Not enough memory to create a database

[bwolfsohn]

do you know the structure ahead of time ?

if so,
use newfile.dbf
append from mycsvfile.csv deliMITED

from the append from help file:

DELIMITED
The option DELIMITED specifies that the source file is an ASCII file with data in each line separated by commas and character fields appearing in double quotes. Fields and records can have variable lengths. ASCII files in the delimited format are managed by the database engine DELDBE.

LOOK AT ROGER'S CSV IMPORT EXAMPLE IN SAMPLES\CSV\CSVIMP.PRG

[Eugene Lutsenko]

Like really no problem. It turns out. Thank you. And could you suggest how to convert this file (in DOS encoding)

[Auge_Ohr]

And could you suggest how to convert this file (in DOS encoding)

i don't think that Xbase++ can convert Cyrillic characters ... same with chinese.
when using OEM DBF Xbase++ will try to convert internal to ANSI when display with GUI XbParts.

you still need right Font / Codepage to display it

if you don't want Xbase++ to convert you must use

Code: Select all

SET CHARSET TO ansi

but than ALL must be ANSI (include your Text Editor) or use

Code: Select all

XPP.EXE /go

[Eugene Lutsenko]

Thank you very much! Well everything turned out! Didn't even think it's that simple

[Eugene Lutsenko]

use newfile.dbf
append from mycsvfile.csv deliMITED

I used this method in one task very successfully. But then decided to standardize and incorporate it into the system. And when I was debugging in different examples, are faced with a problem. Sometimes it works fine and sometimes doesn't work or works incorrectly. Values are separated by delimiter and not entered in different fields and entered all or some in one field. Trying to figure out the reasons. I had the impression that this probably has something to do with different encodings of the source files. He was able to bring all this to the desired result and for a time preserved.

[rdonnay]

Try this:

Use newfile
DC_CSV2WorkArea( 'mycsvfile.csv' )

[Eugene Lutsenko]

Try this:

Use newfile
DC_CSV2WorkArea( 'mycsvfile.csv' )

And could not bring his new text? The fact that the appeal to a CSV file using the NUMTOKEN() and TOKEN() requires a lot of memory. When a CSV file is very large (for example, 4 million records), it just crashes from memory:

Code: Select all

 nHandle := DC_txtOpen( 'test_strings.txt' )
 DO WHILE !DC_TxtEOF( nHandle )                   // Начало цикла по строкам
    mLine = DC_TxtLine( nHandle )                 // Выделить строку из текстового файла
    aStringInp := {}                              // Входная  строка "как есть"
    FOR w=1 TO NUMTOKEN(mLine,",")                // Разделитель между показателями - запятая
        mWord = ALLTRIM(TOKEN(mLine, ",", w))
        AADD(aStringInp, mWord)
    NEXT
    ........
    DC_TxtSkip( nHandle, 1 )
 ENDDO
 DC_TxtClose( nHandle )

[rdonnay]

When a CSV file is very large (for example, 4 million records), it just crashes from memory

That is probably correct.
I do not have an answer for you.

You could possibly break it into multiple files.
You would use FOpen(), FCreate(), FRead() and FWrite() to do this.

[Auge_Ohr]

Code: Select all

 nHandle := DC_txtOpen( 'test_strings.txt' )
        mWord = ALLTRIM(TOKEN(mLine, ",", w))
        AADD(aStringInp, mWord)

why do you AADD() all Words into Array ? this Way you need double space in RAM.
write your Result into DBF after read 1 Line so only 1 Record Length overhead.

[Eugene Lutsenko]

This is just an example, which has a TOKEN() so on...