Skip to content

Solr querying

Docs and sources:

How do send queries to Solr?`

Basic example:

curl "http://localhost:8983/solr/mycore/select?q=*:*"

solr-10.png

About QueryParsers

The query syntax is strictly dependent on the QueryParser being used. Each parser has different syntax and capabilities.

Common query parameters

Full list and description: Common Query Parameters

Some examples:

param description examples defaultValue
defType which QueryParser should be used
  • dismax
  • edismax
  • lucene
 lucene
q main query depending on QueryParser none
fq filter query fq=popularity:[10 TO *]&fq=section:0 none
sort result set sorting order
  • score desc
  • price asc, name desc
  • price asc, div(popularity,price) asc
 score desc
start for pagination: offset of page start 0 0
rows for pagination: number of documents 72 10
fl fields to retrieve
  • id,name,price
  • id na* price
  • *
  • div(popularity,price)
 *
debug return debug information
  • query
  • timing
  • results
  • all
 none
timeAllowed the amount of time, in milliseconds, allowed for a search to complete 2000 none
wt output format
  • json
  • xml
  • csv
 json
echoParams should the query parameters be included in the response
  • all
  • none
  • explicit
 none

Additionally, you can specify:

Lucene Query Parser (default)

Example:

curl "http://localhost:8983/solr/mycore/select?defType=lucene&q=*:*"

Docs: Standard Query Parser

param description defaultValue
q main query (mandatory); see examples below none
q.op AND or OR, relation between tokens OR
df which (single) field should be searched, eg. name none

The main query syntax by example:

  • vw golf - searches for these words in fields defined in df, and joins them using q.op
  • "vw golf" - quotes mark phrases. Tokens must be next to each other.
  • gol?, gol* - wildcards are supported
  • golf~2 - tilde after word (not phrase) - fuzzy searching, finds similar words
  • "vw golf"~3 - tilde after phrase (not word) - proximity search, tokens vw and golf must be within 3 words of each other
  • name:vw - possibility to define explicit fields - <field-name>:<query>
  • name:"vw golf", name:gol?, name:gol*, name:golf~2 - it's all supported
  • name:* - finds documents, which have some value in field name set
  • price:[52 TO 1000] - ranges for numeric fields (including those borders)
  • price:{52 TO 1000} - ranges for numeric fields (EXCLUDING those borders)
  • price:{52 TO 1000], price:[* TO 1000] - also possible
  • vw^4 golf - boosting; token vw is boosted (more important) four times
  • vw OR golf, vw AND golf, vw || golf, vw && golf - explicitly specify operator between tokens
  • "vw golf" OR toyota - also possible
  • (vw AND golf) OR toyota - also possible
  • vw NOT golf, vw ! golf - excluding tokens
  • +vw -golf - must include vw, cannot include golf

DisMax

Example:

curl "http://localhost:8983/solr/mycore/select?defType=dismax&q=vw+golf&qf=brand^2+model"

The main goal of DisMax was to separate the user's query from how the query should be processed. DisMax doesn't support Lucene Query Parser's syntax!

Docs: DixMax Query Parser

param description defaultValue
q main query (mandatory): only basic syntax is supported:
  • quotes for phrases
  • SHOULD MATCH: just word without any modifier (default)
  • MUST MATCH: +word (prefix +)
  • CAN'T MATCH: -word (prefix -)
see examples below		 none
q.alt query which should be used if main query is empty none
qf which fields should be searched with their weights, eg. brand^2.5 model^0.5 none
mm minimum should match; number of SHOULD MATCH words that must match the document; might be absolute or percentage none
pf phrase fields; if the tokens appear in close proximity in this field, the document is boosted, eg title^4 description^1 none
ps phrase slop; the maximum distance between tokens to form a phrase none
tie tie breaker: see the docs none
bq boost query: see the docs none
bf boost function: see the docs none

The main query syntax by example:

  • q=vw golf&qf=brand^2 model - search for words vw (optional) and golf (optional) in fields brand (higher priority) and name (lower priority)
  • q=+vw golf&qf=brand^2 model - search for words vw (mandatory) and golf (optional) in fields brand (higher priority) and name (lower priority)
  • q=+"vw golf"&qf=brand^2 model - search for phrase "vw golf" (mandatory) in fields brand (higher priority) and name (lower priority)
  • q=+vw +golf&qf=brand^2 model - search for words vw (mandatory) and golf (mandatory) in fields brand (higher priority) and name (lower priority)
  • q=+vw +golf -toyota&qf=brand^2 model - search for words vw (mandatory) and golf (mandatory) in fields brand ( higher priority) and name (lower priority); documents cannot contain toyota
  • q=vw golf hatchback&qf=brand^2 model&mm=2 - search for words vw (optional) and golf (optional) and hatchback (optional) in fields brand (higher priority) and name (lower priority); the document is returned if at least two of these optional words have been found
  • q=vw golf hatchback&qf=brand^2 model&mm=50% - search for words vw (optional) and golf (optional) and hatchback (optional) in fields brand (higher priority) and name (lower priority); the document is returned if at least half of the optional words have been found
  • q=vw golf&qf=brand^2 model&pf=name^5 - search for words vw (optional) and golf (optional) in fields brand ( higher priority) and name (lower priority); if the phrase "vw golf" if found in field name, boost the document
  • q=vw golf&qf=brand^2 model&pf=name^5&ps=4 - search for words vw (optional) and golf (optional) in fields brand (higher priority) and name (lower priority); if the phrase "vw golf" if found in field name, boost the document; maximum allowed distance between these words is 4

eDisMax

Example:

curl "http://localhost:8983/solr/mycore/select?defType=edismax&q=vw+AND+golf&qf=brand^2+model"

Extended DisMax, combination DisMax and Lucene Query Parser with add-ons.

As a first approximation, you can think of eDixMax as DisMax, where:

  • you can use AND and OR inside q parameter
  • you can specify a boost function in a better way than in DisMax
  • have some additional enhancements

For better understanding, refer to the docs: Extended DisMax (eDisMax) Query Parser

Comments