ES|QL: Improve support for TEXT fields in functions (elastic#106810)

Re-submitting elastic#106688 after
a revert due to a conflict after merge

docs/changelog/106810.yaml

@@ -0,0 +1,5 @@
+pr: 106810
+summary: "ES|QL: Improve support for TEXT fields in functions"
+area: ES|QL
+type: bug
+issues: []

docs/reference/esql/functions/description/date_format.asciidoc

@@ -0,0 +1,5 @@
+// This is generated by ESQL's AbstractFunctionTestCase. Do no edit it. See ../README.md for how to regenerate it.
+
+*Description*
+
+Returns a string representation of a date, in the provided format.

docs/reference/esql/functions/layout/date_format.asciidoc

@@ -0,0 +1,14 @@
+// This is generated by ESQL's AbstractFunctionTestCase. Do no edit it. See ../README.md for how to regenerate it.
+
+[discrete]
+[[esql-date_format]]
+=== `DATE_FORMAT`
+
+*Syntax*
+
+[.text-center]
+image::esql/functions/signature/date_format.svg[Embedded,opts=inline]
+
+include::../parameters/date_format.asciidoc[]
+include::../description/date_format.asciidoc[]
+include::../types/date_format.asciidoc[]

docs/reference/esql/functions/parameters/date_format.asciidoc

@@ -0,0 +1,7 @@
+*Parameters*
+
+`dateFormat`::
+A valid date pattern
+
+`date`::
+Date expression

docs/reference/esql/functions/types/date_extract.asciidoc

@@ -6,4 +6,5 @@
 |===
 datePart | date | result
 keyword | datetime | long
+text | datetime | long
 |===

docs/reference/esql/functions/types/date_format.asciidoc

@@ -0,0 +1,10 @@
+// This is generated by ESQL's AbstractFunctionTestCase. Do no edit it. See ../README.md for how to regenerate it.
+
+*Supported types*
+
+[%header.monospaced.styled,format=dsv,separator=|]
+|===
+dateFormat | date | result
+keyword | datetime | keyword
+text | datetime | keyword
+|===

docs/reference/esql/functions/types/date_parse.asciidoc

@@ -7,4 +7,5 @@
 datePattern | dateString | result
 keyword | keyword | datetime
 keyword | text | datetime
+text | text | datetime
 |===

x-pack/plugin/esql/qa/testFixtures/src/main/resources/meta.csv-spec

@@ -12,18 +12,18 @@ auto_bucket              |"double|date auto_bucket(field:integer|long|double|dat
 avg                      |"double avg(number:double|integer|long)"                                           |number                     |"double|integer|long"                 |   ""                                               |double                    | "The average of a numeric field."                      | false                | false | true
 case                     |"boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version case(condition:boolean, trueValue...:boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version)"                               |[condition, trueValue]             |["boolean", "boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version"]            |["", ""]                                            |"boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version"                    | "Accepts pairs of conditions and values. The function returns the value that belongs to the first condition that evaluates to true."                      | [false, false]       | true | false
 ceil                     |"double|integer|long|unsigned_long ceil(number:double|integer|long|unsigned_long)"     |number                        |"double|integer|long|unsigned_long"                 | "Numeric expression. If `null`, the function returns `null`." | "double|integer|long|unsigned_long"                    | "Round a number up to the nearest integer."                      | false                | false | false
-cidr_match               |boolean cidr_match(ip:ip, blockX...:keyword)                         |[ip, blockX]             |[ip, keyword]            |["", "CIDR block to test the IP against."]                                            |boolean                    | "Returns true if the provided IP is contained in one of the provided CIDR blocks."                      | [false, false]       | true | false
+cidr_match               |"boolean cidr_match(ip:ip, blockX...:keyword|text)"                         |[ip, blockX]             |[ip, "keyword|text"]            |["", "CIDR block to test the IP against."]                                            |boolean                    | "Returns true if the provided IP is contained in one of the provided CIDR blocks."                      | [false, false]       | true | false
 coalesce                 |"boolean|text|integer|keyword|long coalesce(first:boolean|text|integer|keyword|long, ?rest...:boolean|text|integer|keyword|long)"                           |first             | "boolean|text|integer|keyword|long"            | "Expression to evaluate"                                            |"boolean|text|integer|keyword|long"                    | "Returns the first of its arguments that is not null. If all arguments are null, it returns `null`."                      | false       | true | false
 concat                   |"keyword concat(string1:keyword|text, string2...:keyword|text)"                             |[string1, string2]             |["keyword|text", "keyword|text"]            |["", ""]                                            |keyword                    | "Concatenates two or more strings."                      | [false, false]       | true | false
 cos                      |"double cos(number:double|integer|long|unsigned_long)"      |number             |"double|integer|long|unsigned_long"                 | "An angle, in radians"  |double         | "Returns the trigonometric cosine of an angle"                      | false                | false | false
 cosh                     |"double cosh(number:double|integer|long|unsigned_long)"     |number             |"double|integer|long|unsigned_long"                 | "The number who's hyperbolic cosine is to be returned" |double                    | "Returns the hyperbolic cosine of a number"                      | false                | false | false
 count                    |"long count(?field:boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version)"                                         |field                     |"boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version"                 | "Column or literal for which to count the number of values."                                                 |long                    | "Returns the total number (count) of input values."                      | true                | false | true
 count_distinct           |"long count_distinct(field:boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|version, ?precision:integer)"                        |[field, precision]             |["boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|version, integer"]            |["Column or literal for which to count the number of distinct values.", ""]                                            |long                    | "Returns the approximate number of distinct values."                      | [false, true]       | false | true
 date_diff                |"integer date_diff(unit:keyword|text, startTimestamp:date, endTimestamp:date)"|[unit, startTimestamp, endTimestamp] |["keyword|text", "date", "date"]  |["A valid date unit", "A string representing a start timestamp", "A string representing an end timestamp"] |integer | "Subtract 2 dates and return their difference in multiples of a unit specified in the 1st argument" | [false, false, false] | false | false
-date_extract             |long date_extract(datePart:keyword, date:date)                          |[datePart, date]             |[keyword, date]            |["Part of the date to extract. Can be: aligned_day_of_week_in_month; aligned_day_of_week_in_year; aligned_week_of_month; aligned_week_of_year; ampm_of_day; clock_hour_of_ampm; clock_hour_of_day; day_of_month; day_of_week; day_of_year; epoch_day; era; hour_of_ampm; hour_of_day; instant_seconds; micro_of_day; micro_of_second; milli_of_day; milli_of_second; minute_of_day; minute_of_hour; month_of_year; nano_of_day; nano_of_second; offset_seconds; proleptic_month; second_of_day; second_of_minute; year; or year_of_era.", "Date expression"]                                            |long                    | "Extracts parts of a date, like year, month, day, hour."                      | [false, false]       | false | false
-date_format              |keyword date_format(?dateFormat:keyword, date:date)                           |[dateFormat, date]             |[keyword, date]            |["A valid date pattern", "Date expression"]                                            |keyword                    | "Returns a string representation of a date, in the provided format."                      | [true, false]       | false | false
-date_parse               |"date date_parse(?datePattern:keyword, dateString:keyword|text)"|[datePattern, dateString]|["keyword", "keyword|text"]|["A valid date pattern", "A string representing a date"]|date                 |Parses a string into a date value | [true, false]       | false          | false
-date_trunc               |"date date_trunc(interval:keyword, date:date)"                            |[interval, date]             |[keyword, date]            |["Interval; expressed using the timespan literal syntax.", "Date expression"]                                            |date                    | "Rounds down a date to the closest interval."                      | [false, false]       | false | false
+date_extract             |"long date_extract(datePart:keyword|text, date:date)"                          |[datePart, date]             |["keyword|text", date]            |["Part of the date to extract. Can be: aligned_day_of_week_in_month; aligned_day_of_week_in_year; aligned_week_of_month; aligned_week_of_year; ampm_of_day; clock_hour_of_ampm; clock_hour_of_day; day_of_month; day_of_week; day_of_year; epoch_day; era; hour_of_ampm; hour_of_day; instant_seconds; micro_of_day; micro_of_second; milli_of_day; milli_of_second; minute_of_day; minute_of_hour; month_of_year; nano_of_day; nano_of_second; offset_seconds; proleptic_month; second_of_day; second_of_minute; year; or year_of_era.", "Date expression"]                                            |long                    | "Extracts parts of a date, like year, month, day, hour."                      | [false, false]       | false | false
+date_format              |"keyword date_format(?dateFormat:keyword|text, date:date)"                           |[dateFormat, date]             |["keyword|text", date]            |["A valid date pattern", "Date expression"]                                            |keyword                    | "Returns a string representation of a date, in the provided format."                      | [true, false]       | false | false
+date_parse               |"date date_parse(?datePattern:keyword|text, dateString:keyword|text)"|[datePattern, dateString]|["keyword|text", "keyword|text"]|["A valid date pattern", "A string representing a date"]|date                 |Parses a string into a date value | [true, false]       | false          | false
+date_trunc               |"date date_trunc(interval:keyword, date:date)"                            |[interval, date]             |["keyword", date]            |["Interval; expressed using the timespan literal syntax.", "Date expression"]                                            |date                    | "Rounds down a date to the closest interval."                      | [false, false]       | false | false
 e                        |double e()                                                   | null                    | null             | null                                               |double                    | "Euler’s number."                      | null                 | false | false
 ends_with                |"boolean ends_with(str:keyword|text, suffix:keyword|text)"                             |[str, suffix]             |["keyword|text", "keyword|text"]            |["", ""]                                            |boolean                    | "Returns a boolean that indicates whether a keyword string ends with another string"                      | [false, false]       | false | false
 floor                    |"double|integer|long|unsigned_long floor(number:double|integer|long|unsigned_long)"    |number                       |"double|integer|long|unsigned_long"    | ""                                                 |"double|integer|long|unsigned_long"                    | "Round a number down to the nearest integer."                      | false                | false | false
@@ -116,17 +116,17 @@ synopsis:keyword
 "double avg(number:double|integer|long)"
 "boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version case(condition:boolean, trueValue...:boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version)"
 "double|integer|long|unsigned_long ceil(number:double|integer|long|unsigned_long)"
-boolean cidr_match(ip:ip, blockX...:keyword)
+"boolean cidr_match(ip:ip, blockX...:keyword|text)"
 "boolean|text|integer|keyword|long coalesce(first:boolean|text|integer|keyword|long, ?rest...:boolean|text|integer|keyword|long)"
 "keyword concat(string1:keyword|text, string2...:keyword|text)"
 "double cos(number:double|integer|long|unsigned_long)"
 "double cosh(number:double|integer|long|unsigned_long)"
 "long count(?field:boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|unsigned_long|version)"
 "long count_distinct(field:boolean|cartesian_point|date|double|geo_point|integer|ip|keyword|long|text|version, ?precision:integer)"
 "integer date_diff(unit:keyword|text, startTimestamp:date, endTimestamp:date)"
-long date_extract(datePart:keyword, date:date)
-keyword date_format(?dateFormat:keyword, date:date)
-"date date_parse(?datePattern:keyword, dateString:keyword|text)"
+"long date_extract(datePart:keyword|text, date:date)"
+"keyword date_format(?dateFormat:keyword|text, date:date)"
+"date date_parse(?datePattern:keyword|text, dateString:keyword|text)"
 "date date_trunc(interval:keyword, date:date)"
 double e()
 "boolean ends_with(str:keyword|text, suffix:keyword|text)"

x-pack/plugin/esql/qa/testFixtures/src/main/resources/string.csv-spec

@@ -1160,3 +1160,11 @@ required_feature: esql.agg_values
                                                                                              null | null
 // end::values-grouped-result[]
 ;
+
+
+splitBasedOnField
+from employees | where emp_no == 10001 | eval split = split("fooMbar", gender) | keep gender, split;
+
+gender:keyword | split:keyword 
+M              | [foo, bar]
+;

x-pack/plugin/esql/src/main/java/org/elasticsearch/xpack/esql/expression/EsqlTypeResolutions.java

@@ -27,6 +27,15 @@
 
 public class EsqlTypeResolutions {
 
+    public static Expression.TypeResolution isStringAndExact(Expression e, String operationName, TypeResolutions.ParamOrdinal paramOrd) {
+        Expression.TypeResolution resolution = TypeResolutions.isString(e, operationName, paramOrd);
+        if (resolution.unresolved()) {
+            return resolution;
+        }
+
+        return isExact(e, operationName, paramOrd);
+    }
+
     public static Expression.TypeResolution isExact(Expression e, String operationName, TypeResolutions.ParamOrdinal paramOrd) {
         if (e instanceof FieldAttribute fa) {
             if (DataTypes.isString(fa.dataType())) {

x-pack/plugin/esql/src/main/java/org/elasticsearch/xpack/esql/expression/function/scalar/date/DateExtract.java

@@ -14,6 +14,7 @@
 import org.elasticsearch.xpack.esql.expression.function.FunctionInfo;
 import org.elasticsearch.xpack.esql.expression.function.Param;
 import org.elasticsearch.xpack.esql.expression.function.scalar.EsqlConfigurationFunction;
+import org.elasticsearch.xpack.esql.type.EsqlDataTypes;
 import org.elasticsearch.xpack.ql.InvalidArgumentException;
 import org.elasticsearch.xpack.ql.expression.Expression;
 import org.elasticsearch.xpack.ql.expression.TypeResolutions;
@@ -28,10 +29,10 @@
 import java.util.List;
 import java.util.function.Function;
 
+import static org.elasticsearch.xpack.esql.expression.EsqlTypeResolutions.isStringAndExact;
 import static org.elasticsearch.xpack.esql.type.EsqlDataTypeConverter.EsqlConverter.STRING_TO_CHRONO_FIELD;
 import static org.elasticsearch.xpack.esql.type.EsqlDataTypeConverter.chronoToLong;
 import static org.elasticsearch.xpack.ql.expression.TypeResolutions.isDate;
-import static org.elasticsearch.xpack.ql.expression.TypeResolutions.isStringAndExact;
 
 public class DateExtract extends EsqlConfigurationFunction {
 
@@ -42,7 +43,7 @@ public DateExtract(
         Source source,
         // Need to replace the commas in the description here with semi-colon as there's a bug in the CSV parser
         // used in the CSVTests and fixing it is not trivial
-        @Param(name = "datePart", type = { "keyword" }, description = """
+        @Param(name = "datePart", type = { "keyword", "text" }, description = """
             Part of the date to extract.
             Can be: aligned_day_of_week_in_month; aligned_day_of_week_in_year; aligned_week_of_month;
             aligned_week_of_year; ampm_of_day; clock_hour_of_ampm; clock_hour_of_day; day_of_month; day_of_week;
@@ -76,7 +77,7 @@ private ChronoField chronoField() {
         if (chronoField == null) {
             Expression field = children().get(0);
             try {
-                if (field.foldable() && field.dataType() == DataTypes.KEYWORD) {
+                if (field.foldable() && EsqlDataTypes.isString(field.dataType())) {
                     chronoField = (ChronoField) STRING_TO_CHRONO_FIELD.convert(field.fold());
                 }
             } catch (Exception e) {