[xcl-lens] Change log_entries to set

projects/xcl-lens/src/xcl_lens/parser/rccl.py

@@ -5,8 +5,8 @@ import pandas as pd
 
 class RcclLogParser:
     def __init__(self, verbose=False, hosts=None, ranks=None):
-        # (host, rank, content) -> None
-        self.log_entries = dict()
+        # Deduplicated set of (host, rank, content) tuples
+        self.log_entries: set[tuple[str, int, str]] = set()
 
         # Verbosity flag used by report sections
         self._verbose = verbose
@@ -94,7 +94,7 @@ class RcclLogParser:
                 return
             if self._ranks and rank not in self._ranks:
                 return
-            self.log_entries[(host, rank, content)] = None
+            self.log_entries.add((host, rank, content))
             return
 
         # Backward-compatible fallback for logs without host/pid/tid prefix
@@ -103,13 +103,13 @@ class RcclLogParser:
             rank, content = int(match.group(1)), match.group(2)
             if self._ranks and rank not in self._ranks:
                 return
-            self.log_entries[("-", rank, content)] = None
+            self.log_entries.add(("-", rank, content))
 
     def _report_sys(self):
         """Search patterns and print pre-defined strings if matched"""
         print("===> System Information:\n")
         reported = set()
-        for (_, _, content), _ in self.log_entries.items():
+        for _, _, content in self.log_entries:
             for pattern, out in self.sys_patterns.items():
                 if re.search(pattern, content, re.IGNORECASE):
                     reported.add(out if out is not None else content)
@@ -123,7 +123,7 @@ class RcclLogParser:
         print("===> User-defined Environment Variables:\n")
         env_vars = {}
         pattern = re.compile(r"((?:N|R)CCL_\w+)\s+set(?: by environment)? to\s+(.+)")
-        for (_, _, content), _ in self.log_entries.items():
+        for _, _, content in self.log_entries:
             m = pattern.search(content)
             if m:
                 var_name, var_value = m.group(1), m.group(2)
@@ -143,7 +143,7 @@ class RcclLogParser:
 
         ib_rows = []
         pattern_ib = re.compile(r"NET/IB\s+:\s+GPU Direct RDMA Enabled for HCA\s+(\d+)\s+'([^']+)'")
-        for (host, rank, content), _ in self.log_entries.items():
+        for host, rank, content in self.log_entries:
             m = pattern_ib.search(content)
             if m:
                 hca_no, hca_id = m.groups()
@@ -176,7 +176,7 @@ class RcclLogParser:
             r"GPU Direct RDMA Enabled for GPU\s+(\S+)\s*/\s*"
             r"HCA\s+(\d+)\s*\(distance\s+([^)]*)\),\s*read\s+([01])"
         )
-        for (host, rank, content), _ in self.log_entries.items():
+        for host, rank, content in self.log_entries:
             m = pattern_gpu.search(content)
             if m:
                 gpu, hca_no, distance, read_flag = m.groups()
@@ -245,7 +245,7 @@ class RcclLogParser:
         print(f"===> {title}:\n")
 
         # Filter relevant log lines using the provided filter function
-        data = [(h, r, c) for (h, r, c), _ in self.log_entries.items() if filter_func(c)]
+        data = [(h, r, c) for h, r, c in self.log_entries if filter_func(c)]
         if not data:
             print("  (No data found)\n")
             return
@@ -370,7 +370,7 @@ class RcclLogParser:
             r"(?: \[(\w+)\])?\s+via\s+([\w/]+)"
         )
 
-        for (host, rank, content), _ in self.log_entries.items():
+        for host, rank, content in self.log_entries:
             m = pattern.search(content)
             if m:
                 channel, src, dst, type_, transport = m.groups()