Make headers for batched records shorter (#298)

* make headers for batched records shorter

* sort labels

* update docs

* update CHANGELOG

CHANGELOG.md

@@ -16,6 +16,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `HEAD /api/v1/b/:bucket_name/:entry_name` and `HEAD /api/v1/b/:bucket_name/:entry_name/batch`
   endpoints, [PR-296]https://github.com/reductstore/reductstore/pull/296)
 
+### Changed
+
+- Concise format for headers in `GET /api/v1/:bucket/:entry/batch` response, [PR-298](https://github.com/reductstore/reductstore/pull/298)
+
 ## [1.4.0] - 2023-06-09
 
 ### Fixed

api_tests/entry_api_test.py

@@ -420,8 +420,8 @@ def test_read_batched_records(base_url, session, bucket, method, expected_conten
     assert resp.content == expected_content
     assert resp.headers['content-type'] == 'application/octet-stream'
     assert resp.headers['content-length'] == '20'
-    assert resp.headers['x-reduct-time-1000'] == 'content-length=10,content-type=application/octet-stream'
-    assert resp.headers['x-reduct-time-1100'] == 'content-length=10,content-type=text/plain,label-x="[a,b]"'
+    assert resp.headers['x-reduct-time-1000'] == '10,application/octet-stream'
+    assert resp.headers['x-reduct-time-1100'] == '10,text/plain,x="[a,b]"'
     assert resp.headers['x-reduct-last'] == 'true'
 
 
@@ -437,12 +437,12 @@ def test_read_batched_max_header_size(base_url, session, bucket):
 
     resp = session.get(f'{base_url}/b/{bucket}/entry/batch?q={query_id}')
     assert resp.status_code == 200
-    assert sum(header.startswith('x-reduct-time-') for header in resp.headers) == 84
+    assert sum(header.startswith('x-reduct-time-') for header in resp.headers) == 86
     assert resp.headers['x-reduct-last'] == 'false'
 
     resp = session.get(f'{base_url}/b/{bucket}/entry/batch?q={query_id}')
     assert resp.status_code == 200
-    assert sum(header.startswith('x-reduct-time-') for header in resp.headers) == 16
+    assert sum(header.startswith('x-reduct-time-') for header in resp.headers) == 14
     assert resp.headers['x-reduct-last'] == 'true'
 
     resp = session.get(f'{base_url}/b/{bucket}/entry/batch?q={query_id}')

docs/http-api/entry-api.md

@@ -134,7 +134,7 @@ A UNIX timestamp in microseconds. If it is empty, the latest record is returned.
 
 {% swagger method="head" path="" baseUrl="/api/v1/b/:bucket_name/:entry_name  " summary="Get only meta information about record" %}
 {% swagger-description %}
-The endpoint works exactly as 
+The endpoint works exactly as
 
 `GET /api/v1/b/:bucket_name/:entry_name`
 
@@ -146,7 +146,10 @@ The endpoint works exactly as
 {% swagger-description %}
 Since version 1.5, ReductStore provides a way to read a multiple records in a request. This can improve latency when you have many small records to read. The endpoint sorts all the records by time and concatenates them into a blob and sends it in the body. The meta information is sent for each record as a separate header `x-reduct-time-<timestamp>` which has a value as a CSV row. An example:
 
-`x-reduct-time-192312381273: content-type=text/plain,content-length=100,label-x=y,label-a="[a,b]"`
+`x-reduct-time-192312381273: 100,text/plain,x=y,a="[a,b]"`
+
+The first value is content-length, the second one is content-type, then labels as key=value pairs. If there is a comma in the value, it is escaped with double quotes.
+
 {% endswagger-description %}
 
 {% swagger-parameter in="path" name=":bucket_name" required="true" %}
@@ -190,7 +193,7 @@ Name of entry
 
 {% swagger method="head" path="" baseUrl="/api/v1/b/:bucket_name/:entry_name/batch  " summary="Get only meta information  in bulk" %}
 {% swagger-description %}
-The endpoint works exactly as 
+The endpoint works exactly as
 
 `GET /api/v1/b/:bucket_name/:entry_name/batch`
 

src/http_frontend/entry_api.rs

@@ -120,24 +120,24 @@ fn fetch_and_response_batched_records(
         let name = HeaderName::from_str(&format!("x-reduct-time-{}", reader.timestamp())).unwrap();
 
         let mut meta_data = vec![
-            format!("content-length={}", reader.content_length()),
-            format!("content-type={}", reader.content_type()),
+            reader.content_length().to_string(),
+            reader.content_type().clone(),
         ];
-        meta_data.append(
-            &mut reader
-                .labels()
-                .iter()
-                .map(|(k, v)| {
-                    if v.contains(",") {
-                        format!("label-{}=\"{}\"", k, v)
-                    } else {
-                        format!("label-{}={}", k, v)
-                    }
-                })
-                .collect(),
-        );
-        meta_data.sort();
 
+        let mut labels: Vec<String> = reader
+            .labels()
+            .iter()
+            .map(|(k, v)| {
+                if v.contains(",") {
+                    format!("{}=\"{}\"", k, v)
+                } else {
+                    format!("{}={}", k, v)
+                }
+            })
+            .collect();
+        labels.sort();
+
+        meta_data.append(&mut labels);
         let value: HeaderValue = meta_data.join(",").parse().unwrap();
 
         (name, value)
@@ -819,10 +819,7 @@ mod tests {
         .into_response();
 
         let headers = response.headers();
-        assert_eq!(
-            headers["x-reduct-time-0"],
-            "content-length=6,content-type=text/plain,label-b=\"[a,b]\",label-x=y"
-        );
+        assert_eq!(headers["x-reduct-time-0"], "6,text/plain,b=\"[a,b]\",x=y");
         assert_eq!(headers["content-type"], "application/octet-stream");
         assert_eq!(headers["content-length"], "6");
         assert_eq!(headers["x-reduct-last"], "true");