ABAP CSV Manager

Requirement

☁️ ABAP for Cloud Development edition

Looking for On-Premise edition? Switch to onprem-main branch.

Quick Start

🧑‍💼 I need to read a CSV, how can I do it?

🧙 Create a reader instance and call read method.

DATA(csv_reader) = zcl_tbox_csvman=>get_reader( ).
csv_reader->read(
  EXPORTING i_csv   = csv_data
  IMPORTING e_table = tab_data ).

🧑‍💼 I need to create a CSV, how can I do it?

🧙 Create a writer instance and call write method.

DATA(csv_writer)  = zcl_tbox_csvman=>get_writer( ).
DATA(csv_data)    = csv_reader->write( tab_data ).

* csv_data is a string, if you need the output as a string table:
DATA(csv_data)    = csv_reader->write_to_string_table( tab_data ).

Some remarks about reading a CSV

When you call the read method of a reader instance, csv_data can be:

• a string
• a table of strings
• a table with one-component table line

tab_data must be a flat table: a table whose table line is a structure, whose components are simple elements.

Sometimes you need to read a CSV but you don't know a priori how many and which columns it contains. In these cases you still have two ways to read it:

• you can get a reference variable to a table whose fields (of type string) are named F1, F2, ..., FN and so on;

DATA(csv_ref) = csv_reader->read_with_reference( csv_data ).

FIELD-SYMBOLS <tab> TYPE STANDARD TABLE.
ASSIGN csv_ref->* TO <tab>.

LOOP AT <tab> ASSIGNING FIELD-SYMBOL(<row>).
DATA(row_index) = sy-tabix.
  DO.
    ASSIGN COMPONENT sy-index OF STRUCTURE <row> TO FIELD-SYMBOL(<val>). 
    IF sy-subrc <> 0.
      EXIT.
    ENDIF.
    DATA(info) = |The value at row { row_index } and column { sy-index } is { <val> }| ).
  ENDDO.
ENDLOOP.

* You can always get a statically typed table using dereference method
csv_reader->dereference( EXPORTING i_reference = csv_ref IMPORTING e_table = safe_data ).

• or you can get a transposed version of the CSV, that is, a table with two fields: the row index and a table containing the list of its values, for each column.

DATA(csv_tab_t) = csv_reader->read_with_transposition( csv_data ).

LOOP AT csv_tab_t INTO DATA(csv_row).
  LOOP AT csv_row-row INTO DATA(row_values).
    
    DATA(info) = |The value at row { csv_row-row_index } and column { row_values-column_index } is { row_values-value }| ).
    
  ENDLOOP.
ENDLOOP.

* You can always get a standard table, if you have a suitable variable, using detranspose method
csv_reader->detranspose( EXPORTING i_transposed = csv_tab_t IMPORTING e_table = safe_data ).

General Configuration

🧑‍💼 Can I decide which character to use as delimiter, quotechar, and line terminator? And what about escaping special characters?

🧙 Once instantiated you can configure the csv management object:

• csv_man->delimiter( `;` ). fields delimiter, default is semicolon: ;. Here "delimiter" is a synonym of "separator".
• csv_man->quotechar( `"` ). *to quote fields, default is none.
• csv_man->end_of_line( `|` ). line-terminator char, default is Carriage Return and Line Feed %_CR_LF .
• csv_man->escapechar( `/` ). to escape special characters, both in read and write mode.
• csv_man->doublequote( ). to escape a quotechar character with a quotechar character.
• csv_man->quoting( ztbox_cl_csvman=>c_quote_minimal ). to restrict quoting application, with these options:
  • zcl_tbox_csv_writer=>c_quote_all to apply quotechar character to all fields (this is default behaviour if a quotechar is set);
  • zcl_tbox_csv_writer=>c_quote_minimal to apply quotechar character only to fields containing special characters;
  • zcl_tbox_csv_writer=>c_quote_nonnumeric to apply quotechar character only to non-numeric fields;
  • zcl_tbox_csv_writer=>c_quote_none to never quote fields (this is the default behaviour if no quotechar is set).
• csv_man->header( ). to write/expect an header line in write/read mode. Default is abap_true .
• csv_man->header_desc( ). to use field name as header text field. You can column text by calling ->label( ) method on fields level, see below for details.

Output Format

🧑‍💼 Nice, but I want also control fields output format, especially for date/time/numeric fields.

🧙 Sure, you can use these configurations:

• csv_man->date_format( `yyyy/dd/mm` ). to decide output format for date fields in write mode, or to declare expected format for date fields in read mode. You can use any format containing "dd", "mm", "yy", "yyyy" and an optional separator. Default is dd.mm.yyyy.
• csv_man->time_format( `hh:mm:ss` ). same as above, but for time fields. Default is hh:mm:ss.
• csv_man->number_format( ). three options here:
  • (blank) to output numbers like 1.234.567,89
  • X to output numbers like 1,234,567.89
  • Y to output numbers like 1 234 567,89
• csv_man->country( `US` ). to output date, time and numbers according to a country rules (less specific than previous methods).
• csv_man->decimals( 3 ). to write numerical fields with the specified decimals precision.
• csv_man->condense_values( ). to remove leading and trailing spaces. Default is abap_false.
• csv_man->keep_init( ). to maintain initial values: if set to abap_false a numerical field containing only 0, as well as an initial date or initial time, became blank in write mode. Default is abap_true.
• csv_man->alignment( cl_abap_format=>a_right ). to align fields content according to the following options:
  • cl_abap_format=>a_left to justify text on the left (default option);
  • cl_abap_format=>a_right to justify text on the right.

Fields Properties

🧑‍💼 Ok cool. These are global configuration, valid for all the fields, aren't they? What if I want to set some format property to one field and a different property to another?

🧙 You can restrict the application of the previous methods to a single field by calling field( ) method first, also chaining other methods, e.g.:

csv_man->field( `AMOUNT_USD` )->number_format( `X` ).
csv_man->field( `AMOUNT_EUR` )->country( `IT` )->decimals( 2 ).

You can also exclude some fields from the CSV generation/reading process using exclude( ) method:

csv_man->field( `MANDT` )->exclude( ).

Viceversa, if you work with a table having too many fields, you can generate or reading a CSV considering only a small subset of fields using include( ) method. Once you have called include for a field, only fields for which include has been called will be considered.

csv_man->field( `MATNR` )->include( ).
csv_man->field( `WERKS` )->include( ).

If the order of the fields in the table does not match the columns in the CSV to generate or read, you can map each field with the corresponding csv-column position:

csv_man->header_desc( abap_true ).
csv_man->field( `MATNR` )->position( 2 )->label( `Material!!!` ).
csv_man->field( `WERKS` )->position( 1 )->label( `THE Plant` ).

In this way, the following table:

MATNR	WERKS
AAAA01	US01
BBBB02	US02

gives this CSV as output:

THE Plant,Material!!!
US01,AAAA01
US02,BBBB02

label( ) method set a custom header text when header_desc( ) is used.

Validations

🧑‍💼 Regarding reading the csv, is there an automatism to help identify errors in the data?

🧙 We have it. When a CSV is read, some validation checks are performed according to the data type of the target fields: date fields (if not blank) must be a valid and plausible date; time fields (if not blank) must contain a valid and plausible time, numerical fields (if not blank) must contain a valid number. Whenever any of these check fail, the contents are not transferred. You get a detailed report for the validation fails by calling get_validations_fails( ).

Suppose you read this CSV:

31/02/2023,10:45:19,"1900,20"
28/02/2023,25:00:00,"-894,23"
31/12/2022,00:00:01,"12A4,43"

into a table with structure DATE [D(8)] | TIME [T(6)] | AMOUNT [P(7) DEC(2)]. The output will be

DATE	TIME	AMOUNT
00000000	104519	1900.20
20230228	000000	-894.00
20221231	000001	0.00

And the get_validation_fails( ) output is this table:

ROW	COL	TABLE_FIELD	CSV_VALUE	SAP_VALUE	CHECK_FAILED
1	1	DATE	31/02/2023	20230231	DATE_PLAUSIBILITY
2	2	TIME	25:00:00	250000	TIME_PLAUSIBILITY
3	3	AMOUNT	12A4,43	12A4.43	VALID_NUMB

You can add also custom validation checks: it must be an instance method with the following signature

METHODS sample_check IMPORTING i_value TYPE string RETURNING VALUE(r_fail) TYPE flag.

If, e.g., an object sample_object implements method sample_check, you can add this check to a field:

csv_man->field( `FIELD_NAME` )->add_post_validation(
  check_object  = sample_object
  check_method  = `SAMPLE_CHECK` ).

in two ways: as a pre validation by calling method add_pre_validation( ), and as a post validation by calling method add_post_validation( ).

• Pre-validation checks are applied to raw data: parameter value is a string containing exactly the content of the field in the CSV;
• Post-validation checks are applied to transformed data: parameter value is a string containing the field value interally formatted for the SAP data type field target, according to the format options configured.

fail = abap_true means the check has not been passed.

Installation

Install this project using abapGit