pintos-os.org Git - pspp/blob - tests/data/format-guesser.at

   1 dnl PSPP - a program for statistical analysis.
   2 dnl Copyright (C) 2017 Free Software Foundation, Inc.
   3 dnl
   4 dnl This program is free software: you can redistribute it and/or modify
   5 dnl it under the terms of the GNU General Public License as published by
   6 dnl the Free Software Foundation, either version 3 of the License, or
   7 dnl (at your option) any later version.
   8 dnl
   9 dnl This program is distributed in the hope that it will be useful,
  10 dnl but WITHOUT ANY WARRANTY; without even the implied warranty of
  11 dnl MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12 dnl GNU General Public License for more details.
  13 dnl
  14 dnl You should have received a copy of the GNU General Public License
  15 dnl along with this program.  If not, see <http://www.gnu.org/licenses/>.
  16 dnl
  17 AT_BANNER([format guesser])
  18
  19 AT_SETUP([format guesser])
  20 AT_DATA([format-guesser.txt], [dnl
  21 # No data.
  22 => F8.2
  23 "" => F8.2
  24 "." => F8.2
  25
  26 # Numeric formats.
  27 "1.2" => F3.1
  28 "$1.2" => DOLLAR4.1
  29 "1.2%" => PCT4.1
  30 "$1.2%" => A5
  31 "1e5" => E3.0
  32 "1e+5" => E4.0
  33 "1+5" => E3.0
  34 "1-5" => E3.0
  35 "1.2e5" => E5.1
  36 "1.3e+5" => E6.1
  37 "1.4+5" => E5.1
  38 "1e" => A2
  39 "1e+" => A3
  40 "1+" => A2
  41 "1-" => A2
  42 "1.5-5" => E5.1
  43 "1,123" => COMMA5.0         # Is , is grouping or decimal? Assume grouping.
  44 "1.123" => F5.3             # Ditto.
  45 "1,12" => F4.2              # Not a group of 3, so last delim must be decimal.
  46 "1.12" => F4.2              # Ditto.
  47 "1,1234" => F6.4            # Not a group of 3, so last delim must be decimal.
  48 "1.1234" => F6.4            # Ditto.
  49 "$1.234" => DOLLAR6.3       # Dollar sign means decimal has to be '.'.
  50 "$1,234" => DOLLAR6.0       # Ditto.
  51 "1.234%" => PCT6.3          # Percent sign means decimal has to be '.'.
  52 "1,234%" => PCT6.0          # Ditto.
  53 "1,123.456" => COMMA9.3     # Both '.' and ',', so last delim must be decimal.
  54 "1.123,456" => DOT9.3       # Ditto.
  55 "1,123,456.45" => COMMA12.2 # Ditto.
  56 "1.123.456,45" => DOT12.2   # Ditto.
  57 "1,123,456" => COMMA9.0     # Ditto.
  58 "1.123.456" => DOT9.0       # Ditto.
  59
  60 # Date and time formats.
  61 "01-OCT-1978" => DATE11
  62 "01-x-1978" => EDATE9           # Roman numeral for month.
  63 "01-13-99" => ADATE8
  64 "1-13-99" => ADATE7 (ADATE8)
  65 "13-01-99" => EDATE8
  66 "13-1-99" => EDATE7 (EDATE8)
  67 "32-1-1" => SDATE6 (SDATE8)
  68 "1q01" => QYR4
  69 "1Q01" => QYR4
  70 "1 q 01" => QYR6
  71 "1 Q 01" => QYR6
  72 "1q2001" => QYR6
  73 "1Q2001" => QYR6
  74 "1 q 2001" => QYR8
  75 "1 Q 2001" => QYR8
  76 "oct 05" => MOYR6
  77 "oct 2005" => MOYR8
  78 "1-1-01 1:2" => A10             # Minute needs at least two digits.
  79 "1-1-01 1:02" => DATETIME11.0 (DATETIME17.0)
  80 "1-1-01 1:02:3" => A13          # Second needs at least two digits.
  81 "1-1-01 1:02:03" => DATETIME20.0
  82 "1-1-01 1:02:03.1" => DATETIME20.1 (DATETIME22.1)
  83 "1-1-01 +1:02:03.1" => DATETIME20.1 (DATETIME22.1)
  84 "1-1-01 -1:02:03.1" => DATETIME20.1 (DATETIME22.1)
  85 # XXX YMDHMS formats
  86 # XXX MTIME formats
  87 "1:30" => TIME4.0 (TIME5.0)
  88 "1:30:05" => TIME8.0
  89 "-1:30" => TIME5.0
  90 "+1:30" => TIME5.0
  91 "-1:30:15" => TIME8.0
  92 "+1:30:15" => TIME8.0
  93 "-1:30:15.5" => TIME10.1
  94 "+1:30:15.75" => TIME11.2
  95 "1 1:30" => DTIME6.0 (DTIME8.0)
  96 "+1 1:30" => DTIME7.0 (DTIME8.0)
  97 "-1 1:30" => DTIME7.0 (DTIME8.0)
  98 "-1-13-99" => A8
  99 "+1-13-99" => A8
 100 "1+13+99" => A7
 101 "1:00:01.03" => TIME10.2 (TIME11.2)
 102 "12 1:00:01.3" => DTIME12.1 (DTIME13.1)
 103 "jan" => MONTH3
 104 "Feb" => MONTH3
 105 "MAR" => MONTH3
 106 "i" => A1                       # Not detected as MONTH format.
 107 "v" => A1
 108 "ix" => A2
 109 "x" => A1
 110 "january" => MONTH7
 111 "janaury" => MONTH7             # Only first three letters are significant.
 112 "february" => MONTH8
 113 "febraury" => MONTH8
 114 "march" => MONTH5
 115 "marhc" => MONTH5
 116 "april" => MONTH5
 117 "may" => MONTH3
 118 "june" => MONTH4
 119 "july" => MONTH4
 120 "august" => MONTH6
 121 "september" => MONTH9
 122 "october" => MONTH7
 123 "november" => MONTH8
 124 "decmeber" => MONTH8
 125 "december" => MONTH8
 126 "monady" => WKDAY6
 127 "tuseday" => WKDAY7
 128 "wedensday" => WKDAY9
 129 "thurdsay" => WKDAY8
 130 "fridya" => WKDAY6
 131 "saturady" => WKDAY8
 132 "sudnay" => WKDAY6
 133
 134 # Ambiguous; bias in favor of more sensible DD/MM/YY format:
 135 "1/1/1978" => EDATE8
 136 "01/01/01" => EDATE8
 137
 138 # Several ambiguous dates can be clarified by one unambiguous example:
 139 "1/1/1978" "1/2/1978" "1/3/1978" "1/13/1978" => ADATE9 # MM/DD/YY
 140 "01/01/01" "02/01/01" "03/01/01" "13/01/01" => EDATE8  # DD/MM/YY
 141 "01/01/01" "02/01/01" "03/01/01" "2013/01/01" => SDATE10 # YY/MM/DD
 142 ])
 143 AT_CHECK([[(echo "SET DECIMAL=DOT."
 144            sed -e 's/#.*//
 145 s/[     ]*$//
 146 /^$/d
 147 s,^\(.*\)=> \(.*\)$,DEBUG FORMAT GUESSER \1.,') \
 148            < format-guesser.txt > format-guesser.sps]])
 149 AT_CHECK([pspp --testing-mode -O format=csv format-guesser.sps],
 150   [0], [], [stderr])
 151 AT_CHECK([[sed -e 's/#.*//
 152 s/[     ]*$//
 153 /^$/d' < format-guesser.txt > expout]])
 154 AT_CHECK([cat stderr], [0], [expout])
 155 AT_CLEANUP