provides the desired version for the target database. Any statement in the processed
dump file, that is commented with a version higher than the one supplied, will be stripped.
+The default C<mysql_parser_version> is set to the conservative value of 40000 (MySQL 4.0)
+
Valid version specifiers for C<mysql_parser_version> are listed L<here|SQL::Translator::Utils/parse_mysql_version>
More information about the MySQL comment-syntax: L<http://dev.mysql.com/doc/refman/5.0/en/comments.html>
=cut
use strict;
-use vars qw[ $DEBUG $VERSION $GRAMMAR @EXPORT_OK ];
-$VERSION = '1.59';
+use warnings;
+
+our $VERSION = '1.59';
+
+our $DEBUG;
$DEBUG = 0 unless defined $DEBUG;
use Data::Dumper;
-use Parse::RecDescent;
-use Exporter;
use Storable qw(dclone);
use DBI qw(:sql_types);
-use base qw(Exporter);
+use SQL::Translator::Utils qw/parse_mysql_version ddl_parser_instance/;
-use SQL::Translator::Utils qw/parse_mysql_version/;
+use base qw(Exporter);
+our @EXPORT_OK = qw(parse);
our %type_mapping = ();
-@EXPORT_OK = qw(parse);
+use constant DEFAULT_PARSER_VERSION => 40000;
-# Enable warnings within the Parse::RecDescent module.
-$::RD_ERRORS = 1; # Make sure the parser dies when it encounters an error
-$::RD_WARN = 1; # Enable warnings. This will warn on unused rules &c.
-$::RD_HINT = 1; # Give out hints to help fix problems.
-
-use constant DEFAULT_PARSER_VERSION => 30000;
-
-$GRAMMAR = << 'END_OF_GRAMMAR';
+our $GRAMMAR = << 'END_OF_GRAMMAR';
{
my ( $database_name, %tables, $table_order, @table_comments, %views,
| empty_statement
| <error>
-use : /use/i WORD "$delimiter"
+use : /use/i NAME "$delimiter"
{
$database_name = $item[2];
@table_comments = ();
}
-set : /set/i /[^;]+/ "$delimiter"
+set : /set/i not_delimiter "$delimiter"
{ @table_comments = () }
-drop : /drop/i TABLE /[^;]+/ "$delimiter"
+drop : /drop/i TABLE not_delimiter "$delimiter"
-drop : /drop/i WORD(s) "$delimiter"
+drop : /drop/i NAME(s) "$delimiter"
{ @table_comments = () }
+bit:
+ /(b'[01]{1,64}')/ |
+ /(b"[01]{1,64}")/
+
string :
# MySQL strings, unlike common SQL strings, can be double-quoted or
- # single-quoted, and you can escape the delmiters by doubling (but only the
- # delimiter) or by backslashing.
+ # single-quoted.
- /'(\\.|''|[^\\\'])*'/ |
- /"(\\.|""|[^\\\"])*"/
- # For reference, std sql str: /(?:(?:\')(?:[^\']*(?:(?:\'\')[^\']*)*)(?:\'))//
+ SQSTRING | DQSTRING
nonstring : /[^;\'"]+/
alter_specification : ADD foreign_key_def
{ $return = $item[2] }
-create : CREATE /database/i WORD "$delimiter"
+create : CREATE /database/i NAME "$delimiter"
{ @table_comments = () }
create : CREATE TEMPORARY(?) TABLE opt_if_not_exists(?) table_name '(' create_definition(s /,/) /(,\s*)?\)/ table_option(s?) "$delimiter"
{
my $table_name = $item{'table_name'};
+ die "There is more than one definition for $table_name"
+ if ($tables{$table_name});
+
$tables{ $table_name }{'order'} = ++$table_order;
$tables{ $table_name }{'table_name'} = $table_name;
PROCEDURE : /procedure/i
| /function/i
-create : CREATE replace(?) algorithm(?) /view/i NAME not_delimiter "$delimiter"
+create : CREATE or_replace(?) create_view_option(s?) /view/i NAME /as/i view_select_statement "$delimiter"
{
@table_comments = ();
- my $view_name = $item[5];
- my $sql = join(q{ }, grep { defined and length } $item[1], $item[2]->[0], $item[3]->[0])
- . " $item[4] $item[5] $item[6]";
+ my $view_name = $item{'NAME'};
+ my $select_sql = $item{'view_select_statement'};
+ my $options = $item{'create_view_option(s?)'};
+
+ my $sql = join(q{ },
+ grep { defined and length }
+ map { ref $_ eq 'ARRAY' ? @$_ : $_ }
+ $item{'CREATE'},
+ $item{'or_replace(?)'},
+ $options,
+ $view_name,
+ 'as select',
+ join(', ',
+ map {
+ sprintf('%s%s',
+ $_->{'name'},
+ $_->{'alias'} ? ' as ' . $_->{'alias'} : ''
+ )
+ }
+ @{ $select_sql->{'columns'} || [] }
+ ),
+ ' from ',
+ join(', ',
+ map {
+ sprintf('%s%s',
+ $_->{'name'},
+ $_->{'alias'} ? ' as ' . $_->{'alias'} : ''
+ )
+ }
+ @{ $select_sql->{'from'}{'tables'} || [] }
+ ),
+ $select_sql->{'from'}{'where'}
+ ? 'where ' . $select_sql->{'from'}{'where'}
+ : ''
+ ,
+ );
# Hack to strip database from function calls in SQL
$sql =~ s#`\w+`\.(`\w+`\()##g;
- $views{ $view_name }{'order'} = ++$view_order;
- $views{ $view_name }{'name'} = $view_name;
- $views{ $view_name }{'sql'} = $sql;
+ $views{ $view_name }{'order'} = ++$view_order;
+ $views{ $view_name }{'name'} = $view_name;
+ $views{ $view_name }{'sql'} = $sql;
+ $views{ $view_name }{'options'} = $options;
+ $views{ $view_name }{'select'} = $item{'view_select_statement'};
}
-replace : /or replace/i
+create_view_option : view_algorithm | view_sql_security | view_definer
+
+or_replace : /or replace/i
-algorithm : /algorithm/i /=/ WORD
+view_algorithm : /algorithm/i /=/ WORD
{
$return = "$item[1]=$item[3]";
}
+view_definer : /definer=\S+/i
+
+view_sql_security : /sql \s+ security \s+ (definer|invoker)/ixs
+
not_delimiter : /.*?(?=$delimiter)/is
+view_select_statement : /[(]?/ /select/i view_column_def /from/i view_table_def /[)]?/
+ {
+ $return = {
+ columns => $item{'view_column_def'},
+ from => $item{'view_table_def'},
+ };
+ }
+
+view_column_def : /(.*?)(?=\bfrom\b)/ixs
+ {
+ # split on commas not in parens,
+ # e.g., "concat_ws(\' \', first, last) as first_last"
+ my @tmp = $1 =~ /((?:[^(,]+|\(.*?\))+)/g;
+ my @cols;
+ for my $col ( @tmp ) {
+ my ( $name, $alias ) = map {
+ s/^\s+|\s+$//g;
+ s/[`]//g;
+ $_
+ } split /\s+as\s+/i, $col;
+
+ push @cols, { name => $name, alias => $alias || '' };
+ }
+
+ $return = \@cols;
+ }
+
+not_delimiter : /.*?(?=$delimiter)/is
+
+view_table_def : not_delimiter
+ {
+ my $clause = $item[1];
+ my $where = $1 if $clause =~ s/\bwhere \s+ (.*)//ixs;
+ $clause =~ s/[)]\s*$//;
+
+ my @tables;
+ for my $tbl ( split( /\s*,\s*/, $clause ) ) {
+ my ( $name, $alias ) = split /\s+as\s+/i, $tbl;
+ push @tables, { name => $name, alias => $alias || '' };
+ }
+
+ $return = {
+ tables => \@tables,
+ where => $where || '',
+ };
+ }
+
+view_column_alias : /as/i NAME
+ { $return = $item[2] }
+
create_definition : constraint
| index
| field
$return = $comment;
}
-comment : /\/\*/ /.*?\*\//s
+comment : m{ / \* (?! \!) .*? \* / }xs
{
my $comment = $item[2];
$comment = substr($comment, 0, -2);
$return = $comment;
}
+comment_like_command : m{/\*!(\d+)?}s
+
+comment_end : m{ \* / }xs
+
field_comment : /^\s*(?:#|-{2}).*\n/
{
my $comment = $item[1];
}
-field_comment2 : /comment/i /'.*?'/
- {
- my $comment = $item[2];
- $comment =~ s/^'//;
- $comment =~ s/'$//;
- $return = $comment;
- }
+field_comment2 : /comment/i SQSTRING
+ { $return = $item[2] }
blank : /\s*/
{ $item[2] }
on_update :
- /on update/i 'CURRENT_TIMESTAMP'
+ /on update/i CURRENT_TIMESTAMP
{ $item[2] }
|
/on update/i reference_option
unsigned : /unsigned/i { $return = 0 }
default_val :
- /default/i 'CURRENT_TIMESTAMP'
+ /default/i CURRENT_TIMESTAMP
+ {
+ $return = $item[2];
+ }
+ |
+ /default/i VALUE
{
- $return = \$item[2];
+ $return = $item[2];
+ }
+ |
+ /default/i bit
+ {
+ $item[2] =~ s/b['"]([01]+)['"]/$1/g;
+ $return = $item[2];
}
|
- /default/i /'(?:.*?(?:\\'|''))*.*?'|(?:')?[\w\d:.-]*(?:')?/
+ /default/i /[\w\d:.-]+/
{
- $item[2] =~ s/^\s*'|'\s*$//g;
$return = $item[2];
}
}
}
-foreign_key_def_begin : /constraint/i /foreign key/i WORD
+foreign_key_def_begin : /constraint/i /foreign key/i NAME
{ $return = $item[3] }
|
/constraint/i NAME /foreign key/i
/constraint/i /foreign key/i
{ $return = '' }
|
- /foreign key/i WORD
+ /foreign key/i NAME
{ $return = $item[2] }
|
/foreign key/i
{ $return = '' }
-primary_key_def : primary_key index_name_not_using(?) index_type(?) '(' name_with_opt_paren(s /,/) ')' index_type(?)
+primary_key_def : primary_key index_type(?) '(' name_with_opt_paren(s /,/) ')' index_type(?)
{
$return = {
supertype => 'constraint',
- name => $item[2][0],
type => 'primary_key',
- fields => $item[5],
- options => $item[3][0] || $item[7][0],
+ fields => $item[4],
+ options => $item[2][0] || $item[6][0],
+ };
+ }
+ # In theory, and according to the doc, names should not be allowed here, but
+ # MySQL accept (and ignores) them, so we are not going to be less :)
+ | primary_key index_name_not_using(?) '(' name_with_opt_paren(s /,/) ')' index_type(?)
+ {
+ $return = {
+ supertype => 'constraint',
+ type => 'primary_key',
+ fields => $item[4],
+ options => $item[6][0],
};
}
KEY : /key/i | /index/i
-table_option : /comment/i /=/ /'.*?'/
+table_option : /comment/i /=/ string
{
- my $comment = $item[3];
- $comment =~ s/^'//;
- $comment =~ s/'$//;
- $return = { comment => $comment };
+ $return = { comment => $item[3] };
}
- | /(default )?(charset|character set)/i /\s*=?\s*/ WORD
+ | /(default )?(charset|character set)/i /\s*=?\s*/ NAME
{
$return = { 'CHARACTER SET' => $item[3] };
}
- | /collate/i WORD
+ | /collate/i NAME
{
$return = { 'COLLATE' => $item[2] }
}
{
$return = { $item[1] => $item[4] };
}
- | WORD /\s*=\s*/ MAYBE_QUOTED_WORD
+ | WORD /\s*=\s*/ table_option_value
{
$return = { $item[1] => $item[3] };
}
-MAYBE_QUOTED_WORD: /\w+/
- | /'(\w+)'/
- { $return = $1 }
- | /"(\w+)"/
- { $return = $1 }
+table_option_value : VALUE
+ | NAME
default : /default/i
DOUBLE_QUOTE: '"'
-QUOTED_NAME : BACKTICK /[^`]+/ BACKTICK
- { $item[2] }
- | DOUBLE_QUOTE /[^"]+/ DOUBLE_QUOTE
- { $item[2] }
+SINGLE_QUOTE: "'"
+
+QUOTED_NAME : BQSTRING
+ | SQSTRING
+ | DQSTRING
+
+# MySQL strings, unlike common SQL strings, can have the delmiters
+# escaped either by doubling or by backslashing.
+BQSTRING: BACKTICK <skip: ''> /(?:[^\\`]|``|\\.)*/ BACKTICK
+ { ($return = $item[3]) =~ s/(\\[\\`]|``)/substr($1,1)/ge }
+
+DQSTRING: DOUBLE_QUOTE <skip: ''> /(?:[^\\"]|""|\\.)*/ DOUBLE_QUOTE
+ { ($return = $item[3]) =~ s/(\\[\\"]|"")/substr($1,1)/ge }
+
+SQSTRING: SINGLE_QUOTE <skip: ''> /(?:[^\\']|''|\\.)*/ SINGLE_QUOTE
+ { ($return = $item[3]) =~ s/(\\[\\']|'')/substr($1,1)/ge }
+
NAME: QUOTED_NAME
| /\w+/
-VALUE : /[-+]?\.?\d+(?:[eE]\d+)?/
+VALUE : /[-+]?\d*\.?\d+(?:[eE]\d+)?/
{ $item[1] }
- | /'.*?'/
- {
- # remove leading/trailing quotes
- my $val = $item[1];
- $val =~ s/^['"]|['"]$//g;
- $return = $val;
- }
- | /NULL/
+ | SQSTRING
+ | DQSTRING
+ | /NULL/i
{ 'NULL' }
-CURRENT_TIMESTAMP : /current_timestamp(\(\))?/i
- | /now\(\)/i
- { 'CURRENT_TIMESTAMP' }
+# always a scalar-ref, so that it is treated as a function and not quoted by consumers
+CURRENT_TIMESTAMP :
+ /current_timestamp(\(\))?/i { \'CURRENT_TIMESTAMP' }
+ | /now\(\)/i { \'CURRENT_TIMESTAMP' }
END_OF_GRAMMAR
sub parse {
my ( $translator, $data ) = @_;
- my $parser = Parse::RecDescent->new($GRAMMAR);
+
+ # Enable warnings within the Parse::RecDescent module.
+ # Make sure the parser dies when it encounters an error
+ local $::RD_ERRORS = 1 unless defined $::RD_ERRORS;
+ # Enable warnings. This will warn on unused rules &c.
+ local $::RD_WARN = 1 unless defined $::RD_WARN;
+ # Give out hints to help fix problems.
+ local $::RD_HINT = 1 unless defined $::RD_HINT;
local $::RD_TRACE = $translator->trace ? 1 : undef;
local $DEBUG = $translator->debug;
- unless (defined $parser) {
- return $translator->error("Error instantiating Parse::RecDescent ".
- "instance: Bad grammer");
- }
+ my $parser = ddl_parser_instance('MySQL');
# Preprocess for MySQL-specific and not-before-version comments
# from mysqldump
sql => $result->{procedures}->{$proc_name}->{sql},
);
}
+
my @views = sort {
$result->{views}->{ $a }->{'order'}
<=>
} keys %{ $result->{views} };
for my $view_name ( @views ) {
+ my $view = $result->{'views'}{ $view_name };
+ my @flds = map { $_->{'alias'} || $_->{'name'} }
+ @{ $view->{'select'}{'columns'} || [] };
+ my @from = map { $_->{'alias'} || $_->{'name'} }
+ @{ $view->{'from'}{'tables'} || [] };
+
$schema->add_view(
- name => $view_name,
- sql => $result->{'views'}->{$view_name}->{sql},
+ name => $view_name,
+ sql => $view->{'sql'},
+ order => $view->{'order'},
+ fields => \@flds,
+ tables => \@from,
+ options => $view->{'options'}
);
}
# Takes a field, and returns
sub normalize_field {
my ($field) = @_;
- my ($size, $type, $list, $changed) = @_;
+ my ($size, $type, $list, $unsigned, $changed);
$size = $field->size;
$type = $field->data_type;
$list = $field->extra->{list} || [];
+ $unsigned = defined($field->extra->{unsigned});
if ( !ref $size && $size eq 0 ) {
if ( lc $type eq 'tinyint' ) {
- $changed = $size != 4;
- $size = 4;
+ $changed = $size != 4 - $unsigned;
+ $size = 4 - $unsigned;
}
elsif ( lc $type eq 'smallint' ) {
- $changed = $size != 6;
- $size = 6;
+ $changed = $size != 6 - $unsigned;
+ $size = 6 - $unsigned;
}
elsif ( lc $type eq 'mediumint' ) {
- $changed = $size != 9;
- $size = 9;
+ $changed = $size != 9 - $unsigned;
+ $size = 9 - $unsigned;
}
elsif ( $type =~ /^int(eger)?$/i ) {
- $changed = $size != 11 || $type ne 'int';
+ $changed = $size != 11 - $unsigned || $type ne 'int';
$type = 'int';
- $size = 11;
+ $size = 11 - $unsigned;
}
elsif ( lc $type eq 'bigint' ) {
$changed = $size != 20;