apertium 3.1.0-1.3 / usr / share / apertium / interchunk.dtd

<?xml version="1.0" encoding="ISO-8859-1"?>
<!-- 
    Draft of DTD for the structural transfer rule files 
     
    Sergio Ortiz, Gema Ramírez-Sánchez, Mireia Ginestí, Mikel L. Forcada, 
    2005.07.29. 
-->    

<!ENTITY % condition "(and|or|not|equal|begins-with|begins-with-list|ends-with|ends-with-list|contains-substring|in)">
<!ENTITY % container "(var|clip)">
<!ENTITY % sentence "(let|out|choose|modify-case|call-macro|append)">
<!ENTITY % value "(b|clip|lit|lit-tag|var|get-case-from|case-of|concat)">
<!ENTITY % stringvalue "(clip|lit|var|get-case-from|case-of)">

<!ELEMENT interchunk (section-def-cats, section-def-attrs, section-def-vars, section-def-lists?, section-def-macros?, section-rules)>
<!-- 
     'interchunk' is the root element containing the whole structural
     interchunk rule file.  
-->

<!ELEMENT section-def-cats (def-cat+)>
<!-- 
     The 'def-cats' section defines the categories used to build the
patterns used in rules
 -->

<!ELEMENT def-cat (cat-item+)>
<!ATTLIST def-cat  n ID #REQUIRED>
<!-- 
     Each 'def-cat' defines one category in terms of a list of
     category items and has a unique name 'n', which is mandatory
-->

<!ELEMENT cat-item EMPTY>
<!ATTLIST cat-item lemma CDATA #IMPLIED 
                   tags CDATA #REQUIRED > 
<!-- 
          Each 'cat-item' (category item) represents a set of lexical forms
     and has a mandatory attribute 'tags' whose value is a sequence of
     dot-separated tag names; this sequence is a subsequence of the
     tag sequence defining each possible lexical form. For example,
     tags="n.f" would match all lexical forms containing this tag
     sequence, such as "^casa<n><f><pl>$".

     In addition, an optional attribute, "lemma", may be used to
     define lexical forms having a particular substring in their lemma
-->
 
<!ELEMENT section-def-attrs (def-attr+)>

<!-- 
     The 'def-attrs' section defines the attributes that will be
     identified in matched lexical forms 
-->

<!ELEMENT def-attr (attr-item+)>
<!ATTLIST def-attr n ID #REQUIRED>
<!-- 
     Each def-attr defines one attribute in terms of a list of
     attribute items and has a mandatory unique name n 
-->

<!ELEMENT attr-item EMPTY>
<!ATTLIST attr-item tags CDATA #IMPLIED>
<!-- 
     Each 'attr-item' specifies a subsequence of the tags in
     that lexical form (attribute 'tags')
-->

<!ELEMENT section-def-vars (def-var+)>
<!-- 
     The 'def-vars' section defines the global variables
     that will be used to transfer information between rules
-->

<!ELEMENT def-var EMPTY>
<!ATTLIST def-var n ID #REQUIRED
                  v CDATA #IMPLIED>
<!-- 
     The definition of a global variable has a mandatory unique name 'n' that
     will be used to refer to it. A value of initialization can also be specified
     by means the 'v' attribute.  The default value of the initialization is the
     empty string.
-->

<!ELEMENT section-def-lists (def-list)+>
<!--
     Element 'section-def-lists' encloses a set of list definitions
-->

<!ELEMENT def-list (list-item+)>
<!ATTLIST def-list n ID #REQUIRED>
<!--
     The 'def-list' element defines a named list to search with the 'in' 
     element.  Attribute 'n' sets the name of the list
-->

<!ELEMENT list-item EMPTY>
<!ATTLIST list-item v CDATA #REQUIRED>
<!--
     Attribute 'v' of 'list-item' element contains the value to be added to 
     the list being defined     
-->

<!ELEMENT section-def-macros (def-macro)+>
<!-- 

     The 'def-macros' section defines macros containing portions of
     code frequently used in the action part of rules

-->

<!ELEMENT def-macro (%sentence;)+>
<!ATTLIST def-macro n ID #REQUIRED>
<!ATTLIST def-macro npar CDATA #REQUIRED>
<!-- 
     Macro definition:
     
     A macro has a mandatory name (the value of 'n'), a number of parameters
     (the value of 'npar') and a body containing arguments and statements.  
-->

<!ELEMENT section-rules (rule+)>
<!-- 
     The rules section contains a sequence of one or more rules
-->

<!ELEMENT rule (pattern, action)>
<!ATTLIST rule comment CDATA #IMPLIED>
<!-- 
      Each rule has a pattern and an action 
      * attribute 'comment' allows to put in comments about the purpose of
        the rule being defined
-->

<!ELEMENT pattern (pattern-item+)>
<!-- 
The pattern is specified in terms of pattern items, each one
representing a lexical form in the matched pattern 
-->

<!ELEMENT pattern-item EMPTY>
<!ATTLIST pattern-item n IDREF #REQUIRED>
<!-- 
       Each attribute to be activated is referred to by its name in the def-cats section 
-->

<!ELEMENT action (%sentence;)*>
<!-- 
       Encloses the procedural part of a rule
-->

<!ELEMENT choose (when+,otherwise?)>
<!-- 
     The choose statement is a selection statement (similar to a case
     statement) composed of one or more tested cases and an optional
     otherwise 
-->

<!ELEMENT when (test,(%sentence;)*)>
<!-- 
     Each tested case is a block of zero or more statements 
-->

<!ELEMENT otherwise (%sentence;)+>
<!-- 
     The otherwise case is also a block of one or more statements 
-->

<!ELEMENT test (%condition;)>
<!-- 
     The test in a tested case may be a conjunction, a disjunction, or
     a negation of simpler tests, as well as a simple equality test
-->

<!ELEMENT and ((%condition;),(%condition;)+)>
<!--  
     Each conjuntion test contains two or more simpler tests 
-->

<!ELEMENT or ((%condition;),(%condition;)+)>
<!-- 
     Each disjunction test contains two or more simpler tests 
-->

<!ELEMENT not (%condition;)>
<!-- 
     The negation of a simpler test is a test itself 
-->

<!ELEMENT equal (%value;,%value;)> 
<!ATTLIST equal caseless (no|yes) #IMPLIED>
<!-- 
      The simplest test is an equality test. The right part and the
      left part of the equality may both be a clip (see below), a
      literal string ('lit'), a literal tag ('lit-tag') or the value of 
      a variable ('var') defined in the def-vars section.  When the attribute
      'caseless' is set to 'yes', the comparison is made without attending
      to the case.
-->

<!ELEMENT begins-with (%value;,%value;)> 
<!ATTLIST begins-with caseless (no|yes) #IMPLIED>
<!-- 
      Tests if the left part contains the right part at the beginning.
      Both parts of the test may both be a clip (see below), a
      literal string ('lit'), a literal tag ('lit-tag') or the value of 
      a variable ('var') defined in the def-vars section.  When the attribute
      'caseless' is set to 'yes', the comparison is made without attending
      to the case.
-->

<!ELEMENT ends-with (%value;,%value;)> 
<!ATTLIST ends-with caseless (no|yes) #IMPLIED>
<!-- 
      Tests if the left part contains the right part at the end.
      Both parts of the test may both be a clip (see below), a
      literal string ('lit'), a literal tag ('lit-tag') or the value of 
      a variable ('var') defined in the def-vars section.  When the attribute
      'caseless' is set to 'yes', the comparison is made without attending
      to the case.
-->

<!ELEMENT begins-with-list (%value;,list)> 
<!ATTLIST begins-with-list caseless (no|yes) #IMPLIED>
<!-- 
      Tests if the left part contains the right part at the beginning.
      First parts of the test may be a clip (see below), a
      literal string ('lit'), a literal tag ('lit-tag') or the value of 
      a variable ('var') defined in the def-vars section. The second part
      must be always a list.  When the attribute
      'caseless' is set to 'yes', the comparison is made without attending
      to the case.
-->


<!ELEMENT ends-with-list (%value;,list)> 
<!ATTLIST ends-with-list caseless (no|yes) #IMPLIED>
<!-- 
      Tests if the left part contains the right part at the end.
      First parts of the test may be a clip (see below), a
      literal string ('lit'), a literal tag ('lit-tag') or the value of 
      a variable ('var') defined in the def-vars section. The second part
      must be always a list.  When the attribute
      'caseless' is set to 'yes', the comparison is made without attending
      to the case.
-->


<!ELEMENT contains-substring (%value;,%value;)> 
<!ATTLIST contains-substring caseless (no|yes) #IMPLIED>
<!-- 
      Tests if the left part contains the right part.
      Both parts of the test may both be a clip (see below), a
      literal string ('lit'), a literal tag ('lit-tag') or the value of 
      a variable ('var') defined in the def-vars section.  When the attribute
      'caseless' is set to 'yes', the comparison is made without attending
      to the case.
-->




<!ELEMENT in (%value;, list)>
<!ATTLIST in caseless (no|yes) #IMPLIED>
<!--
    'in' performs a search of a value in a list.  If 'caseless' is set to yes,
    this search is performed without attending to the case
-->

<!ELEMENT list EMPTY>
<!ATTLIST list n IDREF #REQUIRED>
<!--
    'list' refers, with the name in attribute 'n', a list defined before in
    the 'section-def-list' section
-->

<!ELEMENT let (%container;, %value;)>
<!-- 
      An assignment statement ('let') assigns the value of a clip (see
      below), a literal string ('lit'), a literal tag('lit-tag') or the 
      value of a global variable ('var') to either a global variable ('var') 
      or a clip
-->

<!ELEMENT append (%value;)+>
<!ATTLIST append n IDREF #REQUIRED>
<!-- 
      This instruction appends the value of a clip (see
      below), a literal string ('lit'), a literal tag('lit-tag') or the 
      value of a global variable ('var') to either a global variable ('var') 
      or a clip, identified by the "n" attribute
-->


<!ELEMENT out (b|chunk|var)+>
<!-- 
      'out' is an output statement; it may output blanks or chunks
-->

<!ELEMENT modify-case (%container;, %stringvalue;)>
<!--
      The first argument of 'modify-case' copy the case of the second 
      argument.
--> 

<!ELEMENT call-macro (with-param)*>
<!ATTLIST call-macro n IDREF #REQUIRED>
<!-- 
      A macro may be called anywhere by name with one or more
      arguments
-->

<!ELEMENT with-param EMPTY>
<!ATTLIST with-param pos CDATA #REQUIRED>
<!-- 
      The attribute pos in each argument is used to refer to a lexical
      form in the current rule. For example, if a 2-parameter macro
      has been defined to perform noun-adjective agreement operations,
      it may be used with arguments 1 and 2 in a noun-adjective rule,
      with arguments 2, 3 and 1 in a determiner-noun-adjective rule, with
      arguments 1 and 3 in a noun-adverb-adjective rule, and with
      arguments 2 and 1 in an adjective-noun rule 
-->

<!ELEMENT clip EMPTY>
<!ATTLIST clip pos CDATA #REQUIRED
               part CDATA #REQUIRED>
<!-- 
      A 'clip' is a substring of a source-language or target-language
      lexical form, extracted according to an attribute:

      * 'pos' is an index (1, 2, 3...) used to select a lexical form
         inside the rule;
   
      * the value of 'part' is the name of an attribute defined in
        def-attrs, but may take also the values 'lem' (referring to
        the lemma of the lexical form), 'lemh' (lemma head), 'lemq'
        (lemma queue) and 'whole' (referring to the whole lexical form).

-->

<!ELEMENT lit EMPTY>
<!ATTLIST lit v CDATA #REQUIRED>
<!-- 
      A literal string value: the value of the literal is the value of
      the 'v' attribute
-->

<!ELEMENT lit-tag EMPTY>
<!ATTLIST lit-tag v CDATA #REQUIRED>
<!-- 
      A literal string value: the value of the literal is the value of
      the 'v' attribute
-->


<!ELEMENT var EMPTY>
<!ATTLIST var n IDREF #REQUIRED>
<!-- 
     Each 'var' is a variable identifier: the attribute n is the name
     of the variable. When it is in an 'out', a 'test', or the right
     part of a 'let', it represents the value of the variable; when in
     the left part of a 'let' it represents the reference of the
     variable. 
-->

<!ELEMENT get-case-from (clip|lit|var)> 
<!ATTLIST get-case-from pos CDATA #REQUIRED>
<!-- Atención, falta modificar todos los comentarios donde intervenga
get-case-from -->

<!ELEMENT case-of EMPTY>
<!ATTLIST case-of pos CDATA #REQUIRED
               part CDATA #REQUIRED>
<!--
      A 'case-of' is a value representing the case of a "clip".  This value 
      will be "aa" (all lowercase), "Aa" (first uppercase) and "AA",
      (all uppercase).

      * 'pos' is an index (1, 2, 3...) used to select a lexical form
         inside the rule;
   
      * the value of 'part' is the name of an attribute defined in
        def-attrs, but may take also the values 'lem' (referring to
        the lemma of the lexical form), 'lemh' (lemma head), 'lemq'
        (lemma queue) and 'whole' (referring to the whole lexical form).
-->


<!ELEMENT concat (%value;)+>
<!-- Concatenates a sequence of values -->

<!ELEMENT chunk (%value;)+>
<!-- 
     Encloses a chunk inside an 'out' element.      
-->

<!ELEMENT pseudolemma (%value;)>

<!ELEMENT b EMPTY>
<!ATTLIST b pos CDATA #IMPLIED>
<!-- 
     'b' is a [super]blanks item, indexed by pos; for example, a 'b'
     with pos="2" refers to the [super]blanks (including format data
     encapsulated by the de-formatter) between lexical form 2 and
     lexical form 3. Managing [super]blanks explicitly allows for the
     correct placement of format when the result of structural
     transfer has more or less lexical items than the original or has
     been reordered in some way.  If attribute "pos" is not specified, then
     a single blank (ASCII 32) is generated.
-->