inline static grn_cell *
get_op(grn_query *q, grn_operator op, int weight)
{
char *start, *end = q->cur;
int mode, option;
switch (*end) {
case 'S' :
mode = GRN_OP_SIMILAR;
start = ++end;
option = grn_atoi(start, q->str_end, (const char **)&end);
if (start == end) { option = DEFAULT_SIMILARITY_THRESHOLD; }
q->cur = end;
break;
case 'N' :
mode = GRN_OP_NEAR;
start = ++end;
option = grn_atoi(start, q->str_end, (const char **)&end);
if (start == end) { option = DEFAULT_MAX_INTERVAL; }
q->cur = end;
break;
case 'n' :
mode = GRN_OP_NEAR2;
start = ++end;
option = grn_atoi(start, q->str_end, (const char **)&end);
if (start == end) { option = DEFAULT_MAX_INTERVAL; }
q->cur = end;
break;
case 'T' :
mode = GRN_OP_TERM_EXTRACT;
start = ++end;
option = grn_atoi(start, q->str_end, (const char **)&end);
if (start == end) { option = DEFAULT_TERM_EXTRACT_POLICY; }
q->cur = end;
break;
case 'X' : /* force exact mode */
op = GRN_OP_AND;
mode = GRN_OP_EXACT;
option = 0;
start = ++end;
q->cur = end;
break;
default :
return NIL;
}
return op_new(q, op, weight, mode, option);
}
inline static grn_cell *
get_token(grn_ctx *ctx, grn_query *q)
{
grn_cell *token = NIL;
grn_operator op = q->default_op;
{
int weight = DEFAULT_WEIGHT, prefixp = 0, mode = -1, option = 0;
skip_space(ctx, q);
if (q->cur_expr >= q->max_exprs ||
q->cur_cell >= q->max_cells ||
q->cur >= q->str_end) { return NIL; }
switch (*q->cur) {
case '\0' :
return NIL;
case GRN_QUERY_PARENR :
q->cur++;
return NIL;
case GRN_QUERY_QUOTEL :
q->cur++;
if ((token = get_phrase(ctx, q)) == NULL) {
return NIL;
}
break;
case GRN_QUERY_PREFIX :
q->cur++;
token = get_op(q, op, weight);
break;
case GRN_QUERY_AND :
q->cur++;
token = op_new(q, GRN_OP_AND, weight, mode, option);
break;
case GRN_QUERY_BUT :
q->cur++;
token = op_new(q, GRN_OP_BUT, weight, mode, option);
break;
case GRN_QUERY_ADJ_INC :
q->cur++;
if (weight < 127) { weight++; }
token = op_new(q, GRN_OP_ADJUST, weight, mode, option);
break;
case GRN_QUERY_ADJ_DEC :
q->cur++;
if (weight > -128) { weight--; }
token = op_new(q, GRN_OP_ADJUST, weight, mode, option);
break;
case GRN_QUERY_ADJ_NEG :
q->cur++;
token = op_new(q, GRN_OP_ADJUST, -1, mode, option);
break;
case GRN_QUERY_PARENL :
q->cur++;
token = get_expr(ctx, q);
break;
default :
if ((token = get_word(ctx, q, &prefixp)) &&
token->u.b.value[0] == 'O' &&
token->u.b.value[1] == 'R' &&
token->u.b.size == 2) {
cell_del(q);
q->cur_expr--;
token = op_new(q, GRN_OP_OR, weight, mode, option);
}
break;
}
}
return cons(q, token, NIL);
}
int parse_json(uint16_t json_sz) {
int ret = SUCCESS;
char *cursor = json;
char *op = NULL;
char *line_start = NULL, *line_end = NULL, *delim = NULL;
uint16_t remaining = json_sz;
uint16_t line_remaining = 0;
uint16_t consumed = 0;
// There's nothing to do; just return.
if (0 == json_sz) {
#ifdef DEBUG
fprintf(stderr, "[D] parse_json | json_sz = 0x%04x; nothing to do\n", json_sz);
#endif
goto bail;
}
// For each INSN (1 per line)...
while (NULL != (line_start = strnchr(cursor, '\n', (uint32_t)remaining))) {
#ifdef DEBUG
fprintf(stderr, "[D] parse_json | top of loop; line_start = '%s'\n", line_start);
#endif
// Consume the newline.
line_start += 1;
remaining -= 1;
// If there's nothing left, we've parse it all.
if (0 == remaining) {
#ifdef DEBUG
fprintf(stderr, "[D] parse_json | consumed all JSON\n", remaining);
#endif
goto bail;
}
// Find next newline (so we have bounds to this line).
if (NULL == (line_end = strnchr(line_start, '\n', (uint32_t)remaining))) {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | could not find next line ending; line_start is: '%s'\n", line_start);
#endif
ret = ERRNO_MALFORMED_JSON;
goto bail;
}
line_remaining = line_end - line_start;
cursor = line_start;
#ifdef DEBUG
fprintf(stderr, "[D] parse_json | cursor = line_start = '%s'\n", cursor);
#endif
// Decrement from global remaining now, since we'll be consuming this line.
remaining -= line_remaining;
////
// Get OP
////
if (NULL == (delim = strnchr(cursor, ' ', (uint32_t)(line_remaining)))) {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | cannot find OP delim\n");
#endif
ret = ERRNO_MALFORMED_JSON;
goto bail;
}
consumed = delim - cursor;
line_remaining -= consumed;
if (SZ_INSN_OP != consumed) {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | invalid OP length: %d\n", consumed);
#endif
ret = ERRNO_MALFORMED_JSON;
goto bail;
}
////
// Dispatch OP
////
op = cursor;
cursor += SZ_INSN_OP+1;
if (0 == memcmp("NEW", op, SZ_INSN_OP)) {
if (SUCCESS != (ret = op_new(&cursor, &line_remaining))) {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | NEW | non-SUCCESS from op_new()\n");
#endif
goto bail;
}
} else if (0 == memcmp("SET", op, SZ_INSN_OP)) {
if (SUCCESS != (ret = op_set(&cursor, &line_remaining))) {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | NEW | non-SUCCESS from op_set()\n");
#endif
goto bail;
}
} else if (0 == memcmp("DEL", op, SZ_INSN_OP)) {
if (SUCCESS != (ret = op_del(&cursor, &line_remaining))) {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | NEW | non-SUCCESS from op_del()\n");
#endif
goto bail;
}
} else {
#ifdef DEBUG
fprintf(stderr, "[E] parse_json | unknown OP: '%.3s'\n", cursor);
#endif
ret = ERRNO_MALFORMED_JSON;
goto bail;
}
}
bail:
return ret;
}
