tests/data/format-guesser.at

   1 AT_BANNER([format guesser])
   2
   3 AT_SETUP([format guesser])
   4 AT_DATA([format-guesser.txt], [dnl
   5 # No data.
   6 => F8.2
   7 "" => F8.2
   8 "." => F8.2
   9
  10 # Numeric formats.
  11 "1.2" => F3.1
  12 "$1.2" => DOLLAR4.1
  13 "1.2%" => PCT4.1
  14 "$1.2%" => A5
  15 "1e5" => E3.0
  16 "1e+5" => E4.0
  17 "1+5" => E3.0
  18 "1-5" => E3.0
  19 "1.2e5" => E5.1
  20 "1.3e+5" => E6.1
  21 "1.4+5" => E5.1
  22 "1e" => A2
  23 "1e+" => A3
  24 "1+" => A2
  25 "1-" => A2
  26 "1.5-5" => E5.1
  27 "1,123" => COMMA5.0         # Is , is grouping or decimal? Assume grouping.
  28 "1.123" => F5.3             # Ditto.
  29 "1,12" => F4.2              # Not a group of 3, so last delim must be decimal.
  30 "1.12" => F4.2              # Ditto.
  31 "1,1234" => F6.4            # Not a group of 3, so last delim must be decimal.
  32 "1.1234" => F6.4            # Ditto.
  33 "$1.234" => DOLLAR6.3       # Dollar sign means decimal has to be '.'.
  34 "$1,234" => DOLLAR6.0       # Ditto.
  35 "1.234%" => PCT6.3          # Percent sign means decimal has to be '.'.
  36 "1,234%" => PCT6.0          # Ditto.
  37 "1,123.456" => COMMA9.3     # Both '.' and ',', so last delim must be decimal.
  38 "1.123,456" => DOT9.3       # Ditto.
  39 "1,123,456.45" => COMMA12.2 # Ditto.
  40 "1.123.456,45" => DOT12.2   # Ditto.
  41 "1,123,456" => COMMA9.0     # Ditto.
  42 "1.123.456" => DOT9.0       # Ditto.
  43
  44 # Date and time formats.
  45 "01-OCT-1978" => DATE11
  46 "01-x-1978" => EDATE9           # Roman numeral for month.
  47 "01-13-99" => ADATE8
  48 "1-13-99" => ADATE7 (ADATE8)
  49 "13-01-99" => EDATE8
  50 "13-1-99" => EDATE7 (EDATE8)
  51 "32-1-1" => SDATE6 (SDATE8)
  52 "1q01" => QYR4
  53 "1Q01" => QYR4
  54 "1 q 01" => QYR6
  55 "1 Q 01" => QYR6
  56 "1q2001" => QYR6
  57 "1Q2001" => QYR6
  58 "1 q 2001" => QYR8
  59 "1 Q 2001" => QYR8
  60 "oct 05" => MOYR6
  61 "oct 2005" => MOYR8
  62 "1-1-01 1:2" => A10             # Minute needs at least two digits.
  63 "1-1-01 1:02" => DATETIME11.0 (DATETIME17.0)
  64 "1-1-01 1:02:3" => A13          # Second needs at least two digits.
  65 "1-1-01 1:02:03" => DATETIME20.0
  66 "1-1-01 1:02:03.1" => DATETIME20.1 (DATETIME22.1)
  67 "1-1-01 +1:02:03.1" => DATETIME20.1 (DATETIME22.1)
  68 "1-1-01 -1:02:03.1" => DATETIME20.1 (DATETIME22.1)
  69 "1:30" => TIME4.0 (TIME5.0)
  70 "1:30:05" => TIME8.0
  71 "-1:30" => TIME5.0
  72 "+1:30" => TIME5.0
  73 "-1:30:15" => TIME8.0
  74 "+1:30:15" => TIME8.0
  75 "-1:30:15.5" => TIME10.1
  76 "+1:30:15.75" => TIME11.2
  77 "1 1:30" => DTIME6.0 (DTIME8.0)
  78 "+1 1:30" => DTIME7.0 (DTIME8.0)
  79 "-1 1:30" => DTIME7.0 (DTIME8.0)
  80 "-1-13-99" => A8
  81 "+1-13-99" => A8
  82 "1+13+99" => A7
  83 "1:00:01.03" => TIME10.2 (TIME11.2)
  84 "12 1:00:01.3" => DTIME12.1 (DTIME13.1)
  85 "jan" => MONTH3
  86 "Feb" => MONTH3
  87 "MAR" => MONTH3
  88 "i" => A1                       # Not detected as MONTH format.
  89 "v" => A1
  90 "ix" => A2
  91 "x" => A1
  92 "january" => MONTH7
  93 "janaury" => MONTH7             # Only first three letters are significant.
  94 "february" => MONTH8
  95 "febraury" => MONTH8
  96 "march" => MONTH5
  97 "marhc" => MONTH5
  98 "april" => MONTH5
  99 "may" => MONTH3
 100 "june" => MONTH4
 101 "july" => MONTH4
 102 "august" => MONTH6
 103 "september" => MONTH9
 104 "october" => MONTH7
 105 "november" => MONTH8
 106 "decmeber" => MONTH8
 107 "december" => MONTH8
 108 "monady" => WKDAY6
 109 "tuseday" => WKDAY7
 110 "wedensday" => WKDAY9
 111 "thurdsay" => WKDAY8
 112 "fridya" => WKDAY6
 113 "saturady" => WKDAY8
 114 "sudnay" => WKDAY6
 115
 116 # Ambiguous; bias in favor of more sensible DD/MM/YY format:
 117 "1/1/1978" => EDATE8
 118 "01/01/01" => EDATE8
 119
 120 # Several ambiguous dates can be clarified by one unambiguous example:
 121 "1/1/1978" "1/2/1978" "1/3/1978" "1/13/1978" => ADATE9 # MM/DD/YY
 122 "01/01/01" "02/01/01" "03/01/01" "13/01/01" => EDATE8  # DD/MM/YY
 123 "01/01/01" "02/01/01" "03/01/01" "2013/01/01" => SDATE10 # YY/MM/DD
 124 ])
 125 AT_CHECK([[(echo "SET DECIMAL=DOT."
 126            sed -e 's/#.*//
 127 s/[     ]*$//
 128 /^$/d
 129 s,^\(.*\)=> \(.*\)$,DEBUG FORMAT GUESSER \1.,') \
 130            < format-guesser.txt > format-guesser.sps]])
 131 AT_CHECK([pspp --testing-mode -O format=csv format-guesser.sps],
 132   [0], [], [stderr])
 133 AT_CHECK([[sed -e 's/#.*//
 134 s/[     ]*$//
 135 /^$/d' < format-guesser.txt > expout]])
 136 AT_CHECK([cat stderr], [0], [expout])
 137 AT_CLEANUP