cobol-java-v3/benchmark-programs/16-matching-2stage-1-1

16-matching-2stage-1-1: Two-Stage 1:1 Matching

电信业务场景

二级请求书核对。两段式1:1对账处理。第一段:请求书与支付对账→中间文件；第二段:中间文件与调整记录对账→最终结果。

Description

Tests a two-stage pipeline where the output of the first 1:1 match feeds into a second 1:1 match. Three-way chained matching across FILE-A, FILE-B, and FILE-C using an intermediate TEMP-FILE.

• Stage 1: Match FILE-A and FILE-B (1:1), write matched A-side records to TEMP-FILE. Unmatched records (KEY00004 in A, KEY00005 in B) are skipped.
• Stage 2: Match TEMP-FILE and FILE-C (1:1), write matched records to FINAL-OUT. Unmatched C records (KEY00006) are skipped.

Record Layout

Field	Type	Length	Description
STD-KEY	PIC X	10	Record key
STD-DATA-1	PIC X	20	Description text
STD-DATA-2	PIC 9	10	Numeric data (display)
STD-DATA-3	PIC S9(7)V99	05	Numeric data (COMP-3)

Total record length: 45 bytes.

Files

File	Purpose
main-16-matching-2stage-1-1.cbl	Main COBOL program (two stages)
data-gen.sh	Generate three test data files
run.sh	Compile, run, verify
README.md	This file

Data

File	Records
file-a.dat	KEY00001, KEY00002, KEY00003, KEY00004 (unmatched)
file-b.dat	KEY00001, KEY00002, KEY00003, KEY00005 (unmatched)
file-c.dat	KEY00001, KEY00002, KEY00003, KEY00006 (unmatched)

The matching chain: KEY00001, KEY00002, KEY00003 exist in all three files and should survive both stages. The unmatched extras are discarded.

Matching Logic

Each stage implements standard 1:1 sorted merge/match:

1. Compare keys from both input streams.
2. If keys are equal, write a matched record and advance both.
3. If keys differ, advance the stream with the smaller key (skip it).
4. Continue until one stream is exhausted.

Test

Check	Expected
Stage 1 output (temp)	3 records x 45 = 135 bytes
Stage 2 output (final)	3 records x 45 = 135 bytes
Matched chain	KEY00001, KEY00002, KEY00003
Unmatched filtered	KEY00004, KEY00005, KEY00006

Usage

cd 16-matching-2stage-1-1
bash run.sh