Welcome, guest | Sign In | My Account | Store | Cart

Input root of your asp project to "File to analyze" it means some default.asp or index.asp. Select what do you what to find and click start analyze. Classic ASP don't care about duplicate function names, what creates unpredictable mistakes.

Python, 276 lines
  1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
""" Classic ASP script analyzer, finds all functions, includes, duplicate functions  """
#TO-DO: testing, it was tested only on one big project

import string, re, sys, os.path, logging, sqlite3
from Tkinter import *
import tkFileDialog, tkMessageBox

def find_functions(file_name, data, db_cursor, functions):
	""" finds all functions in asp script """
	my_re = re.compile(r'[^\'\"][ ]{0,10}function[ ]{1,3}(?P<fun>[a-z0-9_\-]{1,30})', re.IGNORECASE)
	res = my_re.findall(data)
	if res:
		for line in res:			
			print "function", line.lower()
			tmp = file_name + '\t\t' + line.lower()		#(file_name, line)
			functions.append(tmp)			
			db_cursor.execute("""insert into project_functions
					values ((SELECT max(id) FROM project_functions)+1, '%s', '%s', 'function')""" % (file_name, line.lower()))			

	my_re = re.compile(r'[^\'\"][ ]{0,10}sub[ ]{1,3}(?P<fun>[a-z0-9_\-]{1,30})', re.IGNORECASE)
	res = my_re.findall(data)
	if res:
		for line in res:			
			print "sub", line.lower()
			#tmp = (file_name, line)
			tmp = file_name + '\t\t' + line.lower()		#(file_name, line)
			functions.append(tmp)
			db_cursor.execute("""insert into project_functions
					values ((SELECT max(id) FROM project_functions)+1, '%s', '%s', 'sub')""" % (file_name, line.lower()))				


def find_includes(file_path, recursive_level, dir_before, db_cursor, includes, functions):
	""" find al includes,  recursive call """
	for dr in dir_before:
		os.chdir(dr)
		try: f = open(file_path, "r")
		except: pass

	try:
		data = f.read()
		f.close()
	except:
		print dir_before
		print file_path
		return ([], [])

	
	find_functions(os.path.split(file_path)[1], data, db_cursor, functions)

	if len(os.path.split(file_path)[0]) > 1:	 
		#print "dir_before 1", os.path.split(file_path)[0]
		try:
			os.chdir(os.path.split(file_path)[0]) #change work dir to script home dir
			if not os.getcwd() in dir_before:
				dir_before.append(os.getcwd())
		except:
			pass		

   # print "Current directory", os.getcwd(), dir_before, recursive_level

	include_file = []

	my_re = re.compile(r'[^\'][\s]{0,5}<!--[\s]{0,5}#[\s]{0,5}INCLUDE[\s]{0,5}FILE[\s]{0,5}=(?P<file>[ \d\w\.\\\\/_\-"]{1,50})-->', re.IGNORECASE)
	res = my_re.findall(data)
	if not res:
		#print "No include in %s file Regexp not matched" % (file_path)
		#logger.info("No include in %s file Regexp not matched" % (file_path))
		return 1
	else:
		for line in res:			
			include_file.append(line.replace("\"","").strip())
			#print include_file

	#include_file.sort()
	#include_file.reverse()

	includes.append("Includes in: " + os.path.split(file_path)[0] + "\\" + os.path.split(file_path)[1])
	print "Includes in:", os.path.split(file_path)[0] + "\\" + os.path.split(file_path)[1]
	for inc_file in include_file:
		includes.append(inc_file)
		print inc_file 

	for inc_file in include_file:
		find_includes(inc_file, recursive_level + 1, dir_before, db_cursor, includes, functions)
					
	return (includes, functions)


def main(file_path, db_path):
	""" read arguments and analyze """

	if not os.path.exists(file_path):
		print "File %s not found" % (file_path	)
		return 1

	c = ''	 
	conn = sqlite3.connect(db_path)    
	c = conn.cursor()

	try:
		c.execute('''DROP TABLE project_functions''')
	except:
		pass

	c.execute('''create table project_functions
		(id INTEGER PRIMARY KEY, script_name text, fun_name text, fun_type text)''')	
	
	includes = []
	functions = []


	file_path_dir = os.path.split(file_path)[0]
	os.chdir(file_path_dir) #change work dir to script home dir
	print file_path_dir

	includes, functions = \
		find_includes(file_path, 0, [file_path_dir], c, includes, functions)

	conn.commit()

	doubles_fun = []
	c.execute("""SELECT fun_name FROM project_functions
			group by fun_name
			having count(fun_name) > 1""")
	for fun_name in c:
		doubles_fun.append(fun_name)
		
	duplicites = []		
	for i,fun_name in enumerate(doubles_fun):
		dup = fun_name[0] + '\t'
		print i,fun_name[0]
		#t = (fun_name,)	
		c.execute("select script_name, fun_name, fun_type from project_functions where fun_name='%s'" % (fun_name))
		for row in c:
			print '\t', row[0]
			dup = dup + row[0] + ','
		
		duplicites.append(dup)


	c.close()
	
	return (includes, functions, duplicites)


class App:
	""" GUI """
	def __init__(self, master):
		self.master = master

		Label(master, text="File to analyze:").grid(row=0)
		Label(master, text="File for database:").grid(row=1)
		#Label(master, text="File for export (optional):").grid(row=2)

		self.e1 = Entry(master, width = 35)
		self.e2 = Entry(master, width = 35)
		self.e2.insert(INSERT, "C:\\analyze_webscript.db3")
		#self.e3 = Entry(master, width = 35)
		#self.e3.insert(INSERT, "X:\\temp.web\\analyze_webscript.txt")

		self.e1.grid(row=0, column=1)
		self.e2.grid(row=1, column=1)
		#self.e3.grid(row=2, column=1)

		self.b1 = Button(master, text="Browse", command=self.br1, width = 10)
		self.b2 = Button(master, text="Browse", command=self.br2, width = 10)
		#self.b3 = Button(master, text="Browse", command=self.br3, width = 10)

		self.b1.grid(row=0, column=2)
		self.b2.grid(row=1, column=2)
		#self.b3.grid(row=2, column=2)

		self.v1 = IntVar()
		self.c1 = Checkbutton(master, text="find all includes", variable=self.v1)
		self.c1.var = self.v1
		self.c1.grid(columnspan=2, sticky=W)

		self.v2 = IntVar()
		self.c2 = Checkbutton(master, text="find all functions in all includes", variable=self.v2)
		self.c2.var = self.v2
		self.c2.grid(columnspan=2, sticky=W)

		self.v3 = IntVar()
		self.c3 = Checkbutton(master, text="find duplicite functions", variable=self.v3)
		self.c3.var = self.v3
		self.c3.grid(columnspan=2, sticky=W)


		#self.listbox = Listbox(master, selectmode=SINGLE)
		#for item in ["one", "two", "three", "four"]:
		#	 self.listbox.insert(END, item)		 
		#self.listbox.grid(row=3,column=1)

		self.b4 = Button(master, text="Start Analyze", command = self.start_analyze, width = 15)
		self.b4.grid(row=5, column=2)

		self.text = Text(master,width=70, height=30)
		self.vscroll = Scrollbar(master,orient=VERTICAL)
		self.vscroll.grid(row=6, column=4, sticky=N+S)
		self.vscroll.config(command=self.text.yview)
		#self.vscroll.config(command=self.text.yview)
		#self.text.config(yscrollcommand=self.vscroll.set)
		self.text.grid(row=6, columnspan=3, sticky=W)

	def start_analyze(self):
		print "started"
		print self.v1.get()
		print self.v2.get()
		print self.v3.get()

		if (not self.v1.get()) and (not self.v2.get()) and (not self.v3.get()):
			tkMessageBox.showwarning("Action", "Please choose action")
			return 1	

		analyze_file = self.e1.get()
		db_file = self.e2.get()

		if len(analyze_file) == 0 or len(db_file) == 0:
			tkMessageBox.showwarning("Action", "Please set file location")
			return 1	

		print analyze_file
		print db_file
		#print self.e3.get()
		includes, functions, duplicites = \
				main(analyze_file, db_file)
		#self.frame = Frame(width=768, height=576, bg="", colormap="new")
		#self.ef1 = Text(self.frame, width = 50, heigth = 50)
		#self.ef1.insert(INSERT, includes)
		#self.frame.pack()
	
		self.text.delete(1.0, END)		

		if self.v1.get():
			self.text.insert(END, 'INCLUDES:\n')
			self.text.insert(END, '\n'.join(includes))

		if self.v2.get():
			if self.v1.get():
				self.text.insert(END, '\n\n')
			self.text.insert(END, 'FUNCTIONS:\n')
			self.text.insert(END, '\n'.join(functions))

		if self.v3.get():
			if self.v1.get() or self.v2.get():
				self.text.insert(END, '\n\n')
			self.text.insert(END, 'DUPLICATES:\n')
			self.text.insert(END, '\n'.join(duplicites))

		#self.text.pack(side=LEFT,expand=1,fill=BOTH)
		#self.vscroll = Scrollbar(frame2,orient=VERTICAL)
		#self.vscroll.pack(side=LEFT,fill=Y,anchor=E)
		#self.vscroll.config(command=self.text.yview)
		#self.text.config(yscrollcommand=self.vscroll.set)
		#frame2.pack(expand=1,fill=BOTH)

		

	def br1(self):
		self.e1.delete(0, len(self.e1.get()))
		p = tkFileDialog.askopenfilename(title = "File for analyze", initialdir = "C:\\inetpub\\wwwroot")
		self.e1.insert(INSERT, p)

	def br2(self):
		self.e2.delete(0, len(self.e2.get()))
		p = tkFileDialog.askopenfilename(title = "File for database")
		self.e2.insert(INSERT, p)

	#def br3(self):
#		p = tkFileDialog.askopenfilename(title = "File for export")
#		self.e3.insert(INSERT, p)

root = Tk()
root.title("ASP script analyzer")
App(root)
root.mainloop()