Files in which the programs are stored:

• Parses COBOL data description entries.
• Creates the data dictionary.
• Creates the group data set.

Files in which the documentation is stored:

Files in which test data and their results are stored:

Statement of Purpose

COB2SAS is a tool that can assist you in converting COBOL language entries into equivalent SAS language statements.

COB2SAS, Release 2 is designed to assist you in converting COBOL language data description entries into SAS language statements. It is also designed to be extendible so that you can easily modify it to parse entries other than data description entries.

Overview of COB2SAS, Release 2

COB2SAS, Release 2 uses the information in the data description entries of COBOL language programs to create a data dictionary. Succeeding steps use this data dictionary to create SAS language statements equivalent to the data description entries.

There are several reasons for producing a data dictionary prior to producing any SAS language statements. One important reason is to provide a mechanism that allows the user to reconcile the differing conventions of the COBOL and SAS languages. Once a complete data dictionary is produced, these differences can be reconciled by programs that use the data dictionary as input.

Differences in the conventions of the COBOL and SAS languages

The COBOL language allows variable names to have up to 30 characters, but the SAS language allows only 8 characters for variable names. The COBOL language has provisions for storing character strings whose length exceeds 200, but the SAS language can not store more than 200 characters in a single variable.

Description of the Data Dictionary

A description of each variable in the data dictionary is given here:

Expansion of items and groups occurring more than once

After the data dictionary has been built, succeeding programs use it as input. One of these programs (called, R2COB5) will expand items and groups that occur more than once. This program does not attempt to expand multidimensional tables. It only expands 1 dimensional tables. This program uses the values in the variables OCR_VAL, LEVEL, OCR_BASE and ITM_DISP to create the number of entries specified in OCR_VAL.

The programs used to invoke COB2SAS, Release 2 on each operating system (for example, R2MVS on MVS) are sent with the inclusion of this program commented out. If you wish to invoke it, you must uncomment the line that includes R2COB5 for execution.

SAS language statements produced

COB2SAS, Release 2 is able to produce SAS language INPUT and LABEL statements.

Assumptions, Restrictions, Caveats

COBOL language statements are parsed in the program called R2COB2. R2COB2 does no checking for syntax errors in the COBOL language statements. Furthermore, R2COB2 makes the following assumptions about the data description entries that it will read.

Assumptions regarding the location of entries within margins:

1. The contents of columns 1 through 6 are ignored.
2. Any character in column 7, other than a blank or a hyphen, causes the entire line to be ignored.
3. The level indicators CD, FD, RD, SD, and 01 are in columns 8 through 11.
4. All clauses, other than the division and section identifiers, and the level indicators, are in columns 12 through 72.

Clauses recognized

A list of clauses that R2COB2 is able to parse is given here:

The programs that invoke COB2SAS, Release 2

Included in COB2SAS, Release 2 are 4 programs, one for each operating system for which COB2SAS is distributed, that invoke COB2SAS. These 4 files differ in the conventions used to invoke R2COB1 through R2COB7 and in the conventions used to reference files.

• R2CMS is used to invoke R2COB1 through R2COB7 on CMS.
• R2MVS is used to invoke R2COB1 through R2COB7 on MVS.
• R2VMS is used to invoke R2COB1 through R2COB7 on VMS.
• R2VSE is used to invoke R2COB1 through R2COB7 on VSE.

Any special requirements for the INFILE and FILE statements are provided in thes programs. Also, in these programs, you can provide statements that will direct the data dictionary to a permanent SAS data set and the SAS language statements to a flat file.

Differences between CMS, MVS, VMS and VSE

INFILE & FILE statements

• Under CMS and MVS, there is no need to specify DCB information on the INFILE and FILE statements.
• Under VSE, it necessary to provide complete and explicit DCB information (that is, RECFM, LRECL and BLKSIZE).
• Under VMS, it necessary to explicitly provide RECFM=F LRECL=80 so that the SAS System will treat the file as if it has fixed length records.

R2COB1 through R2COB7

The programs that convert COBOL language data description entries into SAS language statements are stored in 7 files.

1. R2COB1 uses PROC FORMAT to create formats that are used while parsing COBOL language data description entries.
2. R2COB2 uses a data step to parse the data description entries and produce 2 SAS data sets. The first data set contains an observation for each data description entry. The second data set has information about the lengths of groups of data items.
3. R2COB3 uses PROC SORT and a data step to combine the group data set and the data dictionary data set.
4. R2COB4 uses PROC SORT and the data step to compress the lengths of variable names to no more than 8 characters.
5. R2COB5 is an optional step that uses a data step and PROC SORT to expand 1 dimensional tables.

Regardless of the operating system, the contents of R2COB1 through R2COB7 are identical.

The programs that create the data dictionary

R2COB1

Processing performed to build the data dictionary

The data dictionary is a SAS data set that is built by the data step in the R2COB2 file. The following is a description of the algorithm used to build the data dictionary.

• Read a line from the input file.
• Skip blank lines.
• If the variable TRACEPRS has a value of 9, then display a ruler, and the contents of the current and next lines. Return to the top of the data step.
• Evaluate the contents of column 7. If the indicator area has anything other than a blank or a hyphen, return to the top of the data step.
• Check for a line that initiates a new division.
• Check for the beginning of a copy member.
• Depending upon the division, process the input appropriately.
• In the case of the Identification and Environment Divisions, no actions are taken.
• In the case of the Data Division, the file and data description entries are processed.
• In the case of Procedure Division, unless the SEVERAL flag has been set to 'Y', the program performs end of file processing and stops.

• Upon initial entry into the Data Division, any pending processing is finished.
• Check for a line that initiates a new section.
• In the case of sections other than the File Section, no actions are taken.
• In the case of the File Section, data description entries are processed.

Processing within the File Section

Upon initial entry into the File Section, any pending processing is finished.

The current line is evaluated to determine whether it has a level indicator initiating entries in a file, sort-merge file, or data description entry.

Depending on the type of entries, process the input appropriately.

• In the case of file and sort-merge file description entries, any pending processing is completed, initial values are set appropriately, the value of the file name is saved for later processing, and the program returns to the top of the data step.
• In the case of data description entries, any pending processing is completed, each word is parsed and evaluated (the details of this process are described later), and the program returns to the top of the data step.

• Each data description is known as an entry.
• Each entry is made up of one or more clauses.
• Each clause specifies an attribute of the entry.

If the set of attributes is not defined to it, and only if the variable ATTR_ERR is set to 'Y', it will use the SHOCLS format, known as the show clause format to show each type of clause that is in the entry. (It does not have the ability to show the actual clause itself, instead it shows each type of clause in the particular set of attributes that is undefined to it.)

Parsing Algorithm

The syntax of COBOL data description entries are defined in general formats by the American National Standards Institute. In addition to these general formats, each implementor of a COBOL compiler may define extensions to these formats. COB2SAS, Release 2 is designed to recognize the general formats for data description entries in the ANSI specification X3.23-1985, as well as, some of the extensions implemented in IBM(R) OS/VS COBOL.

Overview of the parsing algorithm

Each data description entry consists of one or more attributes. Each attribute is specified by a clause. For example, the entry

Each clause begins with exactly 1 keyword and consists of 1 or more keywords and programmer supplied words. Within the context of several of the clauses, the numbers 1 through 49, 66 and 88 are valid programmer supplied words (for example, 'OCCURS 1 TO 49 TIMES DEPENDING ON VAR1'). These very same numbers can also indicate the beginning of a clause (for example, '01 VAR1'). For this reason, a variable known as Parse Mode is employed. Parse Mode either has the value 'Identify Clause' or the value 'Building Clause'. When it has the value, 'Identify Clause', the keywords encountered are interpreted as the beginning of a new clause. On the other hand, when it has the value, 'Building Clause', the keywords and programmer supplied words are interpreted as being tokens within the clause being built.

Although most of the clauses are relatively simple constructions of a few words, there are several clauses that consist of more than one repetition of the same set of keywords and programmer supplied words. For this reason, a variable known as Clause Mode is employed. For each clause, this variable is given an appropriate value. When parsing data description entries, Clause Mode may have the value 'Picture Clause', 'Value Clause', 'Occurs Clause', '66 Clause', '88 Clause', 'Copy Clause' and 'Simple Clause'.

Detailed example of parsing a data description entry

Valid forms of the 02-49 Clause and the PICTURE Clause

Algorithm used to parse a data description entry

The string in the input buffer is:

The program attempts to appropriately process these attributes.

Actions taken for each clause

The actions taken at the end of each clause is given here:

01 Clause

• Bit 1 is set in the attribute vector.
• Initial values are set for a new record.

Level Nesting

The R2COB2 data step keeps track of the depth of level nesting within file description entries. Since it makes use of a stack for tracking usage values, the index to this stack is the value of the depth.

Usage Stack

In the COBOL language, it is not required that every item have an explicitly stated usage value. For this reason, R2COB2 keeps track of each item's usage value in the STKUSAGE routine.

When an entry has a usage clause, its level number is compared to the previous level number. Depending on the result of the comparison the usage stack is updated with the value in the usage clause. On the other hand, if an entry does not have a usage clause, the usage value is extracted from the usage stack.

Group Stack & Group Data Set

To track groups, it makes use of two data structures. First it maintains a group stack which is used to keep track of the position, within the file, at which a group begins and the total number of bytes in that group. When a group is first encountered, a new entry is pushed onto the group stack. As elementary items within that group are encountered, the total number of bytes in that group is incremented by the number of bytes in that item. When the end of that group is encountered (that is, when a level number that is less than or equal to the level number of the group is encountered), the number of bytes in all groups on the the stack are reevaluated, the group just ended is popped off of the stack and an observation is output to the GROUP data set.

Redefinition Stack

The redefinition stack keeps track of the position and number of bytes in entries with level numbers greater than the current entry. When redefinition is encountered, it uses this information to update the ATBYTE and ITM_DISP variables.

Offset Stack

The offset stack keeps track of the position at which those groups that occur more than once begin and the displacement of each item within those groups.

Conversion from Usage and Picture to Informat

All picture values are converted to a standard format. For alphabetic or alphanumeric strings, the format is either 'A(w)' or 'X(w)'. For numbers, the format is '9(w)V9(f)'; where w and f are numbers of bytes. For example, the picture 'S99999V99' is converted to '9(5)V9(2)' and the picture 'X(48)' remains 'X(48)'.

1. When USAGE is 'DISPLAY':

If PICTURE is 'A' OR 'X', then the informat is '$CHAR'

If PICTURE is '9', then

If BWZ_FLAG = 'Y', then the informat is 'ZDB'

Otherwise the informat is 'ZD'

2. When USAGE is 'COMP' or 'COMP-4':

The PICTURE should be '9' and the informat is 'IB'

3. When USAGE is 'COMP-1':

The PICTURE should be ' ' and the informat is 'RB4.'

4. When USAGE is 'COMP-2' or 'BINARY':

The PICTURE should be ' ' and the informat is 'RB8.'

5. When USAGE is 'COMP-3' or 'PCKDCML':

The PICTURE should be '9' and the informat is 'PD'

6. When USAGE is 'INDEX':

Options available by way of the SWITCHES data set

TRACEPRS (Trace Parse) (Default Value: TRACEPRS='0')

Trace Parse determines the type of tracing information produced while parsing data description entries.

• If you wish to add processing for new sets of attributes within the data description entries, follow these guidelines.
• Create a COBOL data description entry with a set of attributes not yet implemented.
• Set TRACEPRS to '5' and ATTR_ERR to 'Y'.
• Execute COB2SAS, Release 2.
• Inspect the SAS Log for the message:

• This shows you a list of that combination of attributes that is not yet recognized.
• In the line above this message, find the value of AV_SUM.
• You must now edit R2COB1 and implement that combination of attributes.

Modifying the program to parse other entries

If you wish to modify the program to parse entries other than data description entries, follow these guidelines.

1. Build a Token Format for the entry.
• Take the general format from the ANSI Reference Summary.
• Remove any clause that contains either ellipses or more than 1 programmer supplied word.
• Working down from the top of the general format, enter each reserved word in the description token format.
• Items enclosed in braces will produce the same output when formatted (for example, PICTURE and PIC).
• Simply assign a formatted value to each set of reserved words (for example, 'PICTURE',PIC' = '9.').
2. Build an Identify Clause Format for the entry.
• Take the general format from the ANSI reference summary.
• Enter the reserved word or words that identify each clause in the entry. (In other words, enter each token that initiates a clause in the entry.)
• In the case where the clause contains ellipses or more than 1 programmer supplied word, assign a unique value for the formatted output.
• Assign a unique value for the formatted output to those tokens that initiate the entire entry. For example, the level indicators.
• For all tokens that initiate a simple clause (that is, a clause that does not contain an ellipses or more than 1 programmer supplied word), assign the value 'IDENTIFIED'.
• Any other token, when formatted, has the value 'UNIDENTIFIED'.
3. Build a Clause Format for the entry.
• Take the general format from the ANSI Reference Summary.
• Remove any clause that contains either ellipses or more than 1 programmer supplied word.
• Working down from the top of the general format, combine the formatted output from the Token Format to form each valid clause.
• Programmer supplied words are assigned the value 'OTHER' in the Token Format.
• Underlined reserved words are required in the clause.
• At most 1 formatted value is available for items in braces.
• It is up to you to form all possible valid combinations of tokens within each clause.
• The formatted output of the Clause Format is a string. Assign values to this string by choosing sequential numbers starting at the top and working down.
• All other values, when formatted, are assigned the value 'UNDEFINED'.
4. Build an Attribute Vector Format.
• The formatted output from the Clause Format is treated as a number. That number is used as an exponent of 2. The result of raising the base 2 to that exponent value is added to the Attribute Vector Sum (AV_SUM).
• When the end of an entry is found (which is either at the end of the section or the beginning of the next entry) the value in AV_SUM is formatted with Attribute Vector Format.
• This formatted output is used to determine which attributes are in the entry just completed.

Acknowledgements

COBOL is an industry language and is not the property of any company or any group of companies, or of any organization or any group of organizations.

No warranty, expressed or implied, is made by any contributor or by the CODASYL Programming Language Committee as to the accuracy and functioning of the programming system and language. Moreover, no responsibility is assumed by any contributor, or by the committee, in connection therewith.

The authors and copyright holders of the copyrighted material used herein

VMS is a trademark of Digital Equipment Corporation.