If you don't specify an alt parameter, the service returns an Atom feed. The other alternatives are:

alt=rss returns a feed in RSS 2.0 format.

alt=json returns a feed in JSON format.

alt=json-in-script returns a response that wraps JSON in a script tag.

This is a 1-based index of the first result to be retrieved.

For the snippets feed, start-index must be greater than or equal to 1 and less than 1000. The sum of start-index + max-results cannot be greater than 1000.

For the items feed, start-index must be greater than or equal to 1 and less than 10,000. The sum of start-index + max-results cannot be greater than 10,000.

Note that this isn't a general cursoring mechanism.

If you first send a query with ?start-index=1&max-results=10 and then send another query with ?start-index=11&max-results=10, the service cannot guarantee that the last 10 results are equivalent to ?start-index=1&max-results=20, because insertions and deletions could have taken place in between.

Query adjustments are always enabled, and adjusted values are included transparently by default. In some cases you might want to streamline the adjusted values by preventing the gm:adjusted_name and gm:adjusted_value tags from appearing in query results. To do this, add the adjust=true parameter to the feed URL.

For example, the following entry:

<entry><g:item_type>
<gm:adjusted_value>
recipes
</gm:adjusted_value>
recipe
</g:item_type>
<g:cook_time>
<gm:adjusted_name>
cooking time
</gm:adjusted_name>
10 minutes
</g:cook_time>
<g:main_ingredient>
<gm:adjusted_value>
chicken
</gm:adjusted_value>
poultry
</g:main_ingredient>
</entry>

will be transformed as follows if you specify adjust=true:

<entry>
<g:item type>
recipes
</g:item type>
<g:cooking time>
10 minutes
</g:cooking time>
<g:main_ingredient>
chicken
</g:main_ingredient>
</entry>


Be aware that the items returned by a feed on which adjust has been set to true may not appear as the original author intended. Adjustments are added automatically and consequently are subject to error. Specify adjust=true only if you're willing to accept incorrect or incomplete information in exchange for not having to parse <gm:adjusted_name> and <gm:adjusted_value>.

See the Attributes and Queries document for details on query syntax.

If you want to get data about all of the items, you can leave off the ?bq=[query-string] component, as follows:

http://www.google.com/base/feeds/snippets or http://www.google.com/base/feeds/items

content can have the following values for all noted feeds:

• none : specifies that only elements from the atom: namespace should be returned.
• all: specifies that elements from the atom:, g:, and gm: namespaces should be returned. all also returns thumbnails.
• attributes : specifies that elements from the atom: and g: namespaces should be returned.
• meta : specifies that the tags gm:stats and gm:adjusted should be returned.
• adjustments: specifies that gm:adjusted_name and gm:adjusted_value should be included in query results.
• thumbnails: specifies that links to generated thumbnails should be included in query results.
• geocodes: specifies that if an item contains <g:location> data, the geocoded <g:latitude> and <g:longitude> data should also be returned.

In addition, content can have the following value for the Items feed:

• stats : specifies that the gm:stats elements should be returned.

To include more than one value, separate them with a comma (except for none and all, which are shortcuts). For example, to get attributes and statistics but no adjustments, you would specify attributes,stats.

The syntax for the value of crowdby is a comma-separated list of crowding and maxvalue pairs:

crowdby=crowding:maxvalue,crowding:maxvalue, crowding:maxvalue

crowding: is the crowding criteria. You can use any of the following criteria:

• attribute filters results based on the specified attribute.
• url filters results based on the URL.
• customerid filters results based on the customer ID.
• content filters results based on the content of the title and description attributes in the item.
• [crowding expression] uses an expression for crowding.

The itemtype parameter may be used to search for entries that correspond to a specific category, based on the item type.

None

The syntax is http://www.google.com/base/feeds/snippets/-/itemtype .

You can query on multiple categories by listing multiple category parameters, separated by slashes.

To query on a union of categories (OR queries), use a pipe character ( | ), URL-encoded as %7C. For example: http://www.google.com/base/feeds/snippets/-/products%7Cservices returns entries that match either the products category or the services category.

Since every item belongs to exactly one item type, queries on an intersection of categories (AND queries) will not yield any results.

To exclude entries that match a given category, use the form /-categoryname/

When you use a category query on the attributes or itemtypes feed, the response to your query includes a link that points to the snippets feed where more data can be found about this item type. This allows you to write code that automatically takes you to the next level of detail. The link uses a category search, with " /-/" prefix to indicate that the query is to get all items with the item type that follows. For example, your query result might contain the following link:

<link 
rel="related"
type="application/atom+xml" 
href="http://www.google.com/base/feeds/snippets/-/jobs" 
title="Items in jobs"/>

For additional syntax information, refer to the Google Data API /category parameter documentation.

Items, Snippets, Attributes

5

orderby can have the following values:

• relevancy: specifies that query results should be sorted by relevancy. This is the default behavior.
• modification_time: specifies that query results should be sorted by last modification time. The most recently modified item has the highest rank.
• name(type): name refers to an attribute name and type refers to a numerical attribute type (i.e. int or float, possibly with a unit.) The items are ordered by the value of the specified attribute.
• [<ranking criteria>]: This is a custom ranking criteria specified in the ranking query language. The Ranking Language reference explains what you can do with this language and defines the language syntax.

• When creating a query, list search terms separated by spaces, in the form q=term1 term2 term3. (As with all of the query parameter values, the spaces must be URL encoded.) The service returns all entries that match all of the search terms (like using AND between terms). Like Google's web search, a Google Data API service searches on complete words (and related words with the same stem), not substrings.
• To search for an exact phrase, enclose the phrase in quotation marks: q="exact phrase".
• To exclude entries that match a given term, use the form q=-term.
• The search is case-insensitive.
• Example: to search for all entries that contain the exact phrase "Elizabeth Bennet" and the word "Darcy" but don't contain the word "Austen", use the following query: ?q="Elizabeth Bennet" Darcy -Austen

Items, Snippets, Attributes

If refine is set to true, the Google Base data API will rewrite "item type" restricts ITR of the form [item type: term1 ... termN] , [item type: "term1 ... termN"] and [item type == "term1 ... termN"] to ITR [attrib1] ... [attribM] where attribI are names of attributes that we believe are essential for the item type specified by ITR. In other words, this refinement adds additional restrictions to an item type attribute query which guarantees to match items that have a certain number of core attributes.

The purpose of this mechanism is to restrict search results to items that reuse a certain number of attributes we consider to be essential for a given item type. This may improve the quality of certain search results, especially for queries that have many matching items.

On the other hand, setting refine to true might lead to items getting dropped that actually match the specified query. This is, in particular, bad for the customer-specific items feed, because it can give the impression that some attributes are missing, when in fact they simply do not meet the undocumented quality assumptions that this parameter introduces.

Because of this, refine is set to false by default. It should be enabled to improve quality of broad queries.

For example, if refine is set to true,

[item type: recipes] [label:asian]

gets rewritten to:

[item type: recipes] [cooking time] [main ingredient] [label:asian]

On the other hand, [item type: products | services] does not get refined with additional attribute restrictions, because this item type query does not match the specified format.

sortorder can have the following values:

• ascending: specifies that items should be returned in increasing rank.
• descending: specifies that items should be returned in decreasing rank.

If sortorder is specified but orderby is not, the results will be sorted by relevancy in the specified order.