1. Create a full-text index in the data files affected by this search.
2. Create a query to perform the full-text search.
3. Study and display the result of the query.

1. In the data model editor, display the description of the items found in the relevant file ("Description of items" in the context menu).
2. Click . The window for defining the full-text index is displayed. Select the items taking part in the composition of the full-text index. Only the "Text" items and the "Text memo" items found in the file description are proposed. Validate.
3. The "Full-text index" item is displayed in the list of items.
4. Select the item and modify its characteristics if necessary:
   • in the list of items, you have the ability to modify the name of the full-text index.
   • in the right section of the window, select the parameters of the full-text index:
     • minimum size of indexed words (option "Do not index words whose size is less than or equal to"): only words larger than this size will be indexed. This size is set to 2 characters by default.
     • RTF tag management: the "My text may contain RTF (do not index the RTF tags)" option specifies whether the text contains RTF.. In this case, the RTF tags are ignored when indexing.
     • HTML tag management: "My text may contain HTML (don't index the HTML tags)" option specifies whether text contains HTML. In this case, the HTML tags are ignored when indexing.
     • Manage UNICODE punctuation: this option specifies whether the text contains Unicode-specific punctuation characters (curved quotation marks, etc.)..
     • case management: If the "Case sensitive" option is checked, the case of words (upper/lower case) will be taken into account during indexing..
     • accentuation management: If the "Accent sensitive" option is checked, accentuation will be taken into account during indexing.. This option is required if the stemming is requested.
     • Hollow words management: Hollow words management allows you to define hollow words that should not be taken into account by indexing.. Hollow words are usually articles, linking words, etc.. The button "Edit stop words" is used to define the stop words that must be taken into account. For more details, see Defining the stop words.
     • synonym management: Synonym management allows you to define which words should be considered synonyms.. For more details, see Defining the synonyms.
     • root management: If synonym management is not activated, it is possible to activate root management.. The stemming is used to perform searches on a full-text index while taking the variations of a given word into account. This feature is interesting for the knowledge databases and it allows you to find the words in singular or plural form. The stemming depends on the language used.
       If this option is enabled, select the language used.
5. Validate the description window of items.
6. Save and generate the analysis. When generating the analysis, the description of the data files is modified and the full-text index is created (file with a ".FTX" extension).

Remarks

• To directly create a full-text index from a text item or from a text memo item, select the item in the description window of the items and click the link "Create a full-text index on the item".
• It is possible to manage full-text indexes in data files described programmatically: simply use a variable of type FullTextIndex Description to describe the index, then the HDescribeFullTextIndex function to validate index creation.
• You can modify the composition of an existing full-text index at any time: simply click on the "Edit index" button on the left of the heading description in the analysis editor.
• To create a Full-text Index item, a generation of the analysis and an automatic data modification must necessarily be performed.
• Rooting: Due to certain grammatical irregularities (3rd group verbs in French, irregular verbs in English, etc.), some conjugations may not be grouped together (e.g. a search for "peux" will not return records containing "pouvons") or some plurals may not be grouped together with their singular (e.g. feux/fire).

Full-text search: Query editor

• The search value. This value can be entered directly (a wizard is used to build the sought string) or it can correspond to a parameter. For more details, see "Syntax of the search value".
• The sort options of the result.

Full-text search: Query in SQL code

• <Index> corresponds to the list of items found in the index separated by commas (the order of items is not important).
• <Condition> corresponds to the sought string. For more details, see "Syntax of the search value".

QRY_Find is Data Source


MyQuery is string = [
	SELECT * FROM Contacts
	WHERE MATCH(Contacts.LastName, Contacts.FirstName, Contacts.HTMLComment, 
	Contacts.RoughTextComment, Contacts.Comments, 
	Contacts.Phone, Contacts.Office, 
	Contacts.Cell, Contacts.Email, Contacts.MSN, Contacts.Internet_site, 
	Contacts.Country, Contacts.FaxNum, Contacts.City)
	AGAINST ('
]
MyQuery = MyQuery + EDT_Find + [
')  
AND Contacts.UserID =  
]
MyQuery = MyQuery + ConnectedUserID + [
ORDER BY LastName DESC
]

HExecuteSQLQuery(QRY_SRCH, hQueryDefault, MyQuery)
FOR EACH QRY_SRCH
	TableAddLine(TABLE_Contact_by_category, ...
		QRY_SRCH.ContactID, QRY_SRCH.CategoryID, ConnectedUserID, ...
		QRY_SRCH.LastName, QRY_SRCH.FirstName) 
END
CASE ERROR: 
	Error(HErrorInfo())

Syntax of the search value

Specific case of words separated by a dash

• Text contains 'multi-platform': the two words "multi-" and "platform" will be indexed independently..
• During the search, the documents containing "multi" and/or "platform" will be found.
• To search for "multi-platform", you need to enclose the word in quotation marks: "multi-plateforme"..
  Attention: the search will return all documents containing the words "multi" and "platform" consecutively.. The separator between the two words may not be a hyphen: it can be any separator (space, +, period, comma, carriage return, tab, etc.).

• the number of times the sought word is found in the record.
• the number of words in the record and their number of repetitions.
• the ratio between the records that contain the sought words and the records that do not contain the sought words. Indeed, the more the sought word is found in all the records, the less the relevance will be important.
• ...

• Creating a full-text index in a data file created through programming can be done via a FullTextIndex Description variable.
• HIndex is used to re-index full-text indexes.