) ) continue;

      /* Test reverse NEAR chain (i<iPhrase) */
      if( 0==nm_near_chain(-1, pDoc, ii, nPhrase, aPhrase, iPhrase) ) continue;

      /* This is a real match. Increment the counter. */
      nOcc++;
    }
  } 

  return nOcc;
}

/*
** Tclcmd: fts3_near_match DOCUMENT EXPR ?OPTIONS?
*/
static int fts3_near_match_cmd(
  ClientData clientData,
  Tcl_Interp *interp,
  int objc,
  Tcl_Obj *CONST objv[]
){
  int nTotal = 0;
  int rc;
  int ii;
  int nPhrase;
  NearPhrase *aPhrase = 0;
  NearDocument doc = {0, 0};
  Tcl_Obj **apDocToken;
  Tcl_Obj *pRet;
  Tcl_Obj *pPhrasecount = 0;
  
  Tcl_Obj **apExprToken;
  int nExprToken;

  /* Must have 3 or more arguments. */
  if( objc<3 || (objc%2)==0 ){
    Tcl_WrongNumArgs(interp, 1, objv, "DOCUMENT EXPR ?OPTION VALUE?...");
    rc = TCL_ERROR;
    goto near_match_out;
  }

  for(ii=3; ii<objc; ii+=2){
    enum NM_enum { NM_PHRASECOUNTS };
    struct TestnmSubcmd {
      char *zName;
      enum NM_enum eOpt;
    } aOpt[] = {
      { "-phrasecountvar", NM_PHRASECOUNTS },
      { 0, 0 }
    };
    int iOpt;
    if( Tcl_GetIndexFromObjStruct(
        interp, objv[ii], aOpt, sizeof(aOpt[0]), "option", 0, &iOpt) 
    ){
      return TCL_ERROR;
    }

    switch( aOpt[iOpt].eOpt ){
      case NM_PHRASECOUNTS:
        pPhrasecount = objv[ii+1];
        break;
    }
  }

  rc = Tcl_ListObjGetElements(interp, objv[1], &doc.nToken, &apDocToken);
  if( rc!=TCL_OK ) goto near_match_out;
  doc.aToken = (NearToken *)ckalloc(doc.nToken*sizeof(NearToken));
  for(ii=0; ii<doc.nToken; ii++){
    doc.aToken[ii].z = Tcl_GetStringFromObj(apDocToken[ii], &doc.aToken[ii].n);
  }

  rc = Tcl_ListObjGetElements(interp, objv[2], &nExprToken, &apExprToken);
  if( rc!=TCL_OK ) goto near_match_out;

  nPhrase = (nExprToken + 1) / 2;
  aPhrase = (NearPhrase *)ckalloc(nPhrase * sizeof(NearPhrase));
  memset(aPhrase, 0, nPhrase * sizeof(NearPhrase));
  for(ii=0; ii<nPhrase; ii++){
    Tcl_Obj *pPhrase = apExprToken[ii*2];
    Tcl_Obj **apToken;
    int nToken;
    int jj;

    rc = Tcl_ListObjGetElements(interp, pPhrase, &nToken, &apToken);
    if( rc!=TCL_OK ) goto near_match_out;
    if( nToken>NM_MAX_TOKEN ){
      Tcl_AppendResult(interp, "Too many tokens in phrase", 0);
      rc = TCL_ERROR;
      goto near_match_out;
    }
    for(jj=0; jj<nToken; jj++){
      NearToken *pT = &aPhrase[ii].aToken[jj];
      pT->z = Tcl_GetStringFromObj(apToken[jj], &pT->n);
    }
    aPhrase[ii].nToken = nToken;
  }
  for(ii=1; ii<nPhrase; ii++){
    Tcl_Obj *pNear = apExprToken[2*ii-1];
    int nNear;
    rc = Tcl_GetIntFromObj(interp, pNear, &nNear);
    if( rc!=TCL_OK ) goto near_match_out;
    aPhrase[ii].nNear = nNear;
  }

  pRet = Tcl_NewObj();
  Tcl_IncrRefCount(pRet);
  for(ii=0; ii<nPhrase; ii++){
    int nOcc = nm_match_count(&doc, nPhrase, aPhrase, ii);
    Tcl_ListObjAppendElement(interp, pRet, Tcl_NewIntObj(nOcc));
    nTotal += nOcc;
  }
  if( pPhrasecount ){
    Tcl_ObjSetVar2(interp, pPhrasecount, 0, pRet, 0);
  }
  Tcl_DecrRefCount(pRet);
  Tcl_SetObjResult(interp, Tcl_NewBooleanObj(nTotal>0));

 near_match_out: 
  ckfree((char *)aPhrase);
  ckfree((char *)doc.aToken);
  return rc;
}

int Sqlitetestfts3_Init(Tcl_Interp *interp){
  Tcl_CreateObjCommand(interp, "fts3_near_match", fts3_near_match_cmd, 0, 0);
  return TCL_OK;
}

  parse.h \
  sqlite3.h


# Source code to the test files.
#
TESTSRC = \
  $(TOP)/ext/fts3/fts3_test.c \
  $(TOP)/src/test1.c \
  $(TOP)/src/test2.c \
  $(TOP)/src/test3.c \
  $(TOP)/src/test4.c \
  $(TOP)/src/test5.c \
  $(TOP)/src/test6.c \
  $(TOP)/src/test7.c \

    extern int Sqlitetestrtree_Init(Tcl_Interp*);
    extern int Sqlitequota_Init(Tcl_Interp*);
    extern int Sqlitemultiplex_Init(Tcl_Interp*);
    extern int SqliteSuperlock_Init(Tcl_Interp*);
    extern int SqlitetestSyscall_Init(Tcl_Interp*);
    extern int Sqlitetestfuzzer_Init(Tcl_Interp*);
    extern int Sqlitetestwholenumber_Init(Tcl_Interp*);

#ifdef SQLITE_ENABLE_FTS3
    extern int Sqlitetestfts3_Init(Tcl_Interp *interp);
#endif

#ifdef SQLITE_ENABLE_ZIPVFS
    extern int Zipvfs_Init(Tcl_Interp*);
    Zipvfs_Init(interp);
#endif

    Sqliteconfig_Init(interp);

    Sqlitetestrtree_Init(interp);
    Sqlitequota_Init(interp);
    Sqlitemultiplex_Init(interp);
    SqliteSuperlock_Init(interp);
    SqlitetestSyscall_Init(interp);
    Sqlitetestfuzzer_Init(interp);
    Sqlitetestwholenumber_Init(interp);

#ifdef SQLITE_ENABLE_FTS3
    Sqlitetestfts3_Init(interp);
#endif

    Tcl_CreateObjCommand(interp,"load_testfixture_extensions",init_all_cmd,0,0);

#ifdef SQLITE_SSE
    Sqlitetestsse_Init(interp);
#endif
  }

# 2011 June 10
#
#    May you do good and not evil.
#    May you find forgiveness for yourself and forgive others.
#    May you share freely, never taking more than you give.
#
#***********************************************************************
#

set testdir [file dirname $argv0]
source $testdir/tester.tcl

# If this build does not include FTS3, skip the tests in this file.
#
ifcapable !fts3 { finish_test ; return }
source $testdir/fts3_common.tcl
source $testdir/malloc_common.tcl

set testprefix fts3rnd2

proc test_fts3_near_match {tn doc expr res} {
  fts3_near_match $doc $expr -phrasecountvar p
  uplevel do_test [list $tn] [list [list set {} $p]] [list $res]
}

# Simple test cases for C routine [fts3_near_match].
#
test_fts3_near_match 1.1.1 {a b c a b} a                   {2}
test_fts3_near_match 1.1.2 {a b c a b} {a 5 b 6 c}         {2 2 1}
test_fts3_near_match 1.1.3 {a b c a b} {"a b"}             {2}
test_fts3_near_match 1.1.4 {a b c a b} {"b c"}             {1}
test_fts3_near_match 1.1.5 {a b c a b} {"c c"}             {0}

test_fts3_near_match 1.2.1 "a b c d e f g" {b 2 f}         {0 0}
test_fts3_near_match 1.2.2 "a b c d e f g" {b 3 f}         {1 1}
test_fts3_near_match 1.2.3 "a b c d e f g" {f 2 b}         {0 0}
test_fts3_near_match 1.2.4 "a b c d e f g" {f 3 b}         {1 1}
test_fts3_near_match 1.2.5 "a b c d e f g" {"a b" 2 "f g"} {0 0}
test_fts3_near_match 1.2.6 "a b c d e f g" {"a b" 3 "f g"} {1 1}

set A "a b c d e f g h i j k l m n o p q r s t u v w x y z"
test_fts3_near_match 1.3.1 $A {"c d" 5 "i j" 1 "e f"}      {0 0 0}
test_fts3_near_match 1.3.2 $A {"c d" 5 "i j" 2 "e f"}      {1 1 1}

proc mit {blob} {
  set scan(littleEndian) i*
  set scan(bigEndian) I*
  binary scan $blob $scan($::tcl_platform(byteOrder)) r
  return $r
}
db func mit mit

proc fix_near_expr {expr} { 
  set out [list]
  lappend out [lindex $expr 0]
  foreach {a b} [lrange $expr 1 end] {
    if {[string match -nocase near $a]}   { set a 10 }
    if {[string match -nocase near/* $a]} { set a [string range $a 5 end] }
    lappend out $a
    lappend out $b
  }
  return $out
}

proc do_near_test {tn tbl expr} {

  set expr [fix_near_expr $expr]

  # Create the MATCH expression from $expr
  #
  set match [lindex $expr 0]
  if {[llength $match]>1} {
    set match "\"$match\""
  } 
  foreach {nNear phrase} [lrange $expr 1 end] {
    if {[llength $phrase]>1} {
      append match " NEAR/$nNear \"$phrase\""
    } else {
      append match " NEAR/$nNear $phrase"
    }
  }

  # Calculate the expected results using [fts3_near_match]. The following
  # loop populates the "hits" and "counts" arrays as follows:
  # 
  #   1. For each document in the table that matches the NEAR expression,
  #      hits($docid) is set to 1. The set of docids that match the expression
  #      can therefore be found using [array names hits].
  #
  #   2. For each column of each document in the table, counts($docid,$iCol)
  #      is set to the -phrasecountvar output.
  #
  set res [list]
  catch { array unset hits }
  db eval "SELECT docid, * FROM $tbl" d {
    set iCol 0
    foreach col [lrange $d(*) 1 end] {
      set docid $d(docid)
      set hit [fts3_near_match $d($col) $expr -p counts($docid,$iCol)]
      if {$hit} { set hits($docid) 1 }
      incr iCol
    }
  }
  set nPhrase [expr ([llength $expr]+1)/2]
  set nCol $iCol

  # This block populates the nHit and nDoc arrays. For each phrase/column
  # in the query/table, array elements are set as follows:
  #
  #   nHit($iPhrase,$iCol) - Total number of hits for phrase $iPhrase in 
  #                          column $iCol.
  #
  #   nDoc($iPhrase,$iCol) - Number of documents with at least one hit for
  #                          phrase $iPhrase in column $iCol.
  #
  for {set iPhrase 0} {$iPhrase < $nPhrase} {incr iPhrase} {
    for {set iCol 0} {$iCol < $nCol} {incr iCol} {
      set nHit($iPhrase,$iCol) 0
      set nDoc($iPhrase,$iCol) 0
    }
  }
  foreach key [array names counts] {
    set iCol [lindex [split $key ,] 1]
    set iPhrase 0
    foreach c $counts($key) {
      if {$c>0} { incr nHit($iPhrase,$iCol) 1 }
      incr nDoc($iPhrase,$iCol) $c
      incr iPhrase
    }
  }

  # Set up the aMatchinfo array. For each document, set aMatchinfo($docid) to
  # contain the output of matchinfo('x') for the document.
  #
  foreach docid [array names hits] {
    set mi [list]
    for {set iPhrase 0} {$iPhrase<$nPhrase} {incr iPhrase} {
      for {set iCol 0} {$iCol<$nCol} {incr iCol} {
        lappend mi [lindex $counts($docid,$iCol) $iPhrase]
        lappend mi $nDoc($iPhrase,$iCol)
        lappend mi $nHit($iPhrase,$iCol)
      }
    }
    set aMatchinfo($docid) $mi
  }

  set matchinfo_asc [list]
  foreach docid [lsort -integer -incr [array names aMatchinfo]] {
    lappend matchinfo_asc $docid $aMatchinfo($docid)
  }
  set matchinfo_desc [list]
  foreach docid [lsort -integer -decr [array names aMatchinfo]] {
    lappend matchinfo_desc $docid $aMatchinfo($docid)
  }

  set title "(\"$match\" -> [llength [array names hits]] rows)"

  do_execsql_test $tn$title.1 "
    SELECT docid FROM $tbl WHERE $tbl MATCH '$match' ORDER BY docid ASC
  " [lsort -integer -incr [array names hits]] 

  do_execsql_test $tn$title.2 "
    SELECT docid FROM $tbl WHERE $tbl MATCH '$match' ORDER BY docid DESC
  " [lsort -integer -decr [array names hits]] 

  do_execsql_test $tn$title.3 "
    SELECT docid, mit(matchinfo($tbl, 'x')) FROM $tbl 
    WHERE $tbl MATCH '$match' ORDER BY docid DESC
  " $matchinfo_desc

  do_execsql_test $tn$title.4 "
    SELECT docid, mit(matchinfo($tbl, 'x')) FROM $tbl 
    WHERE $tbl MATCH '$match' ORDER BY docid ASC
  " $matchinfo_asc
} 

do_test 2.1 {
  execsql { CREATE VIRTUAL TABLE t1 USING fts3(a, b) }
  for {set i 0} {$i<32} {incr i} {
    set doc [list]
    if {$i&0x01} {lappend doc one}
    if {$i&0x02} {lappend doc two}
    if {$i&0x04} {lappend doc three}
    if {$i&0x08} {lappend doc four}
    if {$i&0x10} {lappend doc five}
    execsql { INSERT INTO t1 VALUES($doc, null) }
  }
} {}
foreach {tn expr} {
  1     {one}
  2     {one NEAR/1 five}
  3     {t*}
  4     {t* NEAR/0 five}
  5     {o* NEAR/1 f*}
  6     {one NEAR five NEAR two NEAR four NEAR three}
} {
  do_near_test 2.2.$tn t1 $expr
}

finish_test

  } {
    set testname "${::testprefix}-$testname"
  }
}
    
proc do_execsql_test {testname sql {result {}}} {
  fix_testname testname
  uplevel do_test [list $testname] [list "execsql {$sql}"] [list [list {*}$result]]
}
proc do_catchsql_test {testname sql result} {
  fix_testname testname
  uplevel do_test [list $testname] [list "catchsql {$sql}"] [list $result]
}
proc do_eqp_test {name sql res} {
  uplevel do_execsql_test $name [list "EXPLAIN QUERY PLAN $sql"] [list $res]
}

#-------------------------------------------------------------------------
#   Usage: do_select_tests PREFIX ?SWITCHES? TESTLIST