Revert lenient start and end anchors quickwit-oss#6089 (quickwit-oss#6098)

* Revert lenient start and end anchors

* Improve api test stability

Author: rdettai-sk

quickwit/quickwit-query/src/query_ast/regex_query.rs

@@ -60,12 +60,6 @@ impl RegexQuery {
         };
         let field_type = field_entry.field_type();
 
-        // tantivy_fst matches `(re)` as if was `^(re)$` and errors if the
-        // anchors are present. We strip them to make the queries more lenient
-        // (matches Elasticsearch behavior).
-        let stripped_regex = self.regex.strip_prefix('^').unwrap_or(&self.regex);
-        let stripped_regex = stripped_regex.strip_suffix('$').unwrap_or(stripped_regex);
-
         match field_type {
             FieldType::Str(text_options) => {
                 text_options.get_indexing_options().ok_or_else(|| {
@@ -75,7 +69,7 @@ impl RegexQuery {
                     ))
                 })?;
 
-                Ok((field, None, stripped_regex.to_string()))
+                Ok((field, None, self.regex.to_string()))
             }
             FieldType::JsonObject(json_options) => {
                 json_options.get_text_indexing_options().ok_or_else(|| {
@@ -96,7 +90,7 @@ impl RegexQuery {
                 // We skip the 1st byte which is a marker to tell this is json. This isn't present
                 // in the dictionary
                 let byte_path_prefix = value.as_serialized()[1..].to_owned();
-                Ok((field, Some(byte_path_prefix), stripped_regex.to_string()))
+                Ok((field, Some(byte_path_prefix), self.regex.to_string()))
             }
             _ => Err(InvalidQuery::SchemaError(
                 "trying to run a regex query on a non-text field".to_string(),

quickwit/rest-api-tests/scenarii/es_compatibility/0020-stats.yaml

@@ -64,12 +64,12 @@ expected:
   _all:
     primaries:
       docs:
-        count: 102
+        count: 104
     total:
       segments:
-        count: 2
+        count: 3
       docs:
-        count: 102
+        count: 104
   indices:
     gharchive:
       primaries:

quickwit/rest-api-tests/scenarii/es_compatibility/0021-cat-indices.yaml

@@ -26,6 +26,8 @@ expected:
   docs.count: '0'
 - index: otel-traces-v0_9
   docs.count: '0'
+- index: simple_es_compat
+  docs.count: '2'
 ---
 method: [GET]
 engines:

quickwit/rest-api-tests/scenarii/es_compatibility/0031-regex.yaml

@@ -61,19 +61,42 @@ expected:
       value: 0
       relation: "eq"
 ---
-# leading ^ and trailing $ are ignored
+# In Elasticsearch, ^ and $ are escaped when they are used as anchors, so
+# ^pushevent$ only matches if the original term is "^pushevent$". In Quickwit
+# this fails (for now) because tantivy-fst returns an error on all zero width
+# assertions.
+engines:
+  - elasticsearch
+endpoint: "simple_es_compat/_search"
 params:
   size: 3
 json:
   track_total_hits: true
   query:
     regexp:
-      type:
-        value: "pushevent"
+      keyword_text:
+        value: "red$"
 expected:
   hits:
     total:
-      value: 60
+      value: 0
+      relation: "eq"
+---
+engines:
+  - elasticsearch
+endpoint: "simple_es_compat/_search"
+params:
+  size: 3
+json:
+  track_total_hits: true
+  query:
+    regexp:
+      keyword_text:
+        value: "gold$"
+expected:
+  hits:
+    total:
+      value: 1
       relation: "eq"
 ---
 # regex in query_string
@@ -87,4 +110,4 @@ expected:
   hits:
     total:
       value: 60
-      relation: "eq"
+      relation: "eq"

quickwit/rest-api-tests/scenarii/es_compatibility/_setup.elasticsearch.yaml

@@ -7,6 +7,10 @@ method: DELETE
 endpoint: empty_index
 status_code: null
 ---
+method: DELETE
+endpoint: simple_es_compat
+status_code: null
+---
 # empty index
 method: PUT
 endpoint: empty_index
@@ -118,10 +122,38 @@ method: PUT
 endpoint: gharchive/_settings
 json: { "number_of_replicas": 0 }
 ---
+# Create index
+method: PUT
+endpoint: simple_es_compat
+json: {
+  "mappings": {
+    "properties": {
+      "keyword_text": {
+        "type": "keyword",
+      }
+    }
+  }
+}
+---
+method: PUT
+endpoint: simple_es_compat/_settings
+json: { "number_of_replicas": 0 }
+---
 # Ingest documents
 method: POST
 endpoint: _bulk
 params:
   refresh: "true"
 headers: {"Content-Type": "application/json", "content-encoding": "gzip"}
 body_from_file: gharchive-bulk.json.gz
+---
+method: POST
+endpoint: _bulk
+params:
+  refresh: "true"
+headers: {"Content-Type": "application/json"}
+ndjson:
+  - {"index":{"_index":"simple_es_compat"}}
+  - {"keyword_text": "red"}
+  - {"index":{"_index":"simple_es_compat"}}
+  - {"keyword_text": "gold$"}

quickwit/rest-api-tests/scenarii/es_compatibility/_setup.quickwit.yaml

@@ -10,6 +10,12 @@ api_root: http://localhost:7280/api/v1/
 endpoint: indexes/empty_index
 status_code: null
 ---
+# Delete possibly remaining index
+method: DELETE
+api_root: http://localhost:7280/api/v1/
+endpoint: indexes/simple_es_compat
+status_code: null
+---
 # Create index
 method: POST
 api_root: http://localhost:7280/api/v1/
@@ -104,3 +110,30 @@ params:
 ndjson:
   - {"fast_text": "abc-123", "obj": {"nested_text": "abc-123"}}
   - {"fast_text": "def-456", "obj": {"nested_text": "ghi-789"}}
+
+---
+# Create simple_es_compat index
+method: POST
+api_root: http://localhost:7280/api/v1/
+endpoint: indexes/
+json:
+  version: "0.7"
+  index_id: simple_es_compat
+  doc_mapping:
+    field_mappings:
+      - name: keyword_text
+        type: text
+        fast: true
+        indexed: true
+        tokenizer: raw
+sleep_after: 1
+---
+# Ingest documents into simple_es_compat
+method: POST
+api_root: http://localhost:7280/api/v1/
+endpoint: simple_es_compat/ingest
+params:
+  commit: force
+ndjson:
+  - {"keyword_text": "red"}
+  - {"keyword_text": "gold$"}

quickwit/rest-api-tests/scenarii/es_compatibility/bulk/0007-illegal-index-name.yaml

@@ -1,3 +1,6 @@
+# allowed characters are different between ES and Quickwit
+engines:
+  - quickwit
 ndjson:
   - index: { "_index": "test-index" }
   - message: Hola, Mundo!

quickwit/rest-api-tests/scenarii/es_compatibility/bulk/_setup.elasticsearch.yaml

@@ -1,10 +1,6 @@
 # Delete possibly remaining index
 method: DELETE
-endpoint: test-index
-status_code: null
----
-method: DELETE
-endpoint: test-index-pattern-777
+endpoint: test-index*
 status_code: null
 ---
 method: PUT

quickwit/rest-api-tests/scenarii/es_compatibility/bulk/_setup.quickwit.yaml

@@ -6,6 +6,11 @@ status_code: null
 ---
 method: DELETE
 api_root: http://localhost:7280/api/v1/
+endpoint: indexes/test-index-pattern-11
+status_code: null
+---
+method: DELETE
+api_root: http://localhost:7280/api/v1/
 endpoint: indexes/test-index-pattern-777
 status_code: null
 ---

quickwit/rest-api-tests/scenarii/es_compatibility/bulk/_teardown.quickwit.yaml

@@ -4,4 +4,12 @@ endpoint: indexes/test-index
 ---
 method: DELETE
 api_root: http://localhost:7280/api/v1/
+endpoint: indexes/test-index-pattern-11
+---
+method: DELETE
+api_root: http://localhost:7280/api/v1/
+endpoint: indexes/test-index-pattern-777
+---
+method: DELETE
+api_root: http://localhost:7280/api/v1/
 endpoint: templates/test-index-template