blog

commit 00097e7b854af12fc4eca332cf5662a76f13a37f
parent dbe00cdbbbfb74d00c60c19fb938f55a73888ba9
Author: Andrew Laack <andrew@laack.co>
Date:   Thu, 18 Sep 2025 23:51:16 -0500

Merge branch 'master' of ssh://brgr:/home/shared/git/public-repos/blog

Diffstat:
A
python/youtube/analysis/size_analysis.py
 | 
105
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
R
python/youtube/analysis/urls.py -> python/youtube/analysis/urls_parsing.py
 | 
0

2 files changed, 105 insertions(+), 0 deletions(-)
diff --git a/python/youtube/analysis/size_analysis.py b/python/youtube/analysis/size_analysis.py
@@ -0,0 +1,105 @@
+import os
+import matplotlib.pyplot as plt
+from pathlib import Path
+import pandas as pd
+
+URLS_DIR = Path("../urls")
+
+
+if URLS_DIR.exists():
+    files = os.listdir(URLS_DIR)
+else:
+    raise Exception(str(URLS_DIR) + "is empty...")
+
+filtered = []
+
+for file in files:
+    if file.endswith(".csv") and file.__contains__("with_size"):
+        filtered.append(URLS_DIR / file)
+
+if not filtered:
+    raise Exception("No applicable files found")
+
+for file in filtered:
+    df = pd.read_csv(file)
+    print(df.describe())
+
+def convert_to_bytes(str_size : str):
+
+    split = str_size.split("M")
+
+    if len(split) == 1:
+        split = str_size.split("G")
+
+    size = 0
+
+    if str_size.__contains__("GiB"):
+        val = float(split[0])
+        size = val * 1073741824
+
+    if str_size.__contains__("MiB"):
+        val = float(split[0])
+        size = val * 1048576
+
+    return size
+
+sizes = df['size'].apply(convert_to_bytes)
+
+sizes = sizes.to_list()
+
+
+def bytes_to_larger(count_bytes : float):
+    cuts = 0
+    while count_bytes > 1000:
+        count_bytes /= 1000
+        cuts += 1
+
+    suffix = "B"
+
+    if cuts == 1:
+        suffix = 'KB'
+    elif cuts == 2:
+        suffix = 'MB'
+    elif cuts == 3:
+        suffix = "GB"
+    elif cuts == 4:
+        suffix = "TB"
+    elif cuts == 5:
+        suffix = "PB"
+    elif cuts == 6:
+        suffix = "EB"
+
+    return str(count_bytes) + " " + suffix
+
+
+
+def mean(sizes):
+    mean_val = sum(sizes) / len(sizes)
+    return mean_val
+
+
+
+sum_bytes = sum(sizes)
+sum_str = bytes_to_larger(sum_bytes)
+
+mean_val = mean(sizes)
+mean_str = bytes_to_larger(mean_val)
+
+
+print("Average video size: " + mean_str)
+print("Sum of all videos queried: " + sum_str)
+
+
+running_means = []
+running_mean = 0
+itrs = []
+itr = 1
+
+for size in sizes:
+    running_mean = running_mean + ((size - running_mean) / itr)
+    running_means.append(running_mean)
+    itrs.append(itr)
+    itr += 1
+
+plt.plot(itrs, running_means)
+plt.show()
diff --git a/python/youtube/analysis/urls.py b/python/youtube/analysis/urls_parsing.py