Hi Rafi, If it were mine to do, I would concentrate on getting the data into a table where I can work with it using SQL.
It sounds like your best bet is to write some simple code to read through your CSV, validate its consistency (ignore the dates), and insert it into a table. Then use Dennis's temp table scenario to process the rows as appropriate. Obviously, it doesn't necessarily need to be a temp table as in "create temp table...". It could easily be a non-temp "scratch", or "pending" table that doesn't go away when the connection closes. That way, you can close and come back later to deal with the rows that crossed-over into past-date state. -Clark ----- Original Message ---- From: Rafi Cohen <[EMAIL PROTECTED]> To: sqlite-users@sqlite.org Sent: Wednesday, March 14, 2007 1:21:21 PM Subject: [sqlite] Format lf csv file RE: [sqlite] date/time implementation question Hi Dennis, the first approach is clear now and I may proceed with it. The second approach is interesting and chalenging, but leaves some issues to clarify and in case I find solutions to those issues I well may adopt it. 1. format of csv file: I have no idea how this csv file is created and which database engine is used. I do know that I receive such a file once or twice a day. While reading the file onto the list of structures, I also validate the consistency of the data in the file. Before the data lines (starting with "d,") there is a header line "hdr,". This line contains the sequential number of the file, number of records in the file and a checksum on a specific field of the data (say coumn 2). As I knew nothing up to now about .import, I wonder if there is a way to include those checings in the second approach? 2. The deletion of the future dates is incorrect. On the contrary, in the first approach, I re-examine the remaining structures each half a minute until any of them becomes past date, then I process it just like any other past date structures and then free it. In case a new .csv file arrives, I add the new list of structures to the remaining ones and continue to examine them every half a minute. I could do the same with the sql3_exec statement in the second approach, but I need another approach for the case of the future records. I hope you have satisfying answers for those 2 issues and then I'll be glad to proceed with the second approach. Thanks, Rafi. -----Original Message----- From: Dennis Cote [mailto:[EMAIL PROTECTED] Sent: Wednesday, March 14, 2007 9:15 PM To: sqlite-users@sqlite.org Subject: Re: [sqlite] date/time implementation question Rafi Cohen wrote: > Dennis, I really appreciate your patience and willingness to help. > Unfortunately, this still did not bring me to the expected solution. I > will give you a small algorithm of what I need to do and I'm sure > after this you'll know how to assist me. 1. I read a .csv file into a > linked list of structures. 2. I examine eacch structure one after the > other: Compare the datetime filed of the structure aginst the current > date. If bigger (future), I skup to the next structure. > If smaller or equal, I check if a row with the same id field already > exists in the table: > Select * from tbl where id = id-in-struct. > If no such row exists, I insert a row according to the structure's > field, otherwise, based on another criteria I EITHER UPDATE THE row with > a new value on the second column or delete the row. > 3. After this process, I free the structure from the linked list and > move to the next structure. > Because of the last section, I thought I need to make the date > comparison in C, but I may be wrong here. > However, if the comparison is made thru sqlite, how can I know if indeed > an insert, update or delete was processed so that I can free the > structures? > All the sql statements I use above are, of course, prepared statements > which I execute with wqlite3_step for the fields of each structure. > I hope I'm clear, this time. > Rafi, I think I have followed your description and I agree that you could do this by iterating in C as you have described. If you want to do that and all you need is an expedient way to get a date and time string for the current time for the comparison in step 2 you have a couple of approaches. One simple way is to let sqlite do it for you by executing a single SQL statement "select datetime('now')". This will return a ISO date and time string that you can compare with the concatenated strings that are in your structure. The other is to use the standard C library routines to build an ISO formatted date and time string. These two examples are shown below in a mix of C an pseudo code comments (you have to fill in the blanks). Get current date and time from sqlite: sqlite3_stmt* get_now; sqlite3_prepare(db, "select datetime'now'", -1, &get_now, NULL); char now_datetime[20]; sqlite3_step(get_now); strcpy(now_datetime, sqlite3_column_text(get_now, 0)); // read csv into list of structures //for each structure in the list char rec_datetime[20]; strcpy(rec_datetime, a_struct.date); strcat(rec_datetime, " "); strcat(rec_datetime, a_struct.time); if (strcmp(rec_datetime, now_datetime) <= 0) { //if row with matching id exists in table //if record should be deleted //delete record //else //update record based on structure //else //insert a new record into table } //free the structure Get current date and time from C library: char now_datetime[20]; time_t now = time(NULL); struct tm *now_tm = localtime(&now); sprintf(now_datetime, "%4d-%02d-%02d %02d:%02d:%02d", now_tm->tm_year + 1900, now_tm->tm_mon + 1, now_tm->tm_mday, now_tm->tm_hour, now_tm->tm_min, now_tm->tm_sec); // read csv into list of structures //for each structure in the list char rec_datetime[20]; strcpy(rec_datetime, a_struct.date); strcat(rec_datetime, " "); strcat(rec_datetime, a_struct.time); if (strcmp(rec_datetime, now_datetime) <= 0) { //if row with matching id exists in table //if record should be deleted //delete record //else //update record based on structure //else //insert a new record into table } //free the structure Both of these approaches use the fact that ISO format date and time strings can be compared using a normal string comparison. I think another approach might be worth considering though. Instead of reading your csv data into a list of structures, import it into an sqlite table. You could use the sqlite shell to do this, or you could copy the code from the import routine in the sqlite shell directly into your application (it's free open source code). With your csv records in a temporary table you can do the manipulations in SQL. // read csv into temporary table csv system("sqlite3 mydb \"create temp table csv(...);.import myfile.csv csv\""); // delete the records with future dates (instead of skiping them) delete from csv where (date || ' ' || time) > datetime('now'); // split the csv table into new and existing records create temp table new_csv as select * from csv where id not in (select id from perm_table); delete from csv where id in (select id from new_csv); // update the existing records in the permanent table update perm_table set field2 = (select field2 from csv where csv.id = perm_table.id) where //update condition is true; // delete the records that we updated from the csv delete from csv where //update condition is true; // delete the remaining records from the permanent table since the // update condition must have been false for these records delete from perm_table where perm_table.id in (select id from csv); // insert the new records into the permanent table insert into perm_table select * from new_csv; // drop temp tables drop table csv; drop table new_csv; Note that all the SQL in this script (after the C system call) can be executed by a single call to sqlite3_exec() from your C code. It will execute each SQL statement in the script in turn before returning. When it returns your table will be updated. In this case all the manipulations are done in "parallel" by SQL statements rather than iterating through a loop as you would in C. HTH Dennis Cote ------------------------------------------------------------------------ ----- To unsubscribe, send email to [EMAIL PROTECTED] ------------------------------------------------------------------------ ----- -- No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.446 / Virus Database: 268.18.11/721 - Release Date: 3/13/2007 4:51 PM ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] ----------------------------------------------------------------------------- ----------------------------------------------------------------------------- To unsubscribe, send email to [EMAIL PROTECTED] -----------------------------------------------------------------------------
