findinsite-ms Search API web service

findinsite-ms has an application programmer interface (API) for searches:

1. Send search text to the findinsite-ms search web service.
2. Receive back the requested results - all or just the first 10 say.
3. Display the results in your desired layout.
4. Reorder or change the results as required.
5. Cache the results locally to avoid further requests.

The Search API at the phdcc web site is at this URL:
     http://www.phdcc.com/findinsite/SearchService.asmx
You can access this URL from your site at any time - it only searches the phdcc web site.
If you are going to use this service URL regularly or as an example, please contact us for permission.

Once you have installed findinsite-ms at your site, the Search API will be available at a similar URL - the exact URL depends on your web site and installation directory.

The Search API web service can be turned off completely in the findinsite-ms Control Panel. Alternatively, you can only allow accesses from your own web site - by disabling external access to the Search API.

If you are using the Free license, these fields in fisSearchResult will be null in results from files after the first 60 files: URL, HighlightURL and Filename.

Search API Methods

• Search() - do a search
• GetSubsets() - get the names of available search database subsets
• GetFieldnames() - get the field names of the specified subset

Note: The GetSubsets() and GetFieldnames() methods will probably be reworked soon to provide more information about the main index and each subset search database.

Examples

The following examples are written as ASP.NET .aspx web form pages. The examples will therefore only work when run on an ASP.NET site. If you are viewing the documentation as a file on your local disk, then the "locally" links will not work. To try out the examples, follow the link to the phdcc web site.

• Simple search, returning results in a Repeater: here or at www.phdcc.com
• Simple search, returning results in a DataGrid: here or at www.phdcc.com
• Full search, returning results in a DataGrid: here or at www.phdcc.com

Making these examples work locally

If you have installed the findinsite-ms development kit, the documentation appears on the Start menu. However, these Search API examples will not work from here.

If you are running IIS (eg if you have Windows XP Pro, Windows Server 2003 or similar) and have .NET installed, then you can make the examples work by making an IIS virtual directory that maps to the documentation directory. To do this, you have to...

• Select Start+All Programs+Administrative Tools+Internet Information Services
• Expand the tree to get the "Default Web Site".
• Right click on "Default Web Site", and select New+Virtual Directory...
• Click Next
• Enter a new directory name in the Alias box, eg findinsite, and click Next
• Select a directory (eg C:\Program Files\PHD\findinsite\), and click Next
• Do not change the permissions, and click Next
• Click Next to finish the wizard
• The findinsite-ms documentation should now be visible at a URL like this: http://localhost/findinsite/.
• Follow the documentation links to run these examples.

Information flows

1. The user enters search text into your search form and presses "Go Search".
   The HTTP request is sent to your web site to be handled, eg by your ASP.NET web form page.
2. The web page processes the form by sending a search request to findinsite-ms, typically using the SOAP protocol.
3. findinsite-ms does the search and returns the results to your web site, again typically using SOAP.
4. Your web page formats the received search results and returns the information to the user using an HTTP response.

When using the findinsite-ms Search API you need to be aware of the information flows. Your web site and findinsite-ms will typically be on the same web site, so SOAP transmission delays will be very short. However if they are on different servers then information transmission times could be significant. In particular, a request that returns many hits could take a long time to transmit.

There are three possible strategies to cope with long transmission delays:

1. Put your web site and findinsite-ms on the same server
2. Once you have received a search result, store it in a Session variable for your web page. If the user returns for the next 10 hits say, then retrieve the results from the Session variable. See example 2 for details of this.
3. Only request the number of hits that you can process quickly. A findinsite-ms search request can set the range of hit numbers required. Example 3 shows how to use this approach.

Note that the findinsite-ms Search API is set up so that a request and its response are cached in the server for 60 seconds. This means that a repeat request (within 60 seconds) does not have to be processed by findinsite-ms. Instead, the result is returned straight away from the ASP.NET cache. However the result will still have to be sent over the network.

Interface technology

The findinsite-ms search API is implemented as an ASP.NET XML Web Service. It responds to SOAP and HTTP GET/POST requests.

The search API can be accessed by any SOAP compliant techology - as well as HTTP GET/POST requests. The Web Services Description Language (WSDL) service description is made available, as at the phdcc web site:
     http://www.phdcc.com/findinsite/SearchService.asmx?WSDL

The search API can also be inspected by a web browser, as at the phdcc web site:
     http://www.phdcc.com/findinsite/SearchService.asmx

Referencing the web service in ASP.NET

1. In your Visual Studio project, add a Web Reference to the Search API web service, eg to use the phdcc web site search API, add a reference to
        http://www.phdcc.com/findinsite/SearchService.asmx

2. You can make a local source file equivalent to a Web Reference using the .NET SDK wsdl.exe tool. For example, this command line generates a C# file SearchService.cs for the phdcc web site:
   wsdl.exe http://www.phdcc.com/findinsite/SearchService.asmx?WSDL
   Include this file in your project.

3. The recommended approach is to use the fisClient class library, supplied in phdcc.fis.fisClient.dll. This library contains a class fisClient.SearchService based on the code generated by the above method, but with the following enhancements:
   • The web service URL is set in the SearchService() constructor.
   • The results classes use properties which make them easy to use in a DataGrid.
   Add a Reference to phdcc.fis.fisClient.dll, copying this file into your project application /bin directory.

Search API - Search() method

The Search API is described in terms of its use in ASP.NET using the supplied fisClient class library in C#. This section describes the Search() method.

First, you need to instantiate a new SearchService object by calling its constructor, passing in the URL string of the search service:

The search API consists of one method:

The fisSearchRequest object has a simple constructor that lets you do a simple text search, returning all results, eg:

The returned fisSearchResult object has a Message string field that is set if there are any errors. The Count integer field is set to the number of hits found. The Hits field is an array of fisHit objects, one for each hit. Each fisHit has the various fields to describe a hit in a file, eg Title and URL.

Search request options specified in fisSearchRequest

namespace fisClient
{
    public class fisSearchRequest
    {
        public string Text = null;
        public string Subsets = null;
        public int HitsFrom = 0;
        public int HitsTo = 0;
        public fisHitAttributes HitElems = null;
        public ArrayList SearchFields;    // fisSearchField-s
        public string UserParam = null;

        public fisSearchRequest(string text) { Text = text; }
    }

    public class fisSearchField
    {
        public string FieldName;
        public string FieldValue;
        public fisSearchField(string FieldName,string FieldValue);
    }

    [FlagsAttribute]
    public enum fisHitAttributes : short
    {
        ResultNo = 1,
        Title = 2,
        URL = 4,
        HighlightURL = 8,
        Abstract = 16,
        Filename = 32,
        WordCount = 64,
        Date = 128,
        Indexed = 256,
        Size = 512,
        Snippet = 1024,
        PageWordCount = 2048,
        All = ResultNo|Title|URL|HighlightURL|Abstract|Filename|WordCount
              |Date|Indexed|Size|Snippet|PageWordCount,
    }
}

Namespace fisClient
Public Class fisSearchRequest
    Public Sub New()
    Public Sub New(ByVal [text] As String)
    Public HitElems As fisHitAttributes
    Public HitsFrom As Integer
    Public HitsTo As Integer
    Public SearchFields As ArrayList
    Public Subsets As String
    Public [Text] As String
    Public UserParam As String
End Class

Public Class fisSearchField
    Public Sub New()
    Public Sub New(ByVal FieldName As String, ByVal FieldValue As String)
    Public FieldName As String
    Public FieldValue As String
End Class

Public Enum fisHitAttributes
    ResultNo = 1
    Title = 2
    URL = 4
    HighlightURL = 8
    Abstract = 16
    Filename = 32
    WordCount = 64
    Date = 128
    Indexed = 256
    Size = 512
    Snippet = 1024
    PageWordCount = 2048
    All = 4095
End Enum
End Namespace

This table summarises the fisSearchRequest fields:

Search results returned in fisSearchResult

If you are using the Free license, these fields in fisSearchResult will be null in results from files after the first 60 files: URL, HighlightURL and Filename.

namespace fisClient
{
    public class fisSearchResult
    {
        public int Count;
        public string Message;
        public object[] Hits;    // fisHit-s
        public string UserParam;
    }
    public class fisHit
    {
        public string Abstract { get; set; }
        public DateTime Date { get; set; }
        public string Filename { get; set; }
        public string HighlightURL { get; set; }
        public DateTime Indexed { get; set; }
        public int PageWordCount = 0 { get; set; }
        public int ResultNo { get; set; }
        public int Size { get; set; }
        public string Snippet { get; set; }
        public string Title { get; set; }
        public string URL { get; set; }
        public int WordCount { get; set; }
    }
}

Namespace fisClient
    Public Class fisSearchResult
        Public Sub New()
        Public Count As Integer
        Public Hits As Object()	' fisHit-s
        Public Message As String
        Public UserParam As String
    End Class
    Public Class fisHit
        Public Sub New()
        Public Property Abstract As String
        Public Property Date As DateTime
        Public Property Filename As String
        Public Property HighlightURL As String
        Public Property Indexed As DateTime
        Public Property PageWordCount As Integer
        Public Property ResultNo As Integer
        Public Property Size As Integer
        Public Property Snippet As String
        Public Property Title As String
        Public Property URL As String
        Public Property WordCount As Integer
    End Class
End Namespace

This table summarises the fisSearchResult fields:

This table summarises the fisHit properties. Note that the property is only set if the appropriate bit in fisSearchRequest.HitElems is set.

Search API - GetSubsets() method

The GetSubsets() method returns a string array containing a list of all the search database subsets supported by findinsite-ms at this site.

Search API - GetFieldnames() method

The GetFieldnames() method returns a string array containing a list of all the field names for the specfied search database subset.

Search API Technical Details

Returned messages are always in English. Only English boolean operators are recognised in the search text, ie AND, OR and NOT. If requested, the Search API could be updated to let you set the language for each request.

As described earlier, all Search API requests and associated responses are cached on the server for 60 seconds. Consequently, an immediate repeat request is honoured without calling the Search API - however the response still has to be sent over the network. Therefore, it is recommended that responses be saved (eg in a Session variable) if they are likely to be used again.

All characters are sent and received using the UTF-8 character set in XML. If the generated XML contains control code characters then it will cause an error in a calling ASP.NET application; therefore these control characters are removed from all returned strings: codes less than \u0020 and codes \uFFFE and \uFFFF.

findinsite-ms checks the search text for cross-site scripting attacks - an error is returned if the validation fails.

The Search API should not normally send any exceptions. However your code should be prepared to cope with any that do arise, eg due to network failures.