cf7bdb743254adfa135e27a93d99193f8cd33d12
[pspp-builds.git] / tests / formats / format-guesser.sh
1 #! /bin/sh
2
3 # Tests guessing of data formats from data.
4
5 TEMPDIR=/tmp/pspp-tst-$$
6
7 # ensure that top_builddir  are absolute
8 if [ -z "$top_builddir" ] ; then top_builddir=. ; fi
9 if [ -z "$top_srcdir" ] ; then top_srcdir=. ; fi
10 top_builddir=`cd $top_builddir; pwd`
11 PSPP=$top_builddir/src/ui/terminal/pspp$EXEEXT
12
13 # ensure that top_srcdir is absolute
14 top_srcdir=`cd $top_srcdir; pwd`
15
16 STAT_CONFIG_PATH=$top_srcdir/config
17 export STAT_CONFIG_PATH
18
19
20 cleanup()
21 {
22      if [ x"$PSPP_TEST_NO_CLEANUP" != x ] ; then 
23         echo "NOT cleaning $TEMPDIR" 
24         return ; 
25      fi
26      cd /
27      rm -rf $TEMPDIR
28 }
29
30
31 fail()
32 {
33     echo $activity
34     echo FAILED
35     cleanup;
36     exit 1;
37 }
38
39
40 no_result()
41 {
42     echo $activity
43     echo NO RESULT;
44     cleanup;
45     exit 2;
46 }
47
48 pass()
49 {
50     cleanup;
51     exit 0;
52 }
53
54 mkdir -p $TEMPDIR
55
56 cd $TEMPDIR
57 activity="create test data"
58 sed -ne 's/#.*//;/^[    ]*$/!p' > $TEMPDIR/test-list <<'EOF'
59 # No data.
60 => F8.2
61 "" => F8.2
62 "." => F8.2
63
64 # Numeric formats.
65 "1.2" => F3.1
66 "$1.2" => DOLLAR4.1
67 "1.2%" => PCT4.1
68 "$1.2%" => A5
69 "1e5" => E3.0
70 "1e+5" => E4.0
71 "1+5" => E3.0
72 "1-5" => E3.0
73 "1.2e5" => E5.1
74 "1.3e+5" => E6.1
75 "1.4+5" => E5.1
76 "1e" => A2
77 "1e+" => A3
78 "1+" => A2
79 "1-" => A2
80 "1.5-5" => E5.1
81 "1,123" => COMMA5.0         # Is , is grouping or decimal? Assume grouping.
82 "1.123" => F5.3             # Ditto.
83 "1,12" => F4.2              # Not a group of 3, so last delim must be decimal.
84 "1.12" => F4.2              # Ditto.
85 "1,1234" => F6.4            # Not a group of 3, so last delim must be decimal.
86 "1.1234" => F6.4            # Ditto.
87 "$1.234" => DOLLAR6.3       # Dollar sign means decimal has to be '.'.
88 "$1,234" => DOLLAR6.0       # Ditto.
89 "1.234%" => PCT6.3          # Percent sign means decimal has to be '.'.
90 "1,234%" => PCT6.0          # Ditto.
91 "1,123.456" => COMMA9.3     # Both '.' and ',', so last delim must be decimal.
92 "1.123,456" => DOT9.3       # Ditto.
93 "1,123,456.45" => COMMA12.2 # Ditto.
94 "1.123.456,45" => DOT12.2   # Ditto.
95 "1,123,456" => COMMA9.0     # Ditto.
96 "1.123.456" => DOT9.0       # Ditto.
97
98 # Date and time formats.
99 "01-OCT-1978" => DATE11
100 "01-x-1978" => EDATE9           # Roman numeral for month.
101 "01-13-99" => ADATE8
102 "1-13-99" => ADATE7 (ADATE8)
103 "13-01-99" => EDATE8
104 "13-1-99" => EDATE7 (EDATE8)
105 "32-1-1" => SDATE6 (SDATE8)
106 "1q01" => QYR4
107 "1Q01" => QYR4
108 "1 q 01" => QYR6
109 "1 Q 01" => QYR6
110 "1q2001" => QYR6
111 "1Q2001" => QYR6
112 "1 q 2001" => QYR8
113 "1 Q 2001" => QYR8
114 "oct 05" => MOYR6
115 "oct 2005" => MOYR8
116 "1-1-01 1:2" => A10             # Minute needs at least two digits.
117 "1-1-01 1:02" => DATETIME11.0 (DATETIME17.0)
118 "1-1-01 1:02:3" => A13          # Second needs at least two digits.
119 "1-1-01 1:02:03" => DATETIME20.0
120 "1-1-01 1:02:03.1" => DATETIME20.1 (DATETIME22.1)
121 "1-1-01 +1:02:03.1" => DATETIME20.1 (DATETIME22.1)
122 "1-1-01 -1:02:03.1" => DATETIME20.1 (DATETIME22.1)
123 "1:30" => TIME4.0 (TIME5.0)
124 "1:30:05" => TIME8.0
125 "-1:30" => TIME5.0
126 "+1:30" => TIME5.0
127 "-1:30:15" => TIME8.0
128 "+1:30:15" => TIME8.0
129 "-1:30:15.5" => TIME10.1
130 "+1:30:15.75" => TIME11.2
131 "1 1:30" => DTIME6.0 (DTIME8.0)
132 "+1 1:30" => DTIME7.0 (DTIME8.0)
133 "-1 1:30" => DTIME7.0 (DTIME8.0)
134 "-1-13-99" => A8
135 "+1-13-99" => A8
136 "1+13+99" => A7
137 "1:00:01.03" => TIME10.2 (TIME11.2)
138 "12 1:00:01.3" => DTIME12.1 (DTIME13.1)
139 "jan" => MONTH3
140 "Feb" => MONTH3
141 "MAR" => MONTH3
142 "i" => A1                       # Not detected as MONTH format.
143 "v" => A1
144 "ix" => A2
145 "x" => A1
146 "january" => MONTH7
147 "janaury" => MONTH7
148 "february" => MONTH8
149 "febraury" => MONTH8
150 "march" => MONTH5
151 "marhc" => MONTH5
152 "april" => MONTH5
153 "may" => MONTH3
154 "june" => MONTH4
155 "july" => MONTH4
156 "august" => MONTH6
157 "september" => MONTH9
158 "october" => MONTH7
159 "november" => MONTH8
160 "decmeber" => MONTH8
161 "december" => MONTH8
162 "monady" => WKDAY6
163 "tuseday" => WKDAY7
164 "wedensday" => WKDAY9
165 "thurdsay" => WKDAY8
166 "fridya" => WKDAY6
167 "saturady" => WKDAY8
168 "sudnay" => WKDAY6
169
170 # Ambiguous; bias in favor of more sensible DD/MM/YY format:
171 "1/1/1978" => EDATE8
172 "01/01/01" => EDATE8
173
174 # Several ambiguous dates can be clarified by one unambiguous example:
175 "1/1/1978" "1/2/1978" "1/3/1978" "1/13/1978" => ADATE9 # MM/DD/YY
176 "01/01/01" "02/01/01" "03/01/01" "13/01/01" => EDATE8  # DD/MM/YY
177 "01/01/01" "02/01/01" "03/01/01" "2013/01/01" => SDATE10 # YY/MM/DD
178 EOF
179 if [ $? -ne 0 ] ; then no_result ; fi
180
181 activity="create syntax file"
182 {
183   echo "SET DECIMAL=DOT." &&
184   sed < $TEMPDIR/test-list -e 's#^\(.*\)=> \(.*\)$#DEBUG FORMAT GUESSER \1.#'
185 } > $TEMPDIR/test.stat
186 if [ $? -ne 0 ] ; then no_result ; fi
187
188 activity="run program"
189 $SUPERVISOR $PSPP --testing-mode -o pspp.csv \
190          $TEMPDIR/test.stat >$TEMPDIR/test.err 2> $TEMPDIR/test.out
191
192 activity="compare output"
193 perl -pi -e 's/^\s*$//g' $TEMPDIR/test-list $TEMPDIR/test.out
194 diff -b $TEMPDIR/test-list $TEMPDIR/test.out
195 if [ $? -ne 0 ] ; then fail ; fi
196
197 pass