Refactor _parse_varvals.

Now it takes two arguments: A regexp and a string to match against.
_parse_varvals itself was split in three:  _preprocess_varvals,
_parse_varvals, and _process_val.

This also modifies the three places that call _parse_varvals; For
two, no real changes were needed, but parse_mysqld() required a fix
to deal with the two final lines of mysqld --help --verbose:

   To see what values a running MySQL server is using, type
   'mysqladmin variables' instead of 'mysqld --verbose --help'.
This commit is contained in:
Brian Fraser
2012-12-04 05:02:56 -03:00
parent 76a010abee
commit a7a0da3bc9
2 changed files with 172 additions and 154 deletions

View File

@@ -259,18 +259,28 @@ sub parse_mysqld {
# help TRUE
# abort-slave-event-count 0
# So we search for that line of hypens.
if ( $output !~ m/^-+ -+$/mg ) {
#
# It also ends with something like
#
# wait_timeout 28800
#
# To see what values a running MySQL server is using, type
# 'mysqladmin variables' instead of 'mysqld --verbose --help'.
#
# So try to find it by locating a double newline, and strip it away
if ( $output !~ m/^-+ -+$(.+?)(?:\n\n.+)?\z/sm ) {
PTDEBUG && _d("mysqld help output doesn't list vars and vals");
return;
}
# Cut off everything before the list of vars and vals.
my $varvals = substr($output, (pos $output) + 1, length $output);
# Grab the varval list.
my $varvals = $1;
# Parse the "var val" lines. 2nd retval is duplicates but there
# shouldn't be any with mysqld.
my ($config, undef) = _parse_varvals(
$varvals =~ m/\G^(\S+)(.*)\n/mg
qr/^(\S+)(.*)$/,
$varvals,
);
return $config, \@opt_files;
@@ -288,9 +298,8 @@ sub parse_my_print_defaults {
# Parse the "--var=val" lines.
my ($config, $dupes) = _parse_varvals(
map { $_ =~ m/^--([^=]+)(?:=(.*))?$/ ? ($1, $2) : () }
grep { $_ !~ m/^\s*$/ } # no empty lines
split("\n", $output)
qr/^--([^=]+)(?:=(.*))?$/,
$output,
);
return $config, $dupes;
@@ -311,113 +320,116 @@ sub parse_option_file {
# Parse the "var=val" lines.
my ($config, $dupes) = _parse_varvals(
map { $_ =~ m/^([^=]+)(?:=(.*))?$/ ? ($1, $2) : () }
grep { $_ !~ m/^\s*$/ } # no empty lines
grep { $_ !~ m/^\s*#/ } # no # comment lines
split("\n", $mysqld_section)
qr/^([^=]+)(?:=(.*))?$/,
$mysqld_section,
);
return $config, $dupes;
}
# Parses a list of variables and their values ("varvals"), returns two
# Called by _parse_varvals(), takes two arguments: a regex, and
# a string to match against. The string will be split in lines,
# and each line will be matched against the regex.
# The regex must return to captures, although the second doesn't
# have to match anything.
# Returns a hashref of arrayrefs ala
# { port => [ 12345, 12346 ], key_buffer_size => [ "16M" ] }
sub _preprocess_varvals {
my ($re, $to_parse) = @_;
my %vars;
LINE:
foreach my $line ( split /\n/, $to_parse ) {
next LINE if $line =~ m/^\s*$/; # no empty lines
next LINE if $line =~ m/^\s*#/; # no # comment lines
if ( $line !~ $re ) {
PTDEBUG && _d("Line <", $line, "> didn't match $re");
next LINE;
}
my ($var, $val) = ($1, $2);
# Variable names are usually specified like "log-bin"
# but in SHOW VARIABLES they're all like "log_bin".
$var =~ tr/-/_/;
if ( !defined $val ) {
$val = '';
}
# Strip leading and trailing whitespace.
for my $item ($var, $val) {
$item =~ s/^\s+//;
$item =~ s/\s+$//;
}
push @{$vars{$var} ||= []}, $val
}
return \%vars;
}
# Parses a string of variables and their values ("varvals"), returns two
# hashrefs: one with normalized variable=>value, the other with duplicate
# vars. The varvals list should start with a var at index 0 and its value
# at index 1 then repeat for the next var-val pair.
# vars.
sub _parse_varvals {
my ( @varvals ) = @_;
my ( $vars ) = _preprocess_varvals(@_);
# Config built from parsing the given varvals.
my %config;
# Discover duplicate vars.
my $duplicate_var = 0;
my %duplicates;
# Keep track if item is var or val because each needs special modifications.
my $var; # current variable (e.g. datadir)
my $val; # value for current variable
ITEM:
foreach my $item ( @varvals ) {
# We were passed an undef for the value, or we're dealing with an empty
# line
if ( !defined($item) ) {
$item = '';
}
if ( $item ) {
# Strip leading and trailing whitespace.
$item =~ s/^\s+//;
$item =~ s/\s+$//;
}
if ( !$var ) {
# No var means this item is (should be) the next var in the list.
$var = $item;
# Variable names are usually specified like "log-bin"
# but in SHOW VARIABLES they're all like "log_bin".
$var =~ s/-/_/g;
while ( my ($var, $vals) = each %$vars ) {
my $val = _process_val( pop @$vals );
# If the variable has duplicates, then @$vals will contain
# the rest of the values
if ( @$vals && !$can_be_duplicate{$var} ) {
# The var is a duplicate (in the bad sense, i.e. where user is
# probably unaware that there's two different values for this var
# but only the last is used) if we've seen it already and it cannot
# be duplicated. We don't have its value yet (next loop iter),
# so we set a flag to indicate that we should save the duplicate value.
if ( exists $config{$var} && !$can_be_duplicate{$var} ) {
PTDEBUG && _d("Duplicate var:", $var);
$duplicate_var = 1; # flag on, save all the var's values
# but only the last is used).
PTDEBUG && _d("Duplicate var:", $var);
foreach my $current_val ( map { _process_val($_) } @$vals ) {
push @{$duplicates{$var} ||= []}, $current_val;
}
}
else {
# $var is set so this item should be its value.
my $val = $item;
PTDEBUG && _d("Var:", $var, "val:", $val);
# Avoid crashing on undef comparison. Also, SHOW VARIABLES uses
# blank strings, not NULL/undef.
if ( !defined $val ) {
$val = '';
}
else {
$val =~ s/
\A # Start of value
(['"]) # Opening quote
(.*) # Value
\1 # Closing quote
[\n\r]*\z # End of value
/$2/x;
if ( my ($num, $factor) = $val =~ m/(\d+)([KMGT])b?$/i ) {
# value is a size like 1k, 16M, etc.
my %factor_for = (
k => 1_024,
m => 1_048_576,
g => 1_073_741_824,
t => 1_099_511_627_776,
);
$val = $num * $factor_for{lc $factor};
}
elsif ( $val =~ m/No default/ ) {
$val = '';
}
}
PTDEBUG && _d("Var:", $var, "val:", $val);
if ( $duplicate_var ) {
# Save the var's last value before we overwrite it with this
# current value.
push @{$duplicates{$var}}, $config{$var};
$duplicate_var = 0; # flag off for next var
}
# Save this var-val.
$config{$var} = $val;
$var = undef; # next item should be a var
}
# Save this var-val.
$config{$var} = $val;
}
return \%config, \%duplicates;
}
sub _process_val {
my ($val) = @_;
$val =~ s/
\A # Start of value
(['"]) # Opening quote
(.*) # Value
\1 # Closing quote
[\n\r]*\z # End of value
/$2/x;
if ( my ($num, $factor) = $val =~ m/(\d+)([KMGT])b?$/i ) {
# value is a size like 1k, 16M, etc.
my %factor_for = (
k => 1_024,
m => 1_048_576,
g => 1_073_741_824,
t => 1_099_511_627_776,
);
$val = $num * $factor_for{lc $factor};
}
elsif ( $val =~ m/No default/ ) {
$val = '';
}
return $val;
}
# Sub: _mimic_show_variables
# Make the variables' values mimic SHOW VARIABLES. Different output formats
# list values differently. To make comparisons easier, outputs are made to