NAME
Data::Stag - Structured Tags datastructuresSYNOPSIS
# PROCEDURAL USAGE
use Data::Stag qw(:all);
$doc = stag_parse($file);
@persons = stag_findnode($doc, "person");
foreach $p (@persons) {
printf "%s, %s phone: %s\n",
stag_sget($p, "family_name"),
stag_sget($p, "given_name"),
stag_sget($p, "phone_no"),
;
}
# OO USAGE
use Data::Stag;
$doc = Data::Stag->new->parse($file);
@persons = $doc->findnode("person");
foreach $p (@person) {
printf "%s, %s phone:%s\n",
$p->sget("family_name"),
$p->sget("given_name"),
$p->sget("phone_no"),
;
}DESCRIPTION
This module is for manipulating data as recursively nested tag/value pairs (Structured TAGs or Simple Tree AGgreggates). These datastructures can be represented as nested arrays, which have the advantage of being native to perl. A simple example is shown below:
[ person=> [ [ family_name => $family_name ],
[ given_name => $given_name ],
[ phone_no => $phone_no ] ] ],Data::Stag uses a subset of XML for import and export. This means the module can also be used as a general XML parser/writer (with certain caveats).
The above set of structured tags can be represented in XML as
<person>
<family_name>...</family_name>
<given_name>...</given_name>
<phone_no>...</phone_no>
</person>Querying is performed by passing functions, for example:
# get all people in dataset with name starting 'A'
@persons =
$document->where('person',
sub {shift->sget('family_name') =~ /^A/});One of the things that marks this module out against other XML modules is this emphasis on a functional approach as opposed to an OO approach (it may appeal to Lisp programmers).
PROCEDURAL VS OBJECT ORIENTED USAGE
Depending on your preference, this module can be used a set of procedural subroutine calls, or as method calls upon Data::Stag objects, or both.
In procedural mode, all the subroutine calls are prefixed "stag_" to avoid namespace clashes. The following two calls are equivalent:
stag_findnode($doc, "person");
$doc->findnode("person");In object mode, you can treat any tree element as if it is an object with automatically defined methods for getting/setting the tag values.
USE OF XML
Nested arrays can be imported and exported as XML, as well as other formats. XML can be slurped into memory all at once (using less memory than an equivalent DOM tree), or a simplified SAX style event handling model can be used. Similarly, data can be exported all at once, or as a series of events.
Although this module can be used as a general XML tool, it is intended primarily as a tool for manipulating complex data using nested tag/value pairs.
By using a simpler subset of XML that can be treated as equivalent to a basic data tree structure, we can write simpler, cleaner code. This simplicity comes at a price - this module is not very suitable for XML with attributes or mixed content.
All attributes are turned into elements. This means that it will not round-trip a piece of xml with attributes in it. For some applications this is acceptable, for others it is not.
Mixed content cannot be represented in a simple tree format, so this is also expanded.
The following piece of XML
<paragraph id="1">
example of <bold>mixed</bold>content
</paragraph>gets parsed as if it were actually:
<paragraph>
<paragraph-id>1</paragraph-id>
<paragraph-text>example of</paragraph-text>
<bold>mixed</bold>
<paragraph-text>content</paragraph-text>
</paragraph>This module is more suited to dealing with complex datamodels than dealing with marked up text
It can also be used as part of a SAX-style event generation / handling framework - see XML::NestedArray::Base
Because nested arrays are native to perl, we can specify an XML datastructure directly in perl without going through multiple object calls.
For example, instead of the lengthy
$obj->startTag("record");
$obj->startTag("field1");
$obj->characters("foo");
$obj->endTag("field1");
$obj->startTag("field2");
$obj->characters("bar");
$obj->endTag("field2");
$obj->end("record");We can instead write
$struct = [ record => [
[ field1 => 'foo'],
[ field2 => 'bar']]];If this appeals to you, then maybe this module is for you.
PARSING
parsing out subsections of a tree and changing sub-elements
use Data::Stag qw(:all);
my $tree = stag_from('xml', $xmlfile);
my ($subtree) = stag_findnode($tree, $element);
stag_set($element, $sub_element, $new_val);
print stag_xml($subtree);OBJECT ORIENTED
the same can be done in a more OO fashion
use Data::Stag qw(:all);
my $tree = Data::Stag->from('xml', $xmlfile);
my ($subtree) = $tree->findnode($element);
$element->set($sub_element, $new_val);
print $subtree->xml;IN A STREAM
use Data::Stag::XMLParser;
use MyTransform; # inherits from XML::NestedArray::Base
my $p = Data::Stag::XMLParser->new;
my $h = MyTransform->new; # create a handler
$p->handler($h);
$p->parse($xmlfile);The above can be simplified like this:
use Data::Stag;
use MyTransform; # inherits from XML::NestedArray::Base
my $h = MyTransform->new;
Data::Stag->new->parse(-file=>$xmlfile, -handler=>$h);see XML::NestedArray::Base for writing handlers
STRUCTURED TAGS TREE DATA STRUCTURE
A tree of structured tags is represented as a recursively nested array, the elements of the array represent nodes in the tree.
A node is a name/data pair, that can represent tags and values. A node is represented using a reference to an array, where the first element of the array is the tagname, or element, and the second element is the data
This can be visualised as a box:
+-----------+
|Name | Data|
+-----------+In perl, we represent this pair as a reference to an array
[ Name => $Data ]The Data can either be a list of child nodes (subtrees), or a data value.
The terminal nodes (leafs of the tree) contain data values; this is represented in perl using primitive scalars.
For example:
[ Name => 'Fred' ]For non-terminal nodes, the Data is a reference to an array, where each element of the the array is a new node.
+-----------+
|Name | Data|
+-----------+
||| +-----------+
||+-->|Name | Data|
|| +-----------+
||
|| +-----------+
|+--->|Name | Data|
| +-----------+
|
| +-----------+
+---->|Name | Data|
+-----------+In perl this would be:
[ Name => [
[Name1 => $Data1],
[Name2 => $Data2],
[Name3 => $Data3],
]
];The extra level of nesting is required to be able to store any node in the tree using a single variable. This representation has lots of advantages over others, eg hashes and mixed hash/array structures.
MANIPULATION AND QUERYING
The following example is taken from molecular biology; we have a list of species (mouse, human, fly) and a list of genes found in that species. These are cross-referenced by an identifier called tax_id. We can do a relational-style natural join on this identifier, as follows -
use Data::Stag qw(:all);
my $tree =
[ 'db' => [
[ 'species_set' => [
[ 'species' => [
[ 'common_name' => 'house mouse' ],
[ 'binomial' => 'Mus musculus' ],
[ 'tax_id' => '10090' ]]],
[ 'species' => [
[ 'common_name' => 'fruit fly' ],
[ 'binomial' => 'Drosophila melanogaster' ],
[ 'tax_id' => '7227' ]]],
[ 'species' => [
[ 'common_name' => 'human' ],
[ 'binomial' => 'Homo sapiens' ],
[ 'tax_id' => '9606' ]]]]],
[ 'gene_set' => [
[ 'gene' => [
[ 'symbol' => 'HGNC' ],
[ 'tax_id' => '9606' ],
[ 'phenotype' => 'Hemochromatosis' ],
[ 'phenotype' => 'Porphyria variegata' ],
[ 'GO_term' => 'iron homeostasis' ],
[ 'map' => '6p21.3' ]]],
[ 'gene' => [
[ 'symbol' => 'Hfe' ],
[ 'synonym' => 'MR2' ],
[ 'tax_id' => '10090' ],
[ 'GO_term' => 'integral membrane protein' ],
[ 'map' => '13 A2-A4' ]]]]]]];
# natural join of species and gene parts of tree,
# based on 'tax_id' element
my ($gene_set) = $tree->findnode("gene_set");
my ($species_set) = $tree->findnode("species_set");
$gene_set->njoin("gene", "tax_id", $species_set);
print $gene_set->xml;
# find all genes starting with H in human
my @genes =
$gene_set->where('gene',
sub { my $g = shift;
$g->get_symbol =~ /^H/ &&
$g->findval("common_name") eq ('human')});S-Expression (Lisp) representation
The data represented using this module can be represented as Lisp-style S-Expressions.
See Data::Stag::SxprParser and Data::Stag::SxprWriter
If we execute this line
print $tree->sxpr;The following S-Expression will be printed:
'(db
(species_set
(species
(common_name "house mouse")
(binomial "Mus musculus")
(tax_id "10090"))
(species
(common_name "fruit fly")
(binomial "Drosophila melanogaster")
(tax_id "7227"))
(species
(common_name "human")
(binomial "Homo sapiens")
(tax_id "9606")))
(gene_set
(gene
(symbol "HGNC")
(tax_id "9606")
(phenotype "Hemochromatosis")
(phenotype "Porphyria variegata")
(GO_term "iron homeostasis")
(map
(cytological
(chromosome "6")
(band "p21.3"))))
(gene
(symbol "Hfe")
(synonym "MR2")
(tax_id "10090")
(GO_term "integral membrane protein")))
(similarity_set
(pair
(symbol "HGNC")
(symbol "Hfe"))
(pair
(symbol "WNT3A")
(symbol "Wnt3a"))))TIPS FOR EMACS USERS AND LISP PROGRAMMERS
If you use emacs, you can save this as a file with the ".el" suffix and get syntax highlighting for editing this file. Quotes around the terminal node data items are optional.
If you know emacs lisp or any other lisp, this also turns out to be a very nice language for manipulating these datastructures. Try copying and pasting the above s-expression to the emacs scratch buffer and playing with it!
I think this module turns out to be a very nice way using my two favourite lnaguages, lisp and perl together.
INDENTED TEXT REPRESENTATION
Data::Stag has its own text format for writing data trees. Again, this is only possible because we are working with a subset of XML (no attributes, no mixed elements). The data structure above can be written as follows -
db:
species_set:
species:
common_name: house mouse
binomial: Mus musculus
tax_id: 10090
species:
common_name: fruit fly
binomial: Drosophila melanogaster
tax_id: 7227
species:
common_name: human
binomial: Homo sapiens
tax_id: 9606
gene_set:
gene:
symbol: HGNC
tax_id: 9606
phenotype: Hemochromatosis
phenotype: Porphyria variegata
GO_term: iron homeostasis
map: 6p21.3
gene:
symbol: Hfe
synonym: MR2
tax_id: 10090
GO_term: integral membrane protein
map: 13 A2-A4
similarity_set:
pair:
symbol: HGNC
symbol: Hfe
pair:
symbol: WNT3A
symbol: Wnt3aSee Data::Stag::ITextParser and Data::Stag::ITextWriter
NESTED ARRAY SPECIFICATION II
To avoid excessive square bracket usage, you can specify a structure like this:
use Data::Stag qw(:all);
*N = \&stag_new;
my $tree =
N(top=>[
N('personset'=>[
N('person'=>[
N('name'=>'davey'),
N('address'=>'here'),
N('description'=>[
N('hair'=>'green'),
N('eyes'=>'two'),
N('teeth'=>5),
]
),
N('pets'=>[
N('petname'=>'igor'),
N('petname'=>'ginger'),
]
),
],
),
N('person'=>[
N('name'=>'shuggy'),
N('address'=>'there'),
N('description'=>[
N('hair'=>'red'),
N('eyes'=>'three'),
N('teeth'=>1),
]
),
N('pets'=>[
N('petname'=>'thud'),
N('petname'=>'spud'),
]
),
]
),
]
),
N('animalset'=>[
N('animal'=>[
N('name'=>'igor'),
N('class'=>'rat'),
N('description'=>[
N('fur'=>'white'),
N('eyes'=>'red'),
N('teeth'=>50),
],
),
],
),
]
),
]
);
# find all people
my @persons = stag_findnode($tree, 'person');
# write xml for all red haired people
foreach my $p (@persons) {
print stag_xml($p)
if stag_tmatch("hair", "red");
} ;
# find all people called shuggy
my @p =
stag_qmatch($tree,
"person",
"name",
"shuggy");NODES AS DATA OBJECTS
As well as the methods listed below, a node can be treated as if it is a data object of a class determined by the element.
For example, the following are equivalent.
$node->get_name;
$node->get('name');
$node->set_name('fred');
$node->set('name', 'fred');This is really just syntactic sugar. The autoloaded methods are not checked against any schema, although this may be added in future.
One addition slated for a future release is the ability to give particular elements certain behaviour, and allow inheritance and all that kind of thing.
fullname: $obj->given_name . ' ' . $obj->family_name;Although it is the module authors preference to avoid this kind of OO paradigm, and instead enforce a cleaner seperation of code from data, utilising a more functional style of programming.
METHODS
All method calls are also available as procedural subroutine calls; unless otherwise noted, the subroutine call is the same as the method call, but with the string stag_ prefixed to the method name. The first argument should be a Data::Stag datastructure.
To import all subroutines into the current namespace, use this idiom:
use Data::Stag qw(:all);If you wish to use this module procedurally, and you are too lazy to prefix all calls with stag_, use this idiom:
use Data::Stag qw(:lazy);MNEMONICS
Most method calls also have a handy short mnemonic. Use of these is optional. Software engineering types prefer longer names, in the belief that this leads to clearer code. Hacker types prefer shorter names, as this requires less keystrokes, and leads to a more compact representation of the code. It is expected that if you do use this module, then its usage will be fairly ubiquitous within your code, and the mnemonics will become familiar, much like the qw and s/ operators in perl. As always with perl, the decision is yours.
INITIALIZATION METHODS
new
Title: new
Args: element str, data ANY
Returns: Data::Stag node
Example: $node = stag_new();
Example: $node = Data::Stag->new;
Example: $node = Data::Stag->new(person => [[name=>$n], [phone=>$p]]);creates a new instance of a Data::Stag node
nodify
Title: nodify
Args: data array-reference
Returns: Data::Stag node
Example: $node = stag_nodify([person => [[name=>$n], [phone=>$p]]]);turns a perl array reference into a Data::Stag node.
similar to new
parse
Title: parse
Args: file str, [format str], [handler obj]
Returns: Data::Stag node
Example: $node = stag_parse($fn);
Example: $node = Data::Stag->parse(-file=>$fn, -handler=>$myhandler);slurps a file or string into a Data::Stag node structure. Will guess the format from the suffix if it is not given.
The format can also be the name of a parsing module, or an actual parser object
from
Title: from
Args: format str, source str
Returns: Data::Stag node
Example: $node = stag_from('xml', $fn);
Example: $node = stag_from('xmlstr', q[<top><x>1</x></top>]);
Example: $node = Data::Stag->from($parser, $fn);Similar to parse
slurps a file or string into a Data::Stag node structure.
The format can also be the name of a parsing module, or an actual parser object
unflatten
Title: unflatten
Args: data array
Returns: Data::Stag node
Example: $node = stag_unflatten(person=>[name=>$n, phone=>$p, address=>[street=>$s, city=>$c]]);Creates a node structure from a semi-flattened representation, in which children of a node are represented as a flat list of data rather than a list of array references.
This means a structure can be specified as:
person=>[name=>$n,
phone=>$p,
address=>[street=>$s,
city=>$c]]Instead of:
[person=>[ [name=>$n],
[phone=>$p],
[address=>[ [street=>$s],
[city=>$c] ] ]
]
]The former gets converted into the latter for the internal representation
RECURSIVE SEARCHING
findnode (fn)
Title: findnode
Synonym: fn
Args: element str
Returns: node[]
Example: @persons = stag_findnode($struct, 'person');
Example: @persons = $struct->findnode('person');recursively searches tree for all elements of the given type, and returns all nodes found.
findval (fv)
Title: findval
Synonym: fv
Args: element str
Returns: ANY
Example: @names = stag_findval($struct, 'name');
Example: @names = $struct->findval('name');recursively searches tree for all elements of the given type, and returns all data values found. the data values could be primitive scalars or nodes.
sfindval (sfv)
Title: sfindval
Synonym: sfv
Args: element str
Returns: ANY
Example: $name = stag_sfindval($struct, 'name');
Example: $name = $struct->sfindval('name');as findval, but returns the first value found
findvallist (fvl)
Title: findvallist
Synonym: fvl
Args: element str[]
Returns: ANY[]
Example: ($name, $phone) = stag_findvallist($personstruct, 'name', 'phone');
Example: ($name, $phone) = $personstruct->findvallist('name', 'phone');recursively searches tree for all elements in the list
DEPRECATED?
DATA ACCESSOR METHODS
these allow getting and setting of elements directly underneath the current one
get (g)
Title: get
Synonym: g
Args: element str
Return: ANY
Example: $name = $person->get('name');
Example: @phone_nos = $person->get('phone_no');gets the data value of an element for any node
the examples above would work on a data structure like this:
[person => [ [name => 'fred'],
[phone_no => '1-800-111-2222'],
[phone_no => '1-415-555-5555']]]will return an array or single value depending on the context
sget (sg)
Title: sget
Synonym: sg
Args: element str
Return: ANY
Example: $name = $person->get('name');
Example: $phone = $person->get('phone_no');as get but always returns a single value
gl (getl getlist)
Title: gl
Synonym: getl
Synonym: getlist
Args: element str[]
Return: ANY[]
Example: ($name, @phone) = $person->get('name', 'phone_no');returns the data values for a list of sub-elements of a node
getn (gn getnode)
Title: getn
Synonym: gn
Synonym: getnode
Args: element str
Return: node[]
Example: $namestruct = $person->getn('name');
Example: @pstructs = $person->getn('phone_no');as get but returns the whole node rather than just the data valie
set (s)
Title: set
Synonym: s
Args: element str, datavalue ANY
Return: ANY
Example: $person->set('name', 'fred');
Example: $person->set('phone_no', $cellphone, $homephone);sets the data value of an element for any node. if the element is multivalued, all the old values will be replaced with the new ones specified.
ordering will be preserved, unless the element specified does not exist, in which case, the new tag/value pair will be placed at the end.
unset (u)
Title: unset
Synonym: u
Args: element str, datavalue ANY
Return: ANY
Example: $person->unset('name');
Example: $person->unset('phone_no');prunes all nodes of the specified element from the current node
add (a)
Title: add
Synonym: a
Args: element str, datavalue ANY[]
Return: ANY
Example: $person->add('phone_no', $cellphone, $homephone);adds a datavalue or list of datavalues. appends if already existing, creates new element value pairs if not already existing.
element (e name)
Title: element
Synonym: e
Synonym: name
Args:
Return: element str
Example: $element = $struct->elementreturns the element name of the current node
kids (k children)
Title: kids
Synonym: k
Synonym: children
Args:
Return: ANY or ANY[]
Example: @nodes = $person->kids
Example: $name = $namestruct->kidsreturns the data value(s) of the current node; if it is a terminal node, returns a single value which is the data. if it is non-terminal, returns an array of nodes
addkid (ak addchild)
Title: addkid
Synonym: ak
Synonym: addchild
Args: kid node
Return: ANY
Example: $person->addkid('job', $job);adds a new child node to a non-terminal node, after all the existing child nodes
subnodes
Title: subnodes
Args:
Return: ANY[]
Example: @nodes = $person->subnodesreturns the non-terminal data value(s) of the current node;
QUERYING AND ADVANCED DATA MANIPULATION
njoin (j)
Title: njoin
Synonym: j
Synonym: nj
Args: element str
Return: undefdoes a relational style natural join - see previous example in this doc
qmatch (qm)
Title: qmatch
Synonym: qm
Args: return-element str, match-element str, match-value str
Return: node[]
Example: @persons = $s->qmatch('name', 'fred');queries the node tree for all elements that satisfy the specified key=val match
tmatch (tm)
Title: tmatch
Synonym: tm
Args: element str, value str
Return: bool
Example: @persons = grep {$_->tmatch('name', 'fred')} @personsreturns true if the the value of the specified element matches
tmatchhash (tmh)
Title: tmatchhash
Synonym: tmh
Args: match hashref
Return: bool
Example: @persons = grep {$_->tmatchhash({name=>'fred', hair_colour=>'green'})} @personsreturns true if the node matches a set of constraints, specified as hash
tmatchnode (tmn)
Title: tmatchnode
Synonym: tmn
Args: match node
Return: bool
Example: @persons = grep {$_->tmatchhash([person=>[[name=>'fred'], [hair_colour=>'green']]])} @personsreturns true if the node matches a set of constraints, specified as node
cmatch (cm)
Title: cmatch
Synonym: cm
Args: element str, value str
Return: bool
Example: $n_freds = $personset->cmatch('name', 'fred');counts the number of matches
where (w)
Title: where
Synonym: w
Args: element str, test CODE
Return: Node[]
Example: @rich_persons = $data->where('person', sub {shift->get_salary > 100000});the tree is queried for all elements of the specified type that satisfy the coderef (must return a boolean)
my @rich_dog_or_cat_owners =
$data->where('person',
sub {my $p = shift;
$p->get_salary > 100000 &&
$p->where('pet',
sub {shift->get_type =~ /(dog|cat)/})});MISCELLANEOUS METHODS
duplicate (d)
Title: duplicate
Synonym: d
Args:
Return: Node
Example: $node2 = $node->duplicate;isanode
Title: isanode
Args:
Return: bool
Example: if (stag_isanode($node)) { ... }really only useful in non OO mode...
hash
Title: hash
Args:
Return: hash
Example: $h = $node->hash;turns a tree into a hash. all data values will be arrayrefs
pairs
Title: pairsturns a tree into a hash. all data values will be scalar (IMPORTANT: this means duplicate values will be lost)
write
Title: write
Args: filename str, format str[optional]
Return:
Example: $node->write("myfile.xml");
Example: $node->write("myfile", "itext");will try and guess the format from the extension if not specified
xml
Title: xml
Args: filename str, format str[optional]
Return:
Example: $node->write("myfile.xml");
Example: $node->write("myfile", "itext");
Args:
Return: xml str
Example: print $node->xml;XML METHODS
sax
Title: sax
Args: saxhandler SAX-CLASS
Return:
Example: $node->sax($mysaxhandler);turns a tree into a series of SAX events
xpath (xp tree2xpath)
Title: xpath
Synonym: xp
Synonym: tree2xpath
Args:
Return: xpath object
Example: $xp = $node->xpath; $q = $xp->find($xpathquerystr);xpquery (xpq xpathquery)
Title: xpquery
Synonym: xpq
Synonym: xpathquery
Args: xpathquery str
Return: Node[]
Example: @nodes = $node->xqp($xpathquerystr);BUGS
none known so far, possibly quite a few undocumented features!
Not a bug, but the underlying default datastructure of nested arrays is more heavyweight than it needs to be. More lightweight implementations are possible. Given time I would like to write the underlying guts in C.
WEBSITE
http://stag.sourceforge.net
AUTHOR
Chris Mungall <cjm@fruitfly.org>
COPYRIGHT
Copyright (c) 2002 Chris Mungall
This module is free software. You may distribute this module under the same terms as perl itself