From 1d0a0a8eb5a4eb0c9ba3415f21180333ccb11e18 Mon Sep 17 00:00:00 2001
From: Alexey Rubaseff <alexey.rubasheff@gmail.com>
Date: Sun, 24 Mar 2024 11:56:15 +0200
Subject: [PATCH] ! chunked all items recursive parsing

---
 rozetka/entities/supercategory.py | 15 +++++++++++----
 rozetka/runners/parse_api.py      | 30 +++++++++++++++++-------------
 2 files changed, 28 insertions(+), 17 deletions(-)

diff --git a/rozetka/entities/supercategory.py b/rozetka/entities/supercategory.py
index 05cb43e..51ce8d7 100644
--- a/rozetka/entities/supercategory.py
+++ b/rozetka/entities/supercategory.py
@@ -69,10 +69,7 @@ def get_fat_menu_categories():
     return output
 
 
-def get_all_items_recursively(loop=False) -> List[Item]:
-    if loop:
-        return list()
-
+def get_all_item_ids_recursively():
     _ = get_super_category_ids()
     categories = list(get_all_categories_recursively())
     all_categories = list(set(categories))
@@ -85,6 +82,16 @@ def get_all_items_recursively(loop=False) -> List[Item]:
     items_ids = tools.fncs_map((_._get_item_ids for _ in all_categories)) or []
     items_ids = list(set(chain(*items_ids)))
     LOG.green(f"Got {len(items_ids)} item ids from {all_categories_len} categories")
+    return items_ids, all_categories_len
+
+
+def get_all_items_recursively(loop=False, items_ids=None, all_categories_len=None) -> List[Item]:
+    if loop:
+        return list()
+
+    if items_ids is None:
+        items_ids, all_categories_len = get_all_item_ids_recursively()
+
     items = Item.parse_multiple(*items_ids, parse_subitems=False)
     LOG.green(f"Got {len(items)} items from {all_categories_len} categories")
 
diff --git a/rozetka/runners/parse_api.py b/rozetka/runners/parse_api.py
index cbb44f4..9b29cd1 100644
--- a/rozetka/runners/parse_api.py
+++ b/rozetka/runners/parse_api.py
@@ -10,7 +10,7 @@
 
 from rozetka.entities.item import Item
 from rozetka.entities.point import Point
-from rozetka.entities.supercategory import get_all_items_recursively
+from rozetka.entities.supercategory import get_all_items_recursively, get_all_item_ids_recursively
 from rozetka.tools import db, constants, tools
 
 LOG = Log.get_logger()
@@ -61,19 +61,23 @@ def _main():
     start = pendulum.now()
     LOG.verbose = constants.VERBOSE
 
-    all_items = get_all_items_recursively()
-
-    LOG.green(f"Building points for {len(all_items)} items")
-    points = list(map(build_item_point, all_items))
-    LOG.green(f"Dumping {len(points)} points")
-    # https://docs.influxdata.com/influxdb/v2.4/write-data/best-practices/optimize-writes/
-    chunked_points = tools.slice_list(points, 5000)
-    for chunked_points_item in Bar(f"Dumping {len(chunked_points)} point chunks").iter(chunked_points):
-        asyncio.run(db.dump_points_async(record=chunked_points_item))
-
+    all_item_ids, _ = get_all_item_ids_recursively()
+    chunked_items_ids = tools.slice_list(all_item_ids, 10000)
+    overal_length = 0
+    for chunked_item_ids in Bar(f"Dumping {len(chunked_items_ids)} point chunks").iter(chunked_items_ids):
+        all_items = get_all_items_recursively(chunked_item_ids)
+        LOG.green(f"Building points for {len(all_items)} items")
+        points = list(map(build_item_point, all_items))
+        LOG.green(f"Dumping {len(points)} points")
+        # https://docs.influxdata.com/influxdb/v2.4/write-data/best-practices/optimize-writes/
+        chunked_points = tools.slice_list(points, 5000)
+        for chunked_points_item in Bar(f"Dumping {len(chunked_points)} point chunks").iter(chunked_points):
+            asyncio.run(db.dump_points_async(record=chunked_points_item))
+
+        overal_length += len(points)
     duration = pendulum.now().diff_for_humans(start)
-    LOG.green(f"Duration: {duration}")
-    return len(points)
+    LOG.green(f"Points: {overal_length}, Duration: {duration}")
+    return overal_length
 
 
 def main():