X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FSQL%2FTranslator%2FParser%2FMySQL.pm;h=4955d02424fc5eebb0fe28a9c29cb938606bc954;hb=478f608d9028508396da37bb5df10b3057b96981;hp=c0a214467f09376264d0908a16ec60b99c32496d;hpb=a82fa2cb89d323c12900a771c8628ff130654d10;p=dbsrgits%2FSQL-Translator.git diff --git a/lib/SQL/Translator/Parser/MySQL.pm b/lib/SQL/Translator/Parser/MySQL.pm index c0a2144..4955d02 100644 --- a/lib/SQL/Translator/Parser/MySQL.pm +++ b/lib/SQL/Translator/Parser/MySQL.pm @@ -1,9 +1,9 @@ package SQL::Translator::Parser::MySQL; # ------------------------------------------------------------------- -# $Id: MySQL.pm,v 1.44 2004-03-01 17:39:22 kycl4rk Exp $ +# $Id$ # ------------------------------------------------------------------- -# Copyright (C) 2002-4 SQLFairy Authors +# Copyright (C) 2002-2009 SQLFairy Authors # # This program is free software; you can redistribute it and/or # modify it under the terms of the GNU General Public License as @@ -101,9 +101,12 @@ Here's the word from the MySQL site table_options: TYPE = {BDB | HEAP | ISAM | InnoDB | MERGE | MRG_MYISAM | MYISAM } + or ENGINE = {BDB | HEAP | ISAM | InnoDB | MERGE | MRG_MYISAM | MYISAM } or AUTO_INCREMENT = # or AVG_ROW_LENGTH = # + or [ DEFAULT ] CHARACTER SET charset_name or CHECKSUM = {0 | 1} + or COLLATE collation_name or COMMENT = "string" or MAX_ROWS = # or MIN_ROWS = # @@ -117,18 +120,49 @@ Here's the word from the MySQL site or DATA DIRECTORY="absolute path to directory" or INDEX DIRECTORY="absolute path to directory" + +A subset of the ALTER TABLE syntax that allows addition of foreign keys: + + ALTER [IGNORE] TABLE tbl_name alter_specification [, alter_specification] ... + + alter_specification: + ADD [CONSTRAINT [symbol]] + FOREIGN KEY [index_name] (index_col_name,...) + [reference_definition] + +A subset of INSERT that we ignore: + + INSERT anything + +=head1 ARGUMENTS + +This parser takes a single optional parser_arg C, which +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. + +Valid version specifiers for C are listed L + +More information about the MySQL comment-syntax: L + + =cut use strict; -use vars qw[ $DEBUG $VERSION $GRAMMAR @EXPORT_OK ]; -$VERSION = sprintf "%d.%02d", q$Revision: 1.44 $ =~ /(\d+)\.(\d+)/; +use vars qw[ $DEBUG $GRAMMAR @EXPORT_OK ]; $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/; + +our %type_mapping = ( +); + @EXPORT_OK = qw(parse); # Enable warnings within the Parse::RecDescent module. @@ -136,10 +170,13 @@ $::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. $::RD_HINT = 1; # Give out hints to help fix problems. -$GRAMMAR = q! +use constant DEFAULT_PARSER_VERSION => 30000; + +$GRAMMAR = << 'END_OF_GRAMMAR'; { - my ( %tables, $table_order, @table_comments ); + my ( $database_name, %tables, $table_order, @table_comments, %views, $view_order, %procedures, $proc_order ); + my $delimiter = ';'; } # @@ -148,7 +185,9 @@ $GRAMMAR = q! # won't cause the failure needed to know that the parse, as a whole, # failed. -ky # -startrule : statement(s) eofile { \%tables } +startrule : statement(s) eofile { + { tables => \%tables, database_name => $database_name, views => \%views, procedures =>\%procedures } +} eofile : /^\Z/ @@ -157,23 +196,64 @@ statement : comment | set | drop | create + | alter + | insert + | delimiter + | empty_statement | -use : /use/i WORD ';' - { @table_comments = () } +use : /use/i WORD "$delimiter" + { + $database_name = $item[2]; + @table_comments = (); + } -set : /set/i /[^;]+/ ';' +set : /set/i /[^;]+/ "$delimiter" { @table_comments = () } -drop : /drop/i TABLE /[^;]+/ ';' +drop : /drop/i TABLE /[^;]+/ "$delimiter" -drop : /drop/i WORD(s) ';' +drop : /drop/i WORD(s) "$delimiter" { @table_comments = () } -create : CREATE /database/i WORD ';' +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. + + /'(\\.|''|[^\\\'])*'/ | + /"(\\.|""|[^\\\"])*"/ + # For reference, std sql str: /(?:(?:\')(?:[^\']*(?:(?:\'\')[^\']*)*)(?:\'))// + +nonstring : /[^;\'"]+/ + +statement_body : (string | nonstring)(s?) + +insert : /insert/i statement_body "$delimiter" + +delimiter : /delimiter/i /[\S]+/ + { $delimiter = $item[2] } + +empty_statement : "$delimiter" + +alter : ALTER TABLE table_name alter_specification(s /,/) "$delimiter" + { + my $table_name = $item{'table_name'}; + die "Cannot ALTER table '$table_name'; it does not exist" + unless $tables{ $table_name }; + for my $definition ( @{ $item[4] } ) { + $definition->{'extra'}->{'alter'} = 1; + push @{ $tables{ $table_name }{'constraints'} }, $definition; + } + } + +alter_specification : ADD foreign_key_def + { $return = $item[2] } + +create : CREATE /database/i WORD "$delimiter" { @table_comments = () } -create : CREATE TEMPORARY(?) TABLE opt_if_not_exists(?) table_name '(' create_definition(s /,/) ')' table_option(s?) ';' +create : CREATE TEMPORARY(?) TABLE opt_if_not_exists(?) table_name '(' create_definition(s /,/) /(,\s*)?\)/ table_option(s?) "$delimiter" { my $table_name = $item{'table_name'}; $tables{ $table_name }{'order'} = ++$table_order; @@ -210,7 +290,15 @@ create : CREATE TEMPORARY(?) TABLE opt_if_not_exists(?) table_name '(' create_de } if ( my @options = @{ $item{'table_option(s?)'} } ) { - $tables{ $table_name }{'table_options'} = \@options; + for my $option ( @options ) { + my ( $key, $value ) = each %$option; + if ( $key eq 'comment' ) { + push @{ $tables{ $table_name }{'comments'} }, $value; + } + else { + push @{ $tables{ $table_name }{'table_options'} }, $option; + } + } } 1; @@ -218,18 +306,60 @@ create : CREATE TEMPORARY(?) TABLE opt_if_not_exists(?) table_name '(' create_de opt_if_not_exists : /if not exists/i -create : CREATE UNIQUE(?) /(index|key)/i index_name /on/i table_name '(' field_name(s /,/) ')' ';' +create : CREATE UNIQUE(?) /(index|key)/i index_name /on/i table_name '(' field_name(s /,/) ')' "$delimiter" { @table_comments = (); push @{ $tables{ $item{'table_name'} }{'indices'} }, { name => $item[4], - type => $item[2] ? 'unique' : 'normal', + type => $item[2][0] ? 'unique' : 'normal', fields => $item[8], } ; } +create : CREATE /trigger/i NAME not_delimiter "$delimiter" + { + @table_comments = (); + } + +create : CREATE PROCEDURE NAME not_delimiter "$delimiter" + { + @table_comments = (); + my $func_name = $item[3]; + my $owner = ''; + my $sql = "$item[1] $item[2] $item[3] $item[4]"; + + $procedures{ $func_name }{'order'} = ++$proc_order; + $procedures{ $func_name }{'name'} = $func_name; + $procedures{ $func_name }{'owner'} = $owner; + $procedures{ $func_name }{'sql'} = $sql; + } + +PROCEDURE : /procedure/i + | /function/i + +create : CREATE algorithm /view/i NAME not_delimiter "$delimiter" + { + @table_comments = (); + my $view_name = $item[4]; + my $sql = "$item[1] $item[2] $item[3] $item[4] $item[5]"; + + # 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; + } + +algorithm : /algorithm/i /=/ WORD + { + $return = "$item[1]=$item[3]"; + } + +not_delimiter : /.*?(?=$delimiter)/is + create_definition : constraint | index | field @@ -242,9 +372,16 @@ comment : /^\s*(?:#|-{2}).*\n/ $comment =~ s/^\s*(#|--)\s*//; $comment =~ s/\s*$//; $return = $comment; - push @table_comments, $comment; } +comment : /\/\*/ /.*?\*\//s + { + my $comment = $item[2]; + $comment = substr($comment, 0, -2); + $comment =~ s/^\s*|\s*$//g; + $return = $comment; + } + field_comment : /^\s*(?:#|-{2}).*\n/ { my $comment = $item[1]; @@ -253,9 +390,18 @@ field_comment : /^\s*(?:#|-{2}).*\n/ $return = $comment; } + +field_comment2 : /comment/i /'.*?'/ + { + my $comment = $item[2]; + $comment =~ s/^'//; + $comment =~ s/'$//; + $return = $comment; + } + blank : /\s*/ -field : field_comment(s?) field_name data_type field_qualifier(s?) reference_definition(?) field_comment(s?) +field : field_comment(s?) field_name data_type field_qualifier(s?) field_comment2(?) reference_definition(?) on_update(?) field_comment(s?) { my %qualifiers = map { %$_ } @{ $item{'field_qualifier(s?)'} || [] }; if ( my @type_quals = @{ $item{'data_type'}{'qualifiers'} || [] } ) { @@ -266,7 +412,7 @@ field : field_comment(s?) field_name data_type field_qualifier(s?) reference_def ? $qualifiers{'not_null'} : 1; delete $qualifiers{'not_null'}; - my @comments = ( @{ $item[1] }, @{ $item[6] } ); + my @comments = ( @{ $item[1] }, @{ $item[5] }, @{ $item[8] } ); $return = { supertype => 'field', @@ -317,22 +463,50 @@ field_qualifier : unsigned } } -field_qualifier : /character set/i WORD +field_qualifier : /character set/i WORD + { + $return = { + 'CHARACTER SET' => $item[2], + } + } + +field_qualifier : /collate/i WORD + { + $return = { + COLLATE => $item[2], + } + } + +field_qualifier : /on update/i CURRENT_TIMESTAMP + { + $return = { + 'ON UPDATE' => $item[2], + } + } + +field_qualifier : /unique/i KEY(?) { $return = { - character_set => $item[2], + is_unique => 1, } } -reference_definition : /references/i table_name parens_field_list(?) match_type(?) on_delete_do(?) on_update_do(?) +field_qualifier : KEY + { + $return = { + has_index => 1, + } + } + +reference_definition : /references/i table_name parens_field_list(?) match_type(?) on_delete(?) on_update(?) { $return = { type => 'foreign_key', reference_table => $item[2], reference_fields => $item[3][0], match_type => $item[4][0], - on_delete_do => $item[5][0], - on_update_do => $item[6][0], + on_delete => $item[5][0], + on_update => $item[6][0], } } @@ -340,10 +514,14 @@ match_type : /match full/i { 'full' } | /match partial/i { 'partial' } -on_delete_do : /on delete/i reference_option +on_delete : /on delete/i reference_option { $item[2] } -on_update_do : /on update/i reference_option +on_update : + /on update/i 'CURRENT_TIMESTAMP' + { $item[2] } + | + /on update/i reference_option { $item[2] } reference_option: /restrict/i | @@ -355,6 +533,7 @@ reference_option: /restrict/i | index : normal_index | fulltext_index + | spatial_index | table_name : NAME @@ -378,42 +557,6 @@ data_type : WORD parens_value_list(s?) type_qualifier(s?) $list = []; } - unless ( @{ $size || [] } ) { - if ( lc $type eq 'tinyint' ) { - $size = 4; - } - elsif ( lc $type eq 'smallint' ) { - $size = 6; - } - elsif ( lc $type eq 'mediumint' ) { - $size = 9; - } - elsif ( $type =~ /^int(eger)?$/ ) { - $type = 'int'; - $size = 11; - } - elsif ( lc $type eq 'bigint' ) { - $size = 20; - } - elsif ( - lc $type =~ /(float|double|decimal|numeric|real|fixed|dec)/ - ) { - $size = [8,2]; - } - } - - if ( $type =~ /^tiny(text|blob)$/i ) { - $size = 255; - } - elsif ( $type =~ /^(blob|text)$/i ) { - $size = 65_535; - } - elsif ( $type =~ /^medium(blob|text)$/i ) { - $size = 16_777_215; - } - elsif ( $type =~ /^long(blob|text)$/i ) { - $size = 4_294_967_295; - } $return = { type => $type, @@ -436,7 +579,11 @@ field_type : WORD create_index : /create/i /index/i -not_null : /not/i /null/i { $return = 0 } +not_null : /not/i /null/i + { $return = 0 } + | + /null/i + { $return = 1 } unsigned : /unsigned/i { $return = 0 } @@ -446,7 +593,13 @@ unsigned : /unsigned/i { $return = 0 } # $return = $item[2]; # } -default_val : /default/i /'(?:.*?\\')*.*?'|(?:')?[\w\d:.-]*(?:')?/ +default_val : + /default/i 'CURRENT_TIMESTAMP' + { + $return = \$item[2]; + } + | + /default/i /'(?:.*?\\')*.*?'|(?:')?[\w\d:.-]*(?:')?/ { $item[2] =~ s/^\s*'|'\s*$//g; $return = $item[2]; @@ -472,10 +625,16 @@ foreign_key_def : foreign_key_def_begin parens_field_list reference_definition } } -foreign_key_def_begin : /constraint/i /foreign key/i +foreign_key_def_begin : /constraint/i /foreign key/i WORD + { $return = $item[3] } + | + /constraint/i NAME /foreign key/i + { $return = $item[2] } + | + /constraint/i /foreign key/i { $return = '' } | - /constraint/i WORD /foreign key/i + /foreign key/i WORD { $return = $item[2] } | /foreign key/i @@ -521,17 +680,52 @@ fulltext_index : /fulltext/i KEY(?) index_name(?) '(' name_with_opt_paren(s /,/) } } +spatial_index : /spatial/i KEY(?) index_name(?) '(' name_with_opt_paren(s /,/) ')' + { + $return = { + supertype => 'index', + type => 'spatial', + name => $item{'index_name(?)'}[0], + fields => $item[5], + } + } + name_with_opt_paren : NAME parens_value_list(s?) { $item[2][0] ? "$item[1]($item[2][0][0])" : $item[1] } -UNIQUE : /unique/i { 1 } +UNIQUE : /unique/i KEY : /key/i | /index/i -table_option : WORD /\s*=\s*/ WORD +table_option : /comment/i /=/ /'.*?'/ + { + my $comment = $item[3]; + $comment =~ s/^'//; + $comment =~ s/'$//; + $return = { comment => $comment }; + } + | /(default )?(charset|character set)/i /\s*=?\s*/ WORD + { + $return = { 'CHARACTER SET' => $item[3] }; + } + | /collate/i WORD + { + $return = { 'COLLATE' => $item[2] } + } + | /union/i /\s*=\s*/ '(' table_name(s /,/) ')' + { + $return = { $item[1] => $item[4] }; + } + | WORD /\s*=\s*/ WORD { $return = { $item[1] => $item[3] }; } + +default : /default/i + +ADD : /add/i + +ALTER : /alter/i CREATE : /create/i @@ -545,7 +739,9 @@ DIGITS : /\d+/ COMMA : ',' -NAME : "`" /\w+/ "`" +BACKTICK : '`' + +NAME : BACKTICK /[^`]+/ BACKTICK { $item[2] } | /\w+/ { $item[1] } @@ -562,13 +758,16 @@ VALUE : /[-+]?\.?\d+(?:[eE]\d+)?/ | /NULL/ { 'NULL' } -!; +CURRENT_TIMESTAMP : /current_timestamp(\(\))?/i + | /now\(\)/i + { 'CURRENT_TIMESTAMP' } + +END_OF_GRAMMAR # ------------------------------------------------------------------- sub parse { my ( $translator, $data ) = @_; my $parser = Parse::RecDescent->new($GRAMMAR); - local $::RD_TRACE = $translator->trace ? 1 : undef; local $DEBUG = $translator->debug; @@ -576,18 +775,28 @@ sub parse { return $translator->error("Error instantiating Parse::RecDescent ". "instance: Bad grammer"); } + + # Preprocess for MySQL-specific and not-before-version comments from mysqldump + my $parser_version = + parse_mysql_version ($translator->parser_args->{mysql_parser_version}, 'mysql') + || DEFAULT_PARSER_VERSION; + while ( $data =~ s#/\*!(\d{5})?(.*?)\*/#($1 && $1 > $parser_version ? '' : $2)#es ) {} my $result = $parser->startrule($data); return $translator->error( "Parse failed." ) unless defined $result; - warn Dumper( $result ) if $DEBUG; + warn "Parse result:".Dumper( $result ) if $DEBUG; my $schema = $translator->schema; + $schema->name($result->{'database_name'}) if $result->{'database_name'}; + my @tables = sort { - $result->{ $a }->{'order'} <=> $result->{ $b }->{'order'} - } keys %{ $result }; + $result->{'tables'}{ $a }{'order'} + <=> + $result->{'tables'}{ $b }{'order'} + } keys %{ $result->{'tables'} }; for my $table_name ( @tables ) { - my $tdata = $result->{ $table_name }; + my $tdata = $result->{tables}{ $table_name }; my $table = $schema->add_table( name => $tdata->{'table_name'}, ) or die $schema->error; @@ -614,20 +823,28 @@ sub parse { $table->primary_key( $field->name ) if $fdata->{'is_primary_key'}; - for my $qual ( qw[ binary unsigned zerofill list ] ) { + for my $qual ( qw[ binary unsigned zerofill list collate ], + 'character set', 'on update' ) { if ( my $val = $fdata->{ $qual } || $fdata->{ uc $qual } ) { next if ref $val eq 'ARRAY' && !@$val; $field->extra( $qual, $val ); } } - if ( $field->data_type =~ /(set|enum)/i && !$field->size ) { - my %extra = $field->extra; - my $longest = 0; - for my $len ( map { length } @{ $extra{'list'} || [] } ) { - $longest = $len if $len > $longest; - } - $field->size( $longest ) if $longest; + if ( $fdata->{'has_index'} ) { + $table->add_index( + name => '', + type => 'NORMAL', + fields => $fdata->{'name'}, + ) or die $table->error; + } + + if ( $fdata->{'is_unique'} ) { + $table->add_constraint( + name => '', + type => 'UNIQUE', + fields => $fdata->{'name'}, + ) or die $table->error; } for my $cdata ( @{ $fdata->{'constraints'} } ) { @@ -635,6 +852,7 @@ sub parse { $cdata->{'fields'} ||= [ $field->name ]; push @{ $tdata->{'constraints'} }, $cdata; } + } for my $idata ( @{ $tdata->{'indices'} || [] } ) { @@ -646,7 +864,19 @@ sub parse { } if ( my @options = @{ $tdata->{'table_options'} || [] } ) { - $table->options( \@options ) or die $table->error; + my @cleaned_options; + my @ignore_opts = $translator->parser_args->{ignore_opts}?split(/,/,$translator->parser_args->{ignore_opts}):(); + if (@ignore_opts) { + my $ignores = { map { $_ => 1 } @ignore_opts }; + foreach my $option (@options) { + # make sure the option isn't in ignore list + my ($option_key) = keys %$option; + push(@cleaned_options, $option) unless (exists $ignores->{$option_key}); + } + } else { + @cleaned_options = @options; + } + $table->options( \@cleaned_options ) or die $table->error; } for my $cdata ( @{ $tdata->{'constraints'} || [] } ) { @@ -657,15 +887,118 @@ sub parse { reference_table => $cdata->{'reference_table'}, reference_fields => $cdata->{'reference_fields'}, match_type => $cdata->{'match_type'} || '', - on_delete => $cdata->{'on_delete_do'}, - on_update => $cdata->{'on_update_do'}, + on_delete => $cdata->{'on_delete'} || $cdata->{'on_delete_do'}, + on_update => $cdata->{'on_update'} || $cdata->{'on_update_do'}, ) or die $table->error; } + + # After the constrains and PK/idxs have been created, we normalize fields + normalize_field($_) for $table->get_fields; + } + + my @procedures = sort { + $result->{procedures}->{ $a }->{'order'} <=> $result->{procedures}->{ $b }->{'order'} + } keys %{ $result->{procedures} }; + foreach my $proc_name (@procedures) { + $schema->add_procedure( + name => $proc_name, + owner => $result->{procedures}->{$proc_name}->{owner}, + sql => $result->{procedures}->{$proc_name}->{sql}, + ); + } + + my @views = sort { + $result->{views}->{ $a }->{'order'} <=> $result->{views}->{ $b }->{'order'} + } keys %{ $result->{views} }; + foreach my $view_name (keys %{ $result->{views} }) { + $schema->add_view( + name => $view_name, + sql => $result->{views}->{$view_name}->{sql}, + ); } return 1; } +# Takes a field, and returns +sub normalize_field { + my ($field) = @_; + my ($size, $type, $list, $changed) = @_; + + $size = $field->size; + $type = $field->data_type; + $list = $field->extra->{list} || []; + + if ( !ref $size && $size eq 0 ) { + if ( lc $type eq 'tinyint' ) { + $changed = $size != 4; + $size = 4; + } + elsif ( lc $type eq 'smallint' ) { + $changed = $size != 6; + $size = 6; + } + elsif ( lc $type eq 'mediumint' ) { + $changed = $size != 9; + $size = 9; + } + elsif ( $type =~ /^int(eger)?$/i ) { + $changed = $size != 11 || $type ne 'int'; + $type = 'int'; + $size = 11; + } + elsif ( lc $type eq 'bigint' ) { + $changed = $size != 20; + $size = 20; + } + elsif ( lc $type =~ /(float|double|decimal|numeric|real|fixed|dec)/ ) { + my $old_size = (ref $size || '') eq 'ARRAY' ? $size : []; + $changed = @$old_size != 2 || $old_size->[0] != 8 || $old_size->[1] != 2; + $size = [8,2]; + } + } + + if ( $type =~ /^tiny(text|blob)$/i ) { + $changed = $size != 255; + $size = 255; + } + elsif ( $type =~ /^(blob|text)$/i ) { + $changed = $size != 65_535; + $size = 65_535; + } + elsif ( $type =~ /^medium(blob|text)$/i ) { + $changed = $size != 16_777_215; + $size = 16_777_215; + } + elsif ( $type =~ /^long(blob|text)$/i ) { + $changed = $size != 4_294_967_295; + $size = 4_294_967_295; + } + if ( $field->data_type =~ /(set|enum)/i && !$field->size ) { + my %extra = $field->extra; + my $longest = 0; + for my $len ( map { length } @{ $extra{'list'} || [] } ) { + $longest = $len if $len > $longest; + } + $changed = 1; + $size = $longest if $longest; + } + + + if ($changed) { + # We only want to clone the field, not *everything* + { local $field->{table} = undef; + $field->parsed_field(dclone($field)); + $field->parsed_field->{table} = $field->table; + } + $field->size($size); + $field->data_type($type); + $field->sql_data_type( $type_mapping{lc $type} ) if exists $type_mapping{lc $type}; + $field->extra->{list} = $list if @$list; + } +} + + 1; # ------------------------------------------------------------------- @@ -677,11 +1010,11 @@ sub parse { =head1 AUTHOR -Ken Y. Clark Ekclark@cpan.orgE, +Ken Youens-Clark Ekclark@cpan.orgE, Chris Mungall Ecjm@fruitfly.orgE. =head1 SEE ALSO -perl(1), Parse::RecDescent, SQL::Translator::Schema. +Parse::RecDescent, SQL::Translator::Schema. =cut