From eca80c624026ad1a7638e4dbea4e59f21b57aa2a Mon Sep 17 00:00:00 2001
From: hmacr <hmac.devo@gmail.com>
Date: Tue, 25 Jul 2023 19:02:53 +0530
Subject: [PATCH] address review comments

---
 src/marqo/tensor_search/constants.py     |  3 +++
 src/marqo/tensor_search/tensor_search.py | 12 +++++++++---
 src/marqo/tensor_search/utils.py         |  7 +++++++
 tests/tensor_search/test_get_stats.py    |  8 ++++++--
 tests/tensor_search/test_utils.py        | 19 +++++++++++++++++++
 5 files changed, 44 insertions(+), 5 deletions(-)

diff --git a/src/marqo/tensor_search/constants.py b/src/marqo/tensor_search/constants.py
index 9314362f0..d34c7b6e7 100644
--- a/src/marqo/tensor_search/constants.py
+++ b/src/marqo/tensor_search/constants.py
@@ -27,3 +27,6 @@
 NON_OFFICIAL_LUCENE_SPECIAL_CHARS = {
     ' '
 }
+
+NUM_BYTES_IN_KB = 1024
+SUPPORTED_SIZES_FOR_STATS = ['B', 'KB', 'MB', 'GB', 'TB', 'PB', 'EB', 'ZB']
diff --git a/src/marqo/tensor_search/tensor_search.py b/src/marqo/tensor_search/tensor_search.py
index 9d848e1af..b9a676c27 100644
--- a/src/marqo/tensor_search/tensor_search.py
+++ b/src/marqo/tensor_search/tensor_search.py
@@ -269,11 +269,17 @@ def _autofill_index_settings(index_settings: dict):
 
 def get_stats(config: Config, index_name: str):
     doc_count = HttpRequests(config).post(path=F"{index_name}/_count")["count"]
-    index_info = HttpRequests(config).get(path=F"_cat/indices/{index_name}?format=json")
-    size = index_info[0]["store.size"]
+    index_stats = HttpRequests(config).get(path=F"{index_name}/_stats")["indices"]
+    size_in_bytes = None
+    try:
+        size_in_bytes = index_stats[index_name]["total"]["store"]["size_in_bytes"]
+    except AttributeError:
+        raise errors.IndexNotFoundError(message="Tried to get a non-existent index: {}".format(index_name))
+
+    formatted_size = utils.convert_bytes_to_human_readable_format(size_in_bytes)
     return {
         "numberOfDocuments": doc_count,
-        "size": size
+        "size": formatted_size
     }
 
 
diff --git a/src/marqo/tensor_search/utils.py b/src/marqo/tensor_search/utils.py
index c6b691e0e..3cc2db091 100644
--- a/src/marqo/tensor_search/utils.py
+++ b/src/marqo/tensor_search/utils.py
@@ -2,6 +2,7 @@
 import typing
 import functools
 import json
+import math
 from timeit import default_timer as timer
 import torch
 from marqo import errors
@@ -349,3 +350,9 @@ def is_tensor_field(field: str,
         return field in tensor_fields
     else:
         return field not in non_tensor_fields
+
+
+def convert_bytes_to_human_readable_format(size_in_bytes: int) -> str:
+    size_factor = math.floor(math.log(size_in_bytes) / math.log(constants.NUM_BYTES_IN_KB))
+    processed_size = size_in_bytes / math.pow(constants.NUM_BYTES_IN_KB, size_factor)
+    return f"{processed_size:.2f} {constants.SUPPORTED_SIZES_FOR_STATS[size_factor]}"
diff --git a/tests/tensor_search/test_get_stats.py b/tests/tensor_search/test_get_stats.py
index 6ce81c693..1012df470 100644
--- a/tests/tensor_search/test_get_stats.py
+++ b/tests/tensor_search/test_get_stats.py
@@ -20,7 +20,9 @@ def test_get_stats_empty(self):
         except IndexNotFoundError as s:
             pass
         tensor_search.create_vector_index(config=self.config, index_name=self.index_name_1)
-        assert tensor_search.get_stats(config=self.config, index_name=self.index_name_1)["numberOfDocuments"] == 0
+        index_stats = tensor_search.get_stats(config=self.config, index_name=self.index_name_1)
+        assert index_stats["numberOfDocuments"] == 0
+        assert len(index_stats["size"]) != 0
 
     def test_get_stats_non_empty(self):
         try:
@@ -35,4 +37,6 @@ def test_get_stats_non_empty(self):
                 auto_refresh=True, device="cpu"
             )
         )
-        assert tensor_search.get_stats(config=self.config, index_name=self.index_name_1)["numberOfDocuments"] == 3
+        index_stats = tensor_search.get_stats(config=self.config, index_name=self.index_name_1)
+        assert index_stats["numberOfDocuments"] == 3
+        assert len(index_stats["size"]) != 0
diff --git a/tests/tensor_search/test_utils.py b/tests/tensor_search/test_utils.py
index ccb4342f3..b556e63ad 100644
--- a/tests/tensor_search/test_utils.py
+++ b/tests/tensor_search/test_utils.py
@@ -398,3 +398,22 @@ def test_is_tensor_field_providing_one_empty(self):
         non_tensor_fields = []
         with self.assertRaises(errors.InternalError):
             utils.is_tensor_field('field1', tensor_fields=tensor_fields, non_tensor_fields=non_tensor_fields)
+
+    def test_convert_bytes_to_human_readable_format(self):
+        size_in_bytes = 1000  # 1000 B
+        assert utils.convert_bytes_to_human_readable_format(size_in_bytes) == "1000.00 B"
+
+        size_in_bytes = 16121  # 15.74 KB
+        assert utils.convert_bytes_to_human_readable_format(size_in_bytes) == "15.74 KB"
+
+        size_in_bytes = 9874321  # 9.42 MB
+        assert utils.convert_bytes_to_human_readable_format(size_in_bytes) == "9.42 MB"
+
+        size_in_bytes = 10000000000  # 9.31 GB
+        assert utils.convert_bytes_to_human_readable_format(size_in_bytes) == "9.31 GB"
+
+        size_in_bytes = 712893712304234  # 648.37 TB
+        assert utils.convert_bytes_to_human_readable_format(size_in_bytes) == "648.37 TB"
+
+        size_in_bytes = 6212893712323224  # 5.52 PB
+        assert utils.convert_bytes_to_human_readable_format(size_in_bytes) == "5.52 PB"