X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2Fmathql%2Fdoc%2Fmathql_introduction_core.tex;h=05f9e9fd93f8836e90b6e90c0a676171557a7a28;hb=4167cea65ca58897d1a3dbb81ff95de5074700cc;hp=fbb2d95f527a8e4f161eab2fcc93d9515e309cee;hpb=468da7af4b52d01451073ff1cca5aa1949b9657f;p=helm.git diff --git a/helm/mathql/doc/mathql_introduction_core.tex b/helm/mathql/doc/mathql_introduction_core.tex index fbb2d95f5..05f9e9fd9 100644 --- a/helm/mathql/doc/mathql_introduction_core.tex +++ b/helm/mathql/doc/mathql_introduction_core.tex @@ -3,65 +3,160 @@ {\MathQL}.4 consists of a core language and of a basic library. Other user-defined libraries can be added at will. The core language includes the \TT{property} operator mentioned in \subsecref{HighAccess} that queries the -underlying {\RDF} database and the infrastructure to post-process the +underlying {\RDF} database, and the infrastructure to post-process the query results. The components of this infrastructure are listed below: \begin{itemize} \item -Explicit sets of attributed values. - +\textbf{Explicit sets of attributed values.} An explicit {\av} set can be placed in a query in two forms: -as a single quoted string, like \verb+"this is a query result"+, that -evaluates in a single {\av} with that value and no attributes, or as a full -{\av} set in the syntax shown in the previous sections but sorrounded by -square brackets, like \verb+["head" attr {"attribute-name" = "contents"}]+. -\newline -In the second form, the contents of an attribute can be the result of a query, -like -\verb+["head" attr {"attribute-name" = property /"metadata" of "resource"}]+. +as a quoted string, like \verb+"this is a query result"+, that evaluates in a +single {\av} with that value and no attributes, or as a full {\av} set in the +syntax shown in the previous sections but surrounded by square brackets, like +\verb+["head" attr {/"attribute" = "contents"}]+. \newline -In this case the contents of the attribute are the head strings of the query -result, whose attributes (if any) are discarded. +In the second form, the contents of an attribute can be the result of a query +and in this case the contents of the attribute are the head strings of the +query result, whose attributes (if any) are discarded. +\end{itemize} -\item -Variable assignment. +\begin{center} +\verb+["head" attr {/"attribute" = property /"metadata" of "resource"}]+ +\end{center} -Variables for {\av} sets (called \emph{set variables}) can be assigned using -a standard \emph{let-in} construction and may appear wherever an {\av} set -({\ie} a query result) is allowed. -\newline +\begin{itemize} + +\item +\textbf{Variable assignment.} +Variables for {\av} sets (preceded by a \TT{\$} sign and called +\emph{set variables}) can be assigned using a standard \emph{let-in} +construction and may appear wherever an {\av} set ({\ie} a query result) is +allowed. The assignment has the form: \TT{let \$}\EM{variable} \TT{=} \EM{av-set} \TT{in} \EM{av-set} so we can write: \newline -\verb+let $var = "contents" in ["head" attr {"attribute-name" = $var}]+. +\verb+let $var = "contents" in ["head" attr {/"attribute" = $var}]+. -The scope rules of {\MathQL} variables are tipical for an imperative -programming language and any case of assignment propagation will be indicated. +The scope of {\MathQL} variables is typical for an imperative programming +language and any case of assignment propagation will be indicated. \item -Sequential composition. - +\textbf{Sequential composition.} This construction has the form: \EM{av-set} \TT{;;} \EM{av-set} and works as follows: the two {\av} sets are evaluated one after the other and the first one is discarded but the variables assigned in the first {\av} set are available to the second one. \item -Bounded iteration. +\textbf{Unbounded iteration.} +This construction comes in two forms: +\TT{while} \EM{av-set-1} \TT{sup} \EM{av-set-2}: +iterates the evaluation of \EM{av-set-2} until \EM{av-set-1} is empty and +returns the {\MathQL} set-theoretic union of all the evaluations of +\EM{av-set-2}. +\TT{while} \EM{av-set-1} \TT{inf} \EM{av-set-2}: +like the former but the set-theoretic intersection is used instead of the +set-theoretic union. -\end{itemize} +In order for \TT{while} to work as expected, both {\av} sets are evaluated in +a common context during the iteration ({\ie} the variables defined in both +are available to both) and this context is also propagated outside the +\TT{while}. + +\item +\textbf{Bounded iteration.} +Also this construction comes in two forms: + +\TT{for @}\EM{variable} \TT{in} \EM{av-set} \TT{sup} \EM{av-set}: +iterates the evaluation of the second \EM{av-set} assigning the \EM{variable} +to each element in the first \EM{av-set} and builds the {\MathQL} +set-theoretic union of the obtained results. -\xcomment { +\TT{for @}\EM{variable} \TT{in} \EM{av-set} \TT{inf} \EM{av-set}: +like the former but the set-theoretic intersection is used instead of the +set-theoretic union. -Attributed values can be used to store any auxiliary information needed during -query execution. -In particular, {\MathQL} provides variables for {\av}'s which, in its textual -syntax, are identifiers% -\footnote{To be understood as in programming languages.} -preceded by the @ sign, as in \TT{@variable}, and that are introduced by the -\TT{for} and \TT{select} constructions to be explained below. +The variables for attributed values (preceded by a \TT{@} sign and called +\emph{element variables}) may appear wherever an {\av} set is allowed and +and in some additional places. +The element variables are kept distinct from the set variables (therefore +\TT{\$variable} and \TT{@variable} may appear in the same query without +ambiguity). +Concerning the scope rules used in these constructions, the variables +assigned by the first {\av} set are available to the second {\av} set during +the iteration and the variables assigned by both {\av} sets are available +outside the \TT{for} as in the previous case. -} % \xcomment +\item +\textbf{Addition of groups.} +\TT{add} \EM{optional-flag} \EM{attribute-groups} \TT{in} \EM{av-set} +builds an {\av} set adding the specified \EM{attribute-groups} to each element +of the given {\av} set. +If no \EM{flag} is specified the addition is set-theoretic, whereas with the +\TT{distr} flag the addition is distributive. +The \EM{attribute-groups} can be given explicitly (in the same syntax used for +explicit {\av} sets) or they can be replaced by an element variable. In the +latter case the attribute groups of the {\av} stored in the variable are +considered. + +\figref{Add} shows how to build a one-element {\av} set using \TT{add}. + +\begin{figure} +\begin{footnotesize} \begin{verbatim} +The set of attributed values given explicitly: + ["head" attr {/"attribute" = property /"metadata" of "resource"}] + +The same set built with the add operator: + add {/"attribute" = property /"metadata" of "resource"} in "head" +\end{verbatim} \end{footnotesize} +\vspace{-1pc} +\caption{A simple use of the add operator} +\label{Add} +\end{figure} + +\item +\textbf{Existential test.} +The existential test has the form \TT{ex} \EM{av-set} where the +specification of the {\av} set contains some instances of the construction +\TT{@}\EM{variable}\TT{.}\EM{attribute-name}, and runs as follows: +the given {\av} set is evaluated replacing each +\TT{@}\EM{variable}\TT{.}\EM{attribute-name} +with the contents of \EM{attribute-name} in an attribute group of the {\av} +stored in \TT{@}\EM{variable} and the evaluation is repeated for every +possible choice of these groups (recall that different groups are allowed to +contain attributes with the same name). If one evaluation gives a non empty +result, the default representation of \emph{true} is returned, in the other +case the empty {\av} set, {\ie} \emph{false}, is returned. + +\item +\textbf{Function invocation.} +The core language allows to invoke two kinds of external functions (with +which a language extension may be provided): the functions of the first kind +return an {\av} set, the functions of the second kind return a piece of +{\MathQL} code representing an {\av} set ({\ie} they interface a {\MathQL} +code generator). In particular: + +\EM{function-name} \verb+{+ +\EM{name} \TT{,} $\cdots$ \TT{,} \EM{name} \verb+} {+ +\EM{av-set} \TT{,} $\cdots$ \TT{,} \EM{av-set} \verb+}+ +invokes a function of the first kind on the given arguments and returns its +result. The \EM{name} arguments are {\MathQL} paths and usually +represent attribute names. + +\TT{gen} \EM{function-name} \verb+{+ +\EM{av-set} \TT{,} $\cdots$ \TT{,} \EM{av-set} \verb+}+ +invokes a function of the second kind on the given arguments and replaces +itself with the function result. + +The function names are {\MathQL} paths exactly as the attribute names and the +graph paths used by the \TT{property} operator. The names of the two kinds of +functions are kept in distinct environments so they do not clash. + +{\MathQL}.4 comes with a basic library of functions of the first kind +(see \subsecref{Basic}) that integrate the core language providing several +facilities. + +\end{itemize}