Soo, I'm having a few places in my program where I just fetch a single row (sometimes even a single column in a single row), for example
SELECT id,title,posted FROM posts WHERE id=4;
Since this will only return a single row (unless I suck at database stuff and somehow manage to have duplicated IDs) I feel it's overkill to go the whole sqlite3_prepare_v2(...)
, sqlite3_bind_int(...)
, sqlite3_step(...)
, sqlite3_finalize(...)
way just to fetch the single row.
I know this can be done with sqlite3_exec(...)
and using a callback function to process the results that also return 1;
's to abort further executions, but this doesn't work in my case because my query looks like this:
SELECT id,title,posted FROM posts WHERE id=?;
And you can't have parameterized queries with sqlite3_exec(...)
. Inserting the ID via some kind of printf(...)
isn't safe here either.
So, this is what I'm doing right now
char *sql;
sqlite3_stmt *stmt;
sql = sqlite3_mprintf(
"SELECT \
title, \
((julianday(posted) 开发者_开发百科- 2440587.5) * 86400), \
text \
FROM \
%s \
WHERE id=?;",
POSTS_TABLE); /* printf safe here, POSTS_TABLE is a #define */
if (sqlite3_prepare_v2(db, sql, strlen(sql), &stmt, NULL) != SQLITE_OK)
show_sqlite_error(db);
else
{
sqlite3_bind_text(stmt, 1, id, strlen(id), SQLITE_TRANSIENT);
sqlite_code = sqlite3_step(stmt); /* This right here seems just ugly */
if (sqlite_code != SQLITE_ROW) /* especially this */
printf("<div id=\"error\">ERROR: No such ID in database</div>\n");
else
{
int i;
char time_string[25];
const unsigned char *title = sqlite3_column_text(stmt, 0);
time_t time = sqlite3_column_int(stmt, 1);
const unsigned char *text = sqlite3_column_text(stmt, 2);
strftime(time_string, 25, DATE_FORMAT, gmtime(&time));
printf("<h2 id=\"post_headline\">%s</h2>\n\n", title);
printf("<h3 id=\"post_time\">%s</h3>\n", time_string);
printf("<p id=\"post_body\">\n");
for (i = 0; i < strlen((char *)text); i++)
if (text[i] == '\n')
printf("<br />");
else
printf("%c", text[i]);
}
}
My issue with this code is, that I call sqlite3_step()
once, error out if everything but SQLITE_ROW
was returned, display whatever needs to be displays if SQLITE_ROW
was returned, and call finalize without actually stepping to SQLITE_DONE
. While this (as I was told) wouldn't do any harm with a SELECT
statement, it feels ugly and overkill.
Another thing is queries which simply return a single value.
SELECT MAX(id) FROM posts;
In my case, this simply returns "5". No more, no less. Especially here it's really overkill to go the whole prepare, bind, step, finalize way, but I HAVE to do it that way, because my query again looks like
SELECT MIN(id) FROM posts WHERE id > ?;
to get the next ID after the current one for example. It just returns "5" when the current ID is "4".
To sum this up: is there a quick way to grab single row results from a SELECT
statement without explicitly looping until SQLITE_DONE
is received from sqlite3_step()
AND how do I grab single row, single column value (skalar basically) without going the whole way again, including a call to sqlite_column_...
?
I'm an advocate of preventing SQL injection, but I think it should be safe to do this:
sql = sqlite3_mprintf("SELECT ... WHERE id=%d;", atoi(id));
This interprets the id
string as an integer via atoi()
, so there's no chance of SQL injection.
As for fetching a single row or a single scalar, I'd write a wrapper function that executes the SQL and returns the contents simply. You may have to use a global to transfer the contents from the callback function to the wrapper function.
精彩评论