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 return $self->header if $self->header;
104 my $header = $self->_csv->getline($self->_io);
107 $self->_csv->error_input,
108 $self->_csv->error_diag,
112 $self->header([ map { lc } @{ $header } ]);
116 my ($self, %params) = @_;
119 $self->_csv->column_names(@{ $self->header });
122 my $row = $self->_csv->getline($self->_io);
125 @hr{@{ $self->header }} = @$row;
128 last if $self->_csv->eof;
130 $self->_csv->error_input,
131 $self->_csv->error_diag,
132 $self->_io->input_line_number,
135 last if $self->_csv->eof;
138 $self->_data(\@data);
139 $self->_push_error(@errors);
145 ':encoding(' . $_[0]->encoding . ')';
149 my ($self, %params) = @_;
152 eval "require " . $self->class;
153 local $::myconfig{numberformat} = $self->numberformat if $self->numberformat;
154 local $::myconfig{dateformat} = $self->dateformat if $self->dateformat;
156 for my $line (@{ $self->_data }) {
157 my $tmp_obj = $self->class->new;
158 $self->dispatcher->dispatch($tmp_obj, $line);
159 push @objs, $tmp_obj;
162 $self->_objects(\@objs);
166 my ($self, %params) = @_;
168 $self->{_dispatcher} ||= $self->_make_dispatcher;
171 sub _make_dispatcher {
172 my ($self, %params) = @_;
174 die 'need a header to make a dispatcher' unless $self->header;
176 return SL::Helper::Csv::Dispatcher->new($self);
179 sub _guess_encoding {
185 my ($self, @errors) = @_;
186 my @new_errors = ($self->errors, map { SL::Helper::Csv::Error->new(@$_) } @errors);
187 $self->_errors(\@new_errors);
199 SL::Helper::Csv - take care of csv file uploads
205 my $csv = SL::Helper::Csv->new(
206 file => \$::form->{upload_file},
207 encoding => 'utf-8', # undef means utf8
208 sep_char => ',', # default ';'
209 quote_char => '\'', # default '"'
210 escape_char => '"', # default '"'
211 header => [qw(id text sellprice word)], # see later
212 profile => { sellprice => 'sellprice_as_number' },
213 class => 'SL::DB::CsvLine', # if present, map lines to this
216 my $status = $csv->parse;
217 my $hrefs = $csv->get_data;
218 my @objects = $csv->get_objects;
220 my @errors = $csv->errors;
226 Text::CSV offeres already good functions to get lines out of a csv file, but in
227 most cases you will want those line to be parsed into hashes or even objects,
228 so this model just skips ahead and gives you objects.
230 Its basic assumptions are:
234 =item You do know what you expect to be in that csv file.
236 This means first and foremost you have knowledge about encoding, number and
237 date format, csv parameters such as quoting and separation characters. You also
238 know what content will be in that csv and what L<Rose::DB> is responsible for
239 it. You provide valid header columns and their mapping to the objects.
241 =item You do NOT know if the csv provider yields to your expectations.
243 Stuff that does not work with what you expect should not crash anything, but
244 give you a hint what went wrong. As a result, if you remeber to check for
245 errors after each step, you should be fine.
247 =item Data does not make sense. It's just data.
249 Almost all data imports have some type of constraints. Some data needs to be
250 unique, other data needs to be connected to existing data sets. This will not
251 happen here. You will receive a plain mapping of the data into the class tree,
262 Standard constructor. You can use this to set most of the data.
266 Do the actual work. Will return true ($self actually) if success, undef if not.
270 Parse the data into objects and return those.
272 This method will return list or arrayref depending on context.
276 Returns an arrayref of the raw lines as hashrefs.
280 Return all errors that came up during parsing. See error handling for detailed
291 The file which contents are to be read. Can be a name of a physical file or a
292 scalar ref for memory data.
296 Encoding of the CSV file. Note that this module does not do any encoding
297 guessing. Know what your data is. Defaults to utf-8.
305 Same as in L<Text::CSV>
307 =item C<header> \@FIELDS
309 Can be an array of columns, in this case the first line is not used as a
310 header. Empty header fields will be ignored in objects.
312 =item C<profile> \%ACCESSORS
314 May be used to map header fields to custom accessors. Example:
316 { listprice => listprice_as_number }
318 In this case C<listprice_as_number> will be used to read in values from the
321 In case of a One-To-One relationsship these can also be set over
322 relationsships by sparating the steps with a dot (C<.>). This will work:
324 { customer => 'customer.name' }
326 And will result in something like this:
328 $obj->customer($obj->meta->relationship('customer')->class->new);
329 $obj->customer->name($csv_line->{customer})
331 But beware, this will not try to look up anything in the database. You will
332 simply receive objects that represent what the profile defined. If some of
333 these information are unique, and should be connected to preexisting data, you
334 will have to do that for yourself. Since you provided the profile, it is
335 assumed you know what to do in this case.
339 If present, the line will be handed to the new sub of this class,
340 and the return value used instead of the line itself.
342 =item C<ignore_unknown_columns>
344 If set, the import will ignore unkown header columns. Useful for lazy imports,
345 but deactivated by default.
347 =item C<strict_profile>
349 If set, all columns to be parsed must be specified in C<profile>. Every header
350 field not listed there will be treated like an unknown column.
354 =head1 ERROR HANDLING
356 After parsing a file all errors will be accumulated into C<errors>.
357 Each entry is an object with the following attributes:
359 raw_input: offending raw input,
360 code: Text::CSV error code if Text:CSV signalled an error, 0 else,
361 diag: error diagnostics,
362 line: position in line,
363 col: estimated line in file,
365 Note that the last entry can be off, but will give an estimate.
373 sep_char, quote_char, and escape_char are passed to Text::CSV on creation.
374 Changing them later has no effect currently.
378 Encoding errors are not dealt with properly.
384 Dispatch to child objects, like this:
386 $csv = SL::Helper::Csv->new(
388 class => SL::DB::Part,
403 Sven Schöling E<lt>s.schoeling@linet-services.deE<gt>