TrigConfFrontier.py

Go to the documentation of this file.

#!/usr/bin/env python
# Copyright (C) 2002-2024 CERN for the benefit of the ATLAS collaboration
 
from AthenaCommon.Logging import logging
import time
import sys
 
log = logging.getLogger( "TrigConfFrontier.py" )
 
def getServerUrls(frontier_servers):
    """
    turns
    '(serverurl=http://atlasfrontier-local.cern.ch:8000/atlr)(serverurl=http://atlasfrontier-ai.cern.ch:8000/atlr)'
    into
    ['http://atlasfrontier-local.cern.ch:8000/atlr','http://atlasfrontier-ai.cern.ch:8000/atlr']
    """
    from re import findall
    return findall(r'\‍(serverurl=(.*?)\‍)',frontier_servers)
 
 
def resolveUrl(url):
    """
    Expects input string to be a URL or $FRONTIER_SERVER
    Returns an accessible URL or None"""
    import re
    if re.match("http://",url): # simple URL specification http://...
        return [url]
 
    if re.match(r'\‍(serverurl=(.*?)\‍)',url): # syntax of FRONTIER_SERVER
        return getServerUrls(url)
 
 
def getFrontierCursor(urls, schema, loglevel = logging.INFO):
    log.setLevel(loglevel)
    url_list = resolveUrl(urls)
    if len(url_list) == 0:
        log.warning("Cannot find a valid frontier connection, will not return a Frontier cursor")
        return None
    else:
        log.info(f"Will use Frontier server at {urls}")
 
    return FrontierCursor( urls = url_list, schema = schema)
        
 
# used by FrontierCursor
def replacebindvars(query, bindvars):
    """Replaces the bound variables with the specified values,
    disables variable binding
    """
    for var,val in list(bindvars.items()):
        if query.find(":%s" % var)<0:
            raise NameError("variable '%s' is not a bound variable in this query: %s" % (var, query) )
        if isinstance (val, int):
            query = query.replace(":%s" % var,"%s" % val)
        else:
            query = query.replace(":%s" % var,"%r" % val)
        log.debug("Resolving bound variable '%s' with %r", var,val)
    log.debug("Resolved query: %s", query)
    return query
 
 
class FrontierCursor(object):
    def __init__(self, urls, schema, refreshFlag=False, doDecode=True, retrieveZiplevel="zip", encoding="utf-8"):
        self.urls = [str(x) + "/Frontier" for x in urls]#Add /Frontier to each URL
        self.schema = schema
        self.refreshFlag = refreshFlag
        self.retrieveZiplevel = retrieveZiplevel
        self.doDecode = doDecode
        self.encoding = encoding
 
    def __str__(self):
        s = "Using Frontier URL: %s\n" % self.urls
        s += "Schema: %s\n" % self.schema
        s += "Refresh cache:  %s" % self.refreshFlag
        return s
 
    def execute(self, query, bindvars={}):
        if len(bindvars)>0:
            query = replacebindvars(query,bindvars)
        
        
        log.debug("Frontier URLs  : %s", self.urls)
        log.debug("Refresh cache : %s", self.refreshFlag)
        log.debug("Query         : %s", query)
        
        import base64, zlib, urllib.request, urllib.error, urllib.parse
 
        self.result = None
 
        for url in self.urls:
            try: 
                compQuery = zlib.compress(query.encode("utf-8"),9)
                base64Query = base64.binascii.b2a_base64(compQuery).decode("utf-8")
                encQuery = base64Query.replace("+", ".").replace("\n","").replace("/","-").replace("=","_")
                log.debug("Frontier Request  : %s", encQuery)
                frontierRequest="%s/type=frontier_request:1:DEFAULT&encoding=BLOB%s&p1=%s" % (url, self.retrieveZiplevel, encQuery)
                request = urllib.request.Request(frontierRequest)
                if self.refreshFlag:
                    request.add_header("pragma", "no-cache")
 
                frontierId = "TrigConfFrontier 1.0"
                request.add_header("X-Frontier-Id", frontierId)
 
                queryStart = time.localtime()
                log.debug("Query started: %s", time.strftime("%m/%d/%y %H:%M:%S %Z", queryStart))
 
                t1 = time.time()
                result = urllib.request.urlopen(request,timeout=10).read().decode('utf-8')
                t2 = time.time()
 
                queryEnd = time.localtime()
                log.debug("Query ended: %s", time.strftime("%m/%d/%y %H:%M:%S %Z", queryEnd))
                log.debug("Query time: %s [seconds]", (t2-t1))
                log.debug("Result size: %i [seconds]", len(result))
                self.result = result
                self.checkResultForErrors()
                return
            except urllib.error.HTTPError:
                log.warning("Problem with Frontier connection to %s trying next server", url)
            except Exception as err:
                log.warning("Problem with the request {0}".format(err))
 
        raise Exception("All servers failed")
                
 
    def fetchall(self):
        if self.doDecode: self.decodeResult()
        return self.result
 
    def checkResultForErrors(self):
        ''' Parse the response, looking for errors '''
        from xml.dom.minidom import parseString
        dom = parseString(self.result)
 
        globalError = dom.getElementsByTagName("global_error")
        for node in globalError:
            raise Exception(node.getAttribute("msg"))
 
        qualityList = dom.getElementsByTagName("quality")
        for node in qualityList:
            if int(node.getAttribute("error")) > 0:
                raise Exception(node.getAttribute("message")) 
 
    def decodeResult(self):
        from xml.dom.minidom import parseString
        import base64, zlib, curses.ascii, re
        #print ("Query result:\n", self.result)
        dom = parseString(self.result)
        dataList = dom.getElementsByTagName("data")
        keepalives = 0
        result = []
        # Control characters represent records, but I won't bother with that now,
        # and will simply replace those by space.
        for data in dataList:
            for node in data.childNodes:
                # <keepalive /> elements may be present, combined with whitespace text
                if node.nodeName == "keepalive":
                    # this is of type Element
                    keepalives += 1
                    continue
                # else assume of type Text
                if node.data.strip() == "":
                    continue
                if keepalives > 0:
                    print (keepalives, "keepalives received\n")
                    keepalives = 0
            
                row = base64.decodebytes(node.data.encode())
                if self.retrieveZiplevel != "":
                    row = zlib.decompress(row).decode(self.encoding)
                
                #Hack to get these lines to work in python 2
                if sys.version_info[0] < 3: 
                    row = row.encode('ascii', 'xmlcharrefreplace')
 
                endFirstRow = row.find('\x07')
                firstRow = row[:endFirstRow]
                for c in firstRow:
                    if curses.ascii.isctrl(c):
                        firstRow = firstRow.replace(c, ' ')
                fields = [x for i,x in enumerate(firstRow.split()) if i%2==0]
                types = [x for i,x in enumerate(firstRow.split()) if i%2==1]
                ptypes = []
                for t in types:
                    if t.startswith("NUMBER"):
                        if ",0" in t:
                            ptypes.append(int)
                        else:
                            ptypes.append(float)
                    else:
                        ptypes.append(str)
 
 
                log.debug("Fields      : %r", fields)
                log.debug("DB Types    : %r", types)
                log.debug("Python Types: %r", ptypes)
                
                row = row[endFirstRow+1:]
 
                row_h = row.rstrip('\x07')
 
                if 'BLOB' in types:
                    row_h = re.sub("^.*?{","{",row_h)
                
                row_h = row_h.replace("\x07\x06",'.nNn.\x06')
 
                #                pattern = re.compile("\x06\x00\x00\x00.",flags=re.S)
                #replace pattern above  more restrictive version, as longerstrings in the results
                #have a size variable in the column separate that becomes visible if the string
                #is large enough - this then broke the prevous  decoding
                pattern = re.compile("\x06\x00\x00..",flags=re.S)
                row_h = pattern.sub('.xXx.',row_h)
                row_h = row_h.replace("\x86", '.xXx.')
 
                row_h = row_h.split('.nNn.')
                row_h = [r.split('.xXx.') for r in row_h]
 
                result = []
                for r in row_h:
                    if r[0]=='': r[0:1]=[]
                    r = tuple([t(v) for t,v in zip(ptypes,r)])
                    result.append( r )
 
        self.result = result
 
 
 
def testQuery(query, bindvars):
    from TrigConfigSvc.TrigConfigSvcUtils import interpretConnection
    connectionParameters = interpretConnection("TRIGGERDBMC")
    cursor = getFrontierCursor( urls = connectionParameters['url'], schema = connectionParameters['schema'])
    cursor.execute(query, bindvars)
    log.info("Raw response:")
    log.info(cursor.result)
    cursor.decodeResult()
    log.info("Decoded response:")
    log.info(cursor.result[0][0])        
    if cursor.result[0][0] != 'MC_pp_v7':
        return 1
    return 0
 
    
if __name__=="__main__":
    log = logging.getLogger( "TrigConfFrontier.py" )
    log.setLevel(logging.DEBUG)
 
    dbalias = "TRIGGERDBMC"
    query = "select distinct HPS.HPS_NAME from ATLAS_CONF_TRIGGER_RUN2_MC.HLT_PRESCALE_SET HPS where HPS.HPS_ID = :psk"
    bindvars = { "psk": 260 }
 
    res = testQuery(query, bindvars)  # pure python frontier query
    sys.exit(res)