Google Hacks, 2nd Edition

Mixing Syntax

There was a time when you couldn’t mix Google’s special syntax elements; you were limited to one per query. Even as Google released ever more powerful special syntax elements, not being able to combine them for their composite power stunted many a search.

This has since changed. While there remain some syntax elements that you just can’t mix, there are plenty to combine in clever and powerful ways. A thoughtful combination can do wonders to narrow a search.

How Not to Mix Syntax

There are some simple rules to follow when mixing syntax elements. These, for the most part, revolve around how not to mix.

• Don’t mix syntax elements that will cancel each other out, such as:

      site:ucla.edu -inurl:ucla

  Here you’re saying you want all results to come from ucla.edu, but that site results should not have the string “ucla” in the results. Obviously, that’s not going to produce many URLs.

• Don’t overuse single syntax elements, as in:

      site:com site:edu

  While you might think you’re asking for results from either .com or .edu sites, what you’re actually saying is that site results should come from both simultaneously. Obviously, a single result can come from only one domain. Take the example perl site:edu site:com. This search will get you exactly zero results. Why? Because a result page cannot come from a .edu domain and a .com domain at the same time. If you want results from .edu and .com domains only, rephrase your search like this:

      perl (site:edu | site:com)

  With the pipe character (|), you’re specifying that you want results to come either from the .edu or the .com domain.

• Don’t use allinurl: or allintitle: when mixing syntax. It takes a careful hand not to misuse these in a mixed search. Instead, stick to inurl: or intitle:. If you don’t put allinurl: in exactly the right place, you’ll create odd search results. Let’s look at this example:

      allinurl:perl intitle:programming

  At first glance, it looks like you’re searching for the string “perl” in the result URL and the word “programming” in the title. And you’re right: this will work fine. But what happens if you move allinurl: to the right of the query?

      intitle:programming allinurl:perl

  This won’t bring any results. Stick to inurl: and intitle:, which are much more forgiving of where you put them in a query.

  The same advice goes for allintext: and allinanchor:.

• Don’t use so much syntax that you get too narrow, like:

      title:agriculture site:ucla.edu inurl:search

  You might find that it’s too narrow to give you any useful results. If you’re trying to find something so specific that you think you’ll need a narrow query, start by building a little bit of the query at a time. Say you want to find plant databases at UCLA. Instead of starting with the query:

      title:plants site:ucla.edu inurl:database

  Try something simpler:

      databases plants site:ucla.edu

  and then try adding syntax to keywords you’ve already established in your search results:

      intitle:plants databases site:ucla.edu

  or:

      intitle:database plants site:ucla.edu

How to Mix Syntax

If you’re trying to narrow down search results, the intitle: and site: syntax elements are your best bet.

Titles and sites

For example, say you want to get an idea of what databases are offered by the state of Texas. Run this search:

intitle:search intitle:records site:tx.us

You’ll find something on the order of 30 very targeted results. And, of course, you can narrow down your search even more by adding keywords:

birth intitle:search intitle:records site:tx.us

It doesn’t seem to matter whether you put plain keywords at the beginning or the end of the search query; I put them at the beginning because they’re easier to keep up with.

The site: syntax, unlike site syntax on other search engines, allows you to get as general as a domain suffix (site:com) or as specific as a domain or subdomain (site:thomas.loc.gov). So if you’re looking for records in El Paso, you can use this query:

intitle:records site:el-paso.tx.us

and you’ll get approximately one result.

Title and URL

Sometimes you want to find a certain type of information, but you don’t want to narrow by type. Instead, you want to narrow by theme of information (e.g., you want help or a search engine). That’s when you need to search in the URL.

The inurl: syntax will search for a string in the URL but won’t count finding it within a larger word. So, for example, if you search for inurl:research, Google will not find pages from http://www.researchbuzz.com, but it will find pages from www.research-councils.ac.uk.

Say you want to find information on neurosurgery, with an emphasis on learning or assistance. Try:

intitle:neurosurgery inurl:help

This returns a more manageable 880 or so results. The whole point is to get a number of results that finds what you need but isn’t so large as to be overwhelming. If you find 880 results overwhelming, you can easily mix the site: syntax into the search and limit your results to universities:

intitle:neurosurgery inurl:help site:edu

Beware, however, of using too much special syntax. As mentioned earlier, you can quickly detail yourself into no results at all.

The Antisocial Syntax Elements

The antisocial syntax elements are the ones that won’t mix and should be used individually for maximum effect. If you try to use them with other syntax elements, you won’t get any results.

The syntax elements that request special information—stocks:, rphonebook:, bphonebook:, and phonebook: are all antisocial. That is, you can’t mix them and expect to get a reasonable result.

The other antisocial syntax is link:, which shows pages that have a link to a specified URL. Wouldn’t it be great if you could specify what domains you want the pages to be from? Sorry, you can’t. The link: syntax does not mix with anything else—not even plain old keywords.

For example, say you want to find out what pages link to O’Reilly Media, Inc., but you don’t want to include pages from the .edu domain. The query link:www.oreilly.com -site:edu will not work because the link: syntax does not work in combination. Well, that’s not quite correct; you will get results, but they’ll be for the phrase “link:www.oreilly.com” from domains that are not .edu.

If you want to search for links and exclude the .edu domain, there’s no single command that will absolutely work. This one’s a good try, though:

inanchor:oreilly -inurl:oreilly -site:edu

This search looks for the word “oreilly” in anchor text, the text that’s used to define links; excludes pages that contain “oreilly” in the search result (e.g., oreilly.com); and, finally, excludes those pages that come from the .edu domain.

But this type of search is nowhere near complete. It finds only those links to O’Reilly that include the string “oreilly”: if someone creates a link such as <a href="http://perl.oreilly.com/">Camel Book</a>, it won’t be found by the preceding query. Furthermore, there are other domains that contain the string “oreilly,” and there may be domains that link to “oreilly” that contain the string “oreilly” but aren’t oreilly.com. You could alter the string slightly, to omit the oreilly.com site itself but not other sites containing the string “oreilly”:

inanchor:oreilly -site:oreilly.com -site:edu

However, you would still be including many O’Reilly sites—XML.com and MacDevCenter.com, for instance—that aren’t at oreilly.com.

All the Possibilities

While it is possible to write down every syntax-mixing combination and briefly explain how they might be useful, there wouldn’t be room for much else in this book.

Experiment. Experiment a lot. Constantly keep in mind that most of these syntax elements do not stand alone, and you can get more done by combining them than by using them individually

Depending on what kind of research you are doing, different patterns will emerge over time. For example, you may discover that focusing on only PDF documents (filetype:pdf) finds you the results you need. You may discover that you should concentrate on specific file types in specific domains (filetype:ppt site:tompeters.com). Mix up the syntax in as many ways as is relevant to your research and see what you get.

As with anything else, the more you use Google’s special syntax, the more natural it will become to you. And Google is constantly adding more, much to the delight of regular web combers.

If, however, you want something more structured and visual than a single query line, Google’s Advanced Search should fit the bill.