int main(int argc, char **argv){
char c, msg[1000];
int colnames = 'y',
rownames = 0,
tab_exists_check = 0;
char **field_names = NULL;
apop_data *field_name_data, *field_name_data_t;
sprintf(msg, "%s [-d delimiters] text_file table_name dbname\n"
"e.g.: %s -d\",|\" infile.txt a_table info.db\n"
"If the input text file name is a single dash, -, then read from STDIN.\n"
"Input must be plain ASCII or UTF-8.\n"
"-d\t\tThe single-character delimiters to use, e.g., -d \" ,\" or -d \"\\t\" (which you \n"
"\t\t\twill almost certainly have to write as -d \"\\\\t\"). Default: \"| ,\\t\", meaning \n"
"\t\t\tthat any of a pipe, space, comma, or tab will delimit separate entries\n"
"-nc\t\tData does not include column names\n"
"-n regex\t\tCase-insensitive regular expression indicating Null values. Default: NaN \n"
"-m\t\tUse a mysql database (default: SQLite)\n"
"-f\t\tfixed width field ends: -f\"3,8,12,17\" (first char is one, not zero)\n"
"-u\t\tmysql username\n"
"-p\t\tmysql password\n"
"-r\t\tData includes row names\n"
"-v\t\tVerbose\n"
"-N\t\tA comma-separated list of column names: -N\"apple,banana,carrot,durian\"\n"
"-O\t\tIf table exists, erase it and write from scratch (i.e., Overwrite)\n"
"-h\t\tPrint this help\n\n"
, argv[0], argv[0]);
int * field_list = NULL;
if(argc<3){
printf("%s", msg);
return 0;
}
while ((c = getopt (argc, argv, "n:d:f:hmp:ru:vN:O")) != -1)
if (c=='n') {
if (optarg[0]=='c')
colnames='n';
else
strcpy(apop_opts.db_nan, optarg);
}
else if (c=='N') {
apop_regex(optarg, " *([^,]*[^ ]) *(,|$) *", &field_name_data);
field_name_data_t = apop_data_transpose(field_name_data);
field_names = field_name_data_t->text[0];
}
else if (c=='d') strcpy(apop_opts.input_delimiters, optarg);
else if (c=='f') field_list = break_down(optarg);
else if (c=='h') printf("%s", msg);
else if (c=='m') apop_opts.db_engine = 'm';
else if (c=='u') strcpy(apop_opts.db_user, optarg);
else if (c=='p') strcpy(apop_opts.db_pass, optarg);
else if (c=='r') rownames++;
else if (c=='v') apop_opts.verbose=2;
else if (c=='O') tab_exists_check++;
apop_db_open(argv[optind + 2]);
if (tab_exists_check) apop_table_exists(argv[optind+1],1);
apop_query("begin;");
apop_text_to_db(argv[optind], argv[optind+1], rownames, colnames, field_names, .field_ends=field_list);
apop_query("commit;");
}
void check_out_impute(char **origin, char **destin, int *imputation_number, char **subset, char **filltabin){
char *filltab = (filltabin && *filltabin) ? *filltabin : "filled";
Tea_stopif(!origin || !*origin, return, 0, "NULL origin table, but I need that.");
char *id_column= get_key_word(NULL, "id");
const char *dest = destin ? *destin : NULL;
int use_rowids = 0;
if (!id_column) {
use_rowids++;
id_column = strdup("rowid");
}
sprintf(apop_opts.db_name_column, "%s", id_column);
begin_transaction();
if (dest && strcmp(*origin, dest)){
apop_table_exists(dest, 'd');
apop_query("create table %s as select %s * from %s %s %s",
dest,
use_rowids ? "rowid as id_col, " : " ", *origin,
(subset && *subset) ? "where" : " ",
(subset && *subset) ? *subset : " "
);
} else dest = *origin;
create_index(dest, use_rowids ? "id_col" : id_column);
Tea_stopif(!apop_table_exists(filltab), return , 0, "No table named '%s'; did you already doMImpute()?", filltab);
apop_data *fills = apop_query_to_text("select %s, field, value from %s where (draw=%i or draw = -1)"
, id_column, filltab, *imputation_number);
Tea_stopif(!fills || fills->error, return, 0, "Expected fill-in table "
"%s, but couldn't query it.", filltab);
for(int i=0; i< *fills->textsize; i++){
_Bool is_null = !strcmp(fills->text[i][1], apop_opts.nan_string);
char tick = is_null ? ' ' : '\'';
apop_query("update %s set %s = %c%s%c "
"where cast(%s as numeric) = %s",
dest, fills->text[i][0],
tick, is_null ? "NULL" : fills->text[i][1], tick,
id_column, fills->names->row[i]);
}
commit_transaction();
apop_data_free(fills);
free(id_column);
}
//these work by checking that K-L divergence shrunk, and that individual margins are correct.
void test_raking_further(){
apop_table_exists("rake_test", 'd');
apop_query("create table rake_test (first, second, weights);"
"insert into rake_test values(1, 1, 10);"
"insert into rake_test values(1, 2, 2);"
"insert into rake_test values(2, 1, 15);"
"insert into rake_test values(2, 2, 5);"
);
//Synthetic data, starting at all ones.
apop_data_print(
apop_rake(.margin_table="rake_test", .count_col="weights",
.contrasts=(char*[]){"first", "second"}, .contrast_ct=2),
void test_check_out_impute(){
apop_table_exists("testcheckoutbase", 'd');
apop_table_exists("testcheckoutfill", 'd');
apop_table_exists("testcheckoutbase_copy", 'd');
apop_table_exists("tcb", 'd');
apop_query("create table testcheckoutbase (id, a, b); "
" insert into testcheckoutbase values(0, 3, 0./0.);"
" insert into testcheckoutbase values(1, 0./0., 3);"
" insert into testcheckoutbase values(2, 3, 3);"
"create table testcheckoutfill (draw, value, id, field); "
" insert into testcheckoutfill values(0, 3, 1, 'a');"
" insert into testcheckoutfill values(0, 3, 0, 'b');"
" insert into testcheckoutfill values(1, 9, 1, 'a');"
" insert into testcheckoutfill values(1, 6, 0, 'b');"
);
char *strings[] = {"testcheckoutbase", //0
"testcheckoutfill", //1
"testcheckoutbase_copy", //2
"tcb" //3
};
set_key_text("id", NULL, "id");
check_out_impute( strings+0, strings+2, (int[]){0}, NULL, strings+1);
void check_out_impute(char **origin, char **destin, int *imputation_number, char **subset, char **filltabin){
char *filltab = (filltabin && *filltabin) ? *filltabin : "filled";
Apop_stopif(!origin || !*origin, return, 0, "NULL origin table, but I need that.");
char *id_column= get_key_word(NULL, "id");
const char *dest = destin ? *destin : NULL;
int use_rowids = 0;
if (!id_column) {
use_rowids++;
id_column = strdup("rowid");
}
sprintf(apop_opts.db_name_column, "%s", id_column);
if (dest){
apop_table_exists(dest, 'd');
apop_query("create table %s as select %s * from %s %s %s",
dest,
use_rowids ? "rowid as id_col, " : " ", *origin,
(subset && *subset) ? "where" : " ",
(subset && *subset) ? *subset : " "
);
} else dest = *origin;
has_sqlite3_index(dest, use_rowids ? "id_col" : id_column, 'y');
Apop_stopif(!apop_table_exists(filltab), return , 0, "No table named '%s'; did you already doMImpute()?", filltab);
apop_data *fills = apop_query_to_text("select %s, field, value from %s where draw+0.0=%i"
, id_column, filltab, *imputation_number);
Apop_stopif(!fills || fills->error, return, 0, "Expected fill-in table "
"%s, but couldn't query it.", filltab);
begin_transaction();
if (fills)
for(int i=0; i< *fills->textsize; i++)
apop_query("update %s set %s = '%s' "
"where %s = %s",
dest, fills->text[i][0], fills->text[i][1],
id_column, fills->names->row[i]);
commit_transaction();
apop_data_free(fills);
free(id_column);
}
/** Print an \ref apop_data set to a file, the database, or the screen,
as determined by the \c .output_type.
\li See \ref apop_prep_output for more on how printing settings are set.
\li See \ref Legi for more details and examples.
\li See \ref sqlsec for notes on writing an \ref apop_data set to the database.
\li This function uses the \ref designated syntax for inputs.
\ingroup all_public
*/
APOP_VAR_HEAD void apop_data_print(const apop_data *data, Output_declares){
const apop_data * apop_varad_var(data, NULL);
Dispatch_output
APOP_VAR_ENDHEAD
if (output_type == 'd'){
if (output_append == 'w') apop_table_exists(output_name, 'd');
apop_data_to_db(data, output_name, output_append);
return;
}
apop_data_print_core(data, output_pipe, output_type);
if (data && data->more) {
output_append='a';
apop_data_print(data->more, Output_vars);
}
if (output_name)
fclose(output_pipe);
}
int check_levenshtein_distances(int max_lev_distance){
int typo_counter=0;
int min_distance;
char *closest;
if (!apop_table_exists("keys")) return 0;
apop_data *userkeys = apop_query_to_text("select key from keys");
for (int i=0; i < *userkeys->textsize; i++){
min_distance = 100;
for (char **keyptr=ok_keys; strlen(*keyptr); keyptr++){
int ld = levenshtein_distance(*keyptr, *userkeys->text[i]);
if (ld < min_distance){
if(ld == 0) {min_distance=0; break;}
min_distance=ld;
closest = *keyptr;
}
}
Apop_stopif(min_distance > 0 && min_distance <= max_lev_distance, typo_counter++ , 0,
"You wrote %s for one of the keys in your spec file. Did you "
"mean to write %s?", *userkeys->text[i], closest);
}
return typo_counter;
}
int main(int argc, char **argv){
char c, msg[1000];
int colnames = 1,
rownames = 0,
tab_exists_check = 0;
sprintf(msg, "%s [-d delimiters] text_file table_name dbname\n"
"e.g.: %s -d\",|\" infile.txt a_table info.db\n"
"If the input text file name is a single dash, -, then read from STDIN.\n"
"Input must be plain ASCII or UTF-8.\n"
"-d\t\tThe single-character delimiters to use, e.g., -d \" ,\" or -d \"\\t\" (which you \n"
"\t\t\twill almost certainly have to write as -d \"\\\\t\"). Default: \"| ,\\t\", meaning \n"
"\t\t\tthat any of a pipe, space, comma, or tab will delimit separate entries\n"
"-nc\t\tData does not include column names\n"
"-n regex\t\tCase-insensitive regular expression indicating Null values. Default: NaN \n"
"-m\t\tUse a mysql database (default: SQLite)\n"
"-f\t\tfixed width field ends: -f\"3,8,12,17\" (first char is one, not zero)\n"
"-u\t\tmysql username\n"
"-p\t\tmysql password\n"
"-r\t\tData includes row names\n"
"-v\t\tVerbose\n"
"-O\t\tIf table exists, erase it and write from scratch (i.e., Overwrite)\n"
"-h\t\tPrint this help\n\n"
, argv[0], argv[0]);
int * field_list = NULL;
if(argc<3){
printf("%s", msg);
return 0;
}
while ((c = getopt (argc, argv, "n:d:f:hmp:ru:vO")) != -1){
switch (c){
case 'n':
if (optarg[0]=='c')
colnames --;
else
strcpy(apop_opts.db_nan, optarg);
break;
case 'd':
strcpy(apop_opts.input_delimiters, optarg);
break;
case 'f':
field_list = break_down(optarg);
break;
case 'h':
printf("%s", msg);
return 0;
case 'm':
apop_opts.db_engine = 'm';
break;
case 'u':
strcpy(apop_opts.db_user, optarg);
break;
case 'p':
strcpy(apop_opts.db_pass, optarg);
break;
case 'r':
rownames ++;
break;
case 'v':
apop_opts.verbose ++;
break;
case 'O':
tab_exists_check ++;
break;
}
}
apop_db_open(argv[optind + 2]);
if (tab_exists_check) apop_table_exists(argv[optind+1],1);
apop_query("begin;");
apop_text_to_db(argv[optind], argv[optind+1], rownames,colnames, NULL, .field_ends=field_list);
apop_query("commit;");
}