Suppose we want to read in data of the form

i, j, value(i,j)

from a file, setting up a dynamic array A(range, range) with just the A(i,j)= value(i,j) for the pairs (i,j) which exist in the file. Here is an example which shows different ways of doing this. We read data from

1. differently formatted text files (example trio.mos described in the 'Mosel User Guide', Section 3.4 Reading sparse data): using 'initializations from', 'readln', or 'diskdata',
2. spreadsheets or databases (examples duo.mos, duosheet.mos, duooci.mos, duoexc.mos, and duodd.mos described in the Xpress Whitepaper 'Using ODBC and other database interfaces with Mosel') ,
3. the application (C/Java) executing the Mosel model (examples datainputc.mos and datainputj.mos

With initializations from data from spreadsheets or databases is read in a similar way as text files are accessed; switching between data sources can be done by replacing the I/O driver name and the data source in the extended filename, such as "mmsheet.excel:data.xls" to access the spreadsheet data.xls or "mmodbc.odbc:data.mdb" to read from the MS Access database data.mdb and the string "mmodbc.odbc:DSN=mysql;DB=data" could be used to access a mysql database named data.

Instead of using an initializations block, we may read data from a database into a Mosel model by writing out the corresponding SQL statements (this format allows fine-tuning of the queries and gives access to a larger range of commands, in particular when working with databases). When developing applications with ODBC/SQL statements, the user is advised to use the following two parameter settings to obtain error messages and debugging information from ODBC:

 setparam("SQLverbose", true)
 setparam("SQLdebug", true)

The module mmoci (requires a separate license) defines a software-specific interface to Oracle databases that is used in a similar way as the ODBC connection.
Using initializations from the extended filename will take a form like "mmoci.oci:myname/mypassword@dbname" and it is also possible to work with PL/SQL statements like the following:

 OCIlogon('myname/mypassword@dbname')
 OCIexecute("select Index_i,Index_j,Value from MyDataTable", A5)
 OCIlogoff

Using initializations from it is also possible to read data into a Mosel model from the calling application (C/C++, C#, Java) using a combination of the I/O drivers mem (data held in memory) and raw driver (data in binary format) in C, drivers dotnet and dotnetraw in C# programs, and drivers java and jraw in Java programs.

Other possibilities of inputting data into a Mosel model include

• Static modules: data exchange between the calling application and a model via custom functions, using the Mosel Native Interface - see the example of embedding the cutting stock problem in the Whitepaper 'Embedding Optimization Algorithms'.
• Custom format/custom data sources: the Mosel Native Interface allows the user to implement his own I/O drivers (to be used with initializations blocks and all other Mosel functionality involving file access, such as matrix export or model compilation). Possible uses of such user-written drivers include encoding/decoding of data on the fly or compression/decompression of data. See the Xpress Whitepaper 'Generalized file handling in Mosel' for the complete code and documentation of an example implementing a compression driver with the zip library.
  Furthermore, the Mosel IO driver pipe may be used to recover the output of an external program. See the section on I/O drivers in the 'Mosel Libraries Reference Manual' for more information.

This example shows different ways of writing out data from a Mosel model, namely to

1. differently formatted text files (example trio_out.mos described in the 'Mosel User Guide', Section 10.2 File output): using 'initializations to', free-format output with 'write'/'writeln', or 'diskdata' format,
2. spreadsheets or databases (examples duo_out.mos, duosheet_out.mos, duooci_out.mos, duoexc_out.mos, and duodd_out.mos described in the Xpress Whitepaper 'Using ODBC and other database interfaces with Mosel'),
3. the application (C/Java) executing the Mosel model (examples dataoutputc.mos and dataoutputj.mos

Using initializations to data is written to spreadsheets or databases through the ODBC connector in a similar way as text files are accessed; all necessary SQL commands are generated automatically. With the extended filename to "mmodbc.odbc:data.mdb" the data will be written to the MS Access database data.mdb and the string "mmodbc.odbc:DSN=mysql;DB=data" could be used to access a mysql database named data.

Instead of using an initializations block, we may write out data from a Mosel model directly with the corresponding SQL statements (this makes it possible, for example, to clear the data table before writing to it - this functionality is not available with spreadsheets)

 SQLconnect("data.sqlite")
 SQLexecute("delete from MyOutTable")
 SQLexecute(
  "insert into MyOutTable (Index1,Index2,AValue) values (?,?,?)", A)
 SQLdisconnect

When working with Excel spreadsheets it is recommended to use one of the dedicated spreadsheet drivers from the mmsheet module: with the excel driver, the spreadsheet may remain open while writing to it from a Mosel model, in this case the data written out to the spreadsheet does not get saved, making it possible to re-run the application with different data sets without causing any lasting modifications to your spreadsheet. All other spreadsheet drivers require the spreadsheet file to be closed. The xls/xlsx drivers do not require an installation of Excel and are available for certain non-Windows platforms.

The module mmoci (requires a separate license) defines a software-specific interface to Oracle databases that is used in a similar way as the ODBC connection provided through mmmodbc. With initializations to an extended filename will take a form like "mmoci.oci:debug;myname/mypassword@dbname" and it is equally possible to use PL/SQL statements directly.

Using initializations to it is also possible to send data from a Mosel model in memory to the calling application (C/C++, C# or Java). This is done using a combination of the I/O drivers mem (data held in memory) and raw driver (data in binary format) in C, drivers dotnet and dotnetraw in C# programs and drivers java and jraw in Java programs.

Other possibilities of outputting data from a Mosel model include

• Use of the Mosel Libraries: information about the model objects is retrieved after the execution of the model is terminated. See the 'Mosel User Guide', Part III Working with the Mosel Libraries and the example dataafterexec.c.
• Redirection of the output and error streams: to a file or to the calling application with the help of a callback function (I/O driver cb).
• Static modules: data exchange between the calling application and a model via custom functions, using the Mosel Native Interface. See the example datainout.c.
• Custom formats/custom data sources: the Mosel Native Interface allows the user to implement his own I/O drivers (to be used with initializations blocks and all other Mosel functionality involving file access, such as matrix export or model compilation). Possible uses of such user-written drivers include encoding/decoding of data on the fly or compression/decompression of data. See the Xpress Whitepaper 'Generalized file handling in Mosel' for the complete code and documentation of an example implementing a compression driver with the zip library.
  Furthermore, the Mosel I/O driver pipe may be used to channel the output of a Mosel model to an external program. See the section on I/O drivers in the 'Mosel Libraries Reference Manual' for more information.

This example demonstrates dataframe-style representation of data. This functionality is provided for CSV format files via the 'diskdata' I/O driver and subroutine of module mmetc, for the spreadsheet I/O drivers 'xls', 'xlsx', and 'csv' of module mmsheet, and for databases accessed through an ODBC connection via SQL commands.

• dataframecsv.mos: reading and writing dataframes for CSV format data via 'diskdata'
• dataframedb.mos: dataframe-style reading of database tables via 'SQLdataframe'
• dataframesht.mos: reading and writing dataframes for spreadsheets and CSV format data via mmsheet

The example 'indexeg' shows how to read data tables with different formats from various data sources:

• a dense 3x2 table
• a dense 2x3 table
• a sparse 2x3 table
• a 2x4 table with partially indexed format

A data may be accessed from Mosel through an ODBC connection (using the odbc driver, indexeg.mos, or with SQL statements, indexeg2.mos), with the software-specific driver excel (indexeg3.mos), or with the portable mmsheet module that can be used to read and write xls, xlsx (indexeg5.mos) and csv (indexeg6.mos) files.
The first two methods (odbc driver and SQL statements) apply to spreadsheets and databases. In addition, for Oracle databases we show how to use the software-specific connection provided by the module mmoci (indexeg4.mos).

This example shows how to read single-dimensional data tables from a spreadsheet or database file that only contains the data values. The data arrays must have an index set of type range that gets populated automatically based on the row count, either starting with the default value 1 or with a specified start value. Several arrays can be populated via a single statement.

• Spreadsheets and databases can be accessed from Mosel through an ODBC connection using the odbc driver (soleg.mos) or with SQL statements (soleg2.mos).
• If Microsoft Excel is installed, the software-specific driver excel (soleg3.mos) can be used.
• For Oracle databases a software-specific connection is provided by the module mmoci (soleg4.mos).
• For generic spreadsheets it is recommended to use the portable mmsheet module that provides subroutines to read and write xls and xlsx (soleg5.mos) and csv (soleg6.mos) files.

1. using the odbc driver,
2. formulated with SQL statements,
3. using the excel driver,
4. using the oci driver,
5. using the xls driver, and
6. using the csv driver.

• Reading several arrays from a single database table/spreadsheet range (multicol.mos)
• Outputting several arrays into a single database table/spreadsheet range (multiout.mos)
• Reading an array from several tables/ranges (multitab.mos)

1. Selection statements using columns in inverted order (ODBC: odbcinv.mos, Oracle: ociinv.mos)
2. SQL with conditional selection statements and functions (ODBC: odbcselfunc.mos, Oracle: ociselfunc.mos)
3. Retrieving information about database table structure: list of database tables, list of field names, primary keys (odbcinspectdb.mos)
4. Implementing 'stop on error' behaviour for SQL statements (odbcchkstatus.mos)

• reading in records (recordin*.mos)
• writing out records (recordout*.mos)
• reading and writing lists (listinout*.mos)

• The model version in unioninout1.mos shows how to work with Mosel's text data file format.
• Databases can be accessed from Mosel through an ODBC connection using the odbc driver (unioninout.mos) or with SQL statements (unioninout2.mos).
• For Oracle databases a software-specific connection is provided by the module mmoci (unioninout4.mos).
• Microsoft Excel spreadsheets can be accessed via the software-specific driver excel (unioninout3.mos). Alternatively, models can use the dedicated I/O drivers xsl/xlsx (unioninout5.mos) from the mmsheet module that support manipulating spreadsheet files on various platforms without having to install Microsoft Excel.
• Module mmsheet also provides an I/O driver for csv files (unioninout6.mos).