Added new script to capture number of symbols typically

benchmarks/bin_symbol_size.py

@@ -1,18 +1,36 @@
 #! /usr/bin/env python
 """Go through /bin and print out the number of symbols in each binary."""
 
+import itertools
 import json
 import os
 import subprocess
 from collections import defaultdict
 
 data = defaultdict(lambda: 0)
 
-for root, dirs, files in os.walk("/bin"):
+
+def is_elf_file(filepath):
+    try:
+        with open(filepath, "rb") as file:
+            # Read the first four bytes of the file
+            magic_number = file.read(4)
+    except FileNotFoundError:
+        return False
+    except IOError:
+        return False
+
+    # Check if the magic number matches the ELF signature
+    return magic_number == b"\x7fELF"
+
+
+for root, dirs, files in itertools.chain(os.walk("/bin"), os.walk("/usr/bin")):
     for file in files:
         full_path_file = os.path.join(root, file)
         if not os.path.isfile(full_path_file) or not os.access(full_path_file, os.X_OK):
             pass
+        if not is_elf_file(full_path_file):
+            continue  # skip non-ELF files
         result = subprocess.run(
             f"readelf -s {full_path_file} | wc -l",
             capture_output=True,

benchmarks/graph_bin_symbol_size.py

@@ -0,0 +1,34 @@
+#! /usr/bin/env python3
+
+import json
+import os
+
+import pandas as pd
+from plotnine import (
+    aes,
+    after_stat,
+    geom_histogram,
+    ggplot,
+    labs,
+    save_as_pdf_pages,
+    theme_minimal,
+)
+
+SCRIPT_DIR = os.path.dirname(__file__)
+JSON_FILE = open(os.path.join(SCRIPT_DIR, "symbols-ubuntu.json"), "r")
+
+# Load JSON data into a Python dictionary
+data_dict = json.load(JSON_FILE)
+
+# Convert to DataFrame
+df = pd.DataFrame(list(data_dict.items()), columns=["File", "Symbols"])
+
+# Create plot
+plot = (
+    ggplot(df, aes(x="Symbols", y=after_stat("ncount")))
+    + geom_histogram(bins=100, fill="skyblue", color="black", alpha=0.7)
+    + labs(title="", x="Number of Symbols", y="Normalized Count")
+    + theme_minimal()
+)
+
+save_as_pdf_pages([plot])