fts2.c

sqlite database for embed linux

typedef enum QueryType {
  QUERY_GENERIC,   /* table scan */
  QUERY_ROWID,     /* lookup by rowid */
  QUERY_FULLTEXT   /* QUERY_FULLTEXT + [i] is a full-text search for column i*/
} QueryType;

typedef enum fulltext_statement {
  CONTENT_INSERT_STMT,
  CONTENT_SELECT_STMT,
  CONTENT_UPDATE_STMT,
  CONTENT_DELETE_STMT,

  BLOCK_INSERT_STMT,
  BLOCK_SELECT_STMT,
  BLOCK_DELETE_STMT,

  SEGDIR_MAX_INDEX_STMT,
  SEGDIR_SET_STMT,
  SEGDIR_SELECT_STMT,
  SEGDIR_SPAN_STMT,
  SEGDIR_DELETE_STMT,
  SEGDIR_SELECT_ALL_STMT,

  MAX_STMT                     /* Always at end! */
} fulltext_statement;

/* These must exactly match the enum above. */
/* TODO(shess): Is there some risk that a statement will be used in two
** cursors at once, e.g.  if a query joins a virtual table to itself?
** If so perhaps we should move some of these to the cursor object.
*/
static const char *const fulltext_zStatement[MAX_STMT] = {
  /* CONTENT_INSERT */ NULL,  /* generated in contentInsertStatement() */
  /* CONTENT_SELECT */ "select * from %_content where rowid = ?",
  /* CONTENT_UPDATE */ NULL,  /* generated in contentUpdateStatement() */
  /* CONTENT_DELETE */ "delete from %_content where rowid = ?",

  /* BLOCK_INSERT */ "insert into %_segments values (?)",
  /* BLOCK_SELECT */ "select block from %_segments where rowid = ?",
  /* BLOCK_DELETE */ "delete from %_segments where rowid between ? and ?",

  /* SEGDIR_MAX_INDEX */ "select max(idx) from %_segdir where level = ?",
  /* SEGDIR_SET */ "insert into %_segdir values (?, ?, ?, ?, ?, ?)",
  /* SEGDIR_SELECT */
  "select start_block, leaves_end_block, root from %_segdir "
  " where level = ? order by idx",
  /* SEGDIR_SPAN */
  "select min(start_block), max(end_block) from %_segdir "
  " where level = ? and start_block <> 0",
  /* SEGDIR_DELETE */ "delete from %_segdir where level = ?",
  /* SEGDIR_SELECT_ALL */ "select root from %_segdir order by level desc, idx",
};

/*
** A connection to a fulltext index is an instance of the following
** structure.  The xCreate and xConnect methods create an instance
** of this structure and xDestroy and xDisconnect free that instance.
** All other methods receive a pointer to the structure as one of their
** arguments.
*/
struct fulltext_vtab {
  sqlite3_vtab base;               /* Base class used by SQLite core */
  sqlite3 *db;                     /* The database connection */
  const char *zDb;                 /* logical database name */
  const char *zName;               /* virtual table name */
  int nColumn;                     /* number of columns in virtual table */
  char **azColumn;                 /* column names.  malloced */
  char **azContentColumn;          /* column names in content table; malloced */
  sqlite3_tokenizer *pTokenizer;   /* tokenizer for inserts and queries */

  /* Precompiled statements which we keep as long as the table is
  ** open.
  */
  sqlite3_stmt *pFulltextStatements[MAX_STMT];

  /* Precompiled statements used for segment merges.  We run a
  ** separate select across the leaf level of each tree being merged.
  */
  sqlite3_stmt *pLeafSelectStmts[MERGE_COUNT];
  /* The statement used to prepare pLeafSelectStmts. */
#define LEAF_SELECT \
  "select block from %_segments where rowid between ? and ? order by rowid"

  /* These buffer pending index updates during transactions.
  ** nPendingData estimates the memory size of the pending data.  It
  ** doesn't include the hash-bucket overhead, nor any malloc
  ** overhead.  When nPendingData exceeds kPendingThreshold, the
  ** buffer is flushed even before the transaction closes.
  ** pendingTerms stores the data, and is only valid when nPendingData
  ** is >=0 (nPendingData<0 means pendingTerms has not been
  ** initialized).  iPrevDocid is the last docid written, used to make
  ** certain we're inserting in sorted order.
  */
  int nPendingData;
#define kPendingThreshold (1*1024*1024)
  sqlite_int64 iPrevDocid;
  fts2Hash pendingTerms;
};

/*
** When the core wants to do a query, it create a cursor using a
** call to xOpen.  This structure is an instance of a cursor.  It
** is destroyed by xClose.
*/
typedef struct fulltext_cursor {
  sqlite3_vtab_cursor base;        /* Base class used by SQLite core */
  QueryType iCursorType;           /* Copy of sqlite3_index_info.idxNum */
  sqlite3_stmt *pStmt;             /* Prepared statement in use by the cursor */
  int eof;                         /* True if at End Of Results */
  Query q;                         /* Parsed query string */
  Snippet snippet;                 /* Cached snippet for the current row */
  int iColumn;                     /* Column being searched */
  DataBuffer result;               /* Doclist results from fulltextQuery */
  DLReader reader;                 /* Result reader if result not empty */
} fulltext_cursor;

static struct fulltext_vtab *cursor_vtab(fulltext_cursor *c){
  return (fulltext_vtab *) c->base.pVtab;
}

static const sqlite3_module fulltextModule;   /* forward declaration */

/* Return a dynamically generated statement of the form
 *   insert into %_content (rowid, ...) values (?, ...)
 */
static const char *contentInsertStatement(fulltext_vtab *v){
  StringBuffer sb;
  int i;

  initStringBuffer(&sb);
  append(&sb, "insert into %_content (rowid, ");
  appendList(&sb, v->nColumn, v->azContentColumn);
  append(&sb, ") values (?");
  for(i=0; i<v->nColumn; ++i)
    append(&sb, ", ?");
  append(&sb, ")");
  return stringBufferData(&sb);
}

/* Return a dynamically generated statement of the form
 *   update %_content set [col_0] = ?, [col_1] = ?, ...
 *                    where rowid = ?
 */
static const char *contentUpdateStatement(fulltext_vtab *v){
  StringBuffer sb;
  int i;

  initStringBuffer(&sb);
  append(&sb, "update %_content set ");
  for(i=0; i<v->nColumn; ++i) {
    if( i>0 ){
      append(&sb, ", ");
    }
    append(&sb, v->azContentColumn[i]);
    append(&sb, " = ?");
  }
  append(&sb, " where rowid = ?");
  return stringBufferData(&sb);
}

/* Puts a freshly-prepared statement determined by iStmt in *ppStmt.
** If the indicated statement has never been prepared, it is prepared
** and cached, otherwise the cached version is reset.
*/
static int sql_get_statement(fulltext_vtab *v, fulltext_statement iStmt,
                             sqlite3_stmt **ppStmt){
  assert( iStmt<MAX_STMT );
  if( v->pFulltextStatements[iStmt]==NULL ){
    const char *zStmt;
    int rc;
    switch( iStmt ){
      case CONTENT_INSERT_STMT:
        zStmt = contentInsertStatement(v); break;
      case CONTENT_UPDATE_STMT:
        zStmt = contentUpdateStatement(v); break;
      default:
        zStmt = fulltext_zStatement[iStmt];
    }
    rc = sql_prepare(v->db, v->zDb, v->zName, &v->pFulltextStatements[iStmt],
                         zStmt);
    if( zStmt != fulltext_zStatement[iStmt]) free((void *) zStmt);
    if( rc!=SQLITE_OK ) return rc;
  } else {
    int rc = sqlite3_reset(v->pFulltextStatements[iStmt]);
    if( rc!=SQLITE_OK ) return rc;
  }

  *ppStmt = v->pFulltextStatements[iStmt];
  return SQLITE_OK;
}

/* Step the indicated statement, handling errors SQLITE_BUSY (by
** retrying) and SQLITE_SCHEMA (by re-preparing and transferring
** bindings to the new statement).
** TODO(adam): We should extend this function so that it can work with
** statements declared locally, not only globally cached statements.
*/
static int sql_step_statement(fulltext_vtab *v, fulltext_statement iStmt,
                              sqlite3_stmt **ppStmt){
  int rc;
  sqlite3_stmt *s = *ppStmt;
  assert( iStmt<MAX_STMT );
  assert( s==v->pFulltextStatements[iStmt] );

  while( (rc=sqlite3_step(s))!=SQLITE_DONE && rc!=SQLITE_ROW ){
    sqlite3_stmt *pNewStmt;

    if( rc==SQLITE_BUSY ) continue;
    if( rc!=SQLITE_ERROR ) return rc;

    rc = sqlite3_reset(s);
    if( rc!=SQLITE_SCHEMA ) return SQLITE_ERROR;

    v->pFulltextStatements[iStmt] = NULL;   /* Still in s */
    rc = sql_get_statement(v, iStmt, &pNewStmt);
    if( rc!=SQLITE_OK ) goto err;
    *ppStmt = pNewStmt;

    rc = sqlite3_transfer_bindings(s, pNewStmt);
    if( rc!=SQLITE_OK ) goto err;

    rc = sqlite3_finalize(s);
    if( rc!=SQLITE_OK ) return rc;
    s = pNewStmt;
  }
  return rc;

 err:
  sqlite3_finalize(s);
  return rc;
}

/* Like sql_step_statement(), but convert SQLITE_DONE to SQLITE_OK.
** Useful for statements like UPDATE, where we expect no results.
*/
static int sql_single_step_statement(fulltext_vtab *v,
                                     fulltext_statement iStmt,
                                     sqlite3_stmt **ppStmt){
  int rc = sql_step_statement(v, iStmt, ppStmt);
  return (rc==SQLITE_DONE) ? SQLITE_OK : rc;
}

/* Like sql_get_statement(), but for special replicated LEAF_SELECT
** statements.
*/
/* TODO(shess) Write version for generic statements and then share
** that between the cached-statement functions.
*/
static int sql_get_leaf_statement(fulltext_vtab *v, int idx,
                                  sqlite3_stmt **ppStmt){
  assert( idx>=0 && idx<MERGE_COUNT );
  if( v->pLeafSelectStmts[idx]==NULL ){
    int rc = sql_prepare(v->db, v->zDb, v->zName, &v->pLeafSelectStmts[idx],
                         LEAF_SELECT);
    if( rc!=SQLITE_OK ) return rc;
  }else{
    int rc = sqlite3_reset(v->pLeafSelectStmts[idx]);
    if( rc!=SQLITE_OK ) return rc;
  }

  *ppStmt = v->pLeafSelectStmts[idx];
  return SQLITE_OK;
}

/* Like sql_step_statement(), but for special replicated LEAF_SELECT
** statements.
*/
/* TODO(shess) Write version for generic statements and then share
** that between the cached-statement functions.
*/
static int sql_step_leaf_statement(fulltext_vtab *v, int idx,
                                   sqlite3_stmt **ppStmt){
  int rc;
  sqlite3_stmt *s = *ppStmt;

  while( (rc=sqlite3_step(s))!=SQLITE_DONE && rc!=SQLITE_ROW ){
    sqlite3_stmt *pNewStmt;

    if( rc==SQLITE_BUSY ) continue;
    if( rc!=SQLITE_ERROR ) return rc;

    rc = sqlite3_reset(s);
    if( rc!=SQLITE_SCHEMA ) return SQLITE_ERROR;

    v->pLeafSelectStmts[idx] = NULL;   /* Still in s */
    rc = sql_get_leaf_statement(v, idx, &pNewStmt);
    if( rc!=SQLITE_OK ) goto err;
    *ppStmt = pNewStmt;

    rc = sqlite3_transfer_bindings(s, pNewStmt);
    if( rc!=SQLITE_OK ) goto err;

    rc = sqlite3_finalize(s);
    if( rc!=SQLITE_OK ) return rc;
    s = pNewStmt;
  }
  return rc;

 err:
  sqlite3_finalize(s);
  return rc;
}

/* insert into %_content (rowid, ...) values ([rowid], [pValues]) */
static int content_insert(fulltext_vtab *v, sqlite3_value *rowid,
                          sqlite3_value **pValues){
  sqlite3_stmt *s;
  int i;
  int rc = sql_get_statement(v, CONTENT_INSERT_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_value(s, 1, rowid);
  if( rc!=SQLITE_OK ) return rc;

  for(i=0; i<v->nColumn; ++i){
    rc = sqlite3_bind_value(s, 2+i, pValues[i]);
    if( rc!=SQLITE_OK ) return rc;
  }

  return sql_single_step_statement(v, CONTENT_INSERT_STMT, &s);
}

/* update %_content set col0 = pValues[0], col1 = pValues[1], ...
 *                  where rowid = [iRowid] */
static int content_update(fulltext_vtab *v, sqlite3_value **pValues,
                          sqlite_int64 iRowid){
  sqlite3_stmt *s;
  int i;
  int rc = sql_get_statement(v, CONTENT_UPDATE_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  for(i=0; i<v->nColumn; ++i){
    rc = sqlite3_bind_value(s, 1+i, pValues[i]);
    if( rc!=SQLITE_OK ) return rc;
  }

  rc = sqlite3_bind_int64(s, 1+v->nColumn, iRowid);
  if( rc!=SQLITE_OK ) return rc;

  return sql_single_step_statement(v, CONTENT_UPDATE_STMT, &s);
}

static void freeStringArray(int nString, const char **pString){
  int i;

  for (i=0 ; i < nString ; ++i) {
    if( pString[i]!=NULL ) free((void *) pString[i]);
  }
  free((void *) pString);
}

/* select * from %_content where rowid = [iRow]
 * The caller must delete the returned array and all strings in it.
 * null fields will be NULL in the returned array.
 *
 * TODO: Perhaps we should return pointer/length strings here for consistency
 * with other code which uses pointer/length. */
static int content_select(fulltext_vtab *v, sqlite_int64 iRow,
                          const char ***pValues){
  sqlite3_stmt *s;
  const char **values;
  int i;
  int rc;

  *pValues = NULL;

  rc = sql_get_statement(v, CONTENT_SELECT_STMT, &s);
  if( rc!=SQLITE_OK ) return rc;

  rc = sqlite3_bind_int64(s, 1, iRow);
  if( rc!=SQLITE_OK ) return rc;

  rc = sql_step_statement(v, CONTENT_SELECT_STMT, &s);
  if( rc!=SQLITE_ROW ) return rc;

  values = (const char **) malloc(v->nColumn * sizeof(const char *));
  for(i=0; i<v->nColumn; ++i){
    if( sqlite3_column_type(s, i)==SQLITE_NULL ){
      values[i] = NULL;
    }else{
      values[i] = string_dup((char*)sqlite3_column_text(s, i));
    }
  }

  /* We expect only one row.  We must execute another sqlite3_step()
   * to complete the iteration; otherwise the table will remain locked. */
  rc = sqlite3_step(s);
  if( rc==SQLITE_DONE ){
    *pValues = values;
    return SQLITE_OK;
  }

  freeStringArray(v->nCo