1 package SL::Helper::Csv;
8 use Params::Validate qw(:all);
10 use Rose::Object::MakeMethods::Generic scalar => [ qw(
11 file encoding sep_char quote_char escape_char header profile class
12 numberformat dateformat ignore_unknown_columns strict_profile _io _csv
13 _objects _parsed _data _errors
16 use SL::Helper::Csv::Dispatcher;
17 use SL::Helper::Csv::Error;
23 my %params = validate(@_, {
24 sep_char => { default => ';' },
25 quote_char => { default => '"' },
26 escape_char => { default => '"' },
27 header => { type => ARRAYREF, optional => 1 },
28 profile => { type => HASHREF, optional => 1 },
34 ignore_unknown_columns => 0,
37 my $self = bless {}, $class;
39 $self->$_($params{$_}) for keys %params;
41 $self->_io(IO::File->new);
42 $self->_csv(Text::CSV_XS->new({
44 sep_char => $self->sep_char,
45 quote_char => $self->quote_char,
46 escape_char => $self->escape_char,
55 my ($self, %params) = @_;
58 return if ! $self->_check_header;
59 return if ! $self->dispatcher->parse_profile;
60 return if ! $self->_parse_data;
71 my ($self, %params) = @_;
72 croak 'no class given' unless $self->class;
73 croak 'must parse first' unless $self->_parsed;
75 $self->_make_objects unless $self->_objects;
76 return wantarray ? @{ $self->_objects } : $self->_objects;
90 my ($self, %params) = @_;
92 $self->encoding($self->_guess_encoding) if !$self->encoding;
94 $self->_io->open($self->file, '<' . $self->_encode_layer)
95 or die "could not open file " . $self->file;
101 my ($self, %params) = @_;
102 my $header = $self->header;
105 $header = $self->_csv->getline($self->_io);
108 $self->_csv->error_input,
109 $self->_csv->error_diag,
114 return unless $header;
115 return $self->header([ map { lc } @$header ]);
119 my ($self, %params) = @_;
122 $self->_csv->column_names(@{ $self->header });
125 my $row = $self->_csv->getline($self->_io);
128 @hr{@{ $self->header }} = @$row;
131 last if $self->_csv->eof;
133 $self->_csv->error_input,
134 $self->_csv->error_diag,
135 $self->_io->input_line_number,
138 last if $self->_csv->eof;
141 $self->_data(\@data);
142 $self->_push_error(@errors);
148 ':encoding(' . $_[0]->encoding . ')';
152 my ($self, %params) = @_;
155 eval "require " . $self->class;
156 local $::myconfig{numberformat} = $self->numberformat if $self->numberformat;
157 local $::myconfig{dateformat} = $self->dateformat if $self->dateformat;
159 for my $line (@{ $self->_data }) {
160 my $tmp_obj = $self->class->new;
161 $self->dispatcher->dispatch($tmp_obj, $line);
162 push @objs, $tmp_obj;
165 $self->_objects(\@objs);
169 my ($self, %params) = @_;
171 $self->{_dispatcher} ||= $self->_make_dispatcher;
174 sub _make_dispatcher {
175 my ($self, %params) = @_;
177 die 'need a header to make a dispatcher' unless $self->header;
179 return SL::Helper::Csv::Dispatcher->new($self);
182 sub _guess_encoding {
188 my ($self, @errors) = @_;
189 my @new_errors = ($self->errors, map { SL::Helper::Csv::Error->new(@$_) } @errors);
190 $self->_errors(\@new_errors);
202 SL::Helper::Csv - take care of csv file uploads
208 my $csv = SL::Helper::Csv->new(
209 file => \$::form->{upload_file},
210 encoding => 'utf-8', # undef means utf8
211 sep_char => ',', # default ';'
212 quote_char => '\'', # default '"'
213 escape_char => '"', # default '"'
214 header => [qw(id text sellprice word)], # see later
215 profile => { sellprice => 'sellprice_as_number' },
216 class => 'SL::DB::CsvLine', # if present, map lines to this
219 my $status = $csv->parse;
220 my $hrefs = $csv->get_data;
221 my @objects = $csv->get_objects;
223 my @errors = $csv->errors;
229 Text::CSV offeres already good functions to get lines out of a csv file, but in
230 most cases you will want those line to be parsed into hashes or even objects,
231 so this model just skips ahead and gives you objects.
233 Its basic assumptions are:
237 =item You do know what you expect to be in that csv file.
239 This means first and foremost you have knowledge about encoding, number and
240 date format, csv parameters such as quoting and separation characters. You also
241 know what content will be in that csv and what L<Rose::DB> is responsible for
242 it. You provide valid header columns and their mapping to the objects.
244 =item You do NOT know if the csv provider yields to your expectations.
246 Stuff that does not work with what you expect should not crash anything, but
247 give you a hint what went wrong. As a result, if you remeber to check for
248 errors after each step, you should be fine.
250 =item Data does not make sense. It's just data.
252 Almost all data imports have some type of constraints. Some data needs to be
253 unique, other data needs to be connected to existing data sets. This will not
254 happen here. You will receive a plain mapping of the data into the class tree,
265 Standard constructor. You can use this to set most of the data.
269 Do the actual work. Will return true ($self actually) if success, undef if not.
273 Parse the data into objects and return those.
275 This method will return list or arrayref depending on context.
279 Returns an arrayref of the raw lines as hashrefs.
283 Return all errors that came up during parsing. See error handling for detailed
294 The file which contents are to be read. Can be a name of a physical file or a
295 scalar ref for memory data.
299 Encoding of the CSV file. Note that this module does not do any encoding
300 guessing. Know what your data is. Defaults to utf-8.
308 Same as in L<Text::CSV>
310 =item C<header> \@FIELDS
312 Can be an array of columns, in this case the first line is not used as a
313 header. Empty header fields will be ignored in objects.
315 =item C<profile> \%ACCESSORS
317 May be used to map header fields to custom accessors. Example:
319 { listprice => listprice_as_number }
321 In this case C<listprice_as_number> will be used to read in values from the
324 In case of a One-To-One relationsship these can also be set over
325 relationsships by sparating the steps with a dot (C<.>). This will work:
327 { customer => 'customer.name' }
329 And will result in something like this:
331 $obj->customer($obj->meta->relationship('customer')->class->new);
332 $obj->customer->name($csv_line->{customer})
334 But beware, this will not try to look up anything in the database. You will
335 simply receive objects that represent what the profile defined. If some of
336 these information are unique, and should be connected to preexisting data, you
337 will have to do that for yourself. Since you provided the profile, it is
338 assumed you know what to do in this case.
342 If present, the line will be handed to the new sub of this class,
343 and the return value used instead of the line itself.
345 =item C<ignore_unknown_columns>
347 If set, the import will ignore unkown header columns. Useful for lazy imports,
348 but deactivated by default.
350 =item C<strict_profile>
352 If set, all columns to be parsed must be specified in C<profile>. Every header
353 field not listed there will be treated like an unknown column.
357 =head1 ERROR HANDLING
359 After parsing a file all errors will be accumulated into C<errors>.
360 Each entry is an object with the following attributes:
362 raw_input: offending raw input,
363 code: Text::CSV error code if Text:CSV signalled an error, 0 else,
364 diag: error diagnostics,
365 line: position in line,
366 col: estimated line in file,
368 Note that the last entry can be off, but will give an estimate.
376 sep_char, quote_char, and escape_char are passed to Text::CSV on creation.
377 Changing them later has no effect currently.
381 Encoding errors are not dealt with properly.
387 Dispatch to child objects, like this:
389 $csv = SL::Helper::Csv->new(
391 class => SL::DB::Part,
406 Sven Schöling E<lt>s.schoeling@linet-services.deE<gt>