QuerySearch < System < BACCHUS Wiki

%STARTINCLUDE%
---+ Query Search
Query searches help you search the contents of forms attached to your topics, as well as the values of other meta-data attached to the topic. Using query searches you can search:
   1 The fields of forms
   1 Parent relationships
   1 File attachment information (but *not* the attached files themselves)

Query searches are defined using a simple query language. The language consists of _field specifiers_ and _constants_ joined with _operators_.

%TOC%

---++ Field specifiers
You use field specifiers to say what value from the topic you are interested
in.

All meta-data in a topic is referenced according to a simple plan.

   * =name= - name of the topic
   * =web= - name of the web the topic is within
   * =text= - the body text of the topic (without embedded meta-data)
   * =META:FILEATTACHMENT=
      * _for each attachment_
         * =name=
         * =attr=
         * =path=
         * =size=
         * =user=
         * =rev=
         * =date=
         * =comment=
   * =META:TOPICPARENT=
      * =name=
   * =META:TOPICINFO=
      * =author=
      * =date=
      * =format=
      * =version= - topic version (integer)
   * =META:TOPICMOVED=
      * =by=
      * =date=
      * =from=
      * =to=
   * =META:FORM= - the main form of the topic
      * =name=
   * =META:FIELD= - the fields in the form.
      * _for each field in the form_
         * =name= - name of the field
         * =title= - title of the field
         * =value= - what is stored in the field
         * =form= - name of the form the field is in (currently always equal to META:FORM.name)
         * =attributes= - string stored in the attributes, like =H= for hidden
   * =META:PREFERENCE=
      * _for each preference in the topic_
         * =name=
         * =value=

See MetaData for details of what all these entries mean. Note that the set
of meta-data types (and the aliases used to refer to them) may be extended
by Foswiki extensions.

Most things at the top level of the plan - =META:TOPICPARENT=, =META:TOPICINFO= etc - are _structures_ which are indexed by _keys_. For example, =META:TOPICINFO= has 4 entries, which are indexed by the keys =author=, =date=, =format= and =version=. =META:FILEATTACHMENT=, =META:FIELD= and =META:PREFERENCE= are all _arrays_, which means they can have any number of records under them. Arrays are indexed by _numbers_ - for example, the first entry in the =META:FIELD= array is entry 0.

It's a bit clumsy having to type =META:FILEATTACHMENT= every time you want to refer to the array of attachments in a topic, so there are some predefined aliases that make it a bit less typing:
   * =attachments= means the same as =META:FILEATTACHMENT=
   * =info= means the same as =META:TOPICINFO=
   * =parent= means the same as =META:TOPICPARENT=. *Note:* =parent= is itself a map; use =parent.name= to access the name of the parent topic
   * =moved= means the same as =META:TOPICMOVED=
   * =form= means the same as =META:FORM=, so to test if a topic has a form named 'UserForm' you test for ="form.name ~ '*.UserForm'"=
   * =fields= means the same as =META:FIELD=, You can also use the name of the form (the value of =form.name= e.g. =PersonForm=)
   * =preferences= means the same as =META:PREFERENCE=
   * extensions may add additional aliases when they register new meta-data types 

Fields in this plan are referenced using a simple _field specifier_ syntax:
| *Syntax* | *Means* | *Examples* |
| =X= | refers to the field named =X=. | =info=, =META:TOPICMOVED=, =attachments=, =name=. |
| =X.Y= | refers to the entry with the key =Y= in the structure named =X=. If =X= is an array of structure, then returns an array made up from the =Y= entry of each member of the array. | =info.date=, =moved.by=, =META:TOPICPARENT.name=, =attachments.name= |
| =X[<i>query</i>]= | refers to all the elements of the array =X= that match _query_. If <i>query</i> is of the form =name='Y'= then you can use the same =X.Y= syntax as is used for accessing structures. | =attachments[size>1024]=, =DocumentContainer[name!='Summary' AND value~'top secret'].value=|
| =X[N]= | where =X= is an array and =N= is an integer number >= 0, gets the Nth element of the array =X=. If N is a floating point number, the integer part will be used as the index. Negative indices can be used to index the array from the end e.g. =attachments[-1]= to get the last attachment. | =attachments[3]= |
| =X/Y= | accesses =Y= from the topic specified by the _value_ of =X=. =X= must evaluate to a topic name | =parent.name/(form.name='ExampleForm')= will evaluate to true if (1) the topic has a parent, (2) the parent topic has the main form type =ExampleForm=. |
| ={X}= | expands to the value of the =configure= setting {X}, if it is accessible, or '' otherwise | only some configuration settings are available: %FORMAT{"%QUERY{"{AccessibleCFG}"}%" type="string" format="=$item=" separator=", "}% |

Note: at some point Foswiki may support multiple forms in the same topic. For this reason you are recommended *not* to use the =fields= shortcut when accessing form fields, but always use the name of the form instead.

If you use the name of a field (for example, =LastName=) in the query without a . before it, that is taken to mean "the value of the field named this". This works if and only if the field name isn't the same as of the top level entry names or their aliases described above. For example, the following expressions will all evaluate to the same thing:
   * =PersonForm[name='Lastname'].value=
   * =Lastname=
   * =PersonForm.Lastname=
If =X= would conflict with the name of an entry or alias (e.g. it's =moved= or maybe =parent=), you can prepend the name of the form followed by a dot, as shown in the last example.

---++ Constants
You use constants for the values that you compare with fields. Constants are either strings, or numbers.

---++ String Constants
String constants are always delimited by single-quotes. You can use backslash =\= to include the following special characters:
| *Code* | *Meaning* |
| =\n= | newline |
| =\t= | tab |
| =\033= | octal character code |
| =\x7f | hexadecimal character code |
| =\x{1234} | heaxadecimal wide character code |
| =\\= | a single =\= | 
All other occurrences of backslashes are carried through into the string, so =\d= means =\d= (unless the string is used as a regular expression, in which case it means any digit).

---++ Numerical constants
Numbers can be any signed or unsigned integer or floating point number using standard scientific notation e.g. -1.2e-3 represents -0.0012

---++ Operators
Field specifiers and constants are combined using _operators_ to create queries.
| *Operator* | *Meaning* |
| <code>=</code> | Left-hand side (LHS) exactly matches the value on the Right-hand side (RHS). Numbers and strings can be compared. |
| <code>!=</code> | Inverse of <code>=</code>. |
| =~= | wildcard match ('*' will match any number of characters, '?' will match any single character e.g. "PersonForm.Surname ~ '*Smit?'") Note: Surname ~ 'Smith' is the same as Surname = 'Smith' |
| <code>=~</code> | regular expression match, see RegularExpressions for details. |
| <code>&lt;</code> | LHS is less than RHS. If both sides are numbers, the order is numeric. Otherwise it is lexical (applies to all comparison operators) |
| <code>&gt;</code> | greater than |
| <code>&lt;=</code> | less than or equal to |
| <code>&gt;=</code> | greater than or equal to |
| =lc(x)= | Converts x to lower case, Use for caseless comparisons. |
| =uc(x)= | Converts x to UPPER CASE. Use for caseless comparisons. |
| =d2n(x)= | Converts a text string representing a date (expressed in [[TimeSpecifications][one of the formats that Foswiki can parse]]) to a number of seconds since 1st Jan 1970. This is the format dates are stored in inside Foswiki, and you have to convert a string date using =d2n= before you can compare it with - for example - the date an attachment was uploaded. Times without a timezone are assumed to be in server local time. If the text string is not recognised as a valid date, then =d2n= will return =undefined=. |
| =NOT= | Invert the result of the subquery |
| =AND= | Combine two subqueries |
| =OR= | Combine two subqueries |
| =()= | Bracketed subquery |

<blockquote class="foswikiHelp">
%I% The same operators are supported by the [[VarIF][%IF]] and [[VarQUERY][%QUERY]] macros.

%I% You can get the current time for date comparisons using SpreadSheetPlugin, thus: =%<nop>CALC{"$TIME()"}%=

%I% If you want to know if a field is undefined (has never been given a value) then you can compare it with =undefined= (this requires that no field called =undefined= exists in the form).

<sticky>
%I% In the operators (<literal><code> = != ~ =~ &lt; &gt; &lt;= &gt;= NOT AND OR</code></literal>) an undefined operand is treated the same as numerical 0. For =lc uc d2n= an undefined operand will give an undefined result. For =length= and undefined operand will give a result of 0.
</sticky>
</blockquote>

---++ Putting it all together
When a query is applied to a topic, the goal is to reduce to a TRUE or FALSE value that indicates whether the topic matches that query or not. If the query returns TRUE, then the topic is included in the search results.

A query matches if the query returns one or more values when it is applied to the topic. So if I have a very simple query, such as ="attachments"=, then this will return TRUE for all topics that have one or more attachments. If I write ="attachments[size>1024 AND name ~ '*.gif']"= then it will return TRUE for all topics that have at least one attachment larger than 1024 bytes with a name ending in =.gif=.

---++ Gotcha
   * Remember that in the query language, topic names are _constants_. You cannot write =%USERSWEB%.UserTopic/UserForm.firstName= because =%USERSWEB%.UserTopic= will be interpreted as a form field name. If you want to refer to topics you *must* enclose the topic name in quotes i.e. ='%USERSWEB%.UserTopic'/UserForm.firstName=

---++ Examples
---+++ Query examples
   * =attachments[name='purdey.gif']= - true if there is an attachment call =purdey.gif= on the topic
   * =(fields[name='Firstname'].value='Emma' OR fields[name='Firstname'].value='John') AND fields[name='Lastname'].value='Peel'= - true for 'Emma Peel' and 'John Peel' but *not* 'Robert Peel' or 'Emma Thompson'
   * =(Firstname='Emma' OR Firstname='John') AND Lastname='Peel'= - shortcut form of the previous query
   * =History<nop>Form[name='Age'].value>2= - true if the topic has a =HistoryForm=, and the form has a field called =Age= with a value > 2
   * =History<nop>Form.Age > 2= - shortcut for the previous query
   * =preferences[name='FaveColour' AND value='Tangerine']= - true if the topic has the given [[%SYSTEMWEB%.PreferenceSettings][preference settings]] and value
   * =Person/(<nop>ClothesForm[name='Headgear'].value ~ '*Bowler*' AND attachments[name~'*hat.gif' AND date < d2n('2007-01-01')])= - true if the form attached to the topic has a field called =Person= that has a value that is the name of a topic, and that topic contains the form =<nop>ClothesForm=, with a field called =Headgear=, and the value of that field contains the string ='Bowler'=, and the topic also has at least one attachment that has a name matching =*hat.gif= and a date before 1st Jan 2007. (Phew!)
   * =length(fields[NOT lc(attributes)=~'h'])= - the number of fields that are not hidden

---+++ Search examples
Find all topics that are children of this topic in the current web
<verbatim class="tml">
%SEARCH{"parent.name = '%TOPIC%'" web="%WEB%" type="query"}%
</verbatim>
Find all topics that have an attachment called 'grunge.gif'
<verbatim class="tml">
%SEARCH{"attachments[name='grunge.gif']" type="query"}%
</verbatim>
Find all topics that have form =ColourForm= where the form field 'Shades' is 'green' or 'yellow' but not 'brown'
<verbatim class="tml">
%SEARCH{"(lc(Shades)='green' OR lc(Shades)='yellow') AND NOT(lc(Shades) ~ 'brown')" type="query"}%
</verbatim>
Find all topics that have PNG attachments that have been added since 26th March 2007
<verbatim class="tml">
%SEARCH{"attachments[name ~ '*.png' AND date >= d2n('2007-03-26')]" type="query"}%
</verbatim>
Find all topics that have a field 'Threat' set to 'Amber' and 'cold virus' somewhere in the topic text.
<verbatim class="tml">
%SEARCH{"Threat='Amber' AND text ~ '*cold virus*'" type="query"}%
</verbatim>
Find all topics newer than one week old
<verbatim class="tml">
%SEARCH{"info.date >= %CALC{"$TIMEADD($TIME(), -7, day)"}%" type="query"}%
</verbatim>
__Related Topics:__ SearchHelp, VarSEARCH, FormattedSearch, Foswiki:System/QuerySearchPatternCookbook
<!-- %JQREQUIRE{"chili"}% -->
This topic: System > WebHome > Macros > QuerySearch
Topic revision: revision 1 (raw view)
Copyright &© by the contributing authors. All material on this site is the property of the contributing authors.
Ideas, requests, problems regarding BACCHUS Wiki? Send feedback