Class FlamegraphCollector

Lib/profiling/sampling/stack_collector.py:66–445

class FlamegraphCollector(StackTraceCollector):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.stats = {}
        self._root = {"samples": 0, "children": {}, "threads": set()}
        self._total_samples = 0
        self._sample_count = 0  # Track actual number of samples (not thread traces)
        self._func_intern = {}
        self._string_table = StringTable()
        self._all_threads = set()

        # Thread status statistics (similar to LiveStatsCollector)
        self.thread_status_counts = {
            "has_gil": 0,
            "on_cpu": 0,
            "gil_requested": 0,
            "unknown": 0,
            "has_exception": 0,
            "total": 0,
        }
        self.samples_with_gc_frames = 0

        # Per-thread statistics
        self.per_thread_stats = {}  # {thread_id: {has_gil, on_cpu, gil_requested, unknown, has_exception, total, gc_samples}}

    def collect(self, stack_frames, timestamps_us=None):
        """Override to track thread status statistics before processing frames."""
        # Weight is number of timestamps (samples with identical stack)
        weight = len(timestamps_us) if timestamps_us else 1

        # Increment sample count by weight
        self._sample_count += weight

        # Collect both aggregate and per-thread statistics using base method
        status_counts, has_gc_frame, per_thread_stats = self._collect_thread_status_stats(stack_frames)

        # Merge aggregate status counts (multiply by weight)
        for key in status_counts:
            self.thread_status_counts[key] += status_counts[key] * weight

        # Update aggregate GC frame count
        if has_gc_frame:
            self.samples_with_gc_frames += weight

        # Merge per-thread statistics (multiply by weight)
        for thread_id, stats in per_thread_stats.items():
            if thread_id not in self.per_thread_stats:
                self.per_thread_stats[thread_id] = {
                    "has_gil": 0,
                    "on_cpu": 0,
                    "gil_requested": 0,
                    "unknown": 0,
                    "has_exception": 0,
                    "total": 0,
                    "gc_samples": 0,
                }
            for key, value in stats.items():
                self.per_thread_stats[thread_id][key] += value * weight