Merge lp:~percona-toolkit-dev/percona-toolkit/check-for-option-typos.

bin/pt-duplicate-key-checker

@@ -4085,7 +4085,7 @@ Print DROP KEY statement for each duplicate key.  By default an ALTER TABLE
 DROP KEY statement is printed below each duplicate key so that, if you want to
 remove the duplicate key, you can copy-paste the statement into MySQL.
 
-To disable printing these statements, specify --nosql.
+To disable printing these statements, specify C<--no-sql>.
 
 =item --[no]summary
 

bin/pt-log-player

@@ -3066,7 +3066,7 @@ L<"--base-dir">.  These session files are played with L<"--play">.
 pt-log-player will L<"--play"> session files in parallel using N number of
 L<"--threads">.  (They're not technically threads, but we call them that
 anyway.)  Each thread will play all the sessions in its given session files.
-The sessions are played as fast as possible--there are no delays--because the
+The sessions are played as fast as possible (there are no delays) because the
 goal is to stress-test and load-test the server.  So be careful using this
 script on a production server!
 

bin/pt-query-digest

@@ -14530,7 +14530,7 @@ and see also L<"FINGERPRINTS">.
 
 A report is printed for each L<"--group-by"> value (unless C<--no-report> is
 given).  Therefore, C<--group-by user,db> means "report on queries with the
-same user and report on queries with the same db"--it does not mean "report
+same user and report on queries with the same db"; it does not mean "report
 on queries with the same user and db."  See also L<"OUTPUT">.
 
 Every value must have a corresponding value in the same position in
@@ -15236,7 +15236,7 @@ The MySQL time expression is wrapped inside a query like
 valid inside this query.  For example, do not use UNIX_TIMESTAMP() because
 UNIX_TIMESTAMP(UNIX_TIMESTAMP()) returns 0.
 
-Events are assumed to be in chronological--older events at the beginning of
+Events are assumed to be in chronological: older events at the beginning of
 the log and newer events at the end of the log.  L<"--since"> is strict: it
 ignores all queries until one is found that is new enough.  Therefore, if
 the query events are not consistently timestamped, some may be ignored which

bin/pt-stalk

@@ -1477,7 +1477,7 @@ L<"--variable"> column matches the L<"--match"> option.  For example, to trigger
 when more than 10 processes are in the "statistics" state, use the following
 options:
 
-  --trigger processlist --variable State \
+  --function processlist --variable State \
     --match statistics --threshold 10
 
 =back

util/check-tool

@@ -23,11 +23,12 @@ my $tool_type;        # perl or bash
 my @check_subs = (qw(
    check_alpha_order
    check_module_usage
-   check_option_types
    check_pod_header_order
    check_pod_formatting
-   check_option_usage
    check_pod_links
+   check_option_usage
+   check_option_types
+   check_option_typos
 ));
 
 TOOL:
@@ -471,6 +472,22 @@ sub check_option_usage {
    return;
 }
 
+sub check_option_typos {
+   my ($fh) = @_;
+
+   my %ops = map { $_=>1 } split /\n/, `awk '/^=item --/ {print \$2}' $tool_file`;
+   my $len = `wc -l $tool_file`;
+   my $doc = `grep '^=pod' -A $len`;
+   while ( $doc =~ m/(--[a-z]+[a-z-]+)/sg ) {
+      my $op  = $1;
+      my $nop = $op;
+      $nop =~ s/^--no-/--[no]/;
+      if ( !$ops{$op} && !$ops{$nop} ) {
+         print "Unknown option in documentation: $op\n"
+      }
+   }
+}
+
 sub check_pod_links {
    my $offset = `cat $tool_file | grep '^=head1 NAME' --byte-offset | cut -d ':' -f 1`;
    if ( !$offset ) {