X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=lib%2FSQL%2FTranslator%2FParser%2FMySQL.pm;h=46085e33b0c1e86527a83f00cf9e212f66a221cd;hb=HEAD;hp=5f8d75a92668296b708c9ae6afc8d4e0caff2970;hpb=936e626bc9111bfbe8319a8558fe4c04a58c1008;p=dbsrgits%2FSQL-Translator.git diff --git a/lib/SQL/Translator/Parser/MySQL.pm b/lib/SQL/Translator/Parser/MySQL.pm index 5f8d75a..46085e3 100644 --- a/lib/SQL/Translator/Parser/MySQL.pm +++ b/lib/SQL/Translator/Parser/MySQL.pm @@ -120,6 +120,8 @@ 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. +The default C is set to the conservative value of 40000 (MySQL 4.0) + Valid version specifiers for C are listed L More information about the MySQL comment-syntax: L @@ -129,31 +131,25 @@ More information about the MySQL comment-syntax: L 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, $table_order, @table_comments, %views, @@ -189,32 +185,29 @@ statement : comment | empty_statement | -use : /use/i WORD "$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) "$delimiter" { @table_comments = () } bit: - /(b'[01]+')/ | - /(b"[01]+")/ + /(b'[01]{1,64}')/ | + /(b"[01]{1,64}")/ string : # MySQL strings, unlike common SQL strings, can be double-quoted or - # single-quoted, and you can escape the delmiters by doubling (but only the - # delimiter) or by backslashing. + # single-quoted. - /'(\\.|''|[^\\\'])*'/ | - /"(\\.|""|[^\\\"])*"/ - # For reference, std sql str: /(?:(?:\')(?:[^\']*(?:(?:\'\')[^\']*)*)(?:\'))// + SQSTRING | DQSTRING nonstring : /[^;\'"]+/ @@ -241,7 +234,7 @@ alter : ALTER TABLE table_name alter_specification(s /,/) "$delimiter" 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?) "$delimiter" @@ -333,30 +326,121 @@ create : CREATE PROCEDURE NAME not_delimiter "$delimiter" 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 -algorithm : /algorithm/i /=/ WORD +or_replace : /or replace/i + +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 @@ -371,7 +455,7 @@ comment : /^\s*(?:#|-{2}).*\n/ $return = $comment; } -comment : /\/\*/ /.*?\*\//s +comment : m{ / \* (?! \!) .*? \* / }xs { my $comment = $item[2]; $comment = substr($comment, 0, -2); @@ -379,6 +463,10 @@ comment : /\/\*/ /.*?\*\//s $return = $comment; } +comment_like_command : m{/\*!(\d+)?}s + +comment_end : m{ \* / }xs + field_comment : /^\s*(?:#|-{2}).*\n/ { my $comment = $item[1]; @@ -388,17 +476,9 @@ field_comment : /^\s*(?:#|-{2}).*\n/ } -field_comment2 : /comment/i /'.*?'/ - { - my $comment = $item[2]; - $comment =~ s/^'//; - $comment =~ s/'$//; - $return = $comment; - } - blank : /\s*/ -field : field_comment(s?) field_name data_type field_qualifier(s?) field_comment2(?) reference_definition(?) on_update(?) field_comment(s?) +field : field_comment(s?) field_name data_type field_qualifier(s?) reference_definition(?) on_update(?) field_comment(s?) { my %qualifiers = map { %$_ } @{ $item{'field_qualifier(s?)'} || [] }; if ( my @type_quals = @{ $item{'data_type'}{'qualifiers'} || [] } ) { @@ -409,7 +489,7 @@ field : field_comment(s?) field_name data_type field_qualifier(s?) field_comment ? $qualifiers{'not_null'} : 1; delete $qualifiers{'not_null'}; - my @comments = ( @{ $item[1] }, @{ $item[5] }, @{ $item[8] } ); + my @comments = ( @{ $item[1] }, (exists $qualifiers{comment} ? delete $qualifiers{comment} : ()) , @{ $item[7] } ); $return = { supertype => 'field', @@ -495,6 +575,13 @@ field_qualifier : KEY } } +field_qualifier : /comment/i string + { + $return = { + comment => $item[2], + } + } + reference_definition : /references/i table_name parens_field_list(?) match_type(?) on_delete(?) on_update(?) { $return = { @@ -515,7 +602,7 @@ on_delete : /on delete/i reference_option { $item[2] } on_update : - /on update/i 'CURRENT_TIMESTAMP' + /on update/i CURRENT_TIMESTAMP { $item[2] } | /on update/i reference_option @@ -585,14 +672,13 @@ not_null : /not/i /null/i unsigned : /unsigned/i { $return = 0 } default_val : - /default/i 'CURRENT_TIMESTAMP' + /default/i CURRENT_TIMESTAMP { - $return = \$item[2]; + $return = $item[2]; } | - /default/i string + /default/i VALUE { - $item[2] =~ s/^\s*'|'\s*$//g or $item[2] =~ s/^\s*"|"\s*$//g; $return = $item[2]; } | @@ -602,7 +688,12 @@ default_val : $return = $item[2]; } | - /default/i /(?:')?[\w\d:.-]*(?:')?/ + /default/i /[\w\d:.-]+/ + { + $return = $item[2]; + } + | + /default/i NAME # column value, allowed in MariaDB { $return = $item[2]; } @@ -614,8 +705,31 @@ primary_key : /primary/i /key/i { 1 } constraint : primary_key_def | unique_key_def | foreign_key_def + | check_def | +expr : /[^)]* \( [^)]+ \) [^)]*/x # parens, balanced one deep + | /[^)]+/ + +check_def : check_def_begin '(' expr ')' + { + $return = { + supertype => 'constraint', + type => 'check', + name => $item[1], + expression => $item[3], + } + } + +check_def_begin : /constraint/i /check/i NAME + { $return = $item[3] } + | + /constraint/i NAME /check/i + { $return = $item[2] } + | + /constraint/i /check/i + { $return = '' } + foreign_key_def : foreign_key_def_begin parens_field_list reference_definition { $return = { @@ -627,7 +741,7 @@ foreign_key_def : foreign_key_def_begin parens_field_list reference_definition } } -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 @@ -636,7 +750,7 @@ foreign_key_def_begin : /constraint/i /foreign key/i WORD /constraint/i /foreign key/i { $return = '' } | - /foreign key/i WORD + /foreign key/i NAME { $return = $item[2] } | /foreign key/i @@ -717,18 +831,15 @@ UNIQUE : /unique/i 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] } } @@ -736,16 +847,13 @@ table_option : /comment/i /=/ /'.*?'/ { $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 @@ -769,42 +877,55 @@ BACKTICK : '`' DOUBLE_QUOTE: '"' -QUOTED_NAME : BACKTICK /[^`]+/ BACKTICK - { $item[2] } - | DOUBLE_QUOTE /[^"]+/ DOUBLE_QUOTE - { $item[2] } +SINGLE_QUOTE: "'" + +QUOTED_NAME : BQSTRING + | SQSTRING + | DQSTRING + +# MySQL strings, unlike common SQL strings, can have the delmiters +# escaped either by doubling or by backslashing. +BQSTRING: BACKTICK /(?:[^\\`]|``|\\.)*/ BACKTICK + { ($return = $item[3]) =~ s/(\\[\\`]|``)/substr($1,1)/ge } + +DQSTRING: DOUBLE_QUOTE /(?:[^\\"]|""|\\.)*/ DOUBLE_QUOTE + { ($return = $item[3]) =~ s/(\\[\\"]|"")/substr($1,1)/ge } + +SQSTRING: SINGLE_QUOTE /(?:[^\\']|''|\\.)*/ SINGLE_QUOTE + { ($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 @@ -924,6 +1045,7 @@ sub parse { name => $cdata->{'name'}, type => $cdata->{'type'}, fields => $cdata->{'fields'}, + expression => $cdata->{'expression'}, reference_table => $cdata->{'reference_table'}, reference_fields => $cdata->{'reference_fields'}, match_type => $cdata->{'match_type'} || '', @@ -952,6 +1074,7 @@ sub parse { sql => $result->{procedures}->{$proc_name}->{sql}, ); } + my @views = sort { $result->{views}->{ $a }->{'order'} <=> @@ -959,9 +1082,19 @@ sub parse { } 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'} ); }