[2] | 1 | #!/usr/bin/env python |
---|
| 2 | """ |
---|
| 3 | check_galaxy can be run by hand, although it is meant to run from cron |
---|
| 4 | via the check_galaxy.sh script in Galaxy's cron/ directory. |
---|
| 5 | """ |
---|
| 6 | |
---|
| 7 | import socket, sys, os, time, tempfile, filecmp, htmllib, formatter, getopt |
---|
| 8 | from user import home |
---|
| 9 | |
---|
| 10 | # options |
---|
| 11 | if os.environ.has_key( "DEBUG" ): |
---|
| 12 | debug = os.environ["DEBUG"] |
---|
| 13 | else: |
---|
| 14 | debug = False |
---|
| 15 | scripts_dir = os.path.abspath( os.path.dirname( sys.argv[0] ) ) |
---|
| 16 | test_data_dir = os.path.join( scripts_dir, "..", "test-data" ) |
---|
| 17 | # what tools to run - not so pretty |
---|
| 18 | tools = { |
---|
| 19 | "gops_intersect_1" : |
---|
| 20 | [ |
---|
| 21 | { |
---|
| 22 | "inputs" : |
---|
| 23 | ( |
---|
| 24 | os.path.join( test_data_dir, "1.bed" ), |
---|
| 25 | os.path.join( test_data_dir, "2.bed" ) |
---|
| 26 | ) |
---|
| 27 | }, |
---|
| 28 | { "check_file" : os.path.join( test_data_dir, "gops_intersect_out.bed" ) }, |
---|
| 29 | { |
---|
| 30 | "tool_run_options" : |
---|
| 31 | { |
---|
| 32 | "input1" : "1.bed", |
---|
| 33 | "input2" : "2.bed", |
---|
| 34 | "min" : "1", |
---|
| 35 | "returntype" : "" |
---|
| 36 | } |
---|
| 37 | } |
---|
| 38 | ] |
---|
| 39 | } |
---|
| 40 | |
---|
| 41 | # handle arg(s) |
---|
| 42 | def usage(): |
---|
| 43 | print "usage: check_galaxy.py <server>" |
---|
| 44 | sys.exit(1) |
---|
| 45 | |
---|
| 46 | try: |
---|
| 47 | opts, args = getopt.getopt( sys.argv[1:], 'n' ) |
---|
| 48 | except getopt.GetoptError, e: |
---|
| 49 | print str(e) |
---|
| 50 | usage() |
---|
| 51 | if len( args ) < 1: |
---|
| 52 | usage() |
---|
| 53 | server = args[0] |
---|
| 54 | if server.endswith(".g2.bx.psu.edu"): |
---|
| 55 | if debug: |
---|
| 56 | print "Checking a PSU Galaxy server, using maint file" |
---|
| 57 | maint = "/errordocument/502/%s/maint" % args[0].split('.', 1)[0] |
---|
| 58 | else: |
---|
| 59 | maint = None |
---|
| 60 | new_history = False |
---|
| 61 | for o, a in opts: |
---|
| 62 | if o == "-n": |
---|
| 63 | if debug: |
---|
| 64 | print "Specified -n, will create a new history" |
---|
| 65 | new_history = True |
---|
| 66 | else: |
---|
| 67 | usage() |
---|
| 68 | |
---|
| 69 | # state information |
---|
| 70 | var_dir = os.path.join( home, ".check_galaxy", server ) |
---|
| 71 | if not os.access( var_dir, os.F_OK ): |
---|
| 72 | os.makedirs( var_dir, 0700 ) |
---|
| 73 | |
---|
| 74 | # get user/pass |
---|
| 75 | login_file = os.path.join( var_dir, "login" ) |
---|
| 76 | try: |
---|
| 77 | f = open( login_file, 'r' ) |
---|
| 78 | except: |
---|
| 79 | print "Please create the file:" |
---|
| 80 | print " ", login_file |
---|
| 81 | print "This should contain a username and password to log in to" |
---|
| 82 | print "Galaxy with, on one line, separated by whitespace, e.g.:" |
---|
| 83 | print "" |
---|
| 84 | print "check_galaxy@example.com password" |
---|
| 85 | print "" |
---|
| 86 | print "If the user does not exist, check_galaxy will create it" |
---|
| 87 | print "for you." |
---|
| 88 | sys.exit(1) |
---|
| 89 | ( username, password ) = f.readline().split() |
---|
| 90 | |
---|
| 91 | # find/import twill |
---|
| 92 | lib_dir = os.path.join( scripts_dir, "..", "lib" ) |
---|
| 93 | sys.path.append( lib_dir ) |
---|
| 94 | from galaxy import eggs |
---|
| 95 | import pkg_resources |
---|
| 96 | pkg_resources.require( "twill" ) |
---|
| 97 | import twill |
---|
| 98 | import twill.commands as tc |
---|
| 99 | |
---|
| 100 | # default timeout for twill browser is never |
---|
| 101 | socket.setdefaulttimeout(300) |
---|
| 102 | |
---|
| 103 | # user-agent |
---|
| 104 | tc.agent("Mozilla/5.0 (compatible; check_galaxy/0.1)") |
---|
| 105 | tc.config('use_tidy', 0) |
---|
| 106 | |
---|
| 107 | class Browser: |
---|
| 108 | |
---|
| 109 | def __init__(self): |
---|
| 110 | self.server = server |
---|
| 111 | self.maint = maint |
---|
| 112 | self.tool = None |
---|
| 113 | self.tool_opts = None |
---|
| 114 | self.id = None |
---|
| 115 | self.status = None |
---|
| 116 | self.check_file = None |
---|
| 117 | self.hid = None |
---|
| 118 | self.cookie_jar = os.path.join( var_dir, "cookie_jar" ) |
---|
| 119 | dprint("cookie jar path: %s" % self.cookie_jar) |
---|
| 120 | if not os.access(self.cookie_jar, os.R_OK): |
---|
| 121 | dprint("no cookie jar at above path, creating") |
---|
| 122 | tc.save_cookies(self.cookie_jar) |
---|
| 123 | tc.load_cookies(self.cookie_jar) |
---|
| 124 | |
---|
| 125 | def get(self, path): |
---|
| 126 | tc.go("http://%s%s" % (self.server, path)) |
---|
| 127 | tc.code(200) |
---|
| 128 | |
---|
| 129 | def reset(self): |
---|
| 130 | self.tool = None |
---|
| 131 | self.tool_opts = None |
---|
| 132 | self.id = None |
---|
| 133 | self.status = None |
---|
| 134 | self.check_file = None |
---|
| 135 | self.delete_datasets() |
---|
| 136 | self.get("/root/history") |
---|
| 137 | p = didParser() |
---|
| 138 | p.feed(tc.browser.get_html()) |
---|
| 139 | if len(p.dids) > 0: |
---|
| 140 | print "Remaining datasets ids:", " ".join( p.dids ) |
---|
| 141 | raise Exception, "History still contains datasets after attempting to delete them" |
---|
| 142 | if new_history: |
---|
| 143 | self.get("/history/delete_current") |
---|
| 144 | tc.save_cookies(self.cookie_jar) |
---|
| 145 | |
---|
| 146 | def check_redir(self, url): |
---|
| 147 | try: |
---|
| 148 | tc.get_browser()._browser.set_handle_redirect(False) |
---|
| 149 | tc.go(url) |
---|
| 150 | tc.code(302) |
---|
| 151 | tc.get_browser()._browser.set_handle_redirect(True) |
---|
| 152 | dprint( "%s is returning redirect (302)" % url ) |
---|
| 153 | return(True) |
---|
| 154 | except twill.errors.TwillAssertionError, e: |
---|
| 155 | tc.get_browser()._browser.set_handle_redirect(True) |
---|
| 156 | dprint( "%s is not returning redirect (302): %s" % (url, e) ) |
---|
| 157 | code = tc.browser.get_code() |
---|
| 158 | if code == 502: |
---|
| 159 | is_maint = self.check_maint() |
---|
| 160 | if is_maint: |
---|
| 161 | dprint( "Galaxy is down, but a maint file was found, so not sending alert" ) |
---|
| 162 | sys.exit(0) |
---|
| 163 | else: |
---|
| 164 | print "Galaxy is down (code 502)" |
---|
| 165 | sys.exit(1) |
---|
| 166 | return(False) |
---|
| 167 | |
---|
| 168 | # checks for a maint file |
---|
| 169 | def check_maint(self): |
---|
| 170 | if self.maint is None: |
---|
| 171 | #dprint( "Warning: unable to check maint file for %s" % self.server ) |
---|
| 172 | return(False) |
---|
| 173 | try: |
---|
| 174 | self.get(self.maint) |
---|
| 175 | return(True) |
---|
| 176 | except twill.errors.TwillAssertionError, e: |
---|
| 177 | return(False) |
---|
| 178 | |
---|
| 179 | def login(self, user, pw): |
---|
| 180 | self.get("/user/login") |
---|
| 181 | tc.fv("1", "email", user) |
---|
| 182 | tc.fv("1", "password", pw) |
---|
| 183 | tc.submit("Login") |
---|
| 184 | tc.code(200) |
---|
| 185 | if len(tc.get_browser().get_all_forms()) > 0: |
---|
| 186 | # uh ohs, fail |
---|
| 187 | p = userParser() |
---|
| 188 | p.feed(tc.browser.get_html()) |
---|
| 189 | if p.no_user: |
---|
| 190 | dprint("user does not exist, will try creating") |
---|
| 191 | self.create_user(user, pw) |
---|
| 192 | elif p.bad_pw: |
---|
| 193 | raise Exception, "Password is incorrect" |
---|
| 194 | else: |
---|
| 195 | raise Exception, "Unknown error logging in" |
---|
| 196 | tc.save_cookies(self.cookie_jar) |
---|
| 197 | |
---|
| 198 | def create_user(self, user, pw): |
---|
| 199 | self.get("/user/create") |
---|
| 200 | tc.fv("1", "email", user) |
---|
| 201 | tc.fv("1", "password", pw) |
---|
| 202 | tc.fv("1", "confirm", pw) |
---|
| 203 | tc.submit("Create") |
---|
| 204 | tc.code(200) |
---|
| 205 | if len(tc.get_browser().get_all_forms()) > 0: |
---|
| 206 | p = userParser() |
---|
| 207 | p.feed(tc.browser.get_html()) |
---|
| 208 | if p.already_exists: |
---|
| 209 | raise Exception, 'The user you were trying to create already exists' |
---|
| 210 | |
---|
| 211 | def upload(self, file): |
---|
| 212 | self.get("/tool_runner/index?tool_id=upload1") |
---|
| 213 | tc.fv("1","file_type", "bed") |
---|
| 214 | tc.formfile("1","file_data", file) |
---|
| 215 | tc.submit("runtool_btn") |
---|
| 216 | tc.code(200) |
---|
| 217 | |
---|
| 218 | def runtool(self): |
---|
| 219 | self.get("/tool_runner/index?tool_id=%s" % self.tool) |
---|
| 220 | for k, v in self.tool_opts.items(): |
---|
| 221 | tc.fv("1", k, v) |
---|
| 222 | tc.submit("runtool_btn") |
---|
| 223 | tc.code(200) |
---|
| 224 | |
---|
| 225 | def wait(self): |
---|
| 226 | sleep_amount = 1 |
---|
| 227 | count = 0 |
---|
| 228 | maxiter = 16 |
---|
| 229 | while count < maxiter: |
---|
| 230 | count += 1 |
---|
| 231 | self.get("/root/history") |
---|
| 232 | page = tc.browser.get_html() |
---|
| 233 | if page.find( '<!-- running: do not change this comment, used by TwillTestCase.wait -->' ) > -1: |
---|
| 234 | time.sleep( sleep_amount ) |
---|
| 235 | sleep_amount += 1 |
---|
| 236 | else: |
---|
| 237 | break |
---|
| 238 | if count == maxiter: |
---|
| 239 | raise Exception, "Tool never finished" |
---|
| 240 | |
---|
| 241 | def check_status(self): |
---|
| 242 | self.get("/root/history") |
---|
| 243 | p = historyParser() |
---|
| 244 | p.feed(tc.browser.get_html()) |
---|
| 245 | if p.status != "ok": |
---|
| 246 | raise Exception, "JOB %s NOT OK: %s" % (p.id, p.status) |
---|
| 247 | self.id = p.id |
---|
| 248 | self.status = p.status |
---|
| 249 | #return((p.id, p.status)) |
---|
| 250 | |
---|
| 251 | def diff(self): |
---|
| 252 | self.get("/datasets/%s/display/index" % self.id) |
---|
| 253 | data = tc.browser.get_html() |
---|
| 254 | tmp = tempfile.mkstemp() |
---|
| 255 | dprint("tmp file: %s" % tmp[1]) |
---|
| 256 | tmpfh = os.fdopen(tmp[0], 'w') |
---|
| 257 | tmpfh.write(data) |
---|
| 258 | tmpfh.close() |
---|
| 259 | if filecmp.cmp(tmp[1], self.check_file): |
---|
| 260 | dprint("Tool output is as expected") |
---|
| 261 | else: |
---|
| 262 | if not debug: |
---|
| 263 | os.remove(tmp[1]) |
---|
| 264 | raise Exception, "Tool output differs from expected" |
---|
| 265 | if not debug: |
---|
| 266 | os.remove(tmp[1]) |
---|
| 267 | |
---|
| 268 | def delete_datasets(self): |
---|
| 269 | self.get("/root/history") |
---|
| 270 | p = didParser() |
---|
| 271 | p.feed(tc.browser.get_html()) |
---|
| 272 | dids = p.dids |
---|
| 273 | for did in dids: |
---|
| 274 | self.get("/root/delete?id=%s" % did) |
---|
| 275 | |
---|
| 276 | def check_if_logged_in(self): |
---|
| 277 | self.get("/user") |
---|
| 278 | p = loggedinParser() |
---|
| 279 | p.feed(tc.browser.get_html()) |
---|
| 280 | return p.logged_in |
---|
| 281 | |
---|
| 282 | class userParser(htmllib.HTMLParser): |
---|
| 283 | def __init__(self): |
---|
| 284 | htmllib.HTMLParser.__init__(self, formatter.NullFormatter()) |
---|
| 285 | self.in_span = False |
---|
| 286 | self.no_user = False |
---|
| 287 | self.bad_pw = False |
---|
| 288 | self.already_exists = False |
---|
| 289 | def start_span(self, attrs): |
---|
| 290 | self.in_span = True |
---|
| 291 | def end_span(self): |
---|
| 292 | self.in_span = False |
---|
| 293 | def handle_data(self, data): |
---|
| 294 | if self.in_span: |
---|
| 295 | if data == "No such user": |
---|
| 296 | self.no_user = True |
---|
| 297 | elif data == "Invalid password": |
---|
| 298 | self.bad_pw = True |
---|
| 299 | elif data == "User with that email already exists": |
---|
| 300 | self.already_exists = True |
---|
| 301 | |
---|
| 302 | class historyParser(htmllib.HTMLParser): |
---|
| 303 | def __init__(self): |
---|
| 304 | htmllib.HTMLParser.__init__(self, formatter.NullFormatter()) |
---|
| 305 | self.status = None |
---|
| 306 | self.id = None |
---|
| 307 | def start_div(self, attrs): |
---|
| 308 | # find the top history item |
---|
| 309 | for i in attrs: |
---|
| 310 | if i[0] == "class" and i[1].startswith("historyItemWrapper historyItem historyItem-"): |
---|
| 311 | self.status = i[1].rsplit("historyItemWrapper historyItem historyItem-", 1)[1] |
---|
| 312 | dprint("status: %s" % self.status) |
---|
| 313 | if i[0] == "id" and i[1].startswith("historyItem-"): |
---|
| 314 | self.id = i[1].rsplit("historyItem-", 1)[1] |
---|
| 315 | dprint("id: %s" % self.id) |
---|
| 316 | if self.status is not None: |
---|
| 317 | self.reset() |
---|
| 318 | |
---|
| 319 | class didParser(htmllib.HTMLParser): |
---|
| 320 | def __init__(self): |
---|
| 321 | htmllib.HTMLParser.__init__(self, formatter.NullFormatter()) |
---|
| 322 | self.dids = [] |
---|
| 323 | def start_div(self, attrs): |
---|
| 324 | for i in attrs: |
---|
| 325 | if i[0] == "id" and i[1].startswith("historyItemContainer-"): |
---|
| 326 | self.dids.append( i[1].rsplit("historyItemContainer-", 1)[1] ) |
---|
| 327 | dprint("got a dataset id: %s" % self.dids[-1]) |
---|
| 328 | |
---|
| 329 | class loggedinParser(htmllib.HTMLParser): |
---|
| 330 | def __init__(self): |
---|
| 331 | htmllib.HTMLParser.__init__(self, formatter.NullFormatter()) |
---|
| 332 | self.in_p = False |
---|
| 333 | self.logged_in = False |
---|
| 334 | def start_p(self, attrs): |
---|
| 335 | self.in_p = True |
---|
| 336 | def end_p(self): |
---|
| 337 | self.in_p = False |
---|
| 338 | def handle_data(self, data): |
---|
| 339 | if self.in_p: |
---|
| 340 | if data == "You are currently not logged in.": |
---|
| 341 | self.logged_in = False |
---|
| 342 | elif data.startswith( "You are currently logged in as " ): |
---|
| 343 | self.logged_in = True |
---|
| 344 | |
---|
| 345 | def dprint(str): |
---|
| 346 | if debug: |
---|
| 347 | print str |
---|
| 348 | |
---|
| 349 | # do stuff here |
---|
| 350 | if __name__ == "__main__": |
---|
| 351 | |
---|
| 352 | dprint("checking %s" % server) |
---|
| 353 | |
---|
| 354 | b = Browser() |
---|
| 355 | |
---|
| 356 | # login (or not) |
---|
| 357 | if b.check_if_logged_in(): |
---|
| 358 | dprint("we are already logged in (via cookies), hooray!") |
---|
| 359 | else: |
---|
| 360 | dprint("not logged in... logging in") |
---|
| 361 | b.login(username, password) |
---|
| 362 | |
---|
| 363 | for tool, params in tools.iteritems(): |
---|
| 364 | |
---|
| 365 | check_file = "" |
---|
| 366 | |
---|
| 367 | # make sure history and state is clean |
---|
| 368 | b.reset() |
---|
| 369 | b.tool = tool |
---|
| 370 | |
---|
| 371 | # get all the tool run conditions |
---|
| 372 | for dict in params: |
---|
| 373 | for k, v in dict.items(): |
---|
| 374 | if k == 'inputs': |
---|
| 375 | for file in v: |
---|
| 376 | b.upload(file) |
---|
| 377 | elif k == 'check_file': |
---|
| 378 | b.check_file = v |
---|
| 379 | elif k == 'tool_run_options': |
---|
| 380 | b.tool_opts = v |
---|
| 381 | else: |
---|
| 382 | raise Exception, "Unknown key in tools dict: %s" % k |
---|
| 383 | |
---|
| 384 | b.runtool() |
---|
| 385 | b.wait() |
---|
| 386 | b.check_status() |
---|
| 387 | b.diff() |
---|
| 388 | b.delete_datasets() |
---|
| 389 | |
---|
| 390 | # by this point, everything else has succeeded. there should be no maint. |
---|
| 391 | is_maint = b.check_maint() |
---|
| 392 | if is_maint: |
---|
| 393 | print "Galaxy is up and fully functional, but a maint file is in place." |
---|
| 394 | sys.exit(1) |
---|
| 395 | |
---|
| 396 | sys.exit(0) |
---|