This directory contains unit tests for Yume.

uniq_test.sh - conformance test

   Check Yume's conformance with uniq.  The test data covers most of
   uniq's operations, but not all of it -- parts of Yume were
   intentionally written to be incompatible, those are noted in the
   documentation.

regression_test.sh - full regression test

   Check output correctness and speed between two different versions
   of Yume.

regression_test3.sh - three-way regression test

   Check output correctness and speed between three different versions
   of Yume.  This was so that Yume compiled by GCC and MSVC can be
   tested simultaneously against a single reference program, using the
   same test data.

regression_test_short.sh - short regression test

   Simplified version of regression_test.sh.  The functionalities are
   reduced to test with Perl version of Yume.


test_manual*.txt - static test data

   1 - empty file.
   2 - small file with some duplicate lines.
   3 - small file with non-uniform end of lines.
   4 - /usr/dict/words: medium sized file with all unique lines.
   5 - MMR thread: Shift-JIS encoded file with random duplicate lines.

test_{generic,longline,i,c,f3,s8,w4}.pl - dynamic test data (pipe based)

   These tests generate text data that can be fed through pipes.

   generic - generic test.
   longline - test with long lines.
   i - ignore case (-i).
   c - count (-c).
   f3 - skip fields (-f 3).
   s8 - skip characters (-s 3).
   w4 - significant characters (-w 4).

test_{marker,random,space,crc}.pl - dynamic test data (file based)

   These tests generate binary data that must be feed through files.

   marker - marker expression test.

      The lines generated should be filtered using the generated
      expressions, resulting in all duplicate lines.

   random - random data test.

      Generates random input and marker expressions for stability
      test.

   space - whitespace test.

      Generates random whitespaces and null characters.

   crc - CRC test.

      Generate lines with colliding CRCs to defeat Yume's CRC-based
      line hashing algorithm.