From a5dfe18e4c360bb6d49b141ad9470d5a49550916 Mon Sep 17 00:00:00 2001
From: Jose Fonseca <jose.r.fonseca@gmail.com>
Date: Sat, 9 Nov 2024 12:09:05 +0000
Subject: [PATCH] Prevent long node IDs.

Fixes https://github.com/jrfonseca/gprof2dot/issues/99
---
 gprof2dot.py | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/gprof2dot.py b/gprof2dot.py
index 1fec633..074dc39 100755
--- a/gprof2dot.py
+++ b/gprof2dot.py
@@ -34,6 +34,7 @@
 import fnmatch
 import codecs
 import io
+import hashlib
 
 assert sys.version_info[0] >= 3
 
@@ -3535,15 +3536,15 @@ def attr(self, what, **attrs):
 
     def node(self, node, **attrs):
         self.write("\t")
-        self.id(node)
+        self.node_id(node)
         self.attr_list(attrs)
         self.write(";\n")
 
     def edge(self, src, dst, **attrs):
         self.write("\t")
-        self.id(src)
+        self.node_id(src)
         self.write(" -> ")
-        self.id(dst)
+        self.node_id(dst)
         self.attr_list(attrs)
         self.write(";\n")
 
@@ -3559,11 +3560,22 @@ def attr_list(self, attrs):
                 first = False
             else:
                 self.write(", ")
-            self.id(name)
+            assert isinstance(name, str)
+            assert name.isidentifier()
+            self.write(name)
             self.write('=')
             self.id(value)
         self.write(']')
 
+    def node_id(self, id):
+        # Node IDs need to be unique (can't be truncated) but dot doesn't allow
+        # IDs longer than 16384 characters, so use an hash instead for the huge
+        # C++ symbols that can arise, as seen in
+        # https://github.com/jrfonseca/gprof2dot/issues/99
+        if isinstance(id, str) and len(id) > 1024:
+            id = '_' + hashlib.sha1(id.encode('utf-8'), usedforsecurity=False).hexdigest()
+        self.id(id)
+
     def id(self, id):
         if isinstance(id, (int, float)):
             s = str(id)