Annotation of ZSQLExtend/importFMPXML.py, revision 1.9
1.1 casties 1: #!/usr/local/bin/python
2: #
3:
4: import string
5: import logging
6: import sys
1.7 dwinter 7: import types
1.8 casties 8: import time
1.1 casties 9:
1.5 casties 10: from xml import sax
11: from amara import saxtools
12:
1.2 casties 13: try:
14: import psycopg2 as psycopg
15: psyco = 2
16: except:
17: import psycopg
18: psyco = 1
19:
1.5 casties 20: fm_ns = 'http://www.filemaker.com/fmpxmlresult'
1.1 casties 21:
1.8 casties 22: version_string = "V0.4 ROC 29.3.2007"
23:
1.1 casties 24: def getTextFromNode(nodename):
25: """get the cdata content of a node"""
26: if nodename is None:
27: return ""
28: nodelist=nodename.childNodes
29: rc = ""
30: for node in nodelist:
31: if node.nodeType == node.TEXT_NODE:
32: rc = rc + node.data
33: return rc
34:
35: def sql_quote(v):
36: # quote dictionary
37: quote_dict = {"\'": "''", "\\": "\\\\"}
38: for dkey in quote_dict.keys():
39: if string.find(v, dkey) >= 0:
40: v=string.join(string.split(v,dkey),quote_dict[dkey])
1.5 casties 41: return "'%s'"%v
1.1 casties 42:
1.8 casties 43: def SimpleSearch(curs,query, args=None, ascii=False):
1.1 casties 44: """execute sql query and return data"""
1.8 casties 45: #logging.debug("executing: "+query)
46: if ascii:
47: # encode all in UTF-8
1.2 casties 48: query = query.encode("UTF-8")
1.8 casties 49: if args is not None:
50: encargs = []
51: for a in args:
52: if a is not None:
53: a = a.encode("UTF-8")
54: encargs.append(a)
55:
56: args = encargs
1.7 dwinter 57:
1.1 casties 58: curs.execute(query, args)
1.8 casties 59: #logging.debug("sql done")
1.4 casties 60: try:
61: return curs.fetchall()
62: except:
63: return None
1.1 casties 64:
65:
1.8 casties 66: class TableColumn:
67: """simple type for storing sql column name and type"""
68:
69: def __init__(self, name, type=None):
70: #print "new tablecolumn(%s,%s)"%(name, type)
71: self.name = name
72: self.type = type
73:
74: def getName(self):
75: return self.name
76:
77: def getType(self):
78: if self.type is not None:
79: return self.type
80: else:
81: return "text"
1.5 casties 82:
1.8 casties 83: def __str__(self):
84: return self.name
85:
86:
1.5 casties 87: class xml_handler:
1.8 casties 88: def __init__(self,options):
1.9 ! dwinter 89: """SAX handler to import FileMaker XML file (FMPXMLRESULT format) into the table.
1.8 casties 90: @param options: dict of options
91: @param options.dsn: database connection string
92: @param options.table: name of the table the xml shall be imported into
93: @param options.filename: xmlfile filename
94: @param options.update_fields: (optional) list of fields to update; default is to create all fields
95: @param options.id_field: (optional) field which uniquely identifies an entry for updating purposes.
96: @param options.sync_mode: (optional) really synchronise, i.e. delete entries not in XML file
97: @param options.lc_names: (optional) lower case and clean up field names from XML
98: @param options.keep_fields: (optional) don't add fields to SQL database
99: @param options.ascii_db: (optional) assume ascii encoding in db
100: @param options.replace_table: (optional) delete and re-insert data
1.9 ! dwinter 101: """
! 102:
1.5 casties 103: # set up parser
104: self.event = None
105: self.top_dispatcher = {
106: (saxtools.START_ELEMENT, fm_ns, u'METADATA'):
107: self.handle_meta_fields,
108: (saxtools.START_ELEMENT, fm_ns, u'RESULTSET'):
1.8 casties 109: self.handle_data_fields,
1.5 casties 110: }
111:
112: # connect database
1.8 casties 113: self.dbCon = psycopg.connect(options.dsn)
1.5 casties 114: self.db = self.dbCon.cursor()
1.8 casties 115: assert self.db, "AIIEE no db cursor for %s!!"%options.dsn
1.1 casties 116:
1.9 ! dwinter 117: self.table = getattr(options,"table",None)
! 118: self.update_fields = getattr(options,"update_fields",None)
! 119: self.id_field = getattr(options,"id_field",None)
! 120: self.sync_mode = getattr(options,"sync_mode",None)
! 121: self.lc_names = getattr(options,"lc_names",None)
! 122: self.keep_fields = getattr(options,"keep_fields",None)
! 123: self.ascii_db = getattr(options,"ascii_db",None)
! 124: self.replace_table = getattr(options,"replace_table",None)
! 125: self.backup_table = getattr(options,"backup_table",None)
1.8 casties 126:
1.9 ! dwinter 127: logging.debug("dsn: "+repr(getattr(options,"dsn",None)))
1.8 casties 128: logging.debug("table: "+repr(self.table))
129: logging.debug("update_fields: "+repr(self.update_fields))
130: logging.debug("id_field: "+repr(self.id_field))
131: logging.debug("sync_mode: "+repr(self.sync_mode))
132: logging.debug("lc_names: "+repr(self.lc_names))
133: logging.debug("keep_fields: "+repr(self.keep_fields))
134: logging.debug("ascii_db: "+repr(self.ascii_db))
135: logging.debug("replace_table: "+repr(self.replace_table))
1.5 casties 136:
137: self.dbIDs = {}
138: self.rowcnt = 0
1.8 casties 139:
140: if self.id_field is not None:
1.5 casties 141: # prepare a list of ids for sync mode
1.8 casties 142: qstr="select %s from %s"%(self.id_field,self.table)
1.5 casties 143: for id in SimpleSearch(self.db, qstr):
144: # value 0: not updated
145: self.dbIDs[id[0]] = 0;
146: self.rowcnt += 1
147:
148: logging.info("%d entries in DB to sync"%self.rowcnt)
149:
1.8 casties 150: # names of fields in XML file
151: self.xml_field_names = []
152: # map XML field names to SQL field names
153: self.xml_field_map = {}
154: # and vice versa
155: self.sql_field_map = {}
1.5 casties 156:
157: return
158:
159: def handle_meta_fields(self, end_condition):
160: dispatcher = {
161: (saxtools.START_ELEMENT, fm_ns, u'FIELD'):
162: self.handle_meta_field,
163: }
164: #First round through the generator corresponds to the
165: #start element event
166: logging.debug("START METADATA")
167: yield None
1.1 casties 168:
1.5 casties 169: #delegate is a generator that handles all the events "within"
170: #this element
171: delegate = None
172: while not self.event == end_condition:
173: delegate = saxtools.tenorsax.event_loop_body(
174: dispatcher, delegate, self.event)
175: yield None
176:
177: #Element closed. Wrap up
178: logging.debug("END METADATA")
1.8 casties 179:
180: # rename table for backup
181: if self.backup_table:
182: self.orig_table = self.table
183: self.table = self.table + "_tmp"
184: # remove old temp table
185: qstr = "DROP TABLE %s"%(self.table)
186: try:
187: self.db.execute(qstr)
188: except:
189: pass
190:
191: self.dbCon.commit()
192:
193: if self.id_field:
194: # sync mode -- copy table
195: logging.info("copy table %s to %s"%(self.orig_table,self.table))
196: qstr = "CREATE TABLE %s AS (SELECT * FROM %s)"%(self.table,self.orig_table)
197:
198: else:
199: # rename table and create empty new one
200: logging.info("create empty table %s"%(self.table))
201: qstr = "CREATE TABLE %s AS (SELECT * FROM %s WHERE 1=0)"%(self.table,self.orig_table)
202:
203: self.db.execute(qstr)
204: self.dbCon.commit()
205:
206: # delete data from table for replace
207: if self.replace_table:
208: logging.info("delete data from table %s"%(self.table))
209: qstr = "TRUNCATE TABLE %s"%(self.table)
210: self.db.execute(qstr)
211: self.dbCon.commit()
212:
213: # try to match date style with XML
214: self.db.execute("set datestyle to 'german'")
215:
216: # translate id_field (SQL-name) to XML-name
217: self.xml_id = self.sql_field_map.get(self.id_field, None)
218:
219: #logging.debug("xml-fieldnames:"+repr(self.xml_field_names))
220: # get list of fields and types of db table
221: 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'"
222: self.sql_fields={}
223: for f in SimpleSearch(self.db, qstr%self.table):
224: n = f[0]
225: t = f[1]
226: #print "SQL fields: %s (%s)"%(n,t)
227: self.sql_fields[n] = TableColumn(n,t)
228:
229: # check fields to update
1.5 casties 230: if self.update_fields is None:
1.8 casties 231: if self.keep_fields:
232: # update existing fields
233: self.update_fields = self.sql_fields
234:
1.9 ! dwinter 235:
1.8 casties 236: else:
237: # update all fields
238: if self.lc_names:
239: # create dict with sql names
240: self.update_fields = {}
241: for f in self.xml_field_map.values():
242: self.update_fields[f.getName()] = f
243:
244: else:
245: self.update_fields = self.xml_field_map
246:
247: # and translate to list of xml fields
248: if self.lc_names:
249: self.xml_update_list = [self.sql_field_map[x] for x in self.update_fields]
250: else:
251: self.xml_update_list = self.update_fields.keys()
1.5 casties 252:
1.8 casties 253: if not self.keep_fields:
254: # adjust db table to fields in XML and update_fields
255: for f in self.xml_field_map.values():
256: logging.debug("sync-fieldname: %s"%f.getName())
257: sf = self.sql_fields.get(f.getName(), None)
258: uf = self.update_fields.get(f.getName(), None)
259: if sf is not None:
260: # name in db -- check type
261: if f.getType() != sf.getType():
262: logging.debug("field %s has different type (%s vs %s)"%(f,f.getType(),sf.getType()))
263: elif uf is not None:
264: # add field to table
265: qstr="alter table %s add %s %s"%(self.table,uf.getName(),uf.getType())
266: logging.info("db add field:"+qstr)
267:
268: if self.ascii_db and type(qstr)==types.UnicodeType:
269: qstr=qstr.encode('utf-8')
270:
271: self.db.execute(qstr)
272: self.dbCon.commit()
1.7 dwinter 273:
1.5 casties 274: # prepare sql statements for update
1.8 casties 275: setStr=string.join(["%s = %%s"%self.xml_field_map[f] for f in self.xml_update_list], ', ')
1.5 casties 276: self.updQuery="UPDATE %s SET %s WHERE %s = %%s"%(self.table,setStr,self.id_field)
277: # and insert
1.8 casties 278: fields=string.join([self.xml_field_map[x].getName() for x in self.xml_update_list], ',')
279: values=string.join(['%s' for f in self.xml_update_list], ',')
1.5 casties 280: self.addQuery="INSERT INTO %s (%s) VALUES (%s)"%(self.table,fields,values)
1.8 casties 281: logging.debug("update-query: "+self.updQuery)
282: logging.debug("add-query: "+self.addQuery)
1.5 casties 283: return
284:
285: def handle_meta_field(self, end_condition):
286: name = self.params.get((None, u'NAME'))
287: yield None
288: #Element closed. Wrap up
1.8 casties 289: if self.lc_names:
290: # clean name
291: sqlname = name.replace(" ","_").lower()
292: else:
293: sqlname = name
294: self.xml_field_names.append(name)
295: # map to sql name and default text type
296: self.xml_field_map[name] = TableColumn(sqlname, 'text')
297: self.sql_field_map[sqlname] = name
1.5 casties 298: logging.debug("FIELD name: "+name)
299: return
300:
1.8 casties 301: def handle_data_fields(self, end_condition):
1.5 casties 302: dispatcher = {
303: (saxtools.START_ELEMENT, fm_ns, u'ROW'):
304: self.handle_row,
305: }
306: #First round through the generator corresponds to the
307: #start element event
308: logging.debug("START RESULTSET")
309: self.rowcnt = 0
310: yield None
1.1 casties 311:
1.5 casties 312: #delegate is a generator that handles all the events "within"
313: #this element
314: delegate = None
315: while not self.event == end_condition:
316: delegate = saxtools.tenorsax.event_loop_body(
317: dispatcher, delegate, self.event)
318: yield None
319:
320: #Element closed. Wrap up
321: logging.debug("END RESULTSET")
322: self.dbCon.commit()
1.1 casties 323:
1.5 casties 324: if self.sync_mode:
325: # delete unmatched entries in db
1.8 casties 326: logging.info("deleting unmatched rows from db")
1.6 casties 327: delQuery = "DELETE FROM %s WHERE %s = %%s"%(self.table,self.id_field)
1.5 casties 328: for id in self.dbIDs.keys():
329: # find all not-updated fields
330: if self.dbIDs[id] == 0:
331: logging.info(" delete:"+id)
1.8 casties 332: SimpleSearch(self.db, delQuery, [id], ascii=self.ascii_db)
1.6 casties 333: sys.exit(1)
1.1 casties 334:
1.5 casties 335: elif self.dbIDs[id] > 1:
1.8 casties 336: logging.info(" sync: ID %s used more than once?"%id)
337:
338: self.dbCon.commit()
1.1 casties 339:
1.8 casties 340: # reinstate backup tables
341: if self.backup_table:
342: backup_name = "%s_%s"%(self.orig_table,time.strftime('%Y_%m_%d_%H_%M_%S'))
343: logging.info("rename backup table %s to %s"%(self.orig_table,backup_name))
344: qstr = "ALTER TABLE %s RENAME TO %s"%(self.orig_table,backup_name)
345: self.db.execute(qstr)
346: logging.info("rename working table %s to %s"%(self.table,self.orig_table))
347: qstr = "ALTER TABLE %s RENAME TO %s"%(self.table,self.orig_table)
348: self.db.execute(qstr)
1.5 casties 349: self.dbCon.commit()
1.1 casties 350:
1.5 casties 351: return
352:
353: def handle_row(self, end_condition):
354: dispatcher = {
355: (saxtools.START_ELEMENT, fm_ns, u'COL'):
356: self.handle_col,
357: }
358: logging.debug("START ROW")
1.8 casties 359: self.xml_data = {}
1.5 casties 360: self.colIdx = 0
361: yield None
1.1 casties 362:
1.5 casties 363: #delegate is a generator that handles all the events "within"
364: #this element
365: delegate = None
366: while not self.event == end_condition:
367: delegate = saxtools.tenorsax.event_loop_body(
368: dispatcher, delegate, self.event)
369: yield None
370:
371: #Element closed. Wrap up
372: logging.debug("END ROW")
373: self.rowcnt += 1
374: # process collected row data
375: update=False
376: id_val=''
377: # synchronize by id_field
378: if self.id_field:
1.8 casties 379: id_val = self.xml_data[self.xml_id]
1.5 casties 380: if id_val in self.dbIDs:
381: self.dbIDs[id_val] += 1
382: update=True
1.8 casties 383:
384: # collect all values
385: args = []
386: for fn in self.xml_update_list:
387: f = self.xml_field_map[fn]
388: val = self.xml_data[fn]
389: type = self.sql_fields[f.getName()].getType()
390: if type == "date" and len(val) == 0:
391: # empty date field
392: val = None
393:
394: elif type == "integer" and len(val) == 0:
395: # empty int field
396: val = None
397:
398: args.append(val)
399:
1.5 casties 400: if update:
401: # update existing row (by id_field)
1.8 casties 402: # last argument is ID match
1.5 casties 403: args.append(id_val)
1.8 casties 404: logging.debug("update: %s = %s"%(id_val, args))
405: SimpleSearch(self.db, self.updQuery, args, ascii=self.ascii_db)
406:
1.5 casties 407: else:
408: # create new row
1.8 casties 409: logging.debug("insert: %s"%args)
410: SimpleSearch(self.db, self.addQuery, args, ascii=self.ascii_db)
1.5 casties 411:
412: #logging.info(" row:"+"%d (%s)"%(self.rowcnt,id_val))
413: if (self.rowcnt % 10) == 0:
414: logging.info(" row:"+"%d (%s)"%(self.rowcnt,id_val))
415: self.dbCon.commit()
416:
417: return
418:
419: def handle_col(self, end_condition):
420: dispatcher = {
421: (saxtools.START_ELEMENT, fm_ns, u'DATA'):
422: self.handle_data_tag,
423: }
424: #print "START COL"
425: yield None
426: #delegate is a generator that handles all the events "within"
427: #this element
428: delegate = None
429: while not self.event == end_condition:
430: delegate = saxtools.tenorsax.event_loop_body(
431: dispatcher, delegate, self.event)
432: yield None
433: #Element closed. Wrap up
434: #print "END COL"
435: self.colIdx += 1
436: return
437:
438: def handle_data_tag(self, end_condition):
439: #print "START DATA"
440: content = u''
441: yield None
442: # gather child elements
443: while not self.event == end_condition:
444: if self.event[0] == saxtools.CHARACTER_DATA:
445: content += self.params
446: yield None
447: #Element closed. Wrap up
1.8 casties 448: fn = self.xml_field_names[self.colIdx]
449: self.xml_data[fn] = content
1.5 casties 450: return
451:
452:
453:
454:
1.1 casties 455:
1.9 ! dwinter 456: if __name__ == "__main__":
! 457: from optparse import OptionParser
1.5 casties 458:
1.9 ! dwinter 459: opars = OptionParser()
! 460: opars.add_option("-f", "--file",
! 461: dest="filename",
! 462: help="FMPXML file name", metavar="FILE")
! 463: opars.add_option("-c", "--dsn",
! 464: dest="dsn",
! 465: help="database connection string")
! 466: opars.add_option("-t", "--table",
! 467: dest="table",
! 468: help="database table name")
! 469: opars.add_option("--fields", default=None,
! 470: dest="update_fields",
! 471: help="list of fields to update (comma separated, sql-names)", metavar="LIST")
! 472: opars.add_option("--id-field", default=None,
! 473: dest="id_field",
! 474: help="name of id field for synchronisation (only appends data otherwise, sql-name)", metavar="NAME")
! 475: opars.add_option("--sync", "--sync-mode", default=False, action="store_true",
! 476: dest="sync_mode",
! 477: help="do full sync based on id field (remove unmatched fields from db)")
! 478: opars.add_option("--lc-names", default=False, action="store_true",
! 479: dest="lc_names",
! 480: help="clean and lower case field names from XML")
! 481: opars.add_option("--keep-fields", default=False, action="store_true",
! 482: dest="keep_fields",
! 483: help="don't add fields from XML to SQL table")
! 484: opars.add_option("--ascii-db", default=False, action="store_true",
! 485: dest="ascii_db",
! 486: help="the SQL database stores ASCII instead of unicode")
! 487: opars.add_option("--replace", default=False, action="store_true",
! 488: dest="replace_table",
! 489: help="replace table i.e. delete and re-insert data")
! 490: opars.add_option("--backup", default=False, action="store_true",
! 491: dest="backup_table",
! 492: help="create backup of old table (breaks indices)")
! 493: opars.add_option("-d", "--debug", default=False, action="store_true",
! 494: dest="debug",
! 495: help="debug mode (more output)")
! 496:
! 497: (options, args) = opars.parse_args()
! 498:
! 499: if len(sys.argv) < 2 or options.filename is None or options.dsn is None:
! 500: print "importFMPXML "+version_string
! 501: opars.print_help()
! 502: sys.exit(1)
! 503:
! 504: if options.debug:
! 505: loglevel = logging.DEBUG
! 506: else:
! 507: loglevel = logging.INFO
! 508:
! 509: logging.basicConfig(level=loglevel,
! 510: format='%(asctime)s %(levelname)s %(message)s',
! 511: datefmt='%H:%M:%S')
! 512:
! 513: importFMPXML(options)
! 514:
! 515: def importFMPXML(options):
! 516: """SAX handler to import FileMaker XML file (FMPXMLRESULT format) into the table.
! 517: @param options: dict of options
! 518: @param options.dsn: database connection string
! 519: @param options.table: name of the table the xml shall be imported into
! 520: @param options.filename: xmlfile filename
! 521: @param options.update_fields: (optional) list of fields to update; default is to create all fields
! 522: @param options.id_field: (optional) field which uniquely identifies an entry for updating purposes.
! 523: @param options.sync_mode: (optional) really synchronise, i.e. delete entries not in XML file
! 524: @param options.lc_names: (optional) lower case and clean up field names from XML
! 525: @param options.keep_fields: (optional) don't add fields to SQL database
! 526: @param options.ascii_db: (optional) assume ascii encoding in db
! 527: @param options.replace_table: (optional) delete and re-insert data
! 528: """
! 529: update_fields = None
! 530:
! 531: if getattr(options,'update_fields',None):
! 532: uf = {}
! 533: for f in options.update_fields.split(','):
! 534: (n,t) = f.split(':')
! 535: uf[n] = TableColumn(n,t)
! 536:
! 537: options.update_fields = uf
! 538:
! 539: if getattr(options,'id_field',None) and getattr(options,'replace_table',None):
! 540: logging.error("ABORT: sorry, you can't do both sync (id_field) and replace")
! 541: sys.exit(1)
! 542:
! 543: parser = sax.make_parser()
! 544: #The "consumer" is our own handler
! 545: consumer = xml_handler(options)
! 546: #Initialize Tenorsax with handler
! 547: handler = saxtools.tenorsax(consumer)
! 548: #Resulting tenorsax instance is the SAX handler
! 549: parser.setContentHandler(handler)
! 550: parser.setFeature(sax.handler.feature_namespaces, 1)
! 551: parser.parse(options.filename)
! 552:
! 553:
1.1 casties 554:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>