|
| 1 | +#!/usr/bin/perl |
| 2 | + |
| 3 | +# Check that the keyword lists in gram.y and kwlist.h are sane. Run from |
| 4 | +# the top directory, or pass a path to a top directory as argument. |
| 5 | +# |
| 6 | +# $PostgreSQL: pgsql/src/tools/check_keywords.pl,v 1.1 2009/04/29 05:05:57 heikki Exp $ |
| 7 | + |
| 8 | +if (@ARGV) { |
| 9 | +$path =$ARGV[0]; |
| 10 | +shift@ARGV; |
| 11 | +} |
| 12 | + |
| 13 | +if ($patheq'') {$path ="."; } |
| 14 | + |
| 15 | +$[ = 1;# set array base to 1 |
| 16 | +$, ='';# set output field separator |
| 17 | +$\ ="\n";# set output record separator |
| 18 | + |
| 19 | +$keyword_categories{'unreserved_keyword'} ='UNRESERVED_KEYWORD'; |
| 20 | +$keyword_categories{'col_name_keyword'} ='COL_NAME_KEYWORD'; |
| 21 | +$keyword_categories{'type_func_name_keyword'} ='TYPE_FUNC_NAME_KEYWORD'; |
| 22 | +$keyword_categories{'reserved_keyword'} ='RESERVED_KEYWORD'; |
| 23 | + |
| 24 | +$gram_filename ="$path/src/backend/parser/gram.y"; |
| 25 | +open(GRAM,$gram_filename) ||die("Could not open$gram_filename!"); |
| 26 | +line:while (<GRAM>) { |
| 27 | +chomp;# strip record separator |
| 28 | +@Fld =split('',$_, -1); |
| 29 | + |
| 30 | +$S =$_; |
| 31 | +# Make sure any braces are split |
| 32 | +$s ='{',$S =~s/$s/ {/g; |
| 33 | +$s ='}',$S =~s/$s/ }/g; |
| 34 | +# Any comments are split |
| 35 | +$s ='[/][*]',$S =~s#$s# /*#g; |
| 36 | +$s ='[*][/]',$S =~s#$s# */#g; |
| 37 | + |
| 38 | +if (!($kcat)) { |
| 39 | +# Is this the beginning of a keyword list? |
| 40 | +foreach$k (keys%keyword_categories) { |
| 41 | +if ($S =~m/^($k):/) { |
| 42 | +$kcat =$k; |
| 43 | +next line; |
| 44 | + } |
| 45 | +} |
| 46 | +next line; |
| 47 | + } |
| 48 | + |
| 49 | +# Now split the line into individual fields |
| 50 | +$n = (@arr =split('',$S)); |
| 51 | + |
| 52 | +# Ok, we're in a keyword list. Go through each field in turn |
| 53 | +for ($fieldIndexer = 1;$fieldIndexer <=$n;$fieldIndexer++) { |
| 54 | +if ($arr[$fieldIndexer]eq'*/' &&$comment) { |
| 55 | +$comment = 0; |
| 56 | +next; |
| 57 | +} |
| 58 | +elsif ($comment) { |
| 59 | +next; |
| 60 | +} |
| 61 | +elsif ($arr[$fieldIndexer]eq'/*') { |
| 62 | +# start of a multiline comment |
| 63 | +$comment = 1; |
| 64 | +next; |
| 65 | +} |
| 66 | +elsif ($arr[$fieldIndexer]eq'//') { |
| 67 | +next line; |
| 68 | +} |
| 69 | + |
| 70 | +if ($arr[$fieldIndexer]eq';') { |
| 71 | +# end of keyword list |
| 72 | +$line =''; |
| 73 | +$kcat =''; |
| 74 | +next; |
| 75 | +} |
| 76 | + |
| 77 | +if ($arr[$fieldIndexer]eq'|') { |
| 78 | +next; |
| 79 | +} |
| 80 | + |
| 81 | +# Put this keyword into the right list |
| 82 | +push @{$keywords{$kcat}},$arr[$fieldIndexer]; |
| 83 | + } |
| 84 | +} |
| 85 | +close GRAM; |
| 86 | + |
| 87 | +# Check that all keywords are in alphabetical order |
| 88 | +foreach$kcat (keys%keyword_categories) { |
| 89 | +$prevkword =''; |
| 90 | + |
| 91 | +foreach$kword (@{$keywords{$kcat}}) { |
| 92 | +# Some keyword have a _P suffix. Remove it for the comparison. |
| 93 | +$bare_kword =$kword; |
| 94 | +$bare_kword =~s/_P$//; |
| 95 | +if ($bare_kwordle$prevkword) { |
| 96 | +print"'$bare_kword' after '$prevkword' in$kcat list is misplaced"; |
| 97 | +} |
| 98 | +$prevkword =$bare_kword; |
| 99 | + } |
| 100 | +} |
| 101 | + |
| 102 | +# Transform the keyword lists into hashes. |
| 103 | +# kwhashes is a hash of hashes, keyed by keyword category id, e.g. |
| 104 | +# UNRESERVED_KEYWORD. Each inner hash is a keyed by keyword id, e.g. ABORT_P |
| 105 | +# with a dummy value. |
| 106 | +while (my ($kcat,$kcat_id) =each(%keyword_categories) ) { |
| 107 | +@arr = @{$keywords{$kcat}}; |
| 108 | + |
| 109 | +my$hash; |
| 110 | +foreachmy$item (@arr) {$hash->{$item} = 1 } |
| 111 | + |
| 112 | +$kwhashes{$kcat_id} =$hash; |
| 113 | +} |
| 114 | + |
| 115 | +# Now read in kwlist.h |
| 116 | + |
| 117 | +$kwlist_filename ="$path/src/include/parser/kwlist.h"; |
| 118 | +open(KWLIST,$kwlist_filename) ||die("Could not open$kwlist_filename!"); |
| 119 | + |
| 120 | +$prevkwstring =''; |
| 121 | +kwlist_line:while (<KWLIST>) { |
| 122 | +my($line) =$_; |
| 123 | + |
| 124 | +if ($line =~/^PG_KEYWORD\(\"(.*)\", (.*), (.*)\)/) |
| 125 | + { |
| 126 | +my($kwstring) =$1; |
| 127 | +my($kwname) =$2; |
| 128 | +my($kwcat_id) =$3; |
| 129 | + |
| 130 | +# Check that the list is in alphabetical order |
| 131 | +if ($kwstringle$prevkwstring) { |
| 132 | +print"'$kwstring' after '$prevkwstring' in kwlist.h is misplaced"; |
| 133 | +} |
| 134 | +$prevkwstring =$kwstring; |
| 135 | + |
| 136 | +# Check that the keyword string is valid: all lower-case ASCII chars |
| 137 | +if ($kwstring !~/^[a-z_]*$/) { |
| 138 | +print"'$kwstring' is not a valid keyword string, must be all lower-case ASCII chars"; |
| 139 | +} |
| 140 | + |
| 141 | +# Check that the keyword name is valid: all upper-case ASCII chars |
| 142 | +if ($kwname !~/^[A-Z_]*$/) { |
| 143 | +print"'$kwname' is not a valid keyword name, must be all upper-case ASCII chars"; |
| 144 | +} |
| 145 | + |
| 146 | +# Check that the keyword string matches keyword name |
| 147 | +$bare_kwname =$kwname; |
| 148 | +$bare_kwname =~s/_P$//; |
| 149 | +if ($bare_kwnameneuc($kwstring)) { |
| 150 | +print"keyword name '$kwname' doesn't match keyword string '$kwstring'"; |
| 151 | +} |
| 152 | + |
| 153 | +# Check that the keyword is present in the grammar |
| 154 | +%kwhash = %{$kwhashes{$kwcat_id}}; |
| 155 | + |
| 156 | +if (!(%kwhash)){ |
| 157 | +#print "Unknown kwcat_id: $kwcat_id"; |
| 158 | +}else { |
| 159 | +if (!($kwhash{$kwname})) { |
| 160 | +print"'$kwname' not present in$kwcat_id section of gram.y"; |
| 161 | + }else { |
| 162 | +# Remove it from the hash, so that we can complain at the end |
| 163 | +# if there's keywords left that were not found in kwlist.h |
| 164 | +delete$kwhashes{$kwcat_id}->{$kwname}; |
| 165 | + } |
| 166 | +} |
| 167 | + } |
| 168 | +} |
| 169 | +close KWLIST; |
| 170 | + |
| 171 | +# Check that we've paired up all keywords from gram.y with lines in kwlist.h |
| 172 | +while (my ($kwcat,$kwcat_id) =each(%keyword_categories) ) { |
| 173 | +%kwhash = %{$kwhashes{$kwcat_id}}; |
| 174 | + |
| 175 | +formy$kw (keys%kwhash ) { |
| 176 | +print"'$kw' found in gram.y$kwcat category, but not in kwlist.h" |
| 177 | + } |
| 178 | +} |