Sunday, August 12, 2007

Find Files Faster by Mastering the Indexing Service's Query Language

Got a hard disk filled with many files, and no easy way to find what you want quickly? Use the Indexing Service and its query language to get what you wantfast.

Packrats like me (and my editor) have a hard time finding exactly what they want on their hard disks. I have thousands of files there, some dating back close to 10 years, which I dutifully copy to a new system every time I upgrade my hardware. After all, who knows when I might need to find the list of books I planned to take out of the library in 1996?

XP's Search Companion is too slow and the kinds of searches it can perform are fairly limited. It can't find files based on properties such as when the file was last printed or the word count of a file, or using a sophisticated search language.

The Indexing Service, first used with Microsoft Internet Information Services (IIS), is a far more powerful tool. It can perform searches hundreds of times faster and includes an exceedingly sophisticated query language you can use for performing searches. It works by indexing the files on your disk, and then, when you do a search, it queries that index rather than searching through your entire hard disk. The indexes the service creates are called catalogs.

By default, the Indexing Service is turned off. To activate it, first run the Search Companion by choosing Start Search[ For Files or Folders]. From the Search Companion, choose Change Preferences With Indexing Service. If the With Indexing Service option isn't available, and instead you see Without Indexing Service, it means the Indexing Service is already turned on.

When you activate the Indexing Service, it won't be available immediately. First it has to build an index, which can take a substantial amount of time, depending on the number of files on your hard disk and your processor speed. It's best to start the Indexing Service and leave your computer on overnight so that it can complete indexing.

To turn off the Indexing Service from the Search Companion, choose Change Preferences Without Indexing Service. When you do that, you'll use the normal Search Companion. The index will remain intact; when you do a search, you just won't search through it. You can always turn the index back on when you want.

3.6.1. Using the Indexing Service's Query Language
The Indexing Service's query language is a sophisticated language, letting you search on file propertiessuch as the author of documents or the number of bytes in a documentusing Boolean operators and other search criteria.

The language uses tags to define search criteria. For example, to search for the phrase "That dog won't hunt," the query would be:

{phrase} That dog won't hunt {/phrase}


You can search for text in the query language using either phrase or freetext. A phrase search searches for the exact words in the exact order, like this:

{phrase} old dog barks backwards {/phrase}


The search results will include only files whose text includes that exact phrase.

A freetext expression search looks for any words in the phrase and returns files that have any one of the words in the phrase. It works like the Boolean OR operator. So, the query:

{freetext} old dog barks backwards {/freetext}


returns many more searches than the phrase query, since it returns results that contain any of the words in the phrase.

3.6.2. Searching Using Properties
The Indexing Service's query language's power is contained in the way it can search not just for text, but also for document properties. The syntax for searching using properties in a query is:

{prop name=property name} query {/prop}


where property name is the name of the property, such as those listed in Table 3-2, and query is the text you're searching for. For example, to search for all documents last edited by Preston Gralla, you would enter:

{prop name=DocLastAuthor} Preston Gralla {/prop}


Queries can use *and ? wildcard characters, as well as Unix-style regular expression queries (for more on regular expressions, see Mastering Regular Expressions from O'Reilly). To use these wildcards, you must use the {regex} tag, like this:

{prop name=filename} {regex} *.xl? {/regex} {/prop}


The Indexing Service indexes not just the text of each document, but also all the summary information associated with each document. (To see summary information for any document, right-click the document and choose Properties Summary.) In addition to searching for properties in the summary, you can also search for the properties found in Table 3-2, which lists the most important properties you can use to search.

Table 3-2. Important properties for searching via the Indexing Service Property
Description


Access
The last time the document was accessed.

All
All available properties. Works with text queries, but not numeric queries.

AllocSize
The total disk space allocated to the document.

Contents
The contents of the document.

Created
The time the document was created.

Directory
The full directory path in which the document is contained.

DocAppName
The name of the application in which the document was created.

DocAuthor
The author of the document.

DocByteCount
The number of bytes in the document.

DocCategory
The type of document.

DocCharCount
The number of characters in the document.

DocComments
Comments made about the document.

DocCompany
The name of the company for which the document was written.

DocCreatedTime
The time spent editing the document.

DocHiddenCount
The number of hidden slides in a PowerPoint document.

DocKeyWords
The keywords in the document.

DocLastAuthor
The name of the person who last edited the document.

DocLastPrinted
The time the document was most recently printed.

DocLineCount
The number of lines contained in the document.

DocLastSavedTm
The time the document was last saved.

DocManager
The name of the manager of the document's author.

DocNoteCount
The number of pages with notes in a PowerPoint document.

DocPageCount
The number of pages in the document.

DocParaCount
The number of paragraphs in the document.

DocPartTitles
The names of document parts, such as spreadsheet names in an Excel document or slide titles in a PowerPoint slide show.

DocRevNumber
The current version number of the document.

DocSlideCount
The number of slides in a PowerPoint document.

DocTemplate
The name of the document's template.

DocTitle
The title of the document.

DocWordCount
The number of words in the document.

FileName
The filename of the document.

Path
The path to the document, including the document filename.

ShortFileName
The 8.3-format name of the document.

Size
The size of the document, in bytes.

Write
The date and time the document was last modified.





3.6.3. Searching Using Operators and Expressions
The query language also lets you use a variety of operators and expressions for both text and numbers:




EQUALS and CONTAINS operators

When you're creating a query using text, you can use the EQUALS and CONTAINS operators to narrow your search. Use the EQUALS operator when you want the exact words matched in the exact order, like this:

{prop name=DocTitle} EQUALS First Draft of Final Novel {/prop}


This query finds all documents with the title "First Draft of Final Novel." The query won't find a document with the title "Final Draft of First Novel" or "First Draft of Novel." The EQUALS operator works like the phrase expression.

Use the CONTAINS operator when you want to find any of the words in the document, in the same way you would use the freetext expression.




Relational operators

Use the following relational operators when you're searching using numbers:

=

Equal to

!=

Not equal to

<

Less than

<=

Less than or equal to

>

Greater than

>=

Greater than or equal to




Date and time expressions

You can use the following formats when searching using dates and times:

yyyy/mm/dd hh:mm:ss
yyyy-mmmm-dd hh:mm:ss


You can also use date and time expressions in combination with relational operatorsfor example, to look for files that were created within the last two days:

{prop name=Created} >-2d {/prop}


Table 3-3 lists the date and time abbreviations you can use.

Table 3-3. Date and time expressions that work with relational operators Abbreviation
Meaning
Abbreviation
Meaning

Y
Year
D
Day

Q
Quarter
H
Hour

M
Month
N
Minute

W
Week
S
Second








Boolean operators

The query language also uses the Boolean operators detailed in Table 3-4.

Table 3-4. Boolean operators used by the Indexing Service's query language Boolean operator
Long form
Short form

AND
&
AND

OR
|
OR

Unary NOT
!
NOT

Binary NOT
&!
AND NOT





Use the unary NOT when you're searching using numbers rather than text. For example, to search for all documents that do not have seven PowerPoint slides, use the query:

{prop name=DocSlideCount} NOT = 7 {/prop}


Use the binary NOT to narrow a search, by combining two properties in a query. For example, to search for all documents with an author of "Preston Gralla" that are not titled Chapter 10, use this query (on one line):

{prop name=DocAuthor} Preston Gralla {/prop} NOT
{prop name=DocTitle} Chapter 10 {/prop}





Alternative verb forms

You can use the double-asterisk wildcard (**) to search for alternative forms of verbs in a document. For example, the query:

{prop name=Contents} run** {/prop}


returns all documents with the word "ran" or the word "run."

3.6.4. Ranking the Order of Search Results
If you're doing a search likely to return many results, you'll want the most-relevant searches to appear at the top of the results, and the least relevant to appear at the bottom. You can determine the relative importance of each term in your search and have the results weighted by that importance by using the weight tag. Note that it does not get a closing tag:

{weight value=n} query


The value parameter ranges between 0.000 and 1.000.

If you are searching for the three terms "fire," "ice," and "slush," and you want to weight "fire" most heavily, "ice" second-most heavily, and "slush" least heavily, you can use this syntax (on a single line) in your query:

{weight value=1.000}fire AND {weight value=.500}ice AND {weight value=.250}slush


3.6.5. Editing the Indexing Service's "Noise" Filter
You can force the Indexing Service to ignore more words when you search, or you can have it ignore fewer words, simply by editing a text file. In a text file called noise.eng, usually found in C:\Windows\System32\, you can find the list of words the Indexing Service ignores. (The extension .eng is for English. You can find noise filters from other languages as wellfor example, noise.deu for German, noise.fra for French, and so on.)

The noise.eng file contains common articles, prepositions, pronouns, conjunctions, various forms of common verbs, and similar words. Open it in Notepad or another text editor, add words you want it to ignore, and delete words you don't want it to ignore. Then save the file, and the Indexing Service will follow your new rules.

AddThis Social Bookmark Button

2 Comments:

Anonymous said...

I really like what you guys are usually up too.
This kind of clever work and exposure! Keep up the fantastic works guys I've added you guys to my blogroll.

Feel free to surf to my web-site: privat krankenversicherung

Anonymous said...

Good post: P

my weblog; all inclusive caribbean vacation