# systemtap tapset for python2
# Copyright (C) 2014 Red Hat, Inc.
#
# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License version 2 as
# published by the Free Software Foundation.

# Global Python variables

global python2_vars_seen			# which vars we have seen
global python2_frame_n = 0			# set by iterate_over_frames
global python2_frame = 0			# set by python2_function_entry

# FUNCTION GET_TYPE
# OBJ: PyObject
# RETURNS: string representation of type

function get_type:string (obj) {
    if (obj == 0)
	return ""
    ob_type = @cast (obj, "PyObject", @PYTHON2_LIBRARY)->ob_type
    if (ob_type == 0)
	return ""
    ob_type_name = @cast (ob_type, "PyTypeObject", @PYTHON2_LIBRARY)->tp_name
    return set_string (ob_type_name)
}

# FUNCTION GET_NAME
# F_CODE: The python frame code whose filename we are retrieving

function get_filename (f_code)
{
    co_filename = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_filename;
    co_filename_s = @cast (co_filename, "PyStringObject", @PYTHON2_LIBRARY)->ob_sval;
    return set_string (co_filename_s)
}


# FUNCTION GET_NAME
# F_CODE: The python frame code whose function name we are retrieving

function get_name (f_code)
{
    co_name = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_name;
    co_name_s = @cast (co_name, "PyStringObject", @PYTHON2_LIBRARY)->ob_sval;
    return set_string (co_name_s)
}

# FUNCTION SET_STRING
# STR_P: pointer to string
# RETURNS: string as a stap string

function set_string (str_p) {
    if (str_p == 0)
	return ""
    else
	return user_string (str_p)
}

# FUNCTION GET_SEQUENCE_ITEM
# TUPLE: array of PyObject
# I: index to retrieve 
# SEQ_TYPE: PyTupleObject or PyListObject
# RETURNS: array entry PyObject

function get_sequence_item (tuple, i, seq_type) {
    if (seq_type == "tuple")
	ob_items = @cast (tuple, "PyTupleObject", @PYTHON2_LIBRARY)->ob_item;
    else if (seq_type == "list")
	ob_items = @cast (tuple, "PyListObject", @PYTHON2_LIBRARY)->ob_item;

    ob_item = user_long (ob_items + (i * uarch_bytes()))
    return ob_item
}

# FUNCTION DISPLAY_VALUE
# VALUE: PyObject
# PREFIX: prefix string to display
# SUFFIX: suffix string to display

function display_value (value, prefix, suffix) {
    if (value == 0) {
	return 0
    }
    value_type_str = get_type (value)
    if (value_type_str == "str") {
	value_s = @cast (value, "PyStringObject", @PYTHON2_LIBRARY)->ob_sval;
	value_str = set_string (value_s)
	printf ("%s\"%s\"%s", prefix, value_str, suffix)
    }
    else if (value_type_str == "int") {
	arg_value_int = @cast (value, "PyIntObject", @PYTHON2_LIBRARY)->ob_ival;
	printf ("%s%d%s", prefix, arg_value_int, suffix)
    }
    else if (value_type_str == "tuple" || value_type_str == "list") {
	n = @cast (value, "PyTupleObject", @PYTHON2_LIBRARY)->ob_size;
	if (value_type_str == "list") printf (" = [")
	else printf (" = ")
	for (i = 0; i < n; i++) 
	    display_value (get_sequence_item (value, i, value_type_str), " ", ",")
	if (value_type_str == "list") printf ("]")
    }
    else if (value_type_str == "set") {
	printf (" = {")
	n = @cast (value, "PySetObject", @PYTHON2_LIBRARY)->used;
	for (i = 0; i <= n; i++)
	    display_value (get_set_item (value, i), " ", ",")
	printf ("}")
    }
    else if (value_type_str == "dict") {
	printf (" = {")
	n = @cast (value, "PyDictObject", @PYTHON2_LIBRARY)->ma_used;
	for (i = 0; i <= n; i++) {
	    dict_hash = get_dict_hash (value, i)      
	    dict_key = get_dict_key (value, i)      
	    dict_value = get_dict_value (value, i)      
	    if (dict_hash == 0 && dict_key == 0)
		continue
	    display_value (dict_key, " ", ":")
	    display_value (dict_value, "", "")
	}
	printf (" }")
    }
}

# FUNCTION GET_SET_ITEM
# SET: PySetObject
# I: set index to retrieve
# RETURNS: set entry PyObject

function get_set_item (set, i) {
    entries = @cast (set, "PySetObject", @PYTHON2_LIBRARY)->table;
    n = @cast (set, "PySetObject", @PYTHON2_LIBRARY)->used;
    if (i > n)
	return 0
  
    return @cast (entries, "setentry", @PYTHON2_LIBRARY)[i]->key
}

# FUNCTION GET_DICT_HASH
# DICT: PySetObject
# I: set index to retrieve
# RETURNS: DICT_HASH

function get_dict_hash (dict, i) {
    entries = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_table;
    n = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_used;
    if (i > n)
	return 0
  
    return @cast (entries, "PyDictEntry", @PYTHON2_LIBRARY)[i]->me_hash
}

# FUNCTION GET_DICT_KEY
# DICT: PySetObject
# I: set index to retrieve
# RETURNS: DICT_KEY

function get_dict_key (dict, i) {
    entries = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_table;
    n = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_used;
    if (i > n)
	return 0
  
    return @cast (entries, "PyDictEntry", @PYTHON2_LIBRARY)[i]->me_key
}

# FUNCTION GET_DICT_VALUE
# DICT: PySetObject
# I: set index to retrieve
# RETURNS: DICT_VALUE

function get_dict_value (dict, i) {
    entries = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_table;
    n = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_used;
    if (i > n)
	return 0
  
    return @cast (entries, "PyDictEntry", @PYTHON2_LIBRARY)[i]->me_value;
}

# FUNCTION DISPLAY_DICT_VARIABLE
# DICT: local or global variable name dictionary
# VARIABLE: variable to retrieve
# NAMESPACE: local or global

function display_dict_variable (dict, variable, namespace, co_name_str, co_filename_str) {
    n = @cast (dict, "PyDictObject", @PYTHON2_LIBRARY)->ma_used;
    for (i = 0; i < n; i++) {
	dict_key = get_dict_key (dict, i)      
	dict_value = get_dict_value (dict, i)      
	if (dict_key == 0 || dict_value == 0) break
	key_str = set_string (@cast (dict_key, "PyStringObject", @PYTHON2_LIBRARY)->ob_sval)
	key_type_str = get_type (dict_key)

	if (wildcard_match (key_str, variable)) {
	    foreach ([var, file] in python2_vars_seen) {
		if (var == key_str
		    && file == co_filename_str)
		    return 0
	    }
	    printf ("%s %s %s", namespace, key_type_str, key_str)
	    printf (" in %s at %s", co_name_str, co_filename_str)
	    display_value (dict_value, " = ", " ")
	    printf ("\n")
	    python2_vars_seen[key_str, co_filename_str] = co_name_str
	}
    }
}

# FUNCTION WILDCARD_MATCH
# STR: String to check for a wildcard
# PATTERN: The wildcard pattern.  Wildcard special characters are limited to '*'

function wildcard_match (str, pattern) {
    if (pattern == "*")
	return 1

    # Does pattern start at the beginning of str or is it '*'?
    if (substr (pattern, 0, 1) != "*" && substr (str, 0, 1) != substr (pattern, 0, 1))
	return 0

    hunk = tokenize (pattern,"*")
    while (hunk != "")
    {
	# find pattern hunk in str
	if (isinstr (str, hunk) == 0) {
	    return 0
	}
	match = 0
	for (i = 0; i < strlen (str) - strlen (hunk) + 1; i++) {
	    if (hunk == substr (str, i, strlen (hunk))) {
		match = 1
		break
	    }
	}	
	if (!match) return 0
	    
	# skip past hunk in str
	str = substr (str, i, strlen (str)-i+1)
	hunk = tokenize ("","*")
    }

    # Does pattern end at the end of str or is it '*'?
    if (substr (pattern, strlen (pattern)-1, 1) != "*"
	&& substr (str, strlen (str)-1, 1) != substr (pattern, strlen (pattern)-1, 1))
	return 0
    else
	return 1
}

# FUNCTION ITERATE_OVER_FRAMES
# FRAME: Pointer to a stack frame
# FRAME_N: The frame number of this stack frame

function python2_iterate_over_frames (frame, frame_n)
{
    if (frame_n != 0) {
	python2_frame_n += 1
	frame =  @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_back
    }
    else
	python2_frame_n = 0
    return frame
}

    
# FUNCTION BACKTRACE_ONE_FRAME
# FRAME: Pointer to stack frame
# SKIP_NAME: Pathnames containing this string are ignored.  Default is "/lib/" to
#     ignore system python libraries

function python2_backtrace_one_frame (frame, skip_name)
{
    if (skip_name == "")
	skip_name = "/lib64/"

    f_code = @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_code;
    co_filename_str = get_filename (f_code)
    if (isinstr (co_filename_str, "importlib") == 1 || isinstr (co_filename_str, "<string>") == 1 || isinstr (co_filename_str, "/lib64/") == 1)
	return 0
    if (skip_name != "" && isinstr (co_filename_str, skip_name) == 1)
	return 0
    co_name_str = get_name (f_code)
    co_firstlineno = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_firstlineno;

    printf ("#%d %s ", python2_frame_n, co_name_str)

    co_varnames = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_varnames;
    co_argcount = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_argcount;
    f_localsplus = @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_localsplus;
    for (i = 0; i < co_argcount; i++) {
	if (i == 0) print ("(");
	arg_name_str = user_string (@cast (get_sequence_item (co_varnames, i, "tuple"), "PyStringObject", @PYTHON2_LIBRARY)->ob_sval)
	arg_value = user_long (f_localsplus + (i * uarch_bytes()))
	arg_type_name_str = get_type (arg_value)
	printf ("%s:%s ", arg_name_str, arg_type_name_str)
    }

    if (co_argcount) printf (")");

    printf (" at %s:%d\n", co_filename_str, co_firstlineno)
}
    
# FUNCTION GETVAR_ONE_FRAME
# FRAME: Pointer to stack frame
# MATCH_ARGS: Match function arguments
# VARIABLE: Variable wildcard to be matched.  Default is "*" to match all variables.
# SKIP_NAME: Pathnames containing this string are ignored.  Default is "/lib/" to
#     ignore system python libraries

function python2_getvar_one_frame (frame, match_args, variable, skip_name)
{
    f_code = @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_code;
    co_filename_str = get_filename (f_code)
    if (isinstr (co_filename_str, "importlib") == 1 || isinstr (co_filename_str, "<string>") == 1 || isinstr (co_filename_str, "/lib64/") == 1)
	return 0
    if (skip_name != "" && isinstr (co_filename_str, skip_name) == 1)
	return 0
    co_name_str = get_name (f_code)
    
    f_globals = @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_globals;
    if (f_globals != 0) {
	display_dict_variable (f_globals, variable, "global", co_name_str, co_filename_str)
    }
    f_locals = @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_locals;
    if (f_locals != 0) {
	display_dict_variable (f_locals, variable, "local", co_name_str, co_filename_str)
    }

    co_varnames = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_varnames;
    co_argcount = @cast (f_code, "PyCodeObject", @PYTHON2_LIBRARY)->co_argcount;
    f_localsplus = @cast (frame, "PyFrameObject", @PYTHON2_LIBRARY)->f_localsplus;
    ob_size = @cast (co_varnames, "PyTypeObject", @PYTHON2_LIBRARY)->ob_size;
    # iterate through the local variable list
    for (i = 0; i < ob_size; i++) {
	arg_name_str = user_string (@cast (get_sequence_item (co_varnames, i, "tuple"), "PyStringObject", @PYTHON2_LIBRARY)->ob_sval)
	if (! wildcard_match (arg_name_str, variable))
	    continue
	arg_value = user_long (f_localsplus + (i * uarch_bytes()))
	arg_type_name_str = get_type (arg_value)
	next_var = 0
	foreach ([var, file] in python2_vars_seen)
	if (var == arg_name_str && file == co_filename_str)
	    next_var = 1
	if (next_var || arg_type_name_str == "")
	    continue
	if (i < co_argcount && match_args)
	    printf ("arg%d ", i + 1)
	python2_vars_seen[arg_name_str, co_filename_str] = co_name_str
	printf ("%s %s", arg_type_name_str, arg_name_str)
	printf (" in %s at %s", co_name_str, co_filename_str)
	display_value (arg_value, " = ", " ")
	printf ("\n")
    }
}

# FUNCTION PYTHON2_BACKTRACE
# SKIP_NAME: Pathnames containing this string are ignored.

function python2_backtrace (skip_name)
{
    frame = python2_frame
    for (i = 0; frame != 0; i++) {
       python2_backtrace_one_frame (frame, skip_name)
       frame = python2_iterate_over_frames (frame, i)
    }
}

# FUNCTION PYTHON2_GET_VARIABLE
# SKIP_NAME: Pathnames containing this string are ignored.

function python2_get_variable (variable, skip_name)
{
    frame = python2_frame
    for (i = 0; frame != 0; i++) {
	python2_getvar_one_frame (frame, 1, variable, skip_name)
	frame = python2_iterate_over_frames (frame, i)
    }
}

probe python2.function_entry = process(@PYTHON2_LIBRARY).mark ("function__entry")
{
  # fedora and rhel python pass function name and line number, not frame pointer
  # so grab the frame pointer using debuginfo
  # python2_frame = $arg1
   python2_frame = $f
}

probe python2.function_return = process(@PYTHON2_LIBRARY).mark ("function__return")
{
  # fedora and rhel python pass function name and line number, not frame pointer
  # unfortunately the frame pointer is not always available in debuginfo
  # python2_frame = $arg1
}

probe python.function_entry = python2.function_entry {}
probe python.function_return = python2.function_return {}