Changeset 4832


Ignore:
Timestamp:
Oct 13, 2015, 3:57:57 PM (2 years ago)
Author:
nmedfort
Message:

Couple test cases for grapheme clusters.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • icGREP/icgrep-devel/QA/greptest.xml

    r4791 r4832  
    460460<grepcase regexp="[A-Z]" datafile="simple1" grepcount="1"/>
    461461<grepcase regexp="fodder|simple" datafile="simple1" grepcount="2"/>
     462<grepcase regexp="(?g)fodder|simple" datafile="simple1" grepcount="2"/>
    462463
    463464<grepcase regexp="[cde]{3}" datafile="bounded_charclass" grepcount="3"/>
     
    492493<grepcase regexp="h.t" datafile="RangeAltSeqMatchStarKplusWhileNotOptAny" grepcount="3" />
    493494<grepcase regexp="do*?c|ez*?t" datafile="RangeAltSeqMatchStarKplusWhileNotOptAny" grepcount="4"/>
    494 <grepcase regexp="^.....\b" datafile="RangeAltSeqMatchStarKplusWhileNotOptAny" grepcount="6"/>>
     495<grepcase regexp="^.....\b" datafile="RangeAltSeqMatchStarKplusWhileNotOptAny" grepcount="6"/>
     496<grepcase regexp="^\X\X\X\X\X\b" datafile="RangeAltSeqMatchStarKplusWhileNotOptAny" grepcount="6"/>
    495497
    496498<grepcase regexp="[]]" datafile="special_characters" grepcount="9"/>
     
    505507
    506508<grepcase regexp="^((([2][5][0-5]|([2][0-4]|[1][0-9]|[0-9])?[0-9])[.]){3})([2][5][0-5]|([2][0-4]|[1][0-9]|[0-9])?[0-9])$" datafile="ips" grepcount="60"/>
     509<grepcase regexp="^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.([a-zA-Z]{2}|com|org|net|edu|gov|mil|biz|info|mobi|name|aero|asia|jobs|museum)$" datafile="emails" grepcount="116"/>
    507510<grepcase regexp="^[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.([a-zA-Z]{2}|com|org|net|edu|gov|mil|biz|info|mobi|name|aero|asia|jobs|museum)$" datafile="emails" grepcount="116"/>
    508511<grepcase regexp="^[-+]?([1-9]0?)+\.?((0*[1-9])+|0)([eE][-+]?([0-9]+)+)?$" datafile="floats" grepcount="26"/>
     
    524527
    525528
    526  <datafile id="codepoints">
    527  A line with 0x89 &#x89;
    528  A line with 0x1234 &#x1234;
    529  A line with 0x1245 &#x1245;
    530  äœ 
    531  å¥œ
    532  A plain line.
    533 </datafile>
    534  <grepcase regexp="[\u{1234}-\u{1245}]" datafile="codepoints" grepcount="2"/>
    535  <grepcase regexp="[\u{086}-\u{9A}]" datafile="codepoints" grepcount="1"/>
    536  <grepcase regexp="[䜠奜]" datafile="codepoints" grepcount="2"/>
    537  <grepcase regexp="\u{4F60}" datafile="codepoints" grepcount="1"/>
     529<datafile id="codepoints">
     530A line with 0x89 &#x89;
     531A line with 0x1234 &#x1234;
     532A line with 0x1245 &#x1245;
     533䜠
     534奜
     535àŠ•
     536àŠ•à§€
     537A plain line.
     538</datafile>
     539<grepcase regexp="[\u{1234}-\u{1245}]" datafile="codepoints" grepcount="2"/>
     540<grepcase regexp="[\u{086}-\u{9A}]" datafile="codepoints" grepcount="1"/>
     541<grepcase regexp="[䜠奜]" datafile="codepoints" grepcount="2"/>
     542<grepcase regexp="^\u{4F60}$" datafile="codepoints" grepcount="1"/>
     543<grepcase regexp="(?g)^\u{4F60}$" datafile="codepoints" grepcount="1"/>
     544 <grepcase regexp="^àŠ•$" datafile="codepoints" grepcount="1"/>
     545 <grepcase regexp="(?g)^àŠ•$" datafile="codepoints" grepcount="2"/>
    538546 
    539547 
    540548 
    541  
    542  <datafile id = "LineBreaking">CRLF1&#13;&#10;CRLF2&#13;&#10;CRLF3&#13;&#10;
     549<datafile id = "LineBreaking">CRLF1&#13;&#10;CRLF2&#13;&#10;CRLF3&#13;&#10;
    543550LS1&#x2028;LS2&#x2028;LS3&#x2028;PS1&#x2029;PS2&#x2029;PS3&#x2029;
    544551PS4&#x2029;CRLF4&#13;&#10;LS4&#x2028;LS5&#x2028;CRLF5&#13;&#10;LS6&#x2028;
    545552Unterminated</datafile>
    546553
    547  <grepcase regexp="^.*$" datafile="LineBreaking" grepcount="19"/>
    548  <grepcase regexp="Unterminated$" datafile="LineBreaking" grepcount="1"/>
    549  <grepcase regexp="^CRLF.$" datafile="LineBreaking" grepcount="5"/>
    550  <grepcase regexp="LS[0-9]*" datafile="LineBreaking" grepcount="6"/>
    551  <grepcase regexp="PS" datafile="LineBreaking" grepcount="4"/>
     554<grepcase regexp="^.*$" datafile="LineBreaking" grepcount="19"/>
     555<grepcase regexp="Unterminated$" datafile="LineBreaking" grepcount="1"/>
     556<grepcase regexp="^CRLF.$" datafile="LineBreaking" grepcount="5"/>
     557<grepcase regexp="LS[0-9]*" datafile="LineBreaking" grepcount="6"/>
     558<grepcase regexp="PS" datafile="LineBreaking" grepcount="4"/>
    552559 
    553560<grepcase regexp="[a-z]{20}" datafile="4KiB-onepage" grepcount="77"/>
Note: See TracChangeset for help on using the changeset viewer.