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 _io _csv _objects _parsed
16 use SL::Helper::Csv::Dispatcher;
22 my %params = validate(@_, {
23 sep_char => { default => ';' },
24 quote_char => { default => '"' },
25 escape_char => { default => '"' },
26 header => { type => ARRAYREF, optional => 1 },
27 profile => { type => HASHREF, optional => 1 },
33 ignore_unknown_columns => 0,
35 my $self = bless {}, $class;
37 $self->$_($params{$_}) for keys %params;
39 $self->_io(IO::File->new);
40 $self->_csv(Text::CSV->new({
42 sep_char => $self->sep_char,
43 quote_char => $self->quote_char,
44 escape_char => $self->escape_char,
53 my ($self, %params) = @_;
56 return if ! $self->_check_header;
57 return if ! $self->dispatcher->parse_profile;
58 return if ! $self->_parse_data;
69 my ($self, %params) = @_;
70 croak 'no class given' unless $self->class;
71 croak 'must parse first' unless $self->_parsed;
73 $self->_make_objects unless $self->_objects;
74 return wantarray ? @{ $self->_objects } : $self->_objects;
88 my ($self, %params) = @_;
90 $self->encoding($self->_guess_encoding) if !$self->encoding;
92 $self->_io->open($self->file, '<' . $self->_encode_layer)
93 or die "could not open file " . $self->file;
99 my ($self, %params) = @_;
100 return $self->header if $self->header;
102 my $header = $self->_csv->getline($self->_io);
105 $self->_csv->error_input,
106 $self->_csv->error_diag,
110 $self->header($header);
114 my ($self, %params) = @_;
117 $self->_csv->column_names(@{ $self->header });
120 my $row = $self->_csv->getline($self->_io);
121 last if $self->_csv->eof;
124 @hr{@{ $self->header }} = @$row;
128 $self->_csv->error_input,
129 $self->_csv->error_diag,
130 $self->_io->input_line_number,
135 $self->_data(\@data);
136 $self->_push_error(@errors);
142 ':encoding(' . $_[0]->encoding . ')';
146 my ($self, %params) = @_;
149 eval "require " . $self->class;
150 local $::myconfig{numberformat} = $self->numberformat if $self->numberformat;
151 local $::myconfig{dateformat} = $self->dateformat if $self->dateformat;
153 for my $line (@{ $self->_data }) {
154 my $tmp_obj = $self->class->new;
155 $self->dispatcher->dispatch($tmp_obj, $line);
156 push @objs, $tmp_obj;
159 $self->_objects(\@objs);
163 my ($self, %params) = @_;
165 $self->{_dispatcher} ||= $self->_make_dispatcher;
168 sub _make_dispatcher {
169 my ($self, %params) = @_;
171 die 'need a header to make a dispatcher' unless $self->header;
173 return SL::Helper::Csv::Dispatcher->new($self);
176 sub _guess_encoding {
182 my ($self, @errors) = @_;
183 my @new_errors = ($self->errors, @errors);
184 $self->_errors(\@new_errors);
196 SL::Helper::Csv - take care of csv file uploads
202 my $csv = SL::Helper::Csv->new(
203 file => \$::form->{upload_file},
204 encoding => 'utf-8', # undef means utf8
205 sep_char => ',', # default ';'
206 quote_char => '\'', # default '"'
207 escape_char => '"', # default '"'
208 header => [qw(id text sellprice word)], # see later
209 profile => { sellprice => 'sellprice_as_number' },
210 class => 'SL::DB::CsvLine', # if present, map lines to this
213 my $status = $csv->parse;
214 my $hrefs = $csv->get_data;
215 my @objects = $csv->get_objects;
217 my @errors = $csv->errors;
223 Text::CSV offeres already good functions to get lines out of a csv file, but in
224 most cases you will want those line to be parsed into hashes or even objects,
225 so this model just skips ahead and gives you objects.
227 Its basic assumptions are:
231 =item You do know what you expect to be in that csv file.
233 This means first and foremost you have knowledge about encoding, number and
234 date format, csv parameters such as quoting and separation characters. You also
235 know what content will be in that csv and what L<Rose::DB> is responsible for
236 it. You provide valid header columns and their mapping to the objects.
238 =item You do NOT know if the csv provider yields to your expectations.
240 Stuff that does not work with what you expect should not crash anything, but
241 give you a hint what went wrong. As a result, if you remeber to check for
242 errors after each step, you should be fine.
244 =item Data does not make sense. It's just data.
246 Almost all data imports have some type of constraints. Some data needs to be
247 unique, other data needs to be connected to existing data sets. This will not
248 happen here. You will receive a plain mapping of the data into the class tree,
259 Standard constructor. You can use this to set most of the data.
263 Do the actual work. Will return true ($self actually) if success, undef if not.
267 Parse the data into objects and return those.
269 This method will return list or arrayref depending on context.
273 Returns an arrayref of the raw lines as hashrefs.
277 Return all errors that came up during parsing. See error handling for detailed
288 The file which contents are to be read. Can be a name of a physical file or a
289 scalar ref for memory data.
293 Encoding of the CSV file. Note that this module does not do any encoding
294 guessing. Know what your data is. Defaults to utf-8.
302 Same as in L<Text::CSV>
304 =item C<header> \@FIELDS
306 Can be an array of columns, in this case the first line is not used as a
307 header. Empty header fields will be ignored in objects.
309 =item C<profile> \%ACCESSORS
311 May be used to map header fields to custom accessors. Example:
313 { listprice => listprice_as_number }
315 In this case C<listprice_as_number> will be used to read in values from the
318 In case of a One-To-One relationsship these can also be set over
319 relationsships by sparating the steps with a dot (C<.>). This will work:
321 { customer => 'customer.name' }
323 And will result in something like this:
325 $obj->customer($obj->meta->relationship('customer')->class->new);
326 $obj->customer->name($csv_line->{customer})
328 But beware, this will not try to look up anything in the database. You will
329 simply receive objects that represent what the profile defined. If some of
330 these information are unique, and should be connected to preexisting data, you
331 will have to do that for yourself. Since you provided the profile, it is
332 assumed you know what to do in this case.
336 If present, the line will be handed to the new sub of this class,
337 and the return value used instead of the line itself.
339 =item C<ignore_unknown_columns>
341 If set, the import will ignore unkown header columns. Useful for lazy imports,
342 but deactivated by default.
346 =head1 ERROR HANDLING
348 After parsing a file all errors will be accumulated into C<errors>.
350 Each entry is an arrayref with the following structure:
353 0 offending raw input,
354 1 Text::CSV error code if Text:CSV signalled an error, 0 else,
357 4 estimated line in file,
360 Note that the last entry can be off, but will give an estimate.
368 sep_char, quote_char, and escape_char are passed to Text::CSV on creation.
369 Changing them later has no effect currently.
373 Encoding errors are not dealt with properly.
379 Dispatch to child objects, like this:
381 $csv = SL::Helper::Csv->new(
383 class => SL::DB::Part,
398 Sven Schöling E<lt>s.schoeling@linet-services.deE<gt>