UCSC %s

") print("

") def printHgcSection(name, rightSideContent, id=None): " print a section header for the hgc page " print("") if id is None: print("

") else: print("

" % id) print("

" % name) print("

") print(rightSideContent) print("

") print("

") def getGbHeader(): " Python cannot include files with the preprocessor, so manually copied kent/src/hg/lib/gbHeader.h for now " return """ %s UCSC %s """ def webStartGbNoBanner(prefix, title): " output the part, largely copied from web.c / webStartGbNoBanner " print (getGbHeader() % (prefix, title)) def runCmd(cmd, mustRun=True): " wrapper around system() that prints error messages. cmd preferably a list, not just a string. " import subprocess ret = subprocess.call(cmd) if ret!=0 and mustRun: errAbort("Could not run command %s" % cmd) return ret def printContentType(contType="text/html", fname=None): " print the HTTP Content-type header line with an optional file name " global contentLineDone contentLineDone = True print("Content-type: %s; charset=utf-8" % contType) if fname is not None: print("Content-Disposition: attachment; filename=%s" % fname) print def queryBottleneck(host, port, ip): " contact UCSC-style bottleneck server to get current delay time. From hg/lib/botDelay.c " # send ip address import socket s = socket.socket() s.connect((host, int(port))) msg = ip d = chr(len(msg))+msg s.send(d) # read delay time expLen = ord(s.recv(1)) totalLen = 0 buf = list() while True: resp = s.recv(1024) buf.append(resp) totalLen+= len(resp) if totalLen==expLen: break return int("".join(buf)) def hgBotDelay(): """ Implement bottleneck delay, get bottleneck server from hg.conf. This behaves similar to the function src/hg/lib/botDelay.c:hgBotDelay It does not use the hgsid, currently it always uses the IP address. Using the hgsid makes little sense. It is more lenient than that C version. """ import time if "DOCUMENT_ROOT" not in os.environ: # skip if not called from Apache return global hgConf hgConf = parseHgConf() if "bottleneck.host" not in hgConf: return ip = os.environ["REMOTE_ADDR"] delay = queryBottleneck(hgConf["bottleneck.host"], hgConf["bottleneck.port"], ip) if delay>10000: time.sleep(delay/1000.0) if delay>20000: errAbort("Too many queries. Your IP has been blocked. Please contact genome-www@soe.ucsc.edu to unblock your IP address.") sys.exit(0) def parseRa(text): " Parse ra-style string and return as dict name -> value " import string lines = text.split("\n") data = dict() for l in lines: if len(l)==0: continue key, val = string.split(l, " ", maxsplit=1) data[key] = val return data def lineFileNextRow(inFile): """ parses tab-sep file with headers in first line. Yields collection.namedtuples. Strips "#"-prefix from header line. Cannot parse headers with non-alpha characters and fields that are not ASCII. Code for these cases is commented out, for performance reasons. """ import gzip, string if isinstance(inFile, str): if inFile.endswith(".gz"): fh = gzip.open(inFile, 'rb') else: fh = open(inFile) else: fh = inFile line1 = fh.readline() line1 = line1.rstrip("\n").lstrip("#") headers = line1.split("\t") #headers = [re.sub("[^a-zA-Z0-9_]","_", h) for h in headers] #headers = [x if x!="" else "noName" for x in headers] #filtHeads = [] #for h in headers: #if h[0].isdigit(): #filtHeads.append("x"+h) #else: #filtHeads.append(h) #headers = filtHeads Record = namedtuple('tsvRec', headers) for line in fh: if line.startswith("#"): continue #line = line.decode("latin1") line = line.rstrip("\n") fields = string.split(line, "\t", maxsplit=len(headers)-1) try: rec = Record(*fields) except Exception, msg: logging.error("Exception occured while parsing line, %s" % msg) logging.error("Filename %s" % fh.name) logging.error("Line was: %s" % line) logging.error("Does number of fields match headers?") logging.error("Headers are: %s" % headers) raise Exception("header count: %d != field count: %d wrong field count in line %s" % (len(headers), len(fields), line)) yield rec def parseDict(fname): """ Parse text file in format keyvalue and return as dict key->val. Does not abort on duplicate keys, for performance reasons. """ import gzip d = {} if fname.endswith(".gz"): fh = gzip.open(fname) else: fh = open(fname) for line in fh: key, val = line.rstrip("\n").split("\t") d[key] = val return d def cgiString(name, default=None): " get named cgi variable as a string, like lib/cheapcgi.c " val = cgiArgs.getfirst(name, default=default) return val def cgiGetAll(): return cgiArgs def makeRandomKey(numBits=128+33): " copied line-by-line from kent/src/lib/htmlshell.c:makeRandomKey " import base64 numBytes = (numBits + 7) / 8 # round up to nearest whole byte. numBytes = ((numBytes+2)/3)*3 # round up to the nearest multiple of 3 to avoid equals-char padding in base64 output f = open("/dev/urandom", "r") # open random system device for read-only access. binaryString = f.read(numBytes) f.close() return base64.b64encode(binaryString, "Aa") # replace + and / with characters that are URL-friendly. def cartDbLoadFromId(conn, table, cartId, oldCart): " Like src/hg/lib/cart.c, opens cart table and parses cart contents given a cartId of the format 123123_csctac " import urlparse if cartId==None: return {} cartFields = cartId.split("_") if len(cartFields)!=2: errAbort("Could not parse identifier %s for cart table %s" % (cgi.escape(cartId), table)) idStr, secureId = cartFields query = "SELECT contents FROM "+table+" WHERE id=%(id)s and sessionKey=%(sessionKey)s" rows = sqlQuery(conn, query, {"id":idStr, "sessionKey":secureId}) if len(rows)==0: # silently ignore invalid cart IDs for now. Future code may want to generate a new cart. return {} cartList = urlparse.parse_qs(rows[0][0]) # by default, python returns a dict with key -> list of vals. We need only the first one for key, vals in cartList.iteritems(): oldCart[key] =vals[0] return oldCart centralConn = None def hConnectCentral(): " similar to src/hg/lib/hdb.c:hConnectCentral. We use a much simpler cache, because we usually read all rows into memory. " global centralConn if centralConn: return centralConn centralDb = cfgOption("central.db") if centralDb is None: errAbort("Could not find central.db in hg.conf. Installation error.") centralUser = cfgOption("central.user") if centralUser is None: errAbort("Could not find central.user in hg.conf. Installation error.") centralPwd = cfgOption("central.password") if centralPwd is None: errAbort("Could not find central.password in hg.conf. Installation error.") centralHost = cfgOption("central.host") if centralHost is None: errAbort("Could not find central.host in hg.conf. Installation error.") conn = sqlConnect(centralDb, host=centralHost, user=centralUser, passwd=centralPwd) centralConn = conn return conn def cartNew(conn, table): " create a new cart and return ID " sessionKey = makeRandomKey() sqlQuery(conn, "INSERT %s VALUES(0,'',0,now(),now(),0,%s" % (table, sessionKey)); sqlLastId = conn.insert_id() newId = "%d_%s" % (sqlLastId, sessionKey) return newId def cartAndCookieSimple(): """ Make the cart from the user cookie (user settings) and the hgsid CGI parameter (session settings, e.g. a browser tab). This is somewhat similar to cartAndCookie from hg/lib/cart.c Two important differences: this cart does not add all CGI arguments automatically. It also does not run cart.c:cartJustify, so track priorities are not applied. Also, if there is no hgsid parameter or no cookie, we do not create a new cart. """ import Cookie # no cgiApoptosis yet - maybe needed in the future. see cart.c / cartNew if "HTTP_COOKIE" in os.environ: cookies = Cookie.SimpleCookie(os.environ["HTTP_COOKIE"]) cookieName = cfgOption("central.cookie", "hguid") hguid = cookies.get(cookieName) if hguid is not None: hguid = hguid.value # cookies have values and other attributes. We only need the value else: hguid = None hgsid = cgiString("hgsid") conn = hConnectCentral() cart = {} userInfo = cartDbLoadFromId(conn, "userDb", hguid, cart) sessionInfo = cartDbLoadFromId(conn, "sessionDb", hgsid, cart) return cart def cartString(cart, default=None): " Get a string from the cart. For better readability for programmers used to the C code. " return cart.get(cart, default) def cgiSetup(): """ do the usual browser CGI setup: parse the hg.conf file, parse the CGI variables, get the cart, do bottleneck delay. Returns the cart. This is not part of the C code (though it maybe should). """ parseHgConf() global cgiArgs cgiArgs = cgi.FieldStorage() # Python has built-in cgiSpoof support: sys.argv[1] is the query string if run from the command line hgBotDelay() if cgiString("debug"): global verboseLevel verboseLevel = int(cgiString("debug")) cart = cartAndCookieSimple() return cart #if __file__=="__main__": #pass