1: #!/usr/local/bin/python
2: #
3:
4: import string
5: import logging
6: import sys
7: import types
8: import time
9:
10: from xml import sax
11: from amara import saxtools
12:
13: try:
14: import psycopg2 as psycopg
15: import psycopg2.extensions
16: # switch to unicode
17: psycopg2.extensions.register_type(psycopg2.extensions.UNICODE)
18: psyco = 2
19: except:
20: import psycopg
21: psyco = 1
22:
23: fm_ns = 'http://www.filemaker.com/fmpxmlresult'
24:
25: version_string = "V0.6 ROC 13.2.2008"
26:
27: def unicodify(text, withNone=False):
28: """decode str (utf-8 or latin-1 representation) into unicode object"""
29: if withNone and text is None:
30: return None
31: if not text:
32: return u""
33: if isinstance(text, str):
34: try:
35: return text.decode('utf-8')
36: except:
37: return text.decode('latin-1')
38: else:
39: return text
40:
41: def utf8ify(text, withNone=False):
42: """encode unicode object or string into byte string in utf-8 representation"""
43: if withNone and text is None:
44: return None
45: if not text:
46: return ""
47: if isinstance(text, unicode):
48: return text.encode('utf-8')
49: else:
50: return text
51:
52: def getTextFromNode(nodename):
53: """get the cdata content of a node"""
54: if nodename is None:
55: return ""
56: nodelist=nodename.childNodes
57: rc = ""
58: for node in nodelist:
59: if node.nodeType == node.TEXT_NODE:
60: rc = rc + node.data
61: return rc
62:
63: def sql_quote(v):
64: # quote dictionary
65: quote_dict = {"\'": "''", "\\": "\\\\"}
66: for dkey in quote_dict.keys():
67: if string.find(v, dkey) >= 0:
68: v=string.join(string.split(v,dkey),quote_dict[dkey])
69: return "'%s'"%v
70:
71: def SimpleSearch(curs,query, args=None, ascii=False):
72: """execute sql query and return data"""
73: #logger.debug("executing: "+query)
74: if ascii:
75: # encode all in UTF-8
76: query = utf8ify(query)
77: if args is not None:
78: encargs = []
79: for a in args:
80: encargs.append(utf8ify(a, withNone=True))
81:
82: args = encargs
83:
84: curs.execute(query, args)
85: #logger.debug("sql done")
86: try:
87: return curs.fetchall()
88: except:
89: return None
90:
91:
92: class TableColumn:
93: """simple type for storing sql column name and type"""
94:
95: def __init__(self, name, type=None):
96: #print "new tablecolumn(%s,%s)"%(name, type)
97: self.name = name
98: self.type = type
99:
100: def getName(self):
101: return self.name
102:
103: def getType(self):
104: if self.type is not None:
105: return self.type
106: else:
107: return "text"
108:
109: def __str__(self):
110: return self.name
111:
112:
113: class xml_handler:
114: def __init__(self,options):
115: """SAX handler to import FileMaker XML file (FMPXMLRESULT format) into the table.
116: @param options: dict of options
117: @param options.dsn: database connection string
118: @param options.table: name of the table the xml shall be imported into
119: @param options.filename: xmlfile filename
120: @param options.update_fields: (optional) list of fields to update; default is to create all fields
121: @param options.id_field: (optional) field which uniquely identifies an entry for updating purposes.
122: @param options.sync_mode: (optional) really synchronise, i.e. delete entries not in XML file
123: @param options.lc_names: (optional) lower case and clean up field names from XML
124: @param options.keep_fields: (optional) don't add fields to SQL database
125: @param options.ascii_db: (optional) assume ascii encoding in db
126: @param options.replace_table: (optional) delete and re-insert data
127: @param options.backup_table: (optional) create backup of old table (breaks indices)
128: @param options.use_logger_instance: (optional) use this instance of a logger
129: """
130:
131: # set up logger
132: if hasattr(options, 'use_logger_instance'):
133: self.logger = options.use_logger_instance
134: else:
135: self.logger = logging.getLogger('db.import.fmpxml')
136:
137:
138: # set up parser
139: self.event = None
140: self.top_dispatcher = {
141: (saxtools.START_ELEMENT, fm_ns, u'METADATA'):
142: self.handle_meta_fields,
143: (saxtools.START_ELEMENT, fm_ns, u'RESULTSET'):
144: self.handle_data_fields,
145: }
146:
147: # connect database
148: self.dbCon = psycopg.connect(options.dsn)
149: logging.debug("DB encoding: %s"%self.dbCon.encoding)
150: self.db = self.dbCon.cursor()
151: assert self.db, "AIIEE no db cursor for %s!!"%options.dsn
152:
153: self.table = getattr(options,"table",None)
154: self.update_fields = getattr(options,"update_fields",None)
155: self.id_field = getattr(options,"id_field",None)
156: self.sync_mode = getattr(options,"sync_mode",None)
157: self.lc_names = getattr(options,"lc_names",None)
158: self.keep_fields = getattr(options,"keep_fields",None)
159: self.ascii_db = getattr(options,"ascii_db",None)
160: self.replace_table = getattr(options,"replace_table",None)
161: self.backup_table = getattr(options,"backup_table",None)
162: self.read_before_update = getattr(options,"read_before_update",None)
163:
164: self.logger.debug("dsn: "+repr(getattr(options,"dsn",None)))
165: self.logger.debug("table: "+repr(self.table))
166: self.logger.debug("update_fields: "+repr(self.update_fields))
167: self.logger.debug("id_field: "+repr(self.id_field))
168: self.logger.debug("sync_mode: "+repr(self.sync_mode))
169: self.logger.debug("lc_names: "+repr(self.lc_names))
170: self.logger.debug("keep_fields: "+repr(self.keep_fields))
171: self.logger.debug("ascii_db: "+repr(self.ascii_db))
172: self.logger.debug("replace_table: "+repr(self.replace_table))
173: self.logger.debug("backup_table: "+repr(self.backup_table))
174: self.logger.debug("read_before_update: "+repr(self.read_before_update))
175:
176: self.dbIDs = {}
177: self.rowcnt = 0
178:
179: if self.id_field is not None:
180: # prepare a list of ids for sync mode
181: qstr="select %s from %s"%(self.id_field,self.table)
182: for id in SimpleSearch(self.db, qstr):
183: # value 0: not updated
184: self.dbIDs[id[0]] = 0;
185: self.rowcnt += 1
186:
187: self.logger.info("%d entries in DB to sync"%self.rowcnt)
188:
189: # names of fields in XML file
190: self.xml_field_names = []
191: # map XML field names to SQL field names
192: self.xml_field_map = {}
193: # and vice versa
194: self.sql_field_map = {}
195:
196: return
197:
198: def handle_meta_fields(self, end_condition):
199: dispatcher = {
200: (saxtools.START_ELEMENT, fm_ns, u'FIELD'):
201: self.handle_meta_field,
202: }
203: #First round through the generator corresponds to the
204: #start element event
205: self.logger.info("reading metadata...")
206: self.logger.debug("START METADATA")
207: yield None
208:
209: #delegate is a generator that handles all the events "within"
210: #this element
211: delegate = None
212: while not self.event == end_condition:
213: delegate = saxtools.tenorsax.event_loop_body(
214: dispatcher, delegate, self.event)
215: yield None
216:
217: #Element closed. Wrap up
218: self.logger.debug("END METADATA")
219:
220: # rename table for backup
221: if self.backup_table:
222: self.orig_table = self.table
223: self.tmp_table = self.table + "_tmp"
224: backup_name = "%s_%s"%(self.table,time.strftime('%Y_%m_%d_%H_%M_%S'))
225:
226: # remove old temp table
227: qstr = "DROP TABLE %s"%(self.tmp_table)
228: try:
229: self.db.execute(qstr)
230: except:
231: pass
232:
233: self.dbCon.commit()
234:
235: if self.id_field:
236: # sync mode -- copy backup table, update current table
237: self.logger.info("copy table %s to %s"%(self.table,backup_name))
238: qstr = "CREATE TABLE %s AS (SELECT * FROM %s)"%(backup_name,self.table)
239:
240: else:
241: # replace mode -- create empty tmp table, insert into tmp table
242: self.table = self.tmp_table
243: self.logger.info("create empty table %s"%(self.table))
244: qstr = "CREATE TABLE %s AS (SELECT * FROM %s WHERE 1=0)"%(self.table,self.orig_table)
245:
246: self.db.execute(qstr)
247: self.dbCon.commit()
248:
249: # delete data from table for replace
250: if self.replace_table:
251: self.logger.info("delete data from table %s"%(self.table))
252: qstr = "TRUNCATE TABLE %s"%(self.table)
253: self.db.execute(qstr)
254: self.dbCon.commit()
255:
256: # try to match date style with XML
257: self.db.execute("set datestyle to 'german'")
258:
259: #self.logger.debug("xml-fieldnames:"+repr(self.xml_field_names))
260: # get list of fields and types of db table
261: qstr="select attname, format_type(pg_attribute.atttypid, pg_attribute.atttypmod) from pg_attribute, pg_class where attrelid = pg_class.oid and pg_attribute.attnum > 0 and relname = '%s'"
262: self.sql_fields={}
263: for f in SimpleSearch(self.db, qstr%self.table):
264: n = f[0]
265: t = f[1]
266: #print "SQL fields: %s (%s)"%(n,t)
267: self.sql_fields[n] = TableColumn(n,t)
268:
269: # translate id_field (SQL-name) to XML-name
270: self.xml_id = self.sql_field_map.get(self.id_field, None)
271: # get type of id_field
272: if self.id_field:
273: self.id_type = self.sql_fields[self.id_field].getType()
274: else:
275: self.id_type = None
276:
277: # check fields to update
278: if self.update_fields is None:
279: if self.keep_fields:
280: # update all existing fields from sql (when they are in the xml file)
281: self.update_fields = {}
282: for f in self.sql_fields.keys():
283: if self.sql_field_map.has_key(f):
284: xf = self.sql_field_map[f]
285: self.update_fields[f] = self.xml_field_map[xf]
286:
287: else:
288: # update all fields
289: if self.lc_names:
290: # create dict with sql names
291: self.update_fields = {}
292: for f in self.xml_field_map.values():
293: self.update_fields[f.getName()] = f
294:
295: else:
296: self.update_fields = self.xml_field_map
297:
298: # and translate to list of xml fields
299: if self.lc_names:
300: self.xml_update_list = [self.sql_field_map[x] for x in self.update_fields]
301: else:
302: self.xml_update_list = self.update_fields.keys()
303:
304: if not self.keep_fields:
305: # adjust db table to fields in XML and update_fields
306: for f in self.xml_field_map.values():
307: self.logger.debug("sync-fieldname: %s"%f.getName())
308: sf = self.sql_fields.get(f.getName(), None)
309: uf = self.update_fields.get(f.getName(), None)
310: if sf is not None:
311: # name in db -- check type
312: if f.getType() != sf.getType():
313: self.logger.debug("field %s has different type (%s vs %s)"%(f,f.getType(),sf.getType()))
314: elif uf is not None:
315: # add field to table
316: qstr="alter table %s add %s %s"%(self.table,uf.getName(),uf.getType())
317: self.logger.info("db add field:"+qstr)
318:
319: if self.ascii_db and type(qstr)==types.UnicodeType:
320: qstr=qstr.encode('utf-8')
321:
322: self.db.execute(qstr)
323: self.dbCon.commit()
324:
325: # prepare sql statements for update (do not update id_field)
326: setStr=string.join(["%s = %%s"%self.xml_field_map[f] for f in self.xml_update_list if f != self.xml_id], ', ')
327: self.updQuery="UPDATE %s SET %s WHERE %s = %%s"%(self.table,setStr,self.id_field)
328: # and select (for update check)
329: selStr=string.join([self.xml_field_map[f].getName() for f in self.xml_update_list if f != self.xml_id], ', ')
330: self.selQuery="SELECT %s FROM %s WHERE %s = %%s"%(selStr,self.table,self.id_field)
331: # and insert
332: fields=string.join([self.xml_field_map[x].getName() for x in self.xml_update_list], ',')
333: values=string.join(['%s' for f in self.xml_update_list], ',')
334: self.addQuery="INSERT INTO %s (%s) VALUES (%s)"%(self.table,fields,values)
335: self.logger.debug("update-query: "+self.updQuery)
336: self.logger.debug("sel-query: "+self.selQuery)
337: self.logger.debug("add-query: "+self.addQuery)
338: return
339:
340: def handle_meta_field(self, end_condition):
341: name = self.params.get((None, u'NAME'))
342: yield None
343: #Element closed. Wrap up
344: if self.lc_names:
345: # clean name
346: sqlname = name.replace(" ","_").lower()
347: else:
348: sqlname = name
349: self.xml_field_names.append(name)
350: # map to sql name and default text type
351: self.xml_field_map[name] = TableColumn(sqlname, 'text')
352: self.sql_field_map[sqlname] = name
353: self.logger.debug("FIELD name: "+name)
354: return
355:
356: def handle_data_fields(self, end_condition):
357: dispatcher = {
358: (saxtools.START_ELEMENT, fm_ns, u'ROW'):
359: self.handle_row,
360: }
361: #First round through the generator corresponds to the
362: #start element event
363: self.logger.info("reading data...")
364: self.logger.debug("START RESULTSET")
365: self.rowcnt = 0
366: yield None
367:
368: #delegate is a generator that handles all the events "within"
369: #this element
370: delegate = None
371: while not self.event == end_condition:
372: delegate = saxtools.tenorsax.event_loop_body(
373: dispatcher, delegate, self.event)
374: yield None
375:
376: #Element closed. Wrap up
377: self.logger.debug("END RESULTSET")
378: self.dbCon.commit()
379:
380: if self.sync_mode:
381: # delete unmatched entries in db
382: self.logger.info("deleting unmatched rows from db")
383: delQuery = "DELETE FROM %s WHERE %s = %%s"%(self.table,self.id_field)
384: for id in self.dbIDs.keys():
385: # find all not-updated fields
386: if self.dbIDs[id] == 0:
387: self.logger.info(" delete:"+id)
388: SimpleSearch(self.db, delQuery, [id], ascii=self.ascii_db)
389:
390: elif self.dbIDs[id] > 1:
391: self.logger.info(" sync: ID %s used more than once?"%id)
392:
393: self.dbCon.commit()
394:
395: # reinstate backup tables
396: if self.backup_table and not self.id_field:
397: backup_name = "%s_%s"%(self.orig_table,time.strftime('%Y_%m_%d_%H_%M_%S'))
398: self.logger.info("rename backup table %s to %s"%(self.orig_table,backup_name))
399: qstr = "ALTER TABLE %s RENAME TO %s"%(self.orig_table,backup_name)
400: self.db.execute(qstr)
401: self.logger.info("rename working table %s to %s"%(self.table,self.orig_table))
402: qstr = "ALTER TABLE %s RENAME TO %s"%(self.table,self.orig_table)
403: self.db.execute(qstr)
404: self.dbCon.commit()
405:
406: return
407:
408: def handle_row(self, end_condition):
409: dispatcher = {
410: (saxtools.START_ELEMENT, fm_ns, u'COL'):
411: self.handle_col,
412: }
413: self.logger.debug("START ROW")
414: self.xml_data = {}
415: self.colIdx = 0
416: yield None
417:
418: #delegate is a generator that handles all the events "within"
419: #this element
420: delegate = None
421: while not self.event == end_condition:
422: delegate = saxtools.tenorsax.event_loop_body(
423: dispatcher, delegate, self.event)
424: yield None
425:
426: #Element closed. Wrap up
427: self.logger.debug("END ROW")
428: self.rowcnt += 1
429: # process collected row data
430: update=False
431: id_val=''
432: # synchronize by id_field
433: if self.id_field:
434: if self.id_type == 'integer':
435: id_val = int(self.xml_data[self.xml_id])
436: else:
437: id_val = self.xml_data[self.xml_id]
438:
439: if id_val in self.dbIDs:
440: self.dbIDs[id_val] += 1
441: update=True
442:
443: # collect all values
444: args = []
445: for fn in self.xml_update_list:
446: # do not update id_field
447: if update and fn == self.xml_id:
448: continue
449:
450: f = self.xml_field_map[fn]
451: val = self.xml_data[fn]
452: type = self.sql_fields[f.getName()].getType()
453: if type == "date" and len(val) == 0:
454: # empty date field
455: val = None
456:
457: elif type == "integer" and len(val) == 0:
458: # empty int field
459: val = None
460:
461: args.append(val)
462:
463: if update:
464: # update existing row (by id_field)
465: if self.read_before_update:
466: # read data
467: self.logger.debug("update check: %s = %s"%(id_val, args))
468: oldrow = SimpleSearch(self.db, self.selQuery, [id_val], ascii=self.ascii_db)
469: #i = 0
470: #for v in oldrow[0]:
471: # logging.debug("v: %s = %s (%s)"%(v,args[i],v==args[i]))
472: # i += 1
473: if tuple(oldrow[0]) != tuple(args):
474: # data has changed -- update
475: self.logger.debug("really update: %s = %s"%(id_val, args))
476: args.append(id_val) # last arg is id
477: SimpleSearch(self.db, self.updQuery, args, ascii=self.ascii_db)
478:
479: else:
480: # always update
481: self.logger.debug("update: %s = %s"%(id_val, args))
482: args.append(id_val) # last arg is id
483: SimpleSearch(self.db, self.updQuery, args, ascii=self.ascii_db)
484:
485: else:
486: # create new row
487: self.logger.debug("insert: %s"%args)
488: SimpleSearch(self.db, self.addQuery, args, ascii=self.ascii_db)
489:
490: #self.logger.info(" row:"+"%d (%s)"%(self.rowcnt,id_val))
491: if (self.rowcnt % 100) == 0:
492: self.logger.info(" row:"+"%d (id:%s)"%(self.rowcnt,id_val))
493: self.dbCon.commit()
494:
495: return
496:
497: def handle_col(self, end_condition):
498: dispatcher = {
499: (saxtools.START_ELEMENT, fm_ns, u'DATA'):
500: self.handle_data_tag,
501: }
502: #print "START COL"
503: yield None
504: #delegate is a generator that handles all the events "within"
505: #this element
506: delegate = None
507: while not self.event == end_condition:
508: delegate = saxtools.tenorsax.event_loop_body(
509: dispatcher, delegate, self.event)
510: yield None
511: #Element closed. Wrap up
512: #print "END COL"
513: self.colIdx += 1
514: return
515:
516: def handle_data_tag(self, end_condition):
517: #print "START DATA"
518: content = u''
519: yield None
520: # gather child elements
521: while not self.event == end_condition:
522: if self.event[0] == saxtools.CHARACTER_DATA:
523: content += self.params
524: yield None
525: #Element closed. Wrap up
526: fn = self.xml_field_names[self.colIdx]
527: self.xml_data[fn] = content
528: return
529:
530:
531: def importFMPXML(options):
532: """import FileMaker XML file (FMPXMLRESULT format) into the table.
533: @param options: dict of options
534: @param options.dsn: database connection string
535: @param options.table: name of the table the xml shall be imported into
536: @param options.filename: xmlfile filename
537: @param options.update_fields: (optional) list of fields to update; default is to create all fields
538: @param options.id_field: (optional) field which uniquely identifies an entry for updating purposes.
539: @param options.sync_mode: (optional) really synchronise, i.e. delete entries not in XML file
540: @param options.lc_names: (optional) lower case and clean up field names from XML
541: @param options.keep_fields: (optional) don't add fields to SQL database
542: @param options.ascii_db: (optional) assume ascii encoding in db
543: @param options.replace_table: (optional) delete and re-insert data
544: @param options.backup_table: (optional) create backup of old table
545: """
546:
547: if getattr(options,'update_fields',None):
548: uf = {}
549: for f in options.update_fields.split(','):
550: if f.find(':') > 0:
551: (n,t) = f.split(':')
552: else:
553: n = f
554: t = None
555: uf[n] = TableColumn(n,t)
556:
557: options.update_fields = uf
558:
559: if getattr(options,'id_field',None) and getattr(options,'replace_table',None):
560: logging.error("ABORT: sorry, you can't do both sync (id_field) and replace")
561: return
562:
563: parser = sax.make_parser()
564: #The "consumer" is our own handler
565: consumer = xml_handler(options)
566: #Initialize Tenorsax with handler
567: handler = saxtools.tenorsax(consumer)
568: #Resulting tenorsax instance is the SAX handler
569: parser.setContentHandler(handler)
570: parser.setFeature(sax.handler.feature_namespaces, 1)
571: parser.parse(options.filename)
572:
573:
574: if __name__ == "__main__":
575: from optparse import OptionParser
576:
577: opars = OptionParser()
578: opars.add_option("-f", "--file",
579: dest="filename",
580: help="FMPXML file name", metavar="FILE")
581: opars.add_option("-c", "--dsn",
582: dest="dsn",
583: help="database connection string")
584: opars.add_option("-t", "--table",
585: dest="table",
586: help="database table name")
587: opars.add_option("--fields", default=None,
588: dest="update_fields",
589: help="list of fields to update (comma separated, sql-names)", metavar="LIST")
590: opars.add_option("--id-field", default=None,
591: dest="id_field",
592: help="name of id field for synchronisation (only appends data otherwise, sql-name)", metavar="NAME")
593: opars.add_option("--sync", "--sync-mode", default=False, action="store_true",
594: dest="sync_mode",
595: help="do full sync based on id field (remove unmatched fields from db)")
596: opars.add_option("--lc-names", default=False, action="store_true",
597: dest="lc_names",
598: help="clean and lower case field names from XML")
599: opars.add_option("--keep-fields", default=False, action="store_true",
600: dest="keep_fields",
601: help="don't add fields from XML to SQL table")
602: opars.add_option("--ascii-db", default=False, action="store_true",
603: dest="ascii_db",
604: help="the SQL database stores ASCII instead of unicode")
605: opars.add_option("--replace", default=False, action="store_true",
606: dest="replace_table",
607: help="replace table i.e. delete and re-insert data")
608: opars.add_option("--backup", default=False, action="store_true",
609: dest="backup_table",
610: help="create backup of old table")
611: opars.add_option("--read-before-update", default=False, action="store_true",
612: dest="read_before_update",
613: help="read all data to check if it really changed")
614: opars.add_option("-d", "--debug", default=False, action="store_true",
615: dest="debug",
616: help="debug mode (more output)")
617:
618: (options, args) = opars.parse_args()
619:
620: if len(sys.argv) < 2 or options.filename is None or options.dsn is None:
621: print "importFMPXML "+version_string
622: opars.print_help()
623: sys.exit(1)
624:
625: if options.debug:
626: loglevel = logging.DEBUG
627: else:
628: loglevel = logging.INFO
629:
630: logging.basicConfig(level=loglevel,
631: format='%(asctime)s %(levelname)s %(message)s',
632: datefmt='%H:%M:%S')
633:
634: importFMPXML(options)
635:
636:
637:
638:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvs/cvsweb/icons/back.gif){kind=icon}
Return to importFMPXML.py CVS log ![[TXT]](/cvs/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvs/cvsweb/icons/dir.gif){kind=icon}
Up to [Repository] / ZSQLExtend