#!/usr/bin/perl -w # mysql2pgsql # MySQL to PostgreSQL dump file converter # # For usage: perl mysql2pgsql.perl --help # # ddl statments are changed but none or only minimal real data # formatting are done. # data consistency is up to the DBA. # # (c) 2004-2007 Jose M Duarte and Joseph Speigle ... gborg # # (c) 2000-2004 Maxim Rudensky # (c) 2000 Valentine Danilchuk # All rights reserved. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions # are met: # 1. Redistributions of source code must retain the above copyright # notice, this list of conditions and the following disclaimer. # 2. Redistributions in binary form must reproduce the above copyright # notice, this list of conditions and the following disclaimer in the # documentation and/or other materials provided with the distribution. # 3. All advertising materials mentioning features or use of this software # must display the following acknowledgement: # This product includes software developed by the Max Rudensky # and its contributors. # 4. Neither the name of the author nor the names of its contributors # may be used to endorse or promote products derived from this software # without specific prior written permission. # THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND # ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE # FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL # DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS # OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) # HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT # LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY # OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF # SUCH DAMAGE. use Getopt::Long; use POSIX; use strict; use warnings; # main sections # ------------- # 1 variable declarations # 2 subroutines # 3 get commandline options and specify help statement # 4 loop through file and process # 5. print_plpgsql function prototype ################################################################# # 1. variable declarations ################################################################# # command line options my( $ENC_IN, $ENC_OUT, $PRESERVE_CASE, $HELP, $DEBUG, $SCHEMA, $LOWERCASE, $CHAR2VARCHAR, $NODROP, $SEP_FILE, $opt_debug, $opt_help, $opt_schema, $opt_preserve_case, $opt_char2varchar, $opt_nodrop, $opt_sepfile, $opt_enc_in, $opt_enc_out ); # variables for constructing pre-create-table entities my $pre_create_sql=''; # comments, 'enum' constraints preceding create table statement my $auto_increment_seq= ''; # so we can easily substitute it if we need a default value my $create_sql=''; # all the datatypes in the create table section my $post_create_sql=''; # create indexes, foreign keys, table comments my $function_create_sql = ''; # for the set (function,trigger) and CURRENT_TIMESTAMP ( function,trigger ) # constraints my ($type, $column_valuesStr, @column_values, $value ); my %constraints=(); # holds values constraints used to emulate mysql datatypes (e.g. year, set) # datatype conversion variables my ( $index,$seq); my ( $column_name, $col, $quoted_column); my ( @year_holder, $year, $constraint_table_name); my $table=""; # table_name for create sql statements my $table_no_quotes=""; # table_name for create sql statements my $sl = '^\s+\w+\s+'; # matches the column name my $tables_first_timestamp_column= 1; # decision to print warnings about default_timestamp not being in postgres my $mysql_numeric_datatypes = "TINYINT|SMALLINT|MEDIUMINT|INT|INTEGER|BIGINT|REAL|DOUBLE|FLOAT|DECIMAL|NUMERIC"; my $mysql_datetime_datatypes = "|DATE|TIME|TIMESTAMP|DATETIME|YEAR"; my $mysql_text_datatypes = "CHAR|VARCHAR|BINARY|VARBINARY|TINYBLOB|BLOB|MEDIUMBLOB|LONGBLOB|TINYTEXT|TEXT|MEDIUMTEXT|LONGTEXT|ENUM|SET"; my $mysql_datatypesStr = $mysql_numeric_datatypes . "|". $mysql_datetime_datatypes . "|". $mysql_text_datatypes ; # handling INSERT INTO statements my $rowRe = qr{ \( # opening parens ( # (start capture) (?: # (start group) ' # string start [^'\\]* # up to string-end or backslash (escape) (?: # (start group) \\. # gobble escaped character [^'\\]* # up to string-end of backslash )* # (end group, repeat zero or more) ' # string end | # (OR) .*? # everything else (not strings) )* # (end group, repeat zero or more) ) # (end capture) \) # closing parent }x; my ($insert_table, $valueString); # ######################################################## # 2. subroutines # # get_identifier # print_post_create_sql() # quote_and_lc() # make_plpgsql($table,$column_name) -- at end of file ######################################################## # returns an identifier with the given suffix doing controlled # truncation if necessary sub get_identifier($$$) { my ($table, $col, $suffix) = @_; my $name = ''; $table=~s/\"//g; # make sure that $table doesn't have quotes so we don't end up with redundant quoting # in the case of multiple columns my @cols = split(/,/,$col); $col =~ s/,//g; # in case all columns together too long we have to truncate them if (length($col) > 55) { my $totaltocut = length($col)-55; my $tocut = ceil($totaltocut / @cols); @cols = map {substr($_,0,abs(length($_)-$tocut))} @cols; $col=""; foreach (@cols){ $col.=$_; } } my $max_table_length = 63 - length("_${col}_$suffix"); if (length($table) > $max_table_length) { $table = substr($table, length($table) - $max_table_length, $max_table_length); } return quote_and_lc("${table}_${col}_${suffix}"); } # # # called when we encounter next CREATE TABLE statement # also called at EOF to print out for last table # prints comments, indexes, foreign key constraints (the latter 2 possibly to a separate file) sub print_post_create_sql() { my ( @create_idx_comments_constraints_commandsArr, $stmts, $table_field_combination); my %stmts; # loop to check for duplicates in $post_create_sql # Needed because of duplicate key declarations ( PRIMARY KEY and KEY), auto_increment columns @create_idx_comments_constraints_commandsArr = split(';\n?', $post_create_sql); if ($SEP_FILE) { open(SEP_FILE, ">>:encoding($ENC_OUT)", $SEP_FILE) or die "Unable to open $SEP_FILE for output: $!\n"; } foreach (@create_idx_comments_constraints_commandsArr) { if (m/CREATE INDEX "*(\S+)"*\s/i) { # CREATE INDEX korean_english_wordsize_idx ON korean_english USING btree (wordsize); $table_field_combination = $1; # if this particular table_field_combination was already used do not print the statement: if ($SEP_FILE) { print SEP_FILE "$_;\n" if !defined($stmts{$table_field_combination}); } else { print OUT "$_;\n" if !defined($stmts{$table_field_combination}); } $stmts{$table_field_combination} = 1; } elsif (m/COMMENT/i) { # COMMENT ON object IS 'text'; but comment may be part of table name so use 'elsif' print OUT "$_;\n" } else { # foreign key constraint or comments (those preceded by -- ) if ($SEP_FILE) { print SEP_FILE "$_;\n"; } else { print OUT "$_;\n" } } } if ($SEP_FILE) { close SEP_FILE; } $post_create_sql=''; # empty %constraints for next " create table" statement } # quotes a string or a multicolumn string (comma separated) # and optionally lowercase (if LOWERCASE is set) # lowercase .... if user wants default postgres behavior # quotes .... to preserve keywords and to preserve case when case-sensitive tables are to be used sub quote_and_lc($) { my $col = shift; if ($LOWERCASE) { $col = lc($col); } if ($col =~ m/,/) { my @cols = split(/,\s?/, $col); @cols = map {"\"$_\""} @cols; return join(', ', @cols); } else { return "\"$col\""; } } ######################################################## # 3. get commandline options and maybe print help ######################################################## GetOptions("help", "debug"=> \$opt_debug, "schema=s" => \$SCHEMA, "preserve_case" => \$opt_preserve_case, "char2varchar" => \$opt_char2varchar, "nodrop" => \$opt_nodrop, "sepfile=s" => \$opt_sepfile, "enc_in=s" => \$opt_enc_in, "enc_out=s" => \$opt_enc_out ); $HELP = $opt_help || 0; $DEBUG = $opt_debug || 0; $PRESERVE_CASE = $opt_preserve_case || 0; if ($PRESERVE_CASE == 1) { $LOWERCASE = 0; } else { $LOWERCASE = 1; } $CHAR2VARCHAR = $opt_char2varchar || 0; $NODROP = $opt_nodrop || 0; $SEP_FILE = $opt_sepfile || 0; $ENC_IN = $opt_enc_in || 'utf8'; $ENC_OUT = $opt_enc_out || 'utf8'; if (($HELP) || ! defined($ARGV[0]) || ! defined($ARGV[1])) { print "\n\nUsage: perl $0 {--help --debug --preserve_case --char2varchar --nodrop --schema --sepfile --enc_in --enc_out } mysql.sql pg.sql\n"; print "\t* OPTIONS WITHOUT ARGS\n"; print "\t--help: prints this message \n"; print "\t--debug: output the commented-out mysql line above the postgres line in pg.sql \n"; print "\t--preserve_case: prevents automatic case-lowering of column and table names\n"; print "\t\tIf you want to preserve case, you must set this flag. For example,\n"; print "\t\tIf your client application quotes table and column-names and they have cases in them, set this flag\n"; print "\t--char2varchar: converts all char fields to varchar\n"; print "\t--nodrop: strips out DROP TABLE statements\n"; print "\t\totherise harmless warnings are printed by psql when the dropped table does not exist\n"; print "\n\t* OPTIONS WITH ARGS\n"; print "\t--schema: outputs a line into the postgres sql file setting search_path \n"; print "\t--sepfile: output foreign key constraints and indexes to a separate file so that it can be\n"; print "\t\timported after large data set is inserted from another dump file\n"; print "\t--enc_in: encoding of mysql in file (default utf8) \n"; print "\t--enc_out: encoding of postgres out file (default utf8) \n"; print "\n\t* REQUIRED ARGUMENTS\n"; if (defined ($ARGV[0])) { print "\tmysql.sql ($ARGV[0])\n"; } else { print "\tmysql.sql (undefined)\n"; } if (defined ($ARGV[1])) { print "\tpg.sql ($ARGV[1])\n"; } else { print "\tpg.sql (undefined)\n"; } print "\n"; exit 1; } ######################################################## # 4. process through mysql_dump.sql file # in a big loop ######################################################## # open in and out files open(IN,"<:encoding($ENC_IN)", $ARGV[0]) || die "can't open mysql dump file $ARGV[0]"; open(OUT,">:encoding($ENC_OUT)", $ARGV[1]) || die "can't open pg dump file $ARGV[1]"; # output header print OUT "--\n"; print OUT "-- Generated from mysql2pgsql.perl\n"; print OUT "-- http://gborg.postgresql.org/project/mysql2psql/\n"; print OUT "-- (c) 2001 - 2007 Jose M. Duarte, Joseph Speigle\n"; print OUT "--\n"; print OUT "\n"; print OUT "-- warnings are printed for drop tables if they do not exist\n"; print OUT "-- please see http://archives.postgresql.org/pgsql-novice/2004-10/msg00158.php\n\n"; print OUT "-- ##############################################################\n"; if ($SCHEMA ) { print OUT "set search_path='" . $SCHEMA . "'\\g\n" ; } # loop through mysql file on a per-line basis while() { ############## flow ######################### # (the lines are directed to different string variables at different times) # # handle drop table , unlock, connect statements # if ( start of create table) { # print out post_create table (indexes, foreign key constraints, comments from previous table) # add drop table statement if !$NODROP to pre_create_sql # next; # } # else if ( inside create table) { # add comments in this portion to create_sql # if ( end of create table) { # delete mysql-unique CREATE TABLE commands # print pre_create_sql # print the constraint tables for set and year datatypes # print create_sql # print function_create_sql (this is for the enum columns only) # next; # } # do substitutions # -- NUMERIC DATATYPES # -- CHARACTER DATATYPES # -- DATE AND TIME DATATYPES # -- KEY AND UNIQUE CREATIONS # and append them to create_sql # } else { # print inserts on-the-spot (this script only changes default timestamp of 0000-00-00) # } # LOOP until EOF # ######################################################## if (!/^\s*insert into/i) { # not inside create table so don't worry about data corruption s/`//g; # '`pgsql uses no backticks to denote table name (CREATE TABLE `sd`) or around field # and table names like mysql # doh! we hope all dashes and special chars are caught by the regular expressions :) } if (/^\s*USE\s*([^;]*);/) { print OUT "\\c ". $1; next; } if (/^(UN)?LOCK TABLES/i || /drop\s+table/i ) { # skip # DROP TABLE is added when we see the CREATE TABLE next; } if (/(create\s+table\s+)([-_\w]+)\s/i) { # example: CREATE TABLE `english_english` print_post_create_sql(); # for last table $tables_first_timestamp_column= 1; # decision to print warnings about default_timestamp not being in postgres $create_sql = ''; $table_no_quotes = $2 ; $table=quote_and_lc($2); if ( !$NODROP ) { # always print drop table if user doesn't explicitly say not to # to drop a table that is referenced by a view or a foreign-key constraint of another table, # CASCADE must be specified. (CASCADE will remove a dependent view entirely, but in the # in the foreign-key case it will only remove the foreign-key constraint, not the other table entirely.) # (source: 8.1.3 docs, section "drop table") warn "table $table will be dropped CASCADE\n"; $pre_create_sql .= "DROP TABLE $table CASCADE\\g\n"; # custom dumps may be missing the 'dump' commands } s/(create\s+table\s+)([-_\w]+)\s/$1 $table /i; if ($DEBUG) { $create_sql .= '-- ' . $_; } $create_sql .= $_; next; } if ($create_sql ne "") { # we are inside create table statement so lets process datatypes # print out comments or empty lines in context if ($DEBUG) { $create_sql .= '-- ' . $_; } if (/^#/ || /^$/ || /^\s*--/) { s/^#/--/; # Two hyphens (--) is the SQL-92 standard indicator for comments $create_sql.=$_; next; } if (/\).*;/i) { # end of create table squence s/INSERT METHOD[=\s+][^;\s]+//i; s/PASSWORD=[^;\s]+//i; s/ROW_FORMAT=(?:DEFAULT|DYNAMIC|FIXED|COMPRESSED|REDUNDANT|COMPACT)+//i; s/DELAY KEY WRITE=[^;\s]+//i; s/INDEX DIRECTORY[=\s+][^;\s]+//i; s/DATA DIRECTORY=[^;\s]+//i; s/CONNECTION=[^;\s]+//i; s/CHECKSUM=[^;\s]+//i; s/Type=[^;\s]+//i; # ISAM , # older versions s/COLLATE=[^;\s]+//i; # table's collate s/COLLATE\s+[^;\s]+//i; # table's collate # possible AUTO_INCREMENT starting index, it is used in mysql 5.0.26, not sure since which version if (/AUTO_INCREMENT=(\d+)/i) { # should take < ---- ) ENGINE=MyISAM AUTO_INCREMENT=16 DEFAULT CHARSET=latin1; # and should ouput ---> CREATE SEQUENCE "rhm_host_info_id_seq" START WITH 16; my $start_value = $1; print $auto_increment_seq . "--\n"; # print $pre_create_sql . "--\n"; $pre_create_sql =~ s/(CREATE SEQUENCE $auto_increment_seq )/$1 START WITH $start_value /; } s/AUTO_INCREMENT=\d+//i; s/PACK_KEYS=\d//i; # mysql 5.0.22 s/DEFAULT CHARSET=[^;\s]+//i; # my mysql version is 4.1.11 s/ENGINE\s*=\s*[^;\s]+//i; # my mysql version is 4.1.11 s/ROW_FORMAT=[^;\s]+//i; # my mysql version is 5.0.22 s/MIN_ROWS=[^;\s]+//i; s/MAX_ROWS=[^;\s]+//i; s/AVG_ROW_LENGTH=[^;\s]+//i; if (/COMMENT='([^']*)'/) { # ) ENGINE=MyISAM DEFAULT CHARSET=utf8 COMMENT='must be country zones'; $post_create_sql.="COMMENT ON TABLE $table IS '$1'\;"; # COMMENT ON table_name IS 'text'; s/COMMENT='[^']*'//i; } $create_sql =~ s/,$//g; # strip last , inside create table # make sure we end in a comma, as KEY statments are turned # into post_create_sql indices # they often are the last line so leaving a 'hanging comma' my @array = split("\n", $create_sql); for (my $a = $#array; $a >= 0; $a--) { #loop backwards if ($a == $#array && $array[$a] =~ m/,\s*$/) { # for last line $array[$a] =~ s/,\s*$//; next; } if ($array[$a] !~ m/create table/i) { # i.e. if there was more than one column in table if ($a != $#array && $array[$a] !~ m/,\s*$/ ) { # for second to last $array[$a] =~ s/$/,/; last; } elsif ($a != $#array && $array[$a] =~ m/,\s*$/ ) { # for second to last last; } } } $create_sql = join("\n", @array) . "\n"; $create_sql .= $_; # put comments out first print OUT $pre_create_sql; # create separate table to reference and to hold mysql's possible set data-type # values. do that table's creation before create table # definition foreach $column_name (keys %constraints) { $type=$constraints{$column_name}{'type'}; $column_valuesStr = $constraints{$column_name}{'values'}; $constraint_table_name = get_identifier(${table},${column_name} ,"constraint_table"); if ($type eq 'set') { print OUT qq~DROP TABLE $constraint_table_name CASCADE\\g\n~ ; print OUT qq~create table $constraint_table_name ( set_values varchar UNIQUE)\\g\n~ ; $function_create_sql .= make_plpgsql($table,$column_name); } elsif ($type eq 'year') { print OUT qq~DROP TABLE $constraint_table_name CASCADE\\g\n~ ; print OUT qq~create table $constraint_table_name ( year_values varchar UNIQUE)\\g\n~ ; } @column_values = split /,/, $column_valuesStr; foreach $value (@column_values) { print OUT qq~insert into $constraint_table_name values ( $value )\\g\n~; # ad ' for ints and varchars } } # print create table and reset create table vars # when moving from each "create table" to "insert" part of dump print OUT $create_sql; print OUT $function_create_sql; $pre_create_sql=""; $auto_increment_seq=""; $create_sql=""; $function_create_sql=''; %constraints=(); # the post_create_sql for this table is output at the beginning of the next table def # in case we want to make indexes after doing inserting next; } if (/^\s*(\w+)\s+.*COMMENT\s*'([^']*)'/) { #`zone_country_id` int(11) COMMENT 'column comment here', $quoted_column=quote_and_lc($1); $post_create_sql.="COMMENT ON COLUMN $table"."."." $quoted_column IS '$2'\;"; # COMMENT ON table_name.column_name IS 'text'; s/COMMENT\s*'[^']*'//i; } # NUMERIC DATATYPES # # auto_increment -> sequences # UNSIGNED conversions # TINYINT # SMALLINT # MEDIUMINT # INT, INTEGER # BIGINT # # DOUBLE [PRECISION], REAL # DECIMAL(M,D), NUMERIC(M,D) # FLOAT(p) # FLOAT s/(\w*int)\(\d+\)/$1/g; # hack of the (n) stuff for e.g. mediumint(2) int(3) if (/^(\s*)(\w+)\s*.*numeric.*auto_increment/i) { # int,auto_increment -> serial $seq = get_identifier($table, $2, 'seq'); $quoted_column=quote_and_lc($2); $pre_create_sql.= "DROP SEQUENCE $seq CASCADE\;\n\n"; # cascade will force drop of table, too $pre_create_sql.= "CREATE SEQUENCE $seq \;\n\n"; $auto_increment_seq = $seq ; # save in case we have the AUTO_INCREMENT=16 to default to # Note: Before PostgreSQL 8.1, the arguments of the sequence functions were of type text, not regclass, # and the above-described conversion from a text string to an OID value would happen at run time during # each call. For backwards compatibility, this facility still exists, but internally it is now handled # as an implicit coercion from text to regclass before the function is invoked. (source: 8.1.3 manual, section 9.12) s/^(\s*)(\w+)\s*.*NUMERIC(.*)auto_increment([^,]*)/$1 $quoted_column numeric $3 DEFAULT nextval('$seq') $4/ig; # MYSQL: data_id mediumint(8) unsigned NOT NULL auto_increment, $create_sql.=$_; next; } if (/^\s*(\w+)\s+.*int.*auto_increment/i) { # example: data_id mediumint(8) unsigned NOT NULL auto_increment, # int,auto_increment -> serial (same as what is done below) # for postgres side see http://www.postgresql.org/docs/7.4/interactive/datatype.html#DATATYPE-SERIAL $seq = get_identifier($table, $1, 'seq'); $quoted_column=quote_and_lc($1); $pre_create_sql.= "DROP SEQUENCE $seq CASCADE \;\n\n"; # cascade will force drop of table, too $auto_increment_seq= $seq ; # save in case we have the AUTO_INCREMENT=16 to default to $pre_create_sql.= "CREATE SEQUENCE $seq \;\n\n"; s/(\s*)(\w+)\s+.*int.*auto_increment[^,]*/$1 $quoted_column integer DEFAULT nextval('$seq') NOT NULL/ig; $create_sql.=$_; next; } # convert UNSIGNED to CHECK constraints if (m/^(\s*)(\w+)\s+((float|double|double precision|real|decimal|numeric))(.*)unsigned/i) { $quoted_column = quote_and_lc($2); s/^(\s*)(\w+)\s+((float|double|double precision|real|decimal|numeric))(.*)unsigned/$1 $quoted_column $3 $4 CHECK ($quoted_column >= 0)/i; } # example: `wordsize` tinyint(3) unsigned default NULL, if (m/^(\s+)(\w+)\s+(\w+)\s+unsigned/i) { $quoted_column=quote_and_lc($2); s/^(\s+)(\w+)\s+(\w+)\s+unsigned/$1 $quoted_column $3 CHECK ($quoted_column >= 0)/i; } if (m/^(\s*)(\w+)\s+(bigint.*)unsigned/) { $quoted_column=quote_and_lc($2); # see http://archives.postgresql.org/pgsql-general/2005-07/msg01178.php # and see http://www.postgresql.org/docs/8.2/interactive/datatype-numeric.html # see http://dev.mysql.com/doc/refman/5.1/en/numeric-types.html max size == 20 digits s/^(\s*)(\w+)\s+bigint(.*)unsigned/$1 $quoted_column NUMERIC (20,0) CHECK ($quoted_column >= 0)/i; } # int type conversion # TINYINT (signed) -128 to 127 (unsigned) 0 255 # SMALLINT A small integer. The signed range is -32768 to 32767. The unsigned range is 0 to 65535. # MEDIUMINT A medium-sized integer. The signed range is -8388608 to 8388607. The unsigned range is 0 to 16777215. # INT A normal-size integer. The signed range is -2147483648 to 2147483647. The unsigned range is 0 to 4294967295. # BIGINT The signed range is -9223372036854775808 to 9223372036854775807. The unsigned range is 0 to 18446744073709551615 # for postgres see http://www.postgresql.org/docs/8.2/static/datatype-numeric.html#DATATYPE-INT s/^(\s+"*\w+"*\s+)tinyint/$1 smallint/i; s/^(\s+"*\w+"*\s+)mediumint/$1 integer/i; # the floating point types # double -> double precision # double(n,m) -> double precision # float - no need for conversion # float(n) - no need for conversion # float(n,m) -> double precision s/(^\s*\w+\s+)double(\(\d+,\d+\))?/$1double precision/i; s/float(\(\d+,\d+\))/double precision/i; # # CHARACTER TYPES # # set # enum # binary(M), VARBINARy(M), tinyblob, tinytext, # bit # char(M), varchar(M) # blob -> text # mediumblob # longblob, longtext # text -> text # mediumtext # longtext # mysql docs: A BLOB is a binary large object that can hold a variable amount of data. # set # For example, a column specified as SET('one', 'two') NOT NULL can have any of these values: # '' # 'one' # 'two' # 'one,two' if (/(\w*)\s+set\(((?:['"]\w+['"]\s*,*)+(?:['"]\w+['"])*)\)(.*)$/i) { # example: `au_auth` set('r','w','d') NOT NULL default '', $column_name = $1; $constraints{$column_name}{'values'} = $2; # 'abc','def', ... $constraints{$column_name}{'type'} = "set"; # 'abc','def', ... $_ = qq~ $column_name varchar , ~; $column_name = quote_and_lc($1); $create_sql.=$_; next; } if (/(\S*)\s+enum\(((?:['"][^'"]+['"]\s*,)+['"][^'"]+['"])\)(.*)$/i) { # enum handling # example: `test` enum('?','+','-') NOT NULL default '?' # $2 is the values of the enum 'abc','def', ... $quoted_column=quote_and_lc($1); # "test" NOT NULL default '?' CONSTRAINT test_test_constraint CHECK ("test" IN ('?','+','-')) $_ = qq~ $quoted_column varchar CHECK ($quoted_column IN ( $2 ))$3\n~; # just assume varchar? $create_sql.=$_; next; } # Take care of "binary" option for char and varchar # (pre-4.1.2, it indicated a byte array; from 4.1.2, indicates # a binary collation) s/(?:var)?char(?:\(\d+\))? (?:byte|binary)/bytea/i; if (m/(?:var)?binary\s*\(\d+\)/i) { # c varBINARY(3) in Mysql warn "WARNING in table '$table' '$_': binary type is converted to bytea (unsized) for Postgres\n"; } s/(?:var)?binary(?:\(\d+\))?/bytea/i; # c varBINARY(3) in Mysql s/bit(?:\(\d+\))?/bytea/i; # bit datatype -> bytea # large datatypes s/\w*blob/bytea/gi; s/tinytext/text/gi; s/mediumtext/text/gi; s/longtext/text/gi; # char -> varchar -- if specified as a command line option # PostgreSQL would otherwise pad with spaces as opposed # to MySQL! Your user interface may depend on this! if ($CHAR2VARCHAR) { s/(^\s+\S+\s+)char/${1}varchar/gi; } # nuke column's collate and character set s/(\S+)\s+character\s+set\s+\w+/$1/gi; s/(\S+)\s+collate\s+\w+/$1/gi; # # DATE AND TIME TYPES # # date time # year # datetime # timestamp # date time # these are the same types in postgres, just do the replacement of 0000-00-00 date if (m/default '(\d+)-(\d+)-(\d+)([^']*)'/i) { # we grab the year, month and day # NOTE: times of 00:00:00 are possible and are okay my $time = ''; my $year=$1; my $month= $2; my $day = $3; if ($4) { $time = $4; } if ($year eq "0000") { $year = '1970'; } if ($month eq "00") { $month = '01'; } if ($day eq "00") { $day = '01'; } s/default '[^']+'/default '$year-$month-$day$time'/i; # finally we replace with $datetime } # convert mysql's year datatype to a constraint if (/(\w*)\s+year\(4\)(.*)$/i) { # can be integer OR string 1901-2155 $constraint_table_name = get_identifier($table,$1 ,"constraint_table"); $column_name=quote_and_lc($1); @year_holder = (); $year=''; for (1901 .. 2155) { $year = "'$_'"; unless ($year =~ /2155/) { $year .= ','; } push( @year_holder, $year); } $constraints{$column_name}{'values'} = join('','',@year_holder); # '1901','1902', ... $constraints{$column_name}{'type'} = "year"; $_ = qq~ $column_name varchar CONSTRAINT ${table}_${column_name}_constraint REFERENCES $constraint_table_name ("year_values") $2\n~; $create_sql.=$_; next; } elsif (/(\w*)\s+year\(2\)(.*)$/i) { # same for a 2-integer string $constraint_table_name = get_identifier($table,$1 ,"constraint_table"); $column_name=quote_and_lc($1); @year_holder = (); $year=''; for (1970 .. 2069) { $year = "'$_'"; if ($year =~ /2069/) { next; } push( @year_holder, $year); } push( @year_holder, '0000'); $constraints{$column_name}{'values'} = join(',',@year_holder); # '1971','1972', ... $constraints{$column_name}{'type'} = "year"; # 'abc','def', ... $_ = qq~ $1 varchar CONSTRAINT ${table}_${column_name}_constraint REFERENCES $constraint_table_name ("year_values") $2\n~; $create_sql.=$_; next; } # datetime # Default on a dump from MySQL 5.0.22 is in the same form as datetime so let it flow down # to the timestamp section and deal with it there s/(${sl})datetime /$1timestamp without time zone /i; # change not null datetime field to null valid ones # (to support remapping of "zero time" to null # s/($sl)datetime not null/$1timestamp without time zone/i; # timestamps # # nuke datetime representation (not supported in PostgreSQL) # change default time of 0000-00-00 to 1970-01-01 # we may possibly need to create a trigger to provide # equal functionality with ON UPDATE CURRENT TIMESTAMP if (m/${sl}timestamp/i) { if ( m/ON UPDATE CURRENT_TIMESTAMP/i ) { # the ... default CURRENT_TIMESTAMP only applies for blank inserts, not updates s/ON UPDATE CURRENT_TIMESTAMP//i ; m/^\s*(\w+)\s+timestamp/i ; # automatic trigger creation $table_no_quotes =~ s/"//g; $function_create_sql .= " CREATE OR REPLACE FUNCTION update_". $table_no_quotes . "() RETURNS trigger AS ' BEGIN NEW.$1 := CURRENT_TIMESTAMP; RETURN NEW; END; ' LANGUAGE 'plpgsql'; -- before INSERT is handled by 'default CURRENT_TIMESTAMP' CREATE TRIGGER add_current_date_to_".$table_no_quotes." BEFORE UPDATE ON ". $table . " FOR EACH ROW EXECUTE PROCEDURE update_".$table_no_quotes."();\n"; } if ($tables_first_timestamp_column && m/DEFAULT NULL/i) { # DEFAULT NULL is the same as DEFAULT CURRENT_TIMESTAMP for the first TIMESTAMP column. (MYSQL manual) s/($sl)(timestamp\s+)default null/$1 $2 DEFAULT CURRENT_TIMESTAMP/i; } $tables_first_timestamp_column= 0; if (m/${sl}timestamp\s*\(\d+\)/i) { # fix for timestamps with width spec not handled (ID: 1628) warn "WARNING for in table '$table' '$_': your default timestamp width is being ignored for table $table \n"; s/($sl)timestamp(?:\(\d+\))/$1datetime/i; } } # end timestamp section # KEY AND UNIQUE CREATIONS # # unique if ( /^\s+unique\s+\(([^(]+)\)/i ) { # example UNIQUE `name` (`name`), same as UNIQUE KEY # POSTGRESQL: treat same as mysql unique $quoted_column = quote_and_lc($1); s/\s+unique\s+\(([^(]+)\)/ unique ($quoted_column) /i; $create_sql.=$_; next; } elsif ( /^\s+unique\s+key\s*(\w+)\s*\(([^(]+)\)/i ) { # example UNIQUE KEY `name` (`name`) # MYSQL: unique key: allows null=YES, allows duplicates=NO (*) # ... new ... UNIQUE KEY `unique_fullname` (`fullname`) in my mysql v. Ver 14.12 Distrib 5.1.7-beta # POSTGRESQL: treat same as mysql unique # just quote columns $quoted_column = quote_and_lc($2); s/\s+unique\s+key\s*(\w+)\s*\(([^(]+)\)/ unique ($quoted_column) /i; $create_sql.=$_; # the index corresponding to the 'key' is automatically created next; } # keys if ( /^\s+fulltext key\s+/i) { # example: FULLTEXT KEY `commenttext` (`commenttext`) # that is key as a word in the first check for a match # the tsvector datatype is made for these types of things # example mysql file: # what is tsvector datatype? # http://www.sai.msu.su/~megera/postgres/gist/tsearch/V2/docs/tsearch-V2-intro.html warn "dba must do fulltext key transformation for $table\n"; next; } if ( /^(\s+)constraint (\S+) foreign key \((\S+)\) references (\S+) \((\S+)\)(.*)/i ) { $quoted_column =quote_and_lc($3); $col=quote_and_lc($5); $post_create_sql .= "ALTER TABLE $table ADD FOREIGN KEY ($quoted_column) REFERENCES " . quote_and_lc($4) . " ($col);\n"; next; } if ( /^\s*primary key\s*\(([^)]+)\)([,\s]+)/i ) { # example PRIMARY KEY (`name`) # MYSQL: primary key: allows null=NO , allows duplicates=NO # POSTGRESQL: When an index is declared unique, multiple table rows with equal indexed values will not be # allowed. Null values are not considered equal. # POSTGRESQL quote's source: 8.1.3 docs section 11.5 "unique indexes" # so, in postgres, we need to add a NOT NULL to the UNIQUE constraint # and, primary key (mysql) == primary key (postgres) so that we *really* don't need change anything $quoted_column = quote_and_lc($1); s/(\s*)primary key\s+\(([^)]+)\)([,\s]+)/$1 primary key ($quoted_column)$3/i; # indexes are automatically created for unique columns $create_sql.=$_; next; } elsif (m/^\s+key\s[-_\s\w]+\((.+)\)/i ) { # example: KEY `idx_mod_english_def_word` (`word`), # regular key: allows null=YES, allows duplicates=YES # MYSQL: KEY is normally a synonym for INDEX. http://dev.mysql.com/doc/refman/5.1/en/create-table.html # # * MySQL: ALTER TABLE {$table} ADD KEY $column ($column) # * PostgreSQL: CREATE INDEX {$table}_$column_idx ON {$table}($column) // Please note the _idx "extension" # PRIMARY KEY (`postid`), # KEY `ownerid` (`ownerid`) # create an index for everything which has a key listed for it. my $col = $1; # TODO we don't have a translation for the substring syntax in text columns in MySQL (e.g. "KEY my_idx (mytextcol(20))") # for now just getting rid of the brackets and numbers (the substring specifier): $col=~s/\(\d+\)//g; $quoted_column = quote_and_lc($col); if ($col =~ m/,/) { $col = s/,/_/; } $index = get_identifier($table, $col, 'idx'); $post_create_sql.="CREATE INDEX $index ON $table USING btree ($quoted_column)\;"; # just create index do not add to create table statement next; } # handle 'key' declared at end of column if (/\w+.*primary key/i) { # mysql: key is normally just a synonym for index # just leave as is ( postgres has primary key type) } elsif (/(\w+\s+(?:$mysql_datatypesStr)\s+.*)key/i) { # mysql: key is normally just a synonym for index # I can't find a reference for 'key' in a postgres command without using the word 'primary key' s/$1key/$1/i ; $index = get_identifier($table, $1, 'idx'); $quoted_column =quote_and_lc($1); $post_create_sql.="CREATE INDEX $index ON $table USING btree ($quoted_column) \;"; $create_sql.=$_; } # do we really need this anymore? # remap colums with names of existing system attribute if (/"oid"/i) { s/"oid"/"_oid"/g; print STDERR "WARNING: table $table uses column \"oid\" which is renamed to \"_oid\"\nYou should fix application manually! Press return to continue."; my $wait=; } s/oid/_oid/i if (/key/i && /oid/i); # fix oid in key # FINAL QUOTING OF ALL COLUMNS # quote column names which were not already quoted # perhaps they were not quoted because they were not explicitly handled if (!/^\s*"(\w+)"(\s+)/i) { /^(\s*)(\w+)(\s+)(.*)$/i ; $quoted_column= quote_and_lc($2); s/^(\s*)(\w+)(\s+)(.*)$/$1 $quoted_column $3 $4 /; } $create_sql.=$_; # END of if ($create_sql ne "") i.e. were inside create table statement so processed datatypes } # add "not in create table" comments or empty lines to pre_create_sql elsif (/^#/ || /^$/ || /^\s*--/) { s/^#/--/; # Two hyphens (--) is the SQL-92 standard indicator for comments $pre_create_sql .= $_ ; # printed above create table statement next; } elsif (/^\s*insert into/i) { # not inside create table and doing insert # fix mysql's zero/null value for timestamps s/'0000-00-00/'1970-01-01/gi; # commented out to fix bug "Field contents interpreted as a timestamp", what was the point of this line anyway? #s/([12]\d\d\d)([01]\d)([0-3]\d)([0-2]\d)([0-6]\d)([0-6]\d)/'$1-$2-$3 $4:$5:$6'/; #---- fix data in inserted data: (from MS world) s!\x96!-!g; # -- s!\x93!"!g; # `` s!\x94!"!g; # '' s!\x85!... !g; # \ldots s!\x92!`!g; print OUT $pre_create_sql; # print comments preceding the insert section $pre_create_sql=""; $auto_increment_seq = ""; s/'((?:.*?(?:\\')?.*?)*)'([),])/E'$1'$2/g; # for the E'' see http://www.postgresql.org/docs/8.2/interactive/release-8-1.html # split 'extended' INSERT INTO statements to something PostgreSQL can understand ( $insert_table, $valueString) = $_ =~ m/^INSERT\s+INTO\s+['`"]*(.*?)['`"]*\s+VALUES\s*(.*)/i; $insert_table = quote_and_lc($insert_table); # parse valueString my @rows = $valueString =~ m/$rowRe/g; s/^INSERT INTO.*?\);//i; # hose the statement which is to be replaced whether a run-on or not # only convert INSERT INTO statements with multiple values if (@rows > 1) { for my $row (@rows) { print OUT qq(INSERT INTO $insert_table VALUES ($row);\n); } # end command print OUT "\n"; } else { # guarantee table names are quoted print OUT qq(INSERT INTO $insert_table VALUES $valueString \n); } } else { # print OUT $_ ; # example: /*!40101 SET @OLD_CHARACTER_SET_CLIENT=@@CHARACTER_SET_CLIENT */; } # keep looping and get next line of IN file } # END while() print_post_create_sql(); # in case there is extra from the last table ################################################################# # 5. print_plgsql function prototype # emulate the set datatype with the following plpgsql function # looks ugly so putting at end of file ################################################################# # sub make_plpgsql { my ($table,$column_name) = ($_[0],$_[1]); $table=~s/\"//g; # make sure that $table doesn't have quotes so we don't end up with redundant quoting my $constraint_table = get_identifier($table,$column_name ,"constraint_table"); return " -- this function is called by the insert/update trigger -- it checks if the INSERT/UPDATE for the 'set' column -- contains members which comprise a valid mysql set -- this TRIGGER function therefore acts like a constraint -- provided limited functionality for mysql's set datatype -- just verifies and matches for string representations of the set at this point -- though the set datatype uses bit comparisons, the only supported arguments to our -- set datatype are VARCHAR arguments -- to add a member to the set add it to the ".$table."_".$column_name." table CREATE OR REPLACE FUNCTION check_".$table."_".$column_name."_set( ) RETURNS TRIGGER AS \$\$\n DECLARE ---- arg_str VARCHAR ; argx VARCHAR := ''; nobreak INT := 1; rec_count INT := 0; psn INT := 0; str_in VARCHAR := NEW.$column_name; ---- BEGIN ---- IF str_in IS NULL THEN RETURN NEW ; END IF; arg_str := REGEXP_REPLACE(str_in, '\\',\\'', ','); -- str_in is CONSTANT arg_str := REGEXP_REPLACE(arg_str, '^\\'', ''); arg_str := REGEXP_REPLACE(arg_str, '\\'\$', ''); -- RAISE NOTICE 'arg_str %',arg_str; psn := POSITION(',' in arg_str); IF psn > 0 THEN psn := psn - 1; -- minus-1 from comma position -- RAISE NOTICE 'psn %',psn; argx := SUBSTRING(arg_str FROM 1 FOR psn); -- get one set member psn := psn + 2; -- go to first starting letter arg_str := SUBSTRING(arg_str FROM psn); -- hack it off ELSE psn := 0; -- minus-1 from comma position argx := arg_str; END IF; -- RAISE NOTICE 'argx %',argx; -- RAISE NOTICE 'new arg_str: %',arg_str; WHILE nobreak LOOP EXECUTE 'SELECT count(*) FROM $constraint_table WHERE set_values = ' || quote_literal(argx) INTO rec_count; IF rec_count = 0 THEN RAISE EXCEPTION 'one of the set values was not found'; END IF; IF psn > 0 THEN psn := psn - 1; -- minus-1 from comma position -- RAISE NOTICE 'psn %',psn; argx := SUBSTRING(arg_str FROM 1 FOR psn); -- get one set member psn := psn + 2; -- go to first starting letter arg_str := SUBSTRING(arg_str FROM psn); -- hack it off psn := POSITION(',' in arg_str); ELSE nobreak = 0; END IF; -- RAISE NOTICE 'next argx % and next arg_str %', argx, arg_str; END LOOP; RETURN NEW; ---- END; \$\$ LANGUAGE 'plpgsql' VOLATILE; drop trigger set_test ON $table; -- make a trigger for each set field -- make trigger and hard-code in column names -- see http://archives.postgresql.org/pgsql-interfaces/2005-02/msg00020.php CREATE TRIGGER set_test BEFORE INSERT OR UPDATE ON $table FOR EACH ROW EXECUTE PROCEDURE check_".$table."_".$column_name."_set();\n"; } # end sub make_plpgsql();