From fffbfc5f085aedbf22fdbf499492f0ef353abf02 Mon Sep 17 00:00:00 2001 From: Deluan Date: Mon, 23 Feb 2026 15:47:30 -0500 Subject: [PATCH] feat(db): statically compile spellfix1 SQLite extension Vendor the spellfix1 extension source and compile it directly into the binary via cgo, avoiding runtime shared library dependencies across platforms. The extension is registered via sqlite3_auto_extension so every new SQLite connection has spellfix1 available. Signed-off-by: Deluan --- .github/workflows/pipeline.yml | 2 +- .golangci.yml | 1 + Dockerfile | 2 +- Makefile | 2 +- conf/buildtags/sqlite_spellfix.go | 8 + db/db.go | 1 + db/spellfix/README.md | 24 + db/spellfix/spellfix.c | 3076 +++++++++++++++++++++++++++++ db/spellfix/spellfix.go | 33 + db/spellfix/spellfix_test.go | 57 + db/spellfix/sqlite3ext.h | 730 +++++++ db/spellfix/update.sh | 31 + main.go | 1 + reflex.conf | 2 +- 14 files changed, 3966 insertions(+), 4 deletions(-) create mode 100644 conf/buildtags/sqlite_spellfix.go create mode 100644 db/spellfix/README.md create mode 100644 db/spellfix/spellfix.c create mode 100644 db/spellfix/spellfix.go create mode 100644 db/spellfix/spellfix_test.go create mode 100644 db/spellfix/sqlite3ext.h create mode 100755 db/spellfix/update.sh diff --git a/.github/workflows/pipeline.yml b/.github/workflows/pipeline.yml index fd8edcd1c..d5fd3cf11 100644 --- a/.github/workflows/pipeline.yml +++ b/.github/workflows/pipeline.yml @@ -117,7 +117,7 @@ jobs: - name: Test run: | pkg-config --define-prefix --cflags --libs taglib # for debugging - go test -shuffle=on -tags netgo,sqlite_fts5 -race ./... -v + go test -shuffle=on -tags netgo,sqlite_fts5,sqlite_spellfix -race ./... -v - name: Test ndpgen run: | diff --git a/.golangci.yml b/.golangci.yml index 1937c2f77..ef22b9ea0 100644 --- a/.golangci.yml +++ b/.golangci.yml @@ -3,6 +3,7 @@ run: build-tags: - netgo - sqlite_fts5 + - sqlite_spellfix linters: enable: - asasalint diff --git a/Dockerfile b/Dockerfile index b32c1df56..50f642934 100644 --- a/Dockerfile +++ b/Dockerfile @@ -109,7 +109,7 @@ RUN --mount=type=bind,source=. \ export EXT=".exe" fi - go build -tags=netgo,sqlite_fts5 -ldflags="${LD_EXTRA} -w -s \ + go build -tags=netgo,sqlite_fts5,sqlite_spellfix -ldflags="${LD_EXTRA} -w -s \ -X github.com/navidrome/navidrome/consts.gitSha=${GIT_SHA} \ -X github.com/navidrome/navidrome/consts.gitTag=${GIT_TAG}" \ -o /out/navidrome${EXT} . diff --git a/Makefile b/Makefile index f7b7b1b05..81d28cd10 100644 --- a/Makefile +++ b/Makefile @@ -1,6 +1,6 @@ GO_VERSION=$(shell grep "^go " go.mod | cut -f 2 -d ' ') NODE_VERSION=$(shell cat .nvmrc) -GO_BUILD_TAGS=netgo,sqlite_fts5 +GO_BUILD_TAGS=netgo,sqlite_fts5,sqlite_spellfix # Set global environment variables, required for most targets export CGO_CFLAGS_ALLOW=--define-prefix diff --git a/conf/buildtags/sqlite_spellfix.go b/conf/buildtags/sqlite_spellfix.go new file mode 100644 index 000000000..99b2d86f6 --- /dev/null +++ b/conf/buildtags/sqlite_spellfix.go @@ -0,0 +1,8 @@ +//go:build sqlite_spellfix + +package buildtags + +// SPELLFIX is required for the spellfix1 virtual table, used for fuzzy/approximate +// string matching. Without this tag, the SQLite driver won't include spellfix1 support. + +var SQLITE_SPELLFIX = true diff --git a/db/db.go b/db/db.go index 0945d1a00..ec6e7336f 100644 --- a/db/db.go +++ b/db/db.go @@ -10,6 +10,7 @@ import ( "github.com/mattn/go-sqlite3" "github.com/navidrome/navidrome/conf" _ "github.com/navidrome/navidrome/db/migrations" + _ "github.com/navidrome/navidrome/db/spellfix" "github.com/navidrome/navidrome/log" "github.com/navidrome/navidrome/utils/hasher" "github.com/navidrome/navidrome/utils/singleton" diff --git a/db/spellfix/README.md b/db/spellfix/README.md new file mode 100644 index 000000000..3f07dd83d --- /dev/null +++ b/db/spellfix/README.md @@ -0,0 +1,24 @@ +# spellfix1 SQLite Extension + +This package statically compiles the [spellfix1](https://sqlite.org/spellfix1.html) SQLite extension +into the Navidrome binary. It is registered via `sqlite3_auto_extension` so that every new SQLite +connection has `spellfix1` available without loading a shared library. + +## Vendored Files + +The C source files are vendored because cgo cannot reference headers from other Go modules: + +- **`spellfix.c`** — from the SQLite source tree: [`ext/misc/spellfix.c`](https://github.com/sqlite/sqlite/blob/master/ext/misc/spellfix.c) +- **`sqlite3ext.h`** — from the SQLite source tree: [`src/sqlite3ext.h`](https://github.com/sqlite/sqlite/blob/master/src/sqlite3ext.h) + +## Updating + +When upgrading `github.com/mattn/go-sqlite3`, run the update script to download +the matching spellfix1 source files for the bundled SQLite version: + +```bash +./db/spellfix/update.sh +``` + +The script reads the SQLite version from go-sqlite3's `sqlite3-binding.h` and +downloads the corresponding files from the [SQLite GitHub mirror](https://github.com/sqlite/sqlite). diff --git a/db/spellfix/spellfix.c b/db/spellfix/spellfix.c new file mode 100644 index 000000000..f51f3bdfe --- /dev/null +++ b/db/spellfix/spellfix.c @@ -0,0 +1,3076 @@ +/* +** 2012 April 10 +** +** 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 module implements the spellfix1 VIRTUAL TABLE that can be used +** to search a large vocabulary for close matches. See separate +** documentation (http://sqlite.org/spellfix1.html) for details. +*/ +#include "sqlite3ext.h" +SQLITE_EXTENSION_INIT1 + +#ifndef SQLITE_AMALGAMATION +# if !defined(NDEBUG) && !defined(SQLITE_DEBUG) +# define NDEBUG 1 +# endif +# if defined(NDEBUG) && defined(SQLITE_DEBUG) +# undef NDEBUG +# endif +# include +# include +# include +# include +# define ALWAYS(X) 1 +# define NEVER(X) 0 + typedef unsigned char u8; + typedef unsigned short u16; +#endif +#include + +#ifndef SQLITE_OMIT_VIRTUALTABLE + +/* +** Character classes for ASCII characters: +** +** 0 '' Silent letters: H W +** 1 'A' Any vowel: A E I O U (Y) +** 2 'B' A bilabeal stop or fricative: B F P V W +** 3 'C' Other fricatives or back stops: C G J K Q S X Z +** 4 'D' Alveolar stops: D T +** 5 'H' Letter H at the beginning of a word +** 6 'L' Glide: L +** 7 'R' Semivowel: R +** 8 'M' Nasals: M N +** 9 'Y' Letter Y at the beginning of a word. +** 10 '9' Digits: 0 1 2 3 4 5 6 7 8 9 +** 11 ' ' White space +** 12 '?' Other. +*/ +#define CCLASS_SILENT 0 +#define CCLASS_VOWEL 1 +#define CCLASS_B 2 +#define CCLASS_C 3 +#define CCLASS_D 4 +#define CCLASS_H 5 +#define CCLASS_L 6 +#define CCLASS_R 7 +#define CCLASS_M 8 +#define CCLASS_Y 9 +#define CCLASS_DIGIT 10 +#define CCLASS_SPACE 11 +#define CCLASS_OTHER 12 + +/* +** The following table gives the character class for non-initial ASCII +** characters. +*/ +static const unsigned char midClass[] = { + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_SPACE, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_SPACE, /* */ CCLASS_SPACE, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_SPACE, + /* ! */ CCLASS_OTHER, /* " */ CCLASS_OTHER, /* # */ CCLASS_OTHER, + /* $ */ CCLASS_OTHER, /* % */ CCLASS_OTHER, /* & */ CCLASS_OTHER, + /* ' */ CCLASS_SILENT, /* ( */ CCLASS_OTHER, /* ) */ CCLASS_OTHER, + /* * */ CCLASS_OTHER, /* + */ CCLASS_OTHER, /* , */ CCLASS_OTHER, + /* - */ CCLASS_OTHER, /* . */ CCLASS_OTHER, /* / */ CCLASS_OTHER, + /* 0 */ CCLASS_DIGIT, /* 1 */ CCLASS_DIGIT, /* 2 */ CCLASS_DIGIT, + /* 3 */ CCLASS_DIGIT, /* 4 */ CCLASS_DIGIT, /* 5 */ CCLASS_DIGIT, + /* 6 */ CCLASS_DIGIT, /* 7 */ CCLASS_DIGIT, /* 8 */ CCLASS_DIGIT, + /* 9 */ CCLASS_DIGIT, /* : */ CCLASS_OTHER, /* ; */ CCLASS_OTHER, + /* < */ CCLASS_OTHER, /* = */ CCLASS_OTHER, /* > */ CCLASS_OTHER, + /* ? */ CCLASS_OTHER, /* @ */ CCLASS_OTHER, /* A */ CCLASS_VOWEL, + /* B */ CCLASS_B, /* C */ CCLASS_C, /* D */ CCLASS_D, + /* E */ CCLASS_VOWEL, /* F */ CCLASS_B, /* G */ CCLASS_C, + /* H */ CCLASS_SILENT, /* I */ CCLASS_VOWEL, /* J */ CCLASS_C, + /* K */ CCLASS_C, /* L */ CCLASS_L, /* M */ CCLASS_M, + /* N */ CCLASS_M, /* O */ CCLASS_VOWEL, /* P */ CCLASS_B, + /* Q */ CCLASS_C, /* R */ CCLASS_R, /* S */ CCLASS_C, + /* T */ CCLASS_D, /* U */ CCLASS_VOWEL, /* V */ CCLASS_B, + /* W */ CCLASS_B, /* X */ CCLASS_C, /* Y */ CCLASS_VOWEL, + /* Z */ CCLASS_C, /* [ */ CCLASS_OTHER, /* \ */ CCLASS_OTHER, + /* ] */ CCLASS_OTHER, /* ^ */ CCLASS_OTHER, /* _ */ CCLASS_OTHER, + /* ` */ CCLASS_OTHER, /* a */ CCLASS_VOWEL, /* b */ CCLASS_B, + /* c */ CCLASS_C, /* d */ CCLASS_D, /* e */ CCLASS_VOWEL, + /* f */ CCLASS_B, /* g */ CCLASS_C, /* h */ CCLASS_SILENT, + /* i */ CCLASS_VOWEL, /* j */ CCLASS_C, /* k */ CCLASS_C, + /* l */ CCLASS_L, /* m */ CCLASS_M, /* n */ CCLASS_M, + /* o */ CCLASS_VOWEL, /* p */ CCLASS_B, /* q */ CCLASS_C, + /* r */ CCLASS_R, /* s */ CCLASS_C, /* t */ CCLASS_D, + /* u */ CCLASS_VOWEL, /* v */ CCLASS_B, /* w */ CCLASS_B, + /* x */ CCLASS_C, /* y */ CCLASS_VOWEL, /* z */ CCLASS_C, + /* { */ CCLASS_OTHER, /* | */ CCLASS_OTHER, /* } */ CCLASS_OTHER, + /* ~ */ CCLASS_OTHER, /* */ CCLASS_OTHER, +}; +/* +** This tables gives the character class for ASCII characters that form the +** initial character of a word. The only difference from midClass is with +** the letters H, W, and Y. +*/ +static const unsigned char initClass[] = { + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_SPACE, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_SPACE, /* */ CCLASS_SPACE, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, + /* */ CCLASS_OTHER, /* */ CCLASS_OTHER, /* */ CCLASS_SPACE, + /* ! */ CCLASS_OTHER, /* " */ CCLASS_OTHER, /* # */ CCLASS_OTHER, + /* $ */ CCLASS_OTHER, /* % */ CCLASS_OTHER, /* & */ CCLASS_OTHER, + /* ' */ CCLASS_OTHER, /* ( */ CCLASS_OTHER, /* ) */ CCLASS_OTHER, + /* * */ CCLASS_OTHER, /* + */ CCLASS_OTHER, /* , */ CCLASS_OTHER, + /* - */ CCLASS_OTHER, /* . */ CCLASS_OTHER, /* / */ CCLASS_OTHER, + /* 0 */ CCLASS_DIGIT, /* 1 */ CCLASS_DIGIT, /* 2 */ CCLASS_DIGIT, + /* 3 */ CCLASS_DIGIT, /* 4 */ CCLASS_DIGIT, /* 5 */ CCLASS_DIGIT, + /* 6 */ CCLASS_DIGIT, /* 7 */ CCLASS_DIGIT, /* 8 */ CCLASS_DIGIT, + /* 9 */ CCLASS_DIGIT, /* : */ CCLASS_OTHER, /* ; */ CCLASS_OTHER, + /* < */ CCLASS_OTHER, /* = */ CCLASS_OTHER, /* > */ CCLASS_OTHER, + /* ? */ CCLASS_OTHER, /* @ */ CCLASS_OTHER, /* A */ CCLASS_VOWEL, + /* B */ CCLASS_B, /* C */ CCLASS_C, /* D */ CCLASS_D, + /* E */ CCLASS_VOWEL, /* F */ CCLASS_B, /* G */ CCLASS_C, + /* H */ CCLASS_SILENT, /* I */ CCLASS_VOWEL, /* J */ CCLASS_C, + /* K */ CCLASS_C, /* L */ CCLASS_L, /* M */ CCLASS_M, + /* N */ CCLASS_M, /* O */ CCLASS_VOWEL, /* P */ CCLASS_B, + /* Q */ CCLASS_C, /* R */ CCLASS_R, /* S */ CCLASS_C, + /* T */ CCLASS_D, /* U */ CCLASS_VOWEL, /* V */ CCLASS_B, + /* W */ CCLASS_B, /* X */ CCLASS_C, /* Y */ CCLASS_Y, + /* Z */ CCLASS_C, /* [ */ CCLASS_OTHER, /* \ */ CCLASS_OTHER, + /* ] */ CCLASS_OTHER, /* ^ */ CCLASS_OTHER, /* _ */ CCLASS_OTHER, + /* ` */ CCLASS_OTHER, /* a */ CCLASS_VOWEL, /* b */ CCLASS_B, + /* c */ CCLASS_C, /* d */ CCLASS_D, /* e */ CCLASS_VOWEL, + /* f */ CCLASS_B, /* g */ CCLASS_C, /* h */ CCLASS_SILENT, + /* i */ CCLASS_VOWEL, /* j */ CCLASS_C, /* k */ CCLASS_C, + /* l */ CCLASS_L, /* m */ CCLASS_M, /* n */ CCLASS_M, + /* o */ CCLASS_VOWEL, /* p */ CCLASS_B, /* q */ CCLASS_C, + /* r */ CCLASS_R, /* s */ CCLASS_C, /* t */ CCLASS_D, + /* u */ CCLASS_VOWEL, /* v */ CCLASS_B, /* w */ CCLASS_B, + /* x */ CCLASS_C, /* y */ CCLASS_Y, /* z */ CCLASS_C, + /* { */ CCLASS_OTHER, /* | */ CCLASS_OTHER, /* } */ CCLASS_OTHER, + /* ~ */ CCLASS_OTHER, /* */ CCLASS_OTHER, +}; + +/* +** Mapping from the character class number (0-13) to a symbol for each +** character class. Note that initClass[] can be used to map the class +** symbol back into the class number. +*/ +static const unsigned char className[] = ".ABCDHLRMY9 ?"; + +/* +** Generate a "phonetic hash" from a string of ASCII characters +** in zIn[0..nIn-1]. +** +** * Map characters by character class as defined above. +** * Omit double-letters +** * Omit vowels beside R and L +** * Omit T when followed by CH +** * Omit W when followed by R +** * Omit D when followed by J or G +** * Omit K in KN or G in GN at the beginning of a word +** +** Space to hold the result is obtained from sqlite3_malloc() +** +** Return NULL if memory allocation fails. +*/ +static unsigned char *phoneticHash(const unsigned char *zIn, int nIn){ + unsigned char *zOut = sqlite3_malloc64( nIn + 1 ); + int i; + int nOut = 0; + char cPrev = 0x77; + char cPrevX = 0x77; + const unsigned char *aClass = initClass; + + if( zOut==0 ) return 0; + if( nIn>2 ){ + switch( zIn[0] ){ + case 'g': + case 'k': { + if( zIn[1]=='n' ){ zIn++; nIn--; } + break; + } + } + } + for(i=0; i=0 ); + if( nOut==0 || c!=zOut[nOut-1] ) zOut[nOut++] = c; + } + zOut[nOut] = 0; + return zOut; +} + +/* +** This is an SQL function wrapper around phoneticHash(). See +** the description of phoneticHash() for additional information. +*/ +static void phoneticHashSqlFunc( + sqlite3_context *context, + int argc, + sqlite3_value **argv +){ + const unsigned char *zIn; + unsigned char *zOut; + + zIn = sqlite3_value_text(argv[0]); + if( zIn==0 ) return; + zOut = phoneticHash(zIn, sqlite3_value_bytes(argv[0])); + if( zOut==0 ){ + sqlite3_result_error_nomem(context); + }else{ + sqlite3_result_text(context, (char*)zOut, -1, sqlite3_free); + } +} + +/* +** Return the character class number for a character given its +** context. +*/ +static char characterClass(char cPrev, char c){ + return cPrev==0 ? initClass[c&0x7f] : midClass[c&0x7f]; +} + +/* +** Return the cost of inserting or deleting character c immediately +** following character cPrev. If cPrev==0, that means c is the first +** character of the word. +*/ +static int insertOrDeleteCost(char cPrev, char c, char cNext){ + char classC = characterClass(cPrev, c); + char classCprev; + + if( classC==CCLASS_SILENT ){ + /* Insert or delete "silent" characters such as H or W */ + return 1; + } + if( cPrev==c ){ + /* Repeated characters, or miss a repeat */ + return 10; + } + if( classC==CCLASS_VOWEL && (cPrev=='r' || cNext=='r') ){ + return 20; /* Insert a vowel before or after 'r' */ + } + classCprev = characterClass(cPrev, cPrev); + if( classC==classCprev ){ + if( classC==CCLASS_VOWEL ){ + /* Remove or add a new vowel to a vowel cluster */ + return 15; + }else{ + /* Remove or add a consonant not in the same class */ + return 50; + } + } + + /* any other character insertion or deletion */ + return 100; +} + +/* +** Divide the insertion cost by this factor when appending to the +** end of the word. +*/ +#define FINAL_INS_COST_DIV 4 + +/* +** Return the cost of substituting cTo in place of cFrom assuming +** the previous character is cPrev. If cPrev==0 then cTo is the first +** character of the word. +*/ +static int substituteCost(char cPrev, char cFrom, char cTo){ + char classFrom, classTo; + if( cFrom==cTo ){ + /* Exact match */ + return 0; + } + if( cFrom==(cTo^0x20) && ((cTo>='A' && cTo<='Z') || (cTo>='a' && cTo<='z')) ){ + /* differ only in case */ + return 0; + } + classFrom = characterClass(cPrev, cFrom); + classTo = characterClass(cPrev, cTo); + if( classFrom==classTo ){ + /* Same character class */ + return 40; + } + if( classFrom>=CCLASS_B && classFrom<=CCLASS_Y + && classTo>=CCLASS_B && classTo<=CCLASS_Y ){ + /* Convert from one consonant to another, but in a different class */ + return 75; + } + /* Any other subsitution */ + return 100; +} + +/* +** Given two strings zA and zB which are pure ASCII, return the cost +** of transforming zA into zB. If zA ends with '*' assume that it is +** a prefix of zB and give only minimal penalty for extra characters +** on the end of zB. +** +** Smaller numbers mean a closer match. +** +** Negative values indicate an error: +** -1 One of the inputs is NULL +** -2 Non-ASCII characters on input +** -3 Unable to allocate memory +** +** If pnMatch is not NULL, then *pnMatch is set to the number of bytes +** of zB that matched the pattern in zA. If zA does not end with a '*', +** then this value is always the number of bytes in zB (i.e. strlen(zB)). +** If zA does end in a '*', then it is the number of bytes in the prefix +** of zB that was deemed to match zA. +*/ +static int editdist1(const char *zA, const char *zB, int *pnMatch){ + unsigned int nA, nB; /* Number of characters in zA[] and zB[] */ + unsigned int xA, xB; /* Loop counters for zA[] and zB[] */ + char cA = 0, cB; /* Current character of zA and zB */ + char cAprev, cBprev; /* Previous character of zA and zB */ + char cAnext, cBnext; /* Next character in zA and zB */ + int d; /* North-west cost value */ + int dc = 0; /* North-west character value */ + int res; /* Final result */ + int *m; /* The cost matrix */ + char *cx; /* Corresponding character values */ + int *toFree = 0; /* Malloced space */ + int nMatch = 0; + int mStack[60+15]; /* Stack space to use if not too much is needed */ + + /* Early out if either input is NULL */ + if( zA==0 || zB==0 ) return -1; + + /* Skip any common prefix */ + while( zA[0] && zA[0]==zB[0] ){ dc = zA[0]; zA++; zB++; nMatch++; } + if( pnMatch ) *pnMatch = nMatch; + if( zA[0]==0 && zB[0]==0 ) return 0; + +#if 0 + printf("A=\"%s\" B=\"%s\" dc=%c\n", zA, zB, dc?dc:' '); +#endif + + /* Verify input strings and measure their lengths */ + for(nA=0; zA[nA]; nA++){ + if( zA[nA]&0x80 ) return -2; + } + for(nB=0; zB[nB]; nB++){ + if( zB[nB]&0x80 ) return -2; + } + + /* Special processing if either string is empty */ + if( nA==0 ){ + cBprev = (char)dc; + for(xB=res=0; (cB = zB[xB])!=0; xB++){ + res += insertOrDeleteCost(cBprev, cB, zB[xB+1])/FINAL_INS_COST_DIV; + cBprev = cB; + } + return res; + } + if( nB==0 ){ + cAprev = (char)dc; + for(xA=res=0; (cA = zA[xA])!=0; xA++){ + res += insertOrDeleteCost(cAprev, cA, zA[xA+1]); + cAprev = cA; + } + return res; + } + + /* A is a prefix of B */ + if( zA[0]=='*' && zA[1]==0 ) return 0; + + /* Allocate and initialize the Wagner matrix */ + if( nB<(sizeof(mStack)*4)/(sizeof(mStack[0])*5) ){ + m = mStack; + }else{ + m = toFree = sqlite3_malloc64( (nB+1)*5*sizeof(m[0])/4 ); + if( m==0 ) return -3; + } + cx = (char*)&m[nB+1]; + + /* Compute the Wagner edit distance */ + m[0] = 0; + cx[0] = (char)dc; + cBprev = (char)dc; + for(xB=1; xB<=nB; xB++){ + cBnext = zB[xB]; + cB = zB[xB-1]; + cx[xB] = cB; + m[xB] = m[xB-1] + insertOrDeleteCost(cBprev, cB, cBnext); + cBprev = cB; + } + cAprev = (char)dc; + for(xA=1; xA<=nA; xA++){ + int lastA = (xA==nA); + cA = zA[xA-1]; + cAnext = zA[xA]; + if( cA=='*' && lastA ) break; + d = m[0]; + dc = cx[0]; + m[0] = d + insertOrDeleteCost(cAprev, cA, cAnext); + cBprev = 0; + for(xB=1; xB<=nB; xB++){ + int totalCost, insCost, delCost, subCost, ncx; + cB = zB[xB-1]; + cBnext = zB[xB]; + + /* Cost to insert cB */ + insCost = insertOrDeleteCost(cx[xB-1], cB, cBnext); + if( lastA ) insCost /= FINAL_INS_COST_DIV; + + /* Cost to delete cA */ + delCost = insertOrDeleteCost(cx[xB], cA, cBnext); + + /* Cost to substitute cA->cB */ + subCost = substituteCost(cx[xB-1], cA, cB); + + /* Best cost */ + totalCost = insCost + m[xB-1]; + ncx = cB; + if( (delCost + m[xB])nLang; i++){ + EditDist3Cost *pCost, *pNext; + pCost = p->a[i].pCost; + while( pCost ){ + pNext = pCost->pNext; + sqlite3_free(pCost); + pCost = pNext; + } + } + sqlite3_free(p->a); + memset(p, 0, sizeof(*p)); +} +static void editDist3ConfigDelete(void *pIn){ + EditDist3Config *p = (EditDist3Config*)pIn; + editDist3ConfigClear(p); + sqlite3_free(p); +} + +/* Compare the FROM values of two EditDist3Cost objects, for sorting. +** Return negative, zero, or positive if the A is less than, equal to, +** or greater than B. +*/ +static int editDist3CostCompare(EditDist3Cost *pA, EditDist3Cost *pB){ + int n = pA->nFrom; + int rc; + if( n>pB->nFrom ) n = pB->nFrom; + rc = strncmp(pA->a, pB->a, n); + if( rc==0 ) rc = pA->nFrom - pB->nFrom; + return rc; +} + +/* +** Merge together two sorted lists of EditDist3Cost objects, in order +** of increasing FROM. +*/ +static EditDist3Cost *editDist3CostMerge( + EditDist3Cost *pA, + EditDist3Cost *pB +){ + EditDist3Cost *pHead = 0; + EditDist3Cost **ppTail = &pHead; + EditDist3Cost *p; + while( pA && pB ){ + if( editDist3CostCompare(pA,pB)<=0 ){ + p = pA; + pA = pA->pNext; + }else{ + p = pB; + pB = pB->pNext; + } + *ppTail = p; + ppTail = &p->pNext; + } + if( pA ){ + *ppTail = pA; + }else{ + *ppTail = pB; + } + return pHead; +} + +/* +** Sort a list of EditDist3Cost objects into order of increasing FROM +*/ +static EditDist3Cost *editDist3CostSort(EditDist3Cost *pList){ + EditDist3Cost *ap[60], *p; + int i; + int mx = 0; + ap[0] = 0; + ap[1] = 0; + while( pList ){ + p = pList; + pList = p->pNext; + p->pNext = 0; + for(i=0; ap[i]; i++){ + p = editDist3CostMerge(ap[i],p); + ap[i] = 0; + } + ap[i] = p; + if( i>mx ){ + mx = i; + ap[i+1] = 0; + } + } + p = 0; + for(i=0; i<=mx; i++){ + if( ap[i] ) p = editDist3CostMerge(p,ap[i]); + } + return p; +} + +/* +** Load all edit-distance weights from a table. +*/ +static int editDist3ConfigLoad( + EditDist3Config *p, /* The edit distance configuration to load */ + sqlite3 *db, /* Load from this database */ + const char *zTable /* Name of the table from which to load */ +){ + sqlite3_stmt *pStmt; + int rc, rc2; + char *zSql; + int iLangPrev = -9999; + EditDist3Lang *pLang = 0; + + zSql = sqlite3_mprintf("SELECT iLang, cFrom, cTo, iCost" + " FROM \"%w\" WHERE iLang>=0 ORDER BY iLang", zTable); + if( zSql==0 ) return SQLITE_NOMEM; + rc = sqlite3_prepare(db, zSql, -1, &pStmt, 0); + sqlite3_free(zSql); + if( rc ) return rc; + editDist3ConfigClear(p); + while( sqlite3_step(pStmt)==SQLITE_ROW ){ + int iLang = sqlite3_column_int(pStmt, 0); + const char *zFrom = (const char*)sqlite3_column_text(pStmt, 1); + int nFrom = zFrom ? sqlite3_column_bytes(pStmt, 1) : 0; + const char *zTo = (const char*)sqlite3_column_text(pStmt, 2); + int nTo = zTo ? sqlite3_column_bytes(pStmt, 2) : 0; + int iCost = sqlite3_column_int(pStmt, 3); + + assert( zFrom!=0 || nFrom==0 ); + assert( zTo!=0 || nTo==0 ); + if( nFrom>100 || nTo>100 ) continue; + if( iCost<0 ) continue; + if( iCost>=10000 ) continue; /* Costs above 10K are considered infinite */ + if( pLang==0 || iLang!=iLangPrev ){ + EditDist3Lang *pNew; + pNew = sqlite3_realloc64(p->a, (p->nLang+1)*sizeof(p->a[0])); + if( pNew==0 ){ rc = SQLITE_NOMEM; break; } + p->a = pNew; + pLang = &p->a[p->nLang]; + p->nLang++; + pLang->iLang = iLang; + pLang->iInsCost = 100; + pLang->iDelCost = 100; + pLang->iSubCost = 150; + pLang->pCost = 0; + iLangPrev = iLang; + } + if( nFrom==1 && zFrom[0]=='?' && nTo==0 ){ + pLang->iDelCost = iCost; + }else if( nFrom==0 && nTo==1 && zTo[0]=='?' ){ + pLang->iInsCost = iCost; + }else if( nFrom==1 && nTo==1 && zFrom[0]=='?' && zTo[0]=='?' ){ + pLang->iSubCost = iCost; + }else{ + EditDist3Cost *pCost; + int nExtra = nFrom + nTo - 4; + if( nExtra<0 ) nExtra = 0; + pCost = sqlite3_malloc64( sizeof(*pCost) + nExtra ); + if( pCost==0 ){ rc = SQLITE_NOMEM; break; } + pCost->nFrom = (u8)nFrom; + pCost->nTo = (u8)nTo; + pCost->iCost = (u16)iCost; + memcpy(pCost->a, zFrom, nFrom); + memcpy(pCost->a + nFrom, zTo, nTo); + pCost->pNext = pLang->pCost; + pLang->pCost = pCost; + } + } + rc2 = sqlite3_finalize(pStmt); + if( rc==SQLITE_OK ) rc = rc2; + if( rc==SQLITE_OK ){ + int iLang; + for(iLang=0; iLangnLang; iLang++){ + p->a[iLang].pCost = editDist3CostSort(p->a[iLang].pCost); + } + } + return rc; +} + +/* +** Return the length (in bytes) of a utf-8 character. Or return a maximum +** of N. +*/ +static int utf8Len(unsigned char c, int N){ + int len = 1; + if( c>0x7f ){ + if( (c&0xe0)==0xc0 ){ + len = 2; + }else if( (c&0xf0)==0xe0 ){ + len = 3; + }else{ + len = 4; + } + } + if( len>N ) len = N; + return len; +} + +/* +** Return TRUE (non-zero) if the To side of the given cost matches +** the given string. +*/ +static int matchTo(EditDist3Cost *p, const char *z, int n){ + assert( n>0 ); + if( p->a[p->nFrom]!=z[0] ) return 0; + if( p->nTo>n ) return 0; + if( strncmp(p->a+p->nFrom, z, p->nTo)!=0 ) return 0; + return 1; +} + +/* +** Return TRUE (non-zero) if the From side of the given cost matches +** the given string. +*/ +static int matchFrom(EditDist3Cost *p, const char *z, int n){ + assert( p->nFrom<=n ); + if( p->nFrom ){ + if( p->a[0]!=z[0] ) return 0; + if( strncmp(p->a, z, p->nFrom)!=0 ) return 0; + } + return 1; +} + +/* +** Return TRUE (non-zero) of the next FROM character and the next TO +** character are the same. +*/ +static int matchFromTo( + EditDist3FromString *pStr, /* Left hand string */ + int n1, /* Index of comparison character on the left */ + const char *z2, /* Right-handl comparison character */ + int n2 /* Bytes remaining in z2[] */ +){ + int b1 = pStr->a[n1].nByte; + if( b1>n2 ) return 0; + assert( b1>0 ); + if( pStr->z[n1]!=z2[0] ) return 0; + if( strncmp(pStr->z+n1, z2, b1)!=0 ) return 0; + return 1; +} + +/* +** Delete an EditDist3FromString objecct +*/ +static void editDist3FromStringDelete(EditDist3FromString *p){ + int i; + if( p ){ + for(i=0; in; i++){ + sqlite3_free(p->a[i].apDel); + sqlite3_free(p->a[i].apSubst); + } + sqlite3_free(p); + } +} + +/* +** Create a EditDist3FromString object. +*/ +static EditDist3FromString *editDist3FromStringNew( + const EditDist3Lang *pLang, + const char *z, + int n +){ + EditDist3FromString *pStr; + EditDist3Cost *p; + int i; + + if( z==0 ) return 0; + if( n<0 ) n = (int)strlen(z); + pStr = sqlite3_malloc64( sizeof(*pStr) + sizeof(pStr->a[0])*n + n + 1 ); + if( pStr==0 ) return 0; + pStr->a = (EditDist3From*)&pStr[1]; + memset(pStr->a, 0, sizeof(pStr->a[0])*n); + pStr->n = n; + pStr->z = (char*)&pStr->a[n]; + memcpy(pStr->z, z, n+1); + if( n && z[n-1]=='*' ){ + pStr->isPrefix = 1; + n--; + pStr->n--; + pStr->z[n] = 0; + }else{ + pStr->isPrefix = 0; + } + + for(i=0; ia[i]; + memset(pFrom, 0, sizeof(*pFrom)); + pFrom->nByte = utf8Len((unsigned char)z[i], n-i); + for(p=pLang->pCost; p; p=p->pNext){ + EditDist3Cost **apNew; + if( i+p->nFrom>n ) continue; + if( matchFrom(p, z+i, n-i)==0 ) continue; + if( p->nTo==0 ){ + apNew = sqlite3_realloc64(pFrom->apDel, + sizeof(*apNew)*(pFrom->nDel+1)); + if( apNew==0 ) break; + pFrom->apDel = apNew; + apNew[pFrom->nDel++] = p; + }else{ + apNew = sqlite3_realloc64(pFrom->apSubst, + sizeof(*apNew)*(pFrom->nSubst+1)); + if( apNew==0 ) break; + pFrom->apSubst = apNew; + apNew[pFrom->nSubst++] = p; + } + } + if( p ){ + editDist3FromStringDelete(pStr); + pStr = 0; + break; + } + } + return pStr; +} + +/* +** Update entry m[i] such that it is the minimum of its current value +** and m[j]+iCost. +*/ +static void updateCost( + unsigned int *m, + int i, + int j, + int iCost +){ + unsigned int b; + assert( iCost>=0 ); + assert( iCost<10000 ); + b = m[j] + iCost; + if( bpCost; p; p=p->pNext){ + EditDist3Cost **apNew; + if( p->nFrom>0 ) break; + if( i2+p->nTo>n2 ) continue; + if( p->a[0]>z2[i2] ) break; + if( matchTo(p, z2+i2, n2-i2)==0 ) continue; + a2[i2].nIns++; + apNew = sqlite3_realloc64(a2[i2].apIns, sizeof(*apNew)*a2[i2].nIns); + if( apNew==0 ){ + res = -1; /* Out of memory */ + goto editDist3Abort; + } + a2[i2].apIns = apNew; + a2[i2].apIns[a2[i2].nIns-1] = p; + } + } + + /* Prepare to compute the minimum edit distance */ + szRow = f.n+1; + memset(m, 0x01, (n2+1)*szRow*sizeof(m[0])); + m[0] = 0; + + /* First fill in the top-row of the matrix with FROM deletion costs */ + for(i1=0; i1iDelCost); + for(k=0; knFrom, i1, p->iCost); + } + } + + /* Fill in all subsequent rows, top-to-bottom, left-to-right */ + for(i2=0; i2iInsCost); + for(k=0; knTo), rxp, p->iCost); + } + for(i1=0; i1iDelCost); + for(k=0; knFrom, cxp, p->iCost); + } + updateCost(m, cx, cxu, pLang->iInsCost); + if( matchFromTo(&f, i1, z2+i2, n2-i2) ){ + updateCost(m, cx, cxd, 0); + } + updateCost(m, cx, cxd, pLang->iSubCost); + for(k=0; knFrom+szRow*p->nTo, cxd, p->iCost); + } + } + } + } + +#if 0 /* Enable for debugging */ + printf(" ^"); + for(i1=0; i19999 ) printf(" ****"); + else printf(" %4d", v); + } + printf("\n"); + for(i2=0; i29999 ) printf(" ****"); + else printf(" %4d", v); + } + printf("\n"); + } +#endif + + /* Free memory allocations and return the result */ + res = (int)m[szRow*(n2+1)-1]; + n = n2; + if( f.isPrefix ){ + for(i2=1; i2<=n2; i2++){ + int b = m[szRow*i2-1]; + if( b<=res ){ + res = b; + n = i2 - 1; + } + } + } + if( pnMatch ){ + int nExtra = 0; + for(k=0; knLang; i++){ + if( pConfig->a[i].iLang==iLang ) return &pConfig->a[i]; + } + return &editDist3Lang; +} + +/* +** Function: editdist3(A,B,iLang) +** editdist3(tablename) +** +** Return the cost of transforming string A into string B using edit +** weights for iLang. +** +** The second form loads edit weights into memory from a table. +*/ +static void editDist3SqlFunc( + sqlite3_context *context, + int argc, + sqlite3_value **argv +){ + EditDist3Config *pConfig = (EditDist3Config*)sqlite3_user_data(context); + sqlite3 *db = sqlite3_context_db_handle(context); + int rc; + if( argc==1 ){ + const char *zTable = (const char*)sqlite3_value_text(argv[0]); + rc = editDist3ConfigLoad(pConfig, db, zTable); + if( rc ) sqlite3_result_error_code(context, rc); + }else{ + const char *zA = (const char*)sqlite3_value_text(argv[0]); + const char *zB = (const char*)sqlite3_value_text(argv[1]); + int nA = sqlite3_value_bytes(argv[0]); + int nB = sqlite3_value_bytes(argv[1]); + int iLang = argc==3 ? sqlite3_value_int(argv[2]) : 0; + const EditDist3Lang *pLang = editDist3FindLang(pConfig, iLang); + EditDist3FromString *pFrom; + int dist; + + pFrom = editDist3FromStringNew(pLang, zA, nA); + if( pFrom==0 ){ + sqlite3_result_error_nomem(context); + return; + } + dist = editDist3Core(pFrom, zB, nB, pLang, 0); + editDist3FromStringDelete(pFrom); + if( dist==(-1) ){ + sqlite3_result_error_nomem(context); + }else{ + sqlite3_result_int(context, dist); + } + } +} + +/* +** Register the editDist3 function with SQLite +*/ +static int editDist3Install(sqlite3 *db){ + int rc; + EditDist3Config *pConfig = sqlite3_malloc64( sizeof(*pConfig) ); + if( pConfig==0 ) return SQLITE_NOMEM; + memset(pConfig, 0, sizeof(*pConfig)); + rc = sqlite3_create_function_v2(db, "editdist3", + 2, SQLITE_UTF8|SQLITE_DETERMINISTIC, pConfig, + editDist3SqlFunc, 0, 0, 0); + if( rc==SQLITE_OK ){ + rc = sqlite3_create_function_v2(db, "editdist3", + 3, SQLITE_UTF8|SQLITE_DETERMINISTIC, pConfig, + editDist3SqlFunc, 0, 0, 0); + } + if( rc==SQLITE_OK ){ + rc = sqlite3_create_function_v2(db, "editdist3", + 1, SQLITE_UTF8|SQLITE_DETERMINISTIC, pConfig, + editDist3SqlFunc, 0, 0, editDist3ConfigDelete); + }else{ + sqlite3_free(pConfig); + } + return rc; +} +/* End configurable cost unicode edit distance routines +****************************************************************************** +****************************************************************************** +** Begin transliterate unicode-to-ascii implementation +*/ + +#if !SQLITE_AMALGAMATION +/* +** This lookup table is used to help decode the first byte of +** a multi-byte UTF8 character. +*/ +static const unsigned char sqlite3Utf8Trans1[] = { + 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, + 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, + 0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17, + 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d, 0x1e, 0x1f, + 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, + 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, + 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, + 0x00, 0x01, 0x02, 0x03, 0x00, 0x01, 0x00, 0x00, +}; +#endif + +/* +** Return the value of the first UTF-8 character in the string. +*/ +static int utf8Read(const unsigned char *z, int n, int *pSize){ + int c, i; + + /* All callers to this routine (in the current implementation) + ** always have n>0. */ + if( NEVER(n==0) ){ + c = i = 0; + }else{ + c = z[0]; + i = 1; + if( c>=0xc0 ){ + c = sqlite3Utf8Trans1[c-0xc0]; + while( i0 ){ + c = utf8Read(zIn, nIn, &sz); + zIn += sz; + nIn -= sz; + if( c<=127 ){ + zOut[nOut++] = (unsigned char)c; + }else{ + int xTop, xBtm, x; + const Transliteration *tbl = spellfixFindTranslit(c, &xTop); + xBtm = 0; + while( xTop>=xBtm ){ + x = (xTop + xBtm)/2; + if( tbl[x].cFrom==c ){ + zOut[nOut++] = tbl[x].cTo0; + if( tbl[x].cTo1 ){ + zOut[nOut++] = tbl[x].cTo1; + if( tbl[x].cTo2 ){ + zOut[nOut++] = tbl[x].cTo2; + if( tbl[x].cTo3 ){ + zOut[nOut++] = tbl[x].cTo3; +#ifdef SQLITE_SPELLFIX_5BYTE_MAPPINGS + if( tbl[x].cTo4 ){ + zOut[nOut++] = tbl[x].cTo4; + } +#endif /* SQLITE_SPELLFIX_5BYTE_MAPPINGS */ + } + } + } + c = 0; + break; + }else if( tbl[x].cFrom>c ){ + xTop = x-1; + }else{ + xBtm = x+1; + } + } + if( c ) zOut[nOut++] = '?'; + } + } + zOut[nOut] = 0; + return zOut; +} + +/* +** Return the number of characters in the shortest prefix of the input +** string that transliterates to an ASCII string nTrans bytes or longer. +** Or, if the transliteration of the input string is less than nTrans +** bytes in size, return the number of characters in the input string. +*/ +static int translen_to_charlen(const char *zIn, int nIn, int nTrans){ + int i, c, sz, nOut; + int nChar; + + i = nOut = 0; + for(nChar=0; i=128 ){ + int xTop, xBtm, x; + const Transliteration *tbl = spellfixFindTranslit(c, &xTop); + xBtm = 0; + while( xTop>=xBtm ){ + x = (xTop + xBtm)/2; + if( tbl[x].cFrom==c ){ + if( tbl[x].cTo1 ){ + nOut++; + if( tbl[x].cTo2 ){ + nOut++; + if( tbl[x].cTo3 ){ + nOut++; + } + } + } + break; + }else if( tbl[x].cFrom>c ){ + xTop = x-1; + }else{ + xBtm = x+1; + } + } + } + } + + return nChar; +} + + +/* +** spellfix1_translit(X) +** +** Convert a string that contains non-ASCII Roman characters into +** pure ASCII. +*/ +static void transliterateSqlFunc( + sqlite3_context *context, + int argc, + sqlite3_value **argv +){ + const unsigned char *zIn = sqlite3_value_text(argv[0]); + int nIn = sqlite3_value_bytes(argv[0]); + unsigned char *zOut = transliterate(zIn, nIn); + if( zOut==0 ){ + sqlite3_result_error_nomem(context); + }else{ + sqlite3_result_text(context, (char*)zOut, -1, sqlite3_free); + } +} + +/* +** spellfix1_scriptcode(X) +** +** Try to determine the dominant script used by the word X and return +** its ISO 15924 numeric code. +** +** The current implementation only understands the following scripts: +** +** 215 (Latin) +** 220 (Cyrillic) +** 200 (Greek) +** +** This routine will return 998 if the input X contains characters from +** two or more of the above scripts or 999 if X contains no characters +** from any of the above scripts. +*/ +static void scriptCodeSqlFunc( + sqlite3_context *context, + int argc, + sqlite3_value **argv +){ + const unsigned char *zIn = sqlite3_value_text(argv[0]); + int nIn = sqlite3_value_bytes(argv[0]); + int c, sz; + int scriptMask = 0; + int res; + int seenDigit = 0; +# define SCRIPT_LATIN 0x0001 +# define SCRIPT_CYRILLIC 0x0002 +# define SCRIPT_GREEK 0x0004 +# define SCRIPT_HEBREW 0x0008 +# define SCRIPT_ARABIC 0x0010 + + while( nIn>0 ){ + c = utf8Read(zIn, nIn, &sz); + zIn += sz; + nIn -= sz; + if( c<0x02af ){ + if( c>=0x80 || midClass[c&0x7f]='0' && c<='9' ){ + seenDigit = 1; + } + }else if( c>=0x0400 && c<=0x04ff ){ + scriptMask |= SCRIPT_CYRILLIC; + }else if( c>=0x0386 && c<=0x03ce ){ + scriptMask |= SCRIPT_GREEK; + }else if( c>=0x0590 && c<=0x05ff ){ + scriptMask |= SCRIPT_HEBREW; + }else if( c>=0x0600 && c<=0x06ff ){ + scriptMask |= SCRIPT_ARABIC; + } + } + if( scriptMask==0 && seenDigit ) scriptMask = SCRIPT_LATIN; + switch( scriptMask ){ + case 0: res = 999; break; + case SCRIPT_LATIN: res = 215; break; + case SCRIPT_CYRILLIC: res = 220; break; + case SCRIPT_GREEK: res = 200; break; + case SCRIPT_HEBREW: res = 125; break; + case SCRIPT_ARABIC: res = 160; break; + default: res = 998; break; + } + sqlite3_result_int(context, res); +} + +/* End transliterate +****************************************************************************** +****************************************************************************** +** Begin spellfix1 virtual table. +*/ + +/* Maximum length of a phonehash used for querying the shadow table */ +#define SPELLFIX_MX_HASH 32 + +/* Maximum number of hash strings to examine per query */ +#define SPELLFIX_MX_RUN 1 + +typedef struct spellfix1_vtab spellfix1_vtab; +typedef struct spellfix1_cursor spellfix1_cursor; + +/* Fuzzy-search virtual table object */ +struct spellfix1_vtab { + sqlite3_vtab base; /* Base class - must be first */ + sqlite3 *db; /* Database connection */ + char *zDbName; /* Name of database holding this table */ + char *zTableName; /* Name of the virtual table */ + char *zCostTable; /* Table holding edit-distance cost numbers */ + EditDist3Config *pConfig3; /* Parsed edit distance costs */ +}; + +/* Fuzzy-search cursor object */ +struct spellfix1_cursor { + sqlite3_vtab_cursor base; /* Base class - must be first */ + spellfix1_vtab *pVTab; /* The table to which this cursor belongs */ + char *zPattern; /* rhs of MATCH clause */ + int idxNum; /* idxNum value passed to xFilter() */ + int nRow; /* Number of rows of content */ + int nAlloc; /* Number of allocated rows */ + int iRow; /* Current row of content */ + int iLang; /* Value of the langid= constraint */ + int iTop; /* Value of the top= constraint */ + int iScope; /* Value of the scope= constraint */ + int nSearch; /* Number of vocabulary items checked */ + sqlite3_stmt *pFullScan; /* Shadow query for a full table scan */ + struct spellfix1_row { /* For each row of content */ + sqlite3_int64 iRowid; /* Rowid for this row */ + char *zWord; /* Text for this row */ + int iRank; /* Rank for this row */ + int iDistance; /* Distance from pattern for this row */ + int iScore; /* Score for sorting */ + int iMatchlen; /* Value of matchlen column (or -1) */ + char zHash[SPELLFIX_MX_HASH]; /* the phonehash used for this match */ + } *a; +}; + +/* +** Construct one or more SQL statements from the format string given +** and then evaluate those statements. The success code is written +** into *pRc. +** +** If *pRc is initially non-zero then this routine is a no-op. +*/ +static void spellfix1DbExec( + int *pRc, /* Success code */ + sqlite3 *db, /* Database in which to run SQL */ + const char *zFormat, /* Format string for SQL */ + ... /* Arguments to the format string */ +){ + va_list ap; + char *zSql; + if( *pRc ) return; + va_start(ap, zFormat); + zSql = sqlite3_vmprintf(zFormat, ap); + va_end(ap); + if( zSql==0 ){ + *pRc = SQLITE_NOMEM; + }else{ + *pRc = sqlite3_exec(db, zSql, 0, 0, 0); + sqlite3_free(zSql); + } +} + +/* +** xDisconnect/xDestroy method for the fuzzy-search module. +*/ +static int spellfix1Uninit(int isDestroy, sqlite3_vtab *pVTab){ + spellfix1_vtab *p = (spellfix1_vtab*)pVTab; + int rc = SQLITE_OK; + if( isDestroy ){ + sqlite3 *db = p->db; + spellfix1DbExec(&rc, db, "DROP TABLE IF EXISTS \"%w\".\"%w_vocab\"", + p->zDbName, p->zTableName); + } + if( rc==SQLITE_OK ){ + sqlite3_free(p->zTableName); + editDist3ConfigDelete(p->pConfig3); + sqlite3_free(p->zCostTable); + sqlite3_free(p); + } + return rc; +} +static int spellfix1Disconnect(sqlite3_vtab *pVTab){ + return spellfix1Uninit(0, pVTab); +} +static int spellfix1Destroy(sqlite3_vtab *pVTab){ + return spellfix1Uninit(1, pVTab); +} + +/* +** Make a copy of a string. Remove leading and trailing whitespace +** and dequote it. +*/ +static char *spellfix1Dequote(const char *zIn){ + char *zOut; + int i, j; + char c; + while( isspace((unsigned char)zIn[0]) ) zIn++; + zOut = sqlite3_mprintf("%s", zIn); + if( zOut==0 ) return 0; + i = (int)strlen(zOut); +#if 0 /* The parser will never leave spaces at the end */ + while( i>0 && isspace(zOut[i-1]) ){ i--; } +#endif + zOut[i] = 0; + c = zOut[0]; + if( c=='\'' || c=='"' ){ + for(i=1, j=0; ALWAYS(zOut[i]); i++){ + zOut[j++] = zOut[i]; + if( zOut[i]==c ){ + if( zOut[i+1]==c ){ + i++; + }else{ + zOut[j-1] = 0; + break; + } + } + } + } + return zOut; +} + + +/* +** xConnect/xCreate method for the spellfix1 module. Arguments are: +** +** argv[0] -> module name ("spellfix1") +** argv[1] -> database name +** argv[2] -> table name +** argv[3].. -> optional arguments (i.e. "edit_cost_table" parameter) +*/ +static int spellfix1Init( + int isCreate, + sqlite3 *db, + void *pAux, + int argc, const char *const*argv, + sqlite3_vtab **ppVTab, + char **pzErr +){ + spellfix1_vtab *pNew = 0; + /* const char *zModule = argv[0]; // not used */ + const char *zDbName = argv[1]; + const char *zTableName = argv[2]; + int nDbName; + int rc = SQLITE_OK; + int i; + + nDbName = (int)strlen(zDbName); + pNew = sqlite3_malloc64( sizeof(*pNew) + nDbName + 1); + if( pNew==0 ){ + rc = SQLITE_NOMEM; + }else{ + memset(pNew, 0, sizeof(*pNew)); + pNew->zDbName = (char*)&pNew[1]; + memcpy(pNew->zDbName, zDbName, nDbName+1); + pNew->zTableName = sqlite3_mprintf("%s", zTableName); + pNew->db = db; + if( pNew->zTableName==0 ){ + rc = SQLITE_NOMEM; + }else{ + sqlite3_vtab_config(db, SQLITE_VTAB_INNOCUOUS); + rc = sqlite3_declare_vtab(db, + "CREATE TABLE x(word,rank,distance,langid, " + "score, matchlen, phonehash HIDDEN, " + "top HIDDEN, scope HIDDEN, srchcnt HIDDEN, " + "soundslike HIDDEN, command HIDDEN)" + ); +#define SPELLFIX_COL_WORD 0 +#define SPELLFIX_COL_RANK 1 +#define SPELLFIX_COL_DISTANCE 2 +#define SPELLFIX_COL_LANGID 3 +#define SPELLFIX_COL_SCORE 4 +#define SPELLFIX_COL_MATCHLEN 5 +#define SPELLFIX_COL_PHONEHASH 6 +#define SPELLFIX_COL_TOP 7 +#define SPELLFIX_COL_SCOPE 8 +#define SPELLFIX_COL_SRCHCNT 9 +#define SPELLFIX_COL_SOUNDSLIKE 10 +#define SPELLFIX_COL_COMMAND 11 + } + if( rc==SQLITE_OK && isCreate ){ + spellfix1DbExec(&rc, db, + "CREATE TABLE IF NOT EXISTS \"%w\".\"%w_vocab\"(\n" + " id INTEGER PRIMARY KEY,\n" + " rank INT,\n" + " langid INT,\n" + " word TEXT,\n" + " k1 TEXT,\n" + " k2 TEXT\n" + ");\n", + zDbName, zTableName + ); + spellfix1DbExec(&rc, db, + "CREATE INDEX IF NOT EXISTS \"%w\".\"%w_vocab_index_langid_k2\" " + "ON \"%w_vocab\"(langid,k2);", + zDbName, zTableName, zTableName + ); + } + for(i=3; rc==SQLITE_OK && ibase); + }else{ + *ppVTab = (sqlite3_vtab *)pNew; + } + return rc; +} + +/* +** The xConnect and xCreate methods +*/ +static int spellfix1Connect( + sqlite3 *db, + void *pAux, + int argc, const char *const*argv, + sqlite3_vtab **ppVTab, + char **pzErr +){ + return spellfix1Init(0, db, pAux, argc, argv, ppVTab, pzErr); +} +static int spellfix1Create( + sqlite3 *db, + void *pAux, + int argc, const char *const*argv, + sqlite3_vtab **ppVTab, + char **pzErr +){ + return spellfix1Init(1, db, pAux, argc, argv, ppVTab, pzErr); +} + +/* +** Clear all of the content from a cursor. +*/ +static void spellfix1ResetCursor(spellfix1_cursor *pCur){ + int i; + for(i=0; inRow; i++){ + sqlite3_free(pCur->a[i].zWord); + } + pCur->nRow = 0; + pCur->iRow = 0; + pCur->nSearch = 0; + if( pCur->pFullScan ){ + sqlite3_finalize(pCur->pFullScan); + pCur->pFullScan = 0; + } +} + +/* +** Resize the cursor to hold up to N rows of content +*/ +static void spellfix1ResizeCursor(spellfix1_cursor *pCur, int N){ + struct spellfix1_row *aNew; + assert( N>=pCur->nRow ); + aNew = sqlite3_realloc64(pCur->a, sizeof(pCur->a[0])*N); + if( aNew==0 && N>0 ){ + spellfix1ResetCursor(pCur); + sqlite3_free(pCur->a); + pCur->nAlloc = 0; + pCur->a = 0; + }else{ + pCur->nAlloc = N; + pCur->a = aNew; + } +} + + +/* +** Close a fuzzy-search cursor. +*/ +static int spellfix1Close(sqlite3_vtab_cursor *cur){ + spellfix1_cursor *pCur = (spellfix1_cursor *)cur; + spellfix1ResetCursor(pCur); + spellfix1ResizeCursor(pCur, 0); + sqlite3_free(pCur->zPattern); + sqlite3_free(pCur); + return SQLITE_OK; +} + +#define SPELLFIX_IDXNUM_MATCH 0x01 /* word MATCH $str */ +#define SPELLFIX_IDXNUM_LANGID 0x02 /* langid == $langid */ +#define SPELLFIX_IDXNUM_TOP 0x04 /* top = $top */ +#define SPELLFIX_IDXNUM_SCOPE 0x08 /* scope = $scope */ +#define SPELLFIX_IDXNUM_DISTLT 0x10 /* distance < $distance */ +#define SPELLFIX_IDXNUM_DISTLE 0x20 /* distance <= $distance */ +#define SPELLFIX_IDXNUM_ROWID 0x40 /* rowid = $rowid */ +#define SPELLFIX_IDXNUM_DIST (0x10|0x20) /* DISTLT and DISTLE */ + +/* +** +** The plan number is a bitmask of the SPELLFIX_IDXNUM_* values defined +** above. +** +** filter.argv[*] values contains $str, $langid, $top, $scope and $rowid +** if specified and in that order. +*/ +static int spellfix1BestIndex(sqlite3_vtab *tab, sqlite3_index_info *pIdxInfo){ + int iPlan = 0; + int iLangTerm = -1; + int iTopTerm = -1; + int iScopeTerm = -1; + int iDistTerm = -1; + int iRowidTerm = -1; + int i; + const struct sqlite3_index_constraint *pConstraint; + pConstraint = pIdxInfo->aConstraint; + for(i=0; inConstraint; i++, pConstraint++){ + if( pConstraint->usable==0 ) continue; + + /* Terms of the form: word MATCH $str */ + if( (iPlan & SPELLFIX_IDXNUM_MATCH)==0 + && pConstraint->iColumn==SPELLFIX_COL_WORD + && pConstraint->op==SQLITE_INDEX_CONSTRAINT_MATCH + ){ + iPlan |= SPELLFIX_IDXNUM_MATCH; + pIdxInfo->aConstraintUsage[i].argvIndex = 1; + pIdxInfo->aConstraintUsage[i].omit = 1; + } + + /* Terms of the form: langid = $langid */ + if( (iPlan & SPELLFIX_IDXNUM_LANGID)==0 + && pConstraint->iColumn==SPELLFIX_COL_LANGID + && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ + ){ + iPlan |= SPELLFIX_IDXNUM_LANGID; + iLangTerm = i; + } + + /* Terms of the form: top = $top */ + if( (iPlan & SPELLFIX_IDXNUM_TOP)==0 + && pConstraint->iColumn==SPELLFIX_COL_TOP + && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ + ){ + iPlan |= SPELLFIX_IDXNUM_TOP; + iTopTerm = i; + } + + /* Terms of the form: scope = $scope */ + if( (iPlan & SPELLFIX_IDXNUM_SCOPE)==0 + && pConstraint->iColumn==SPELLFIX_COL_SCOPE + && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ + ){ + iPlan |= SPELLFIX_IDXNUM_SCOPE; + iScopeTerm = i; + } + + /* Terms of the form: distance < $dist or distance <= $dist */ + if( (iPlan & SPELLFIX_IDXNUM_DIST)==0 + && pConstraint->iColumn==SPELLFIX_COL_DISTANCE + && (pConstraint->op==SQLITE_INDEX_CONSTRAINT_LT + || pConstraint->op==SQLITE_INDEX_CONSTRAINT_LE) + ){ + if( pConstraint->op==SQLITE_INDEX_CONSTRAINT_LT ){ + iPlan |= SPELLFIX_IDXNUM_DISTLT; + }else{ + iPlan |= SPELLFIX_IDXNUM_DISTLE; + } + iDistTerm = i; + } + + /* Terms of the form: distance < $dist or distance <= $dist */ + if( (iPlan & SPELLFIX_IDXNUM_ROWID)==0 + && pConstraint->iColumn<0 + && pConstraint->op==SQLITE_INDEX_CONSTRAINT_EQ + ){ + iPlan |= SPELLFIX_IDXNUM_ROWID; + iRowidTerm = i; + } + } + if( iPlan&SPELLFIX_IDXNUM_MATCH ){ + int idx = 2; + pIdxInfo->idxNum = iPlan; + if( pIdxInfo->nOrderBy==1 + && pIdxInfo->aOrderBy[0].iColumn==SPELLFIX_COL_SCORE + && pIdxInfo->aOrderBy[0].desc==0 + ){ + pIdxInfo->orderByConsumed = 1; /* Default order by iScore */ + } + if( iPlan&SPELLFIX_IDXNUM_LANGID ){ + pIdxInfo->aConstraintUsage[iLangTerm].argvIndex = idx++; + pIdxInfo->aConstraintUsage[iLangTerm].omit = 1; + } + if( iPlan&SPELLFIX_IDXNUM_TOP ){ + pIdxInfo->aConstraintUsage[iTopTerm].argvIndex = idx++; + pIdxInfo->aConstraintUsage[iTopTerm].omit = 1; + } + if( iPlan&SPELLFIX_IDXNUM_SCOPE ){ + pIdxInfo->aConstraintUsage[iScopeTerm].argvIndex = idx++; + pIdxInfo->aConstraintUsage[iScopeTerm].omit = 1; + } + if( iPlan&SPELLFIX_IDXNUM_DIST ){ + pIdxInfo->aConstraintUsage[iDistTerm].argvIndex = idx++; + pIdxInfo->aConstraintUsage[iDistTerm].omit = 1; + } + pIdxInfo->estimatedCost = 1e5; + }else if( (iPlan & SPELLFIX_IDXNUM_ROWID) ){ + pIdxInfo->idxNum = SPELLFIX_IDXNUM_ROWID; + pIdxInfo->aConstraintUsage[iRowidTerm].argvIndex = 1; + pIdxInfo->aConstraintUsage[iRowidTerm].omit = 1; + pIdxInfo->estimatedCost = 5; + }else{ + pIdxInfo->idxNum = 0; + pIdxInfo->estimatedCost = 1e50; + } + return SQLITE_OK; +} + +/* +** Open a new fuzzy-search cursor. +*/ +static int spellfix1Open(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCursor){ + spellfix1_vtab *p = (spellfix1_vtab*)pVTab; + spellfix1_cursor *pCur; + pCur = sqlite3_malloc64( sizeof(*pCur) ); + if( pCur==0 ) return SQLITE_NOMEM; + memset(pCur, 0, sizeof(*pCur)); + pCur->pVTab = p; + *ppCursor = &pCur->base; + return SQLITE_OK; +} + +/* +** Adjust a distance measurement by the words rank in order to show +** preference to common words. +*/ +static int spellfix1Score(int iDistance, int iRank){ + int iLog2; + for(iLog2=0; iRank>0; iLog2++, iRank>>=1){} + return iDistance + 32 - iLog2; +} + +/* +** Compare two spellfix1_row objects for sorting purposes in qsort() such +** that they sort in order of increasing distance. +*/ +static int SQLITE_CDECL spellfix1RowCompare(const void *A, const void *B){ + const struct spellfix1_row *a = (const struct spellfix1_row*)A; + const struct spellfix1_row *b = (const struct spellfix1_row*)B; + return a->iScore - b->iScore; +} + +/* +** A structure used to pass information from spellfix1FilterForMatch() +** into spellfix1RunQuery(). +*/ +typedef struct MatchQuery { + spellfix1_cursor *pCur; /* The cursor being queried */ + sqlite3_stmt *pStmt; /* shadow table query statment */ + char zHash[SPELLFIX_MX_HASH]; /* The current phonehash for zPattern */ + const char *zPattern; /* Transliterated input string */ + int nPattern; /* Length of zPattern */ + EditDist3FromString *pMatchStr3; /* Original unicode string */ + EditDist3Config *pConfig3; /* Edit-distance cost coefficients */ + const EditDist3Lang *pLang; /* The selected language coefficients */ + int iLang; /* The language id */ + int iScope; /* Default scope */ + int iMaxDist; /* Maximum allowed edit distance, or -1 */ + int rc; /* Error code */ + int nRun; /* Number of prior runs for the same zPattern */ + char azPrior[SPELLFIX_MX_RUN][SPELLFIX_MX_HASH]; /* Prior hashes */ +} MatchQuery; + +/* +** Run a query looking for the best matches against zPattern using +** zHash as the character class seed hash. +*/ +static void spellfix1RunQuery(MatchQuery *p, const char *zQuery, int nQuery){ + const char *zK1; + const char *zWord; + int iDist; + int iRank; + int iScore; + int iWorst = 0; + int idx; + int idxWorst = -1; + int i; + int iScope = p->iScope; + spellfix1_cursor *pCur = p->pCur; + sqlite3_stmt *pStmt = p->pStmt; + char zHash1[SPELLFIX_MX_HASH]; + char zHash2[SPELLFIX_MX_HASH]; + char *zClass; + int nClass; + int rc; + + if( pCur->a==0 || p->rc ) return; /* Prior memory allocation failure */ + zClass = (char*)phoneticHash((unsigned char*)zQuery, nQuery); + if( zClass==0 ){ + p->rc = SQLITE_NOMEM; + return; + } + nClass = (int)strlen(zClass); + if( nClass>SPELLFIX_MX_HASH-2 ){ + nClass = SPELLFIX_MX_HASH-2; + zClass[nClass] = 0; + } + if( nClass<=iScope ){ + if( nClass>2 ){ + iScope = nClass-1; + }else{ + iScope = nClass; + } + } + memcpy(zHash1, zClass, iScope); + sqlite3_free(zClass); + zHash1[iScope] = 0; + memcpy(zHash2, zHash1, iScope); + zHash2[iScope] = 'Z'; + zHash2[iScope+1] = 0; +#if SPELLFIX_MX_RUN>1 + for(i=0; inRun; i++){ + if( strcmp(p->azPrior[i], zHash1)==0 ) return; + } +#endif + assert( p->nRunazPrior[p->nRun++], zHash1, iScope+1); + if( sqlite3_bind_text(pStmt, 1, zHash1, -1, SQLITE_STATIC)==SQLITE_NOMEM + || sqlite3_bind_text(pStmt, 2, zHash2, -1, SQLITE_STATIC)==SQLITE_NOMEM + ){ + p->rc = SQLITE_NOMEM; + return; + } +#if SPELLFIX_MX_RUN>1 + for(i=0; inRow; i++){ + if( pCur->a[i].iScore>iWorst ){ + iWorst = pCur->a[i].iScore; + idxWorst = i; + } + } +#endif + while( sqlite3_step(pStmt)==SQLITE_ROW ){ + int iMatchlen = -1; + iRank = sqlite3_column_int(pStmt, 2); + if( p->pMatchStr3 ){ + int nWord = sqlite3_column_bytes(pStmt, 1); + zWord = (const char*)sqlite3_column_text(pStmt, 1); + iDist = editDist3Core(p->pMatchStr3, zWord, nWord, p->pLang, &iMatchlen); + }else{ + zK1 = (const char*)sqlite3_column_text(pStmt, 3); + if( zK1==0 ) continue; + iDist = editdist1(p->zPattern, zK1, 0); + } + if( iDist<0 ){ + p->rc = SQLITE_NOMEM; + break; + } + pCur->nSearch++; + + /* If there is a "distance < $dist" or "distance <= $dist" constraint, + ** check if this row meets it. If not, jump back up to the top of the + ** loop to process the next row. Otherwise, if the row does match the + ** distance constraint, check if the pCur->a[] array is already full. + ** If it is and no explicit "top = ?" constraint was present in the + ** query, grow the array to ensure there is room for the new entry. */ + assert( (p->iMaxDist>=0)==((pCur->idxNum & SPELLFIX_IDXNUM_DIST) ? 1 : 0) ); + if( p->iMaxDist>=0 ){ + if( iDist>p->iMaxDist ) continue; + if( pCur->nRow>=pCur->nAlloc && (pCur->idxNum & SPELLFIX_IDXNUM_TOP)==0 ){ + spellfix1ResizeCursor(pCur, pCur->nAlloc*2 + 10); + if( pCur->a==0 ) break; + } + } + + iScore = spellfix1Score(iDist,iRank); + if( pCur->nRownAlloc ){ + idx = pCur->nRow; + }else if( iScorea[idx].zWord); + }else{ + continue; + } + + pCur->a[idx].zWord = sqlite3_mprintf("%s", sqlite3_column_text(pStmt, 1)); + if( pCur->a[idx].zWord==0 ){ + p->rc = SQLITE_NOMEM; + break; + } + pCur->a[idx].iRowid = sqlite3_column_int64(pStmt, 0); + pCur->a[idx].iRank = iRank; + pCur->a[idx].iDistance = iDist; + pCur->a[idx].iScore = iScore; + pCur->a[idx].iMatchlen = iMatchlen; + memcpy(pCur->a[idx].zHash, zHash1, iScope+1); + if( pCur->nRownAlloc ) pCur->nRow++; + if( pCur->nRow==pCur->nAlloc ){ + iWorst = pCur->a[0].iScore; + idxWorst = 0; + for(i=1; inRow; i++){ + iScore = pCur->a[i].iScore; + if( iWorstrc = rc; +} + +/* +** This version of the xFilter method work if the MATCH term is present +** and we are doing a scan. +*/ +static int spellfix1FilterForMatch( + spellfix1_cursor *pCur, + int argc, + sqlite3_value **argv +){ + int idxNum = pCur->idxNum; + const unsigned char *zMatchThis; /* RHS of the MATCH operator */ + EditDist3FromString *pMatchStr3 = 0; /* zMatchThis as an editdist string */ + char *zPattern; /* Transliteration of zMatchThis */ + int nPattern; /* Length of zPattern */ + int iLimit = 20; /* Max number of rows of output */ + int iScope = 3; /* Use this many characters of zClass */ + int iLang = 0; /* Language code */ + char *zSql; /* SQL of shadow table query */ + sqlite3_stmt *pStmt = 0; /* Shadow table query */ + int rc; /* Result code */ + int idx = 1; /* Next available filter parameter */ + spellfix1_vtab *p = pCur->pVTab; /* The virtual table that owns pCur */ + MatchQuery x; /* For passing info to RunQuery() */ + + /* Load the cost table if we have not already done so */ + if( p->zCostTable!=0 && p->pConfig3==0 ){ + p->pConfig3 = sqlite3_malloc64( sizeof(p->pConfig3[0]) ); + if( p->pConfig3==0 ) return SQLITE_NOMEM; + memset(p->pConfig3, 0, sizeof(p->pConfig3[0])); + rc = editDist3ConfigLoad(p->pConfig3, p->db, p->zCostTable); + if( rc ) return rc; + } + memset(&x, 0, sizeof(x)); + x.iScope = 3; /* Default scope if none specified by "WHERE scope=N" */ + x.iMaxDist = -1; /* Maximum allowed edit distance */ + + if( idxNum&2 ){ + iLang = sqlite3_value_int(argv[idx++]); + } + if( idxNum&4 ){ + iLimit = sqlite3_value_int(argv[idx++]); + if( iLimit<1 ) iLimit = 1; + } + if( idxNum&8 ){ + x.iScope = sqlite3_value_int(argv[idx++]); + if( x.iScope<1 ) x.iScope = 1; + if( x.iScope>SPELLFIX_MX_HASH-2 ) x.iScope = SPELLFIX_MX_HASH-2; + } + if( idxNum&(16|32) ){ + x.iMaxDist = sqlite3_value_int(argv[idx++]); + if( idxNum&16 ) x.iMaxDist--; + if( x.iMaxDist<0 ) x.iMaxDist = 0; + } + spellfix1ResetCursor(pCur); + spellfix1ResizeCursor(pCur, iLimit); + zMatchThis = sqlite3_value_text(argv[0]); + if( zMatchThis==0 ) return SQLITE_OK; + if( p->pConfig3 ){ + x.pLang = editDist3FindLang(p->pConfig3, iLang); + pMatchStr3 = editDist3FromStringNew(x.pLang, (const char*)zMatchThis, -1); + if( pMatchStr3==0 ){ + x.rc = SQLITE_NOMEM; + goto filter_exit; + } + }else{ + x.pLang = 0; + } + zPattern = (char*)transliterate(zMatchThis, sqlite3_value_bytes(argv[0])); + sqlite3_free(pCur->zPattern); + pCur->zPattern = zPattern; + if( zPattern==0 ){ + x.rc = SQLITE_NOMEM; + goto filter_exit; + } + nPattern = (int)strlen(zPattern); + if( zPattern[nPattern-1]=='*' ) nPattern--; + zSql = sqlite3_mprintf( + "SELECT id, word, rank, coalesce(k1,word)" + " FROM \"%w\".\"%w_vocab\"" + " WHERE langid=%d AND k2>=?1 AND k2zDbName, p->zTableName, iLang + ); + if( zSql==0 ){ + x.rc = SQLITE_NOMEM; + pStmt = 0; + goto filter_exit; + } + rc = sqlite3_prepare_v2(p->db, zSql, -1, &pStmt, 0); + sqlite3_free(zSql); + pCur->iLang = iLang; + x.pCur = pCur; + x.pStmt = pStmt; + x.zPattern = zPattern; + x.nPattern = nPattern; + x.pMatchStr3 = pMatchStr3; + x.iLang = iLang; + x.rc = rc; + x.pConfig3 = p->pConfig3; + if( x.rc==SQLITE_OK ){ + spellfix1RunQuery(&x, zPattern, nPattern); + } + + if( pCur->a ){ + qsort(pCur->a, pCur->nRow, sizeof(pCur->a[0]), spellfix1RowCompare); + pCur->iTop = iLimit; + pCur->iScope = iScope; + }else{ + x.rc = SQLITE_NOMEM; + } + +filter_exit: + sqlite3_finalize(pStmt); + editDist3FromStringDelete(pMatchStr3); + return x.rc; +} + +/* +** This version of xFilter handles a full-table scan case +*/ +static int spellfix1FilterForFullScan( + spellfix1_cursor *pCur, + int argc, + sqlite3_value **argv +){ + int rc = SQLITE_OK; + int idxNum = pCur->idxNum; + char *zSql; + spellfix1_vtab *pVTab = pCur->pVTab; + spellfix1ResetCursor(pCur); + assert( idxNum==0 || idxNum==64 ); + zSql = sqlite3_mprintf( + "SELECT word, rank, NULL, langid, id FROM \"%w\".\"%w_vocab\"%s", + pVTab->zDbName, pVTab->zTableName, + ((idxNum & 64) ? " WHERE rowid=?" : "") + ); + if( zSql==0 ) return SQLITE_NOMEM; + rc = sqlite3_prepare_v2(pVTab->db, zSql, -1, &pCur->pFullScan, 0); + sqlite3_free(zSql); + if( rc==SQLITE_OK && (idxNum & 64) ){ + assert( argc==1 ); + rc = sqlite3_bind_value(pCur->pFullScan, 1, argv[0]); + } + pCur->nRow = pCur->iRow = 0; + if( rc==SQLITE_OK ){ + rc = sqlite3_step(pCur->pFullScan); + if( rc==SQLITE_ROW ){ pCur->iRow = -1; rc = SQLITE_OK; } + if( rc==SQLITE_DONE ){ rc = SQLITE_OK; } + }else{ + pCur->iRow = 0; + } + return rc; +} + + +/* +** Called to "rewind" a cursor back to the beginning so that +** it starts its output over again. Always called at least once +** prior to any spellfix1Column, spellfix1Rowid, or spellfix1Eof call. +*/ +static int spellfix1Filter( + sqlite3_vtab_cursor *cur, + int idxNum, const char *idxStr, + int argc, sqlite3_value **argv +){ + spellfix1_cursor *pCur = (spellfix1_cursor *)cur; + int rc; + pCur->idxNum = idxNum; + if( idxNum & 1 ){ + rc = spellfix1FilterForMatch(pCur, argc, argv); + }else{ + rc = spellfix1FilterForFullScan(pCur, argc, argv); + } + return rc; +} + + +/* +** Advance a cursor to its next row of output +*/ +static int spellfix1Next(sqlite3_vtab_cursor *cur){ + spellfix1_cursor *pCur = (spellfix1_cursor *)cur; + int rc = SQLITE_OK; + if( pCur->iRow < pCur->nRow ){ + if( pCur->pFullScan ){ + rc = sqlite3_step(pCur->pFullScan); + if( rc!=SQLITE_ROW ) pCur->iRow = pCur->nRow; + if( rc==SQLITE_ROW || rc==SQLITE_DONE ) rc = SQLITE_OK; + }else{ + pCur->iRow++; + } + } + return rc; +} + +/* +** Return TRUE if we are at the end-of-file +*/ +static int spellfix1Eof(sqlite3_vtab_cursor *cur){ + spellfix1_cursor *pCur = (spellfix1_cursor *)cur; + return pCur->iRow>=pCur->nRow; +} + +/* +** Return columns from the current row. +*/ +static int spellfix1Column( + sqlite3_vtab_cursor *cur, + sqlite3_context *ctx, + int i +){ + spellfix1_cursor *pCur = (spellfix1_cursor*)cur; + if( pCur->pFullScan ){ + if( i<=SPELLFIX_COL_LANGID ){ + sqlite3_result_value(ctx, sqlite3_column_value(pCur->pFullScan, i)); + }else{ + sqlite3_result_null(ctx); + } + return SQLITE_OK; + } + switch( i ){ + case SPELLFIX_COL_WORD: { + sqlite3_result_text(ctx, pCur->a[pCur->iRow].zWord, -1, SQLITE_STATIC); + break; + } + case SPELLFIX_COL_RANK: { + sqlite3_result_int(ctx, pCur->a[pCur->iRow].iRank); + break; + } + case SPELLFIX_COL_DISTANCE: { + sqlite3_result_int(ctx, pCur->a[pCur->iRow].iDistance); + break; + } + case SPELLFIX_COL_LANGID: { + sqlite3_result_int(ctx, pCur->iLang); + break; + } + case SPELLFIX_COL_SCORE: { + sqlite3_result_int(ctx, pCur->a[pCur->iRow].iScore); + break; + } + case SPELLFIX_COL_MATCHLEN: { + int iMatchlen = pCur->a[pCur->iRow].iMatchlen; + if( iMatchlen<0 ){ + int nPattern = (int)strlen(pCur->zPattern); + char *zWord = pCur->a[pCur->iRow].zWord; + int nWord = (int)strlen(zWord); + + if( nPattern>0 && pCur->zPattern[nPattern-1]=='*' ){ + char *zTranslit; + int res; + zTranslit = (char *)transliterate((unsigned char *)zWord, nWord); + if( !zTranslit ) return SQLITE_NOMEM; + res = editdist1(pCur->zPattern, zTranslit, &iMatchlen); + sqlite3_free(zTranslit); + if( res<0 ) return SQLITE_NOMEM; + iMatchlen = translen_to_charlen(zWord, nWord, iMatchlen); + }else{ + iMatchlen = utf8Charlen(zWord, nWord); + } + } + + sqlite3_result_int(ctx, iMatchlen); + break; + } + case SPELLFIX_COL_PHONEHASH: { + sqlite3_result_text(ctx, pCur->a[pCur->iRow].zHash, -1, SQLITE_STATIC); + break; + } + case SPELLFIX_COL_TOP: { + sqlite3_result_int(ctx, pCur->iTop); + break; + } + case SPELLFIX_COL_SCOPE: { + sqlite3_result_int(ctx, pCur->iScope); + break; + } + case SPELLFIX_COL_SRCHCNT: { + sqlite3_result_int(ctx, pCur->nSearch); + break; + } + default: { + sqlite3_result_null(ctx); + break; + } + } + return SQLITE_OK; +} + +/* +** The rowid. +*/ +static int spellfix1Rowid(sqlite3_vtab_cursor *cur, sqlite_int64 *pRowid){ + spellfix1_cursor *pCur = (spellfix1_cursor*)cur; + if( pCur->pFullScan ){ + *pRowid = sqlite3_column_int64(pCur->pFullScan, 4); + }else{ + *pRowid = pCur->a[pCur->iRow].iRowid; + } + return SQLITE_OK; +} + +/* +** This function is called by the xUpdate() method. It returns a string +** containing the conflict mode that xUpdate() should use for the current +** operation. One of: "ROLLBACK", "IGNORE", "ABORT" or "REPLACE". +*/ +static const char *spellfix1GetConflict(sqlite3 *db){ + static const char *azConflict[] = { + /* Note: Instead of "FAIL" - "ABORT". */ + "ROLLBACK", "IGNORE", "ABORT", "ABORT", "REPLACE" + }; + int eConflict = sqlite3_vtab_on_conflict(db); + + assert( eConflict==SQLITE_ROLLBACK || eConflict==SQLITE_IGNORE + || eConflict==SQLITE_FAIL || eConflict==SQLITE_ABORT + || eConflict==SQLITE_REPLACE + ); + assert( SQLITE_ROLLBACK==1 ); + assert( SQLITE_IGNORE==2 ); + assert( SQLITE_FAIL==3 ); + assert( SQLITE_ABORT==4 ); + assert( SQLITE_REPLACE==5 ); + + return azConflict[eConflict-1]; +} + +/* +** The xUpdate() method. +*/ +static int spellfix1Update( + sqlite3_vtab *pVTab, + int argc, + sqlite3_value **argv, + sqlite_int64 *pRowid +){ + int rc = SQLITE_OK; + sqlite3_int64 rowid, newRowid; + spellfix1_vtab *p = (spellfix1_vtab*)pVTab; + sqlite3 *db = p->db; + + if( argc==1 ){ + /* A delete operation on the rowid given by argv[0] */ + rowid = *pRowid = sqlite3_value_int64(argv[0]); + spellfix1DbExec(&rc, db, "DELETE FROM \"%w\".\"%w_vocab\" " + " WHERE id=%lld", + p->zDbName, p->zTableName, rowid); + }else{ + const unsigned char *zWord = sqlite3_value_text(argv[SPELLFIX_COL_WORD+2]); + int nWord = sqlite3_value_bytes(argv[SPELLFIX_COL_WORD+2]); + int iLang = sqlite3_value_int(argv[SPELLFIX_COL_LANGID+2]); + int iRank = sqlite3_value_int(argv[SPELLFIX_COL_RANK+2]); + const unsigned char *zSoundslike = + sqlite3_value_text(argv[SPELLFIX_COL_SOUNDSLIKE+2]); + int nSoundslike = sqlite3_value_bytes(argv[SPELLFIX_COL_SOUNDSLIKE+2]); + char *zK1, *zK2; + int i; + char c; + const char *zConflict = spellfix1GetConflict(db); + + if( zWord==0 ){ + /* Inserts of the form: INSERT INTO table(command) VALUES('xyzzy'); + ** cause zWord to be NULL, so we look at the "command" column to see + ** what special actions to take */ + const char *zCmd = + (const char*)sqlite3_value_text(argv[SPELLFIX_COL_COMMAND+2]); + if( zCmd==0 ){ + pVTab->zErrMsg = sqlite3_mprintf("NOT NULL constraint failed: %s.word", + p->zTableName); + return SQLITE_CONSTRAINT_NOTNULL; + } + if( strcmp(zCmd,"reset")==0 ){ + /* Reset the edit cost table (if there is one). */ + editDist3ConfigDelete(p->pConfig3); + p->pConfig3 = 0; + return SQLITE_OK; + } + if( strncmp(zCmd,"edit_cost_table=",16)==0 ){ + editDist3ConfigDelete(p->pConfig3); + p->pConfig3 = 0; + sqlite3_free(p->zCostTable); + p->zCostTable = spellfix1Dequote(zCmd+16); + if( p->zCostTable==0 ) return SQLITE_NOMEM; + if( p->zCostTable[0]==0 || sqlite3_stricmp(p->zCostTable,"null")==0 ){ + sqlite3_free(p->zCostTable); + p->zCostTable = 0; + } + return SQLITE_OK; + } + pVTab->zErrMsg = sqlite3_mprintf("unknown value for %s.command: \"%w\"", + p->zTableName, zCmd); + return SQLITE_ERROR; + } + if( iRank<1 ) iRank = 1; + if( zSoundslike ){ + zK1 = (char*)transliterate(zSoundslike, nSoundslike); + }else{ + zK1 = (char*)transliterate(zWord, nWord); + } + if( zK1==0 ) return SQLITE_NOMEM; + for(i=0; (c = zK1[i])!=0; i++){ + if( c>='A' && c<='Z' ) zK1[i] += 'a' - 'A'; + } + zK2 = (char*)phoneticHash((const unsigned char*)zK1, i); + if( zK2==0 ){ + sqlite3_free(zK1); + return SQLITE_NOMEM; + } + if( sqlite3_value_type(argv[0])==SQLITE_NULL ){ + if( sqlite3_value_type(argv[1])==SQLITE_NULL ){ + spellfix1DbExec(&rc, db, + "INSERT INTO \"%w\".\"%w_vocab\"(rank,langid,word,k1,k2) " + "VALUES(%d,%d,%Q,nullif(%Q,%Q),%Q)", + p->zDbName, p->zTableName, + iRank, iLang, zWord, zK1, zWord, zK2 + ); + }else{ + newRowid = sqlite3_value_int64(argv[1]); + spellfix1DbExec(&rc, db, + "INSERT OR %s INTO \"%w\".\"%w_vocab\"(id,rank,langid,word,k1,k2) " + "VALUES(%lld,%d,%d,%Q,nullif(%Q,%Q),%Q)", + zConflict, p->zDbName, p->zTableName, + newRowid, iRank, iLang, zWord, zK1, zWord, zK2 + ); + } + *pRowid = sqlite3_last_insert_rowid(db); + }else{ + rowid = sqlite3_value_int64(argv[0]); + newRowid = *pRowid = sqlite3_value_int64(argv[1]); + spellfix1DbExec(&rc, db, + "UPDATE OR %s \"%w\".\"%w_vocab\" SET id=%lld, rank=%d, langid=%d," + " word=%Q, k1=nullif(%Q,%Q), k2=%Q WHERE id=%lld", + zConflict, p->zDbName, p->zTableName, newRowid, iRank, iLang, + zWord, zK1, zWord, zK2, rowid + ); + } + sqlite3_free(zK1); + sqlite3_free(zK2); + } + return rc; +} + +/* +** Rename the spellfix1 table. +*/ +static int spellfix1Rename(sqlite3_vtab *pVTab, const char *zNew){ + spellfix1_vtab *p = (spellfix1_vtab*)pVTab; + sqlite3 *db = p->db; + int rc = SQLITE_OK; + char *zNewName = sqlite3_mprintf("%s", zNew); + if( zNewName==0 ){ + return SQLITE_NOMEM; + } + spellfix1DbExec(&rc, db, + "ALTER TABLE \"%w\".\"%w_vocab\" RENAME TO \"%w_vocab\"", + p->zDbName, p->zTableName, zNewName + ); + if( rc==SQLITE_OK ){ + sqlite3_free(p->zTableName); + p->zTableName = zNewName; + }else{ + sqlite3_free(zNewName); + } + return rc; +} + + +/* +** A virtual table module that provides fuzzy search. +*/ +static sqlite3_module spellfix1Module = { + 0, /* iVersion */ + spellfix1Create, /* xCreate - handle CREATE VIRTUAL TABLE */ + spellfix1Connect, /* xConnect - reconnected to an existing table */ + spellfix1BestIndex, /* xBestIndex - figure out how to do a query */ + spellfix1Disconnect, /* xDisconnect - close a connection */ + spellfix1Destroy, /* xDestroy - handle DROP TABLE */ + spellfix1Open, /* xOpen - open a cursor */ + spellfix1Close, /* xClose - close a cursor */ + spellfix1Filter, /* xFilter - configure scan constraints */ + spellfix1Next, /* xNext - advance a cursor */ + spellfix1Eof, /* xEof - check for end of scan */ + spellfix1Column, /* xColumn - read data */ + spellfix1Rowid, /* xRowid - read data */ + spellfix1Update, /* xUpdate */ + 0, /* xBegin */ + 0, /* xSync */ + 0, /* xCommit */ + 0, /* xRollback */ + 0, /* xFindMethod */ + spellfix1Rename, /* xRename */ + 0, /* xSavepoint */ + 0, /* xRelease */ + 0, /* xRollbackTo */ + 0, /* xShadowName */ + 0 /* xIntegrity */ +}; + +/* +** Register the various functions and the virtual table. +*/ +static int spellfix1Register(sqlite3 *db){ + int rc = SQLITE_OK; + unsigned int i; + rc = sqlite3_create_function(db, "spellfix1_translit", 1, + SQLITE_UTF8|SQLITE_DETERMINISTIC, 0, + transliterateSqlFunc, 0, 0); + if( rc==SQLITE_OK ){ + rc = sqlite3_create_function(db, "spellfix1_editdist", 2, + SQLITE_UTF8|SQLITE_DETERMINISTIC, 0, + editdistSqlFunc, 0, 0); + } + if( rc==SQLITE_OK ){ + rc = sqlite3_create_function(db, "spellfix1_phonehash", 1, + SQLITE_UTF8|SQLITE_DETERMINISTIC, 0, + phoneticHashSqlFunc, 0, 0); + } + if( rc==SQLITE_OK ){ + rc = sqlite3_create_function(db, "spellfix1_scriptcode", 1, + SQLITE_UTF8|SQLITE_DETERMINISTIC, 0, + scriptCodeSqlFunc, 0, 0); + } + if( rc==SQLITE_OK ){ + rc = sqlite3_create_module(db, "spellfix1", &spellfix1Module, 0); + } + if( rc==SQLITE_OK ){ + rc = editDist3Install(db); + } + + /* Verify sanity of the translit[] table */ + for(i=0; iaggregate_context +#ifndef SQLITE_OMIT_DEPRECATED +#define sqlite3_aggregate_count sqlite3_api->aggregate_count +#endif +#define sqlite3_bind_blob sqlite3_api->bind_blob +#define sqlite3_bind_double sqlite3_api->bind_double +#define sqlite3_bind_int sqlite3_api->bind_int +#define sqlite3_bind_int64 sqlite3_api->bind_int64 +#define sqlite3_bind_null sqlite3_api->bind_null +#define sqlite3_bind_parameter_count sqlite3_api->bind_parameter_count +#define sqlite3_bind_parameter_index sqlite3_api->bind_parameter_index +#define sqlite3_bind_parameter_name sqlite3_api->bind_parameter_name +#define sqlite3_bind_text sqlite3_api->bind_text +#define sqlite3_bind_text16 sqlite3_api->bind_text16 +#define sqlite3_bind_value sqlite3_api->bind_value +#define sqlite3_busy_handler sqlite3_api->busy_handler +#define sqlite3_busy_timeout sqlite3_api->busy_timeout +#define sqlite3_changes sqlite3_api->changes +#define sqlite3_close sqlite3_api->close +#define sqlite3_collation_needed sqlite3_api->collation_needed +#define sqlite3_collation_needed16 sqlite3_api->collation_needed16 +#define sqlite3_column_blob sqlite3_api->column_blob +#define sqlite3_column_bytes sqlite3_api->column_bytes +#define sqlite3_column_bytes16 sqlite3_api->column_bytes16 +#define sqlite3_column_count sqlite3_api->column_count +#define sqlite3_column_database_name sqlite3_api->column_database_name +#define sqlite3_column_database_name16 sqlite3_api->column_database_name16 +#define sqlite3_column_decltype sqlite3_api->column_decltype +#define sqlite3_column_decltype16 sqlite3_api->column_decltype16 +#define sqlite3_column_double sqlite3_api->column_double +#define sqlite3_column_int sqlite3_api->column_int +#define sqlite3_column_int64 sqlite3_api->column_int64 +#define sqlite3_column_name sqlite3_api->column_name +#define sqlite3_column_name16 sqlite3_api->column_name16 +#define sqlite3_column_origin_name sqlite3_api->column_origin_name +#define sqlite3_column_origin_name16 sqlite3_api->column_origin_name16 +#define sqlite3_column_table_name sqlite3_api->column_table_name +#define sqlite3_column_table_name16 sqlite3_api->column_table_name16 +#define sqlite3_column_text sqlite3_api->column_text +#define sqlite3_column_text16 sqlite3_api->column_text16 +#define sqlite3_column_type sqlite3_api->column_type +#define sqlite3_column_value sqlite3_api->column_value +#define sqlite3_commit_hook sqlite3_api->commit_hook +#define sqlite3_complete sqlite3_api->complete +#define sqlite3_complete16 sqlite3_api->complete16 +#define sqlite3_create_collation sqlite3_api->create_collation +#define sqlite3_create_collation16 sqlite3_api->create_collation16 +#define sqlite3_create_function sqlite3_api->create_function +#define sqlite3_create_function16 sqlite3_api->create_function16 +#define sqlite3_create_module sqlite3_api->create_module +#define sqlite3_create_module_v2 sqlite3_api->create_module_v2 +#define sqlite3_data_count sqlite3_api->data_count +#define sqlite3_db_handle sqlite3_api->db_handle +#define sqlite3_declare_vtab sqlite3_api->declare_vtab +#define sqlite3_enable_shared_cache sqlite3_api->enable_shared_cache +#define sqlite3_errcode sqlite3_api->errcode +#define sqlite3_errmsg sqlite3_api->errmsg +#define sqlite3_errmsg16 sqlite3_api->errmsg16 +#define sqlite3_exec sqlite3_api->exec +#ifndef SQLITE_OMIT_DEPRECATED +#define sqlite3_expired sqlite3_api->expired +#endif +#define sqlite3_finalize sqlite3_api->finalize +#define sqlite3_free sqlite3_api->free +#define sqlite3_free_table sqlite3_api->free_table +#define sqlite3_get_autocommit sqlite3_api->get_autocommit +#define sqlite3_get_auxdata sqlite3_api->get_auxdata +#define sqlite3_get_table sqlite3_api->get_table +#ifndef SQLITE_OMIT_DEPRECATED +#define sqlite3_global_recover sqlite3_api->global_recover +#endif +#define sqlite3_interrupt sqlite3_api->interruptx +#define sqlite3_last_insert_rowid sqlite3_api->last_insert_rowid +#define sqlite3_libversion sqlite3_api->libversion +#define sqlite3_libversion_number sqlite3_api->libversion_number +#define sqlite3_malloc sqlite3_api->malloc +#define sqlite3_mprintf sqlite3_api->mprintf +#define sqlite3_open sqlite3_api->open +#define sqlite3_open16 sqlite3_api->open16 +#define sqlite3_prepare sqlite3_api->prepare +#define sqlite3_prepare16 sqlite3_api->prepare16 +#define sqlite3_prepare_v2 sqlite3_api->prepare_v2 +#define sqlite3_prepare16_v2 sqlite3_api->prepare16_v2 +#define sqlite3_profile sqlite3_api->profile +#define sqlite3_progress_handler sqlite3_api->progress_handler +#define sqlite3_realloc sqlite3_api->realloc +#define sqlite3_reset sqlite3_api->reset +#define sqlite3_result_blob sqlite3_api->result_blob +#define sqlite3_result_double sqlite3_api->result_double +#define sqlite3_result_error sqlite3_api->result_error +#define sqlite3_result_error16 sqlite3_api->result_error16 +#define sqlite3_result_int sqlite3_api->result_int +#define sqlite3_result_int64 sqlite3_api->result_int64 +#define sqlite3_result_null sqlite3_api->result_null +#define sqlite3_result_text sqlite3_api->result_text +#define sqlite3_result_text16 sqlite3_api->result_text16 +#define sqlite3_result_text16be sqlite3_api->result_text16be +#define sqlite3_result_text16le sqlite3_api->result_text16le +#define sqlite3_result_value sqlite3_api->result_value +#define sqlite3_rollback_hook sqlite3_api->rollback_hook +#define sqlite3_set_authorizer sqlite3_api->set_authorizer +#define sqlite3_set_auxdata sqlite3_api->set_auxdata +#define sqlite3_snprintf sqlite3_api->xsnprintf +#define sqlite3_step sqlite3_api->step +#define sqlite3_table_column_metadata sqlite3_api->table_column_metadata +#define sqlite3_thread_cleanup sqlite3_api->thread_cleanup +#define sqlite3_total_changes sqlite3_api->total_changes +#define sqlite3_trace sqlite3_api->trace +#ifndef SQLITE_OMIT_DEPRECATED +#define sqlite3_transfer_bindings sqlite3_api->transfer_bindings +#endif +#define sqlite3_update_hook sqlite3_api->update_hook +#define sqlite3_user_data sqlite3_api->user_data +#define sqlite3_value_blob sqlite3_api->value_blob +#define sqlite3_value_bytes sqlite3_api->value_bytes +#define sqlite3_value_bytes16 sqlite3_api->value_bytes16 +#define sqlite3_value_double sqlite3_api->value_double +#define sqlite3_value_int sqlite3_api->value_int +#define sqlite3_value_int64 sqlite3_api->value_int64 +#define sqlite3_value_numeric_type sqlite3_api->value_numeric_type +#define sqlite3_value_text sqlite3_api->value_text +#define sqlite3_value_text16 sqlite3_api->value_text16 +#define sqlite3_value_text16be sqlite3_api->value_text16be +#define sqlite3_value_text16le sqlite3_api->value_text16le +#define sqlite3_value_type sqlite3_api->value_type +#define sqlite3_vmprintf sqlite3_api->vmprintf +#define sqlite3_vsnprintf sqlite3_api->xvsnprintf +#define sqlite3_overload_function sqlite3_api->overload_function +#define sqlite3_prepare_v2 sqlite3_api->prepare_v2 +#define sqlite3_prepare16_v2 sqlite3_api->prepare16_v2 +#define sqlite3_clear_bindings sqlite3_api->clear_bindings +#define sqlite3_bind_zeroblob sqlite3_api->bind_zeroblob +#define sqlite3_blob_bytes sqlite3_api->blob_bytes +#define sqlite3_blob_close sqlite3_api->blob_close +#define sqlite3_blob_open sqlite3_api->blob_open +#define sqlite3_blob_read sqlite3_api->blob_read +#define sqlite3_blob_write sqlite3_api->blob_write +#define sqlite3_create_collation_v2 sqlite3_api->create_collation_v2 +#define sqlite3_file_control sqlite3_api->file_control +#define sqlite3_memory_highwater sqlite3_api->memory_highwater +#define sqlite3_memory_used sqlite3_api->memory_used +#define sqlite3_mutex_alloc sqlite3_api->mutex_alloc +#define sqlite3_mutex_enter sqlite3_api->mutex_enter +#define sqlite3_mutex_free sqlite3_api->mutex_free +#define sqlite3_mutex_leave sqlite3_api->mutex_leave +#define sqlite3_mutex_try sqlite3_api->mutex_try +#define sqlite3_open_v2 sqlite3_api->open_v2 +#define sqlite3_release_memory sqlite3_api->release_memory +#define sqlite3_result_error_nomem sqlite3_api->result_error_nomem +#define sqlite3_result_error_toobig sqlite3_api->result_error_toobig +#define sqlite3_sleep sqlite3_api->sleep +#define sqlite3_soft_heap_limit sqlite3_api->soft_heap_limit +#define sqlite3_vfs_find sqlite3_api->vfs_find +#define sqlite3_vfs_register sqlite3_api->vfs_register +#define sqlite3_vfs_unregister sqlite3_api->vfs_unregister +#define sqlite3_threadsafe sqlite3_api->xthreadsafe +#define sqlite3_result_zeroblob sqlite3_api->result_zeroblob +#define sqlite3_result_error_code sqlite3_api->result_error_code +#define sqlite3_test_control sqlite3_api->test_control +#define sqlite3_randomness sqlite3_api->randomness +#define sqlite3_context_db_handle sqlite3_api->context_db_handle +#define sqlite3_extended_result_codes sqlite3_api->extended_result_codes +#define sqlite3_limit sqlite3_api->limit +#define sqlite3_next_stmt sqlite3_api->next_stmt +#define sqlite3_sql sqlite3_api->sql +#define sqlite3_status sqlite3_api->status +#define sqlite3_backup_finish sqlite3_api->backup_finish +#define sqlite3_backup_init sqlite3_api->backup_init +#define sqlite3_backup_pagecount sqlite3_api->backup_pagecount +#define sqlite3_backup_remaining sqlite3_api->backup_remaining +#define sqlite3_backup_step sqlite3_api->backup_step +#define sqlite3_compileoption_get sqlite3_api->compileoption_get +#define sqlite3_compileoption_used sqlite3_api->compileoption_used +#define sqlite3_create_function_v2 sqlite3_api->create_function_v2 +#define sqlite3_db_config sqlite3_api->db_config +#define sqlite3_db_mutex sqlite3_api->db_mutex +#define sqlite3_db_status sqlite3_api->db_status +#define sqlite3_extended_errcode sqlite3_api->extended_errcode +#define sqlite3_log sqlite3_api->log +#define sqlite3_soft_heap_limit64 sqlite3_api->soft_heap_limit64 +#define sqlite3_sourceid sqlite3_api->sourceid +#define sqlite3_stmt_status sqlite3_api->stmt_status +#define sqlite3_strnicmp sqlite3_api->strnicmp +#define sqlite3_unlock_notify sqlite3_api->unlock_notify +#define sqlite3_wal_autocheckpoint sqlite3_api->wal_autocheckpoint +#define sqlite3_wal_checkpoint sqlite3_api->wal_checkpoint +#define sqlite3_wal_hook sqlite3_api->wal_hook +#define sqlite3_blob_reopen sqlite3_api->blob_reopen +#define sqlite3_vtab_config sqlite3_api->vtab_config +#define sqlite3_vtab_on_conflict sqlite3_api->vtab_on_conflict +/* Version 3.7.16 and later */ +#define sqlite3_close_v2 sqlite3_api->close_v2 +#define sqlite3_db_filename sqlite3_api->db_filename +#define sqlite3_db_readonly sqlite3_api->db_readonly +#define sqlite3_db_release_memory sqlite3_api->db_release_memory +#define sqlite3_errstr sqlite3_api->errstr +#define sqlite3_stmt_busy sqlite3_api->stmt_busy +#define sqlite3_stmt_readonly sqlite3_api->stmt_readonly +#define sqlite3_stricmp sqlite3_api->stricmp +#define sqlite3_uri_boolean sqlite3_api->uri_boolean +#define sqlite3_uri_int64 sqlite3_api->uri_int64 +#define sqlite3_uri_parameter sqlite3_api->uri_parameter +#define sqlite3_uri_vsnprintf sqlite3_api->xvsnprintf +#define sqlite3_wal_checkpoint_v2 sqlite3_api->wal_checkpoint_v2 +/* Version 3.8.7 and later */ +#define sqlite3_auto_extension sqlite3_api->auto_extension +#define sqlite3_bind_blob64 sqlite3_api->bind_blob64 +#define sqlite3_bind_text64 sqlite3_api->bind_text64 +#define sqlite3_cancel_auto_extension sqlite3_api->cancel_auto_extension +#define sqlite3_load_extension sqlite3_api->load_extension +#define sqlite3_malloc64 sqlite3_api->malloc64 +#define sqlite3_msize sqlite3_api->msize +#define sqlite3_realloc64 sqlite3_api->realloc64 +#define sqlite3_reset_auto_extension sqlite3_api->reset_auto_extension +#define sqlite3_result_blob64 sqlite3_api->result_blob64 +#define sqlite3_result_text64 sqlite3_api->result_text64 +#define sqlite3_strglob sqlite3_api->strglob +/* Version 3.8.11 and later */ +#define sqlite3_value_dup sqlite3_api->value_dup +#define sqlite3_value_free sqlite3_api->value_free +#define sqlite3_result_zeroblob64 sqlite3_api->result_zeroblob64 +#define sqlite3_bind_zeroblob64 sqlite3_api->bind_zeroblob64 +/* Version 3.9.0 and later */ +#define sqlite3_value_subtype sqlite3_api->value_subtype +#define sqlite3_result_subtype sqlite3_api->result_subtype +/* Version 3.10.0 and later */ +#define sqlite3_status64 sqlite3_api->status64 +#define sqlite3_strlike sqlite3_api->strlike +#define sqlite3_db_cacheflush sqlite3_api->db_cacheflush +/* Version 3.12.0 and later */ +#define sqlite3_system_errno sqlite3_api->system_errno +/* Version 3.14.0 and later */ +#define sqlite3_trace_v2 sqlite3_api->trace_v2 +#define sqlite3_expanded_sql sqlite3_api->expanded_sql +/* Version 3.18.0 and later */ +#define sqlite3_set_last_insert_rowid sqlite3_api->set_last_insert_rowid +/* Version 3.20.0 and later */ +#define sqlite3_prepare_v3 sqlite3_api->prepare_v3 +#define sqlite3_prepare16_v3 sqlite3_api->prepare16_v3 +#define sqlite3_bind_pointer sqlite3_api->bind_pointer +#define sqlite3_result_pointer sqlite3_api->result_pointer +#define sqlite3_value_pointer sqlite3_api->value_pointer +/* Version 3.22.0 and later */ +#define sqlite3_vtab_nochange sqlite3_api->vtab_nochange +#define sqlite3_value_nochange sqlite3_api->value_nochange +#define sqlite3_vtab_collation sqlite3_api->vtab_collation +/* Version 3.24.0 and later */ +#define sqlite3_keyword_count sqlite3_api->keyword_count +#define sqlite3_keyword_name sqlite3_api->keyword_name +#define sqlite3_keyword_check sqlite3_api->keyword_check +#define sqlite3_str_new sqlite3_api->str_new +#define sqlite3_str_finish sqlite3_api->str_finish +#define sqlite3_str_appendf sqlite3_api->str_appendf +#define sqlite3_str_vappendf sqlite3_api->str_vappendf +#define sqlite3_str_append sqlite3_api->str_append +#define sqlite3_str_appendall sqlite3_api->str_appendall +#define sqlite3_str_appendchar sqlite3_api->str_appendchar +#define sqlite3_str_reset sqlite3_api->str_reset +#define sqlite3_str_errcode sqlite3_api->str_errcode +#define sqlite3_str_length sqlite3_api->str_length +#define sqlite3_str_value sqlite3_api->str_value +/* Version 3.25.0 and later */ +#define sqlite3_create_window_function sqlite3_api->create_window_function +/* Version 3.26.0 and later */ +#define sqlite3_normalized_sql sqlite3_api->normalized_sql +/* Version 3.28.0 and later */ +#define sqlite3_stmt_isexplain sqlite3_api->stmt_isexplain +#define sqlite3_value_frombind sqlite3_api->value_frombind +/* Version 3.30.0 and later */ +#define sqlite3_drop_modules sqlite3_api->drop_modules +/* Version 3.31.0 and later */ +#define sqlite3_hard_heap_limit64 sqlite3_api->hard_heap_limit64 +#define sqlite3_uri_key sqlite3_api->uri_key +#define sqlite3_filename_database sqlite3_api->filename_database +#define sqlite3_filename_journal sqlite3_api->filename_journal +#define sqlite3_filename_wal sqlite3_api->filename_wal +/* Version 3.32.0 and later */ +#define sqlite3_create_filename sqlite3_api->create_filename +#define sqlite3_free_filename sqlite3_api->free_filename +#define sqlite3_database_file_object sqlite3_api->database_file_object +/* Version 3.34.0 and later */ +#define sqlite3_txn_state sqlite3_api->txn_state +/* Version 3.36.1 and later */ +#define sqlite3_changes64 sqlite3_api->changes64 +#define sqlite3_total_changes64 sqlite3_api->total_changes64 +/* Version 3.37.0 and later */ +#define sqlite3_autovacuum_pages sqlite3_api->autovacuum_pages +/* Version 3.38.0 and later */ +#define sqlite3_error_offset sqlite3_api->error_offset +#define sqlite3_vtab_rhs_value sqlite3_api->vtab_rhs_value +#define sqlite3_vtab_distinct sqlite3_api->vtab_distinct +#define sqlite3_vtab_in sqlite3_api->vtab_in +#define sqlite3_vtab_in_first sqlite3_api->vtab_in_first +#define sqlite3_vtab_in_next sqlite3_api->vtab_in_next +/* Version 3.39.0 and later */ +#ifndef SQLITE_OMIT_DESERIALIZE +#define sqlite3_deserialize sqlite3_api->deserialize +#define sqlite3_serialize sqlite3_api->serialize +#endif +#define sqlite3_db_name sqlite3_api->db_name +/* Version 3.40.0 and later */ +#define sqlite3_value_encoding sqlite3_api->value_encoding +/* Version 3.41.0 and later */ +#define sqlite3_is_interrupted sqlite3_api->is_interrupted +/* Version 3.43.0 and later */ +#define sqlite3_stmt_explain sqlite3_api->stmt_explain +/* Version 3.44.0 and later */ +#define sqlite3_get_clientdata sqlite3_api->get_clientdata +#define sqlite3_set_clientdata sqlite3_api->set_clientdata +/* Version 3.50.0 and later */ +#define sqlite3_setlk_timeout sqlite3_api->setlk_timeout +/* Version 3.51.0 and later */ +#define sqlite3_set_errmsg sqlite3_api->set_errmsg +#define sqlite3_db_status64 sqlite3_api->db_status64 +#endif /* !defined(SQLITE_CORE) && !defined(SQLITE_OMIT_LOAD_EXTENSION) */ + +#if !defined(SQLITE_CORE) && !defined(SQLITE_OMIT_LOAD_EXTENSION) + /* This case when the file really is being compiled as a loadable + ** extension */ +# define SQLITE_EXTENSION_INIT1 const sqlite3_api_routines *sqlite3_api=0; +# define SQLITE_EXTENSION_INIT2(v) sqlite3_api=v; +# define SQLITE_EXTENSION_INIT3 \ + extern const sqlite3_api_routines *sqlite3_api; +#else + /* This case when the file is being statically linked into the + ** application */ +# define SQLITE_EXTENSION_INIT1 /*no-op*/ +# define SQLITE_EXTENSION_INIT2(v) (void)v; /* unused parameter */ +# define SQLITE_EXTENSION_INIT3 /*no-op*/ +#endif + +#endif /* SQLITE3EXT_H */ diff --git a/db/spellfix/update.sh b/db/spellfix/update.sh new file mode 100755 index 000000000..2b2f1d69c --- /dev/null +++ b/db/spellfix/update.sh @@ -0,0 +1,31 @@ +#!/usr/bin/env bash +# +# Updates the vendored spellfix1 source files to match the SQLite version +# bundled with the current go-sqlite3 dependency. +# +set -euo pipefail + +cd "$(dirname "$0")" + +SQLITE_VERSION=$(grep '#define SQLITE_VERSION ' \ + "$(go env GOMODCACHE)/$(go list -m -f '{{.Path}}@{{.Version}}' github.com/mattn/go-sqlite3)/sqlite3-binding.h" \ + | awk '{gsub(/"/, "", $3); print $3}') + +if [ -z "$SQLITE_VERSION" ]; then + echo "ERROR: Could not determine SQLite version from go-sqlite3" >&2 + exit 1 +fi + +TAG="version-${SQLITE_VERSION}" +BASE_URL="https://raw.githubusercontent.com/sqlite/sqlite/${TAG}" + +echo "SQLite version from go-sqlite3: ${SQLITE_VERSION}" +echo "Downloading from tag: ${TAG}" + +curl -sfL "${BASE_URL}/ext/misc/spellfix.c" -o spellfix.c +echo " Updated spellfix.c" + +curl -sfL "${BASE_URL}/src/sqlite3ext.h" -o sqlite3ext.h +echo " Updated sqlite3ext.h" + +echo "Done." diff --git a/main.go b/main.go index b5fb508b4..fc41089c7 100644 --- a/main.go +++ b/main.go @@ -15,6 +15,7 @@ func main() { // To avoid these kind of errors, you should use `make build` to compile the project. _ = buildtags.NETGO _ = buildtags.SQLITE_FTS5 + _ = buildtags.SQLITE_SPELLFIX cmd.Execute() } diff --git a/reflex.conf b/reflex.conf index 47dd775ab..d805e27b0 100644 --- a/reflex.conf +++ b/reflex.conf @@ -1 +1 @@ --s -r "(\.go$$|\.cpp$$|\.h$$|navidrome.toml|resources|token_received.html)" -R "(^ui|^data|^db/migrations)" -R "_test\.go$$" -- go run -race -tags netgo,sqlite_fts5 . +-s -r "(\.go$$|\.cpp$$|\.h$$|navidrome.toml|resources|token_received.html)" -R "(^ui|^data|^db/migrations)" -R "_test\.go$$" -- go run -race -tags netgo,sqlite_fts5,sqlite_spellfix .