fix: detect python apps better (#3712)

ebpf/python/procinfo.go

@@ -3,6 +3,7 @@ package python
 import (
 	"bufio"
 	"fmt"
+	"os"
 	"path/filepath"
 	"regexp"
 	"strconv"
@@ -21,6 +22,21 @@ type ProcInfo struct {
 
 var rePython = regexp.MustCompile("/.*/((?:lib)?python)(\\d+)\\.(\\d+)(?:[mu]?(-pyston\\d.\\d)?(?:\\.so)?)?(?:.1.0)?$")
 
+func GetProcInfoForPid(pid uint32) (ProcInfo, error) {
+	mapsFD, err := os.Open(fmt.Sprintf("/proc/%d/maps", pid))
+	if err != nil {
+		return ProcInfo{}, fmt.Errorf("reading proc maps %d: %w", pid, err)
+	}
+	defer mapsFD.Close()
+
+	info, err := GetProcInfo(bufio.NewScanner(mapsFD))
+
+	if err != nil {
+		return ProcInfo{}, fmt.Errorf("GetPythonProcInfo error %s: %w", fmt.Sprintf("/proc/%d/maps", pid), err)
+	}
+	return info, nil
+}
+
 // GetProcInfo parses /proc/pid/map of a python process.
 func GetProcInfo(s *bufio.Scanner) (ProcInfo, error) {
 	res := ProcInfo{}

ebpf/python/pyperf_pid_data.go

@@ -1,7 +1,6 @@
 package python
 
 import (
-	"bufio"
 	"debug/elf"
 	"fmt"
 	"os"
@@ -12,16 +11,9 @@ import (
 )
 
 func GetPyPerfPidData(l log.Logger, pid uint32, collectKernel bool) (*PerfPyPidData, error) {
-	mapsFD, err := os.Open(fmt.Sprintf("/proc/%d/maps", pid))
+	info, err := GetProcInfoForPid(pid)
 	if err != nil {
-		return nil, fmt.Errorf("reading proc maps %d: %w", pid, err)
-	}
-	defer mapsFD.Close()
-
-	info, err := GetProcInfo(bufio.NewScanner(mapsFD))
-
-	if err != nil {
-		return nil, fmt.Errorf("GetPythonProcInfo error %s: %w", fmt.Sprintf("/proc/%d/maps", pid), err)
+		return nil, err
 	}
 	var pythonMeat []*symtab.ProcMap
 	if info.LibPythonMaps == nil {

ebpf/sd/target.go

@@ -41,6 +41,7 @@ const (
 	OptionCollectKernel            = labelMetaPyroscopeOptionsPrefix + "collect_kernel"
 	OptionPythonFullFilePath       = labelMetaPyroscopeOptionsPrefix + "python_full_file_path"
 	OptionPythonEnabled            = labelMetaPyroscopeOptionsPrefix + "python_enabled"
+	OptionPythonProfilingType      = labelMetaPyroscopeOptionsPrefix + "python_profiling_type"
 	OptionPythonBPFDebugLogEnabled = labelMetaPyroscopeOptionsPrefix + "python_bpf_debug_log"
 	OptionPythonBPFErrorLogEnabled = labelMetaPyroscopeOptionsPrefix + "python_bpf_error_log"
 	OptionDemangle                 = labelMetaPyroscopeOptionsPrefix + "demangle"

ebpf/session.go

@@ -691,7 +691,7 @@ func (s *session) selectProfilingType(pid uint32, target *sd.Target) procInfoLit
 	}
 	exe := filepath.Base(exePath)
 
-	if s.pythonEnabled(target) && strings.HasPrefix(exe, "python") || exe == "uwsgi" {
+	if s.pythonEnabled(target) && isPythonProfilingType(pid, exe, target) {
 		return procInfoLite{pid: pid, comm: string(comm), typ: pyrobpf.ProfilingTypePython}
 	}
 	return procInfoLite{pid: pid, comm: string(comm), typ: pyrobpf.ProfilingTypeFramepointers}
@@ -903,6 +903,29 @@ func (s *session) pythonEnabled(target *sd.Target) bool {
 	return enabled
 }
 
+func isPythonProfilingType(pid uint32, exe string, target *sd.Target) bool {
+	if v, present := target.GetFlag(sd.OptionPythonProfilingType); present {
+		return v
+	}
+	if strings.HasPrefix(exe, "python") {
+		return true
+	}
+	if exe == "uwsgi" {
+		return true
+	}
+	if exe == "streamlit" {
+		return true
+	}
+	pi, err := python.GetProcInfoForPid(pid) // todo(korniltsev): reorganize code to not do this twice
+	if err != nil {
+		return false
+	}
+	if pi.PythonMaps != nil || pi.LibPythonMaps != nil {
+		return true
+	}
+	return false
+}
+
 func (s *session) pythonBPFDebugLogEnabled(target *sd.Target) bool {
 	enabled := s.options.PythonBPFDebugLogEnabled
 	if v, present := target.GetFlag(sd.OptionPythonBPFDebugLogEnabled); present {