trunk/contrib/diff_formed.py

# -*- coding: utf-8 -*-
#!/usr/bin/python
"""\n
Diffs to given formedtrees.  Note that currently no checks on external choice
lists are done.  Furter checks on Repeatgroups are very simple for now.
Warning: If a field with the same name is moved from master_tbl into a RG, it will not be
noticed, as we do not diff strutural changes.

Usage: diff_formed.py old new 

Options:
    -h / --help         Print this message and exit.
    -e / --elements     comma-seperated list of elements to check
    --check-attributes  Check changes in attributes (changed descriptions, target...)

"""
# 20100810 Torsten Irlaender <[email protected]>
#
# This program is free software under the GNU GPL (>=v2)
# Read the file COPYING coming with the software for details.


import sys 
import getopt
import logging

from lxml import etree

logging.basicConfig(level=logging.ERROR)
log = logging.getLogger('diff_formed')

class AttributeChangeException(Exception):
    def __init__(self, n, c, d):
        self.new = n
        self.changed = c
        self.deleted = d

        msg = []
        if self.new:
            msg.append('New: %s' % ",".join(self.new))
        if self.changed:
            msg.append('Changed: %s' % ",".join(self.changed))
        if self.deleted:
            msg.append('Deleted: %s' % ",".join(self.deleted))
        self.message = " | ".join(msg)

class Inspector():

    def __init__(self, filename_old, filename_new):

        self.old_xml_doc = None
        self.new_xml_doc = None
        self.elements_old = {}
        self.elements_new = {}
        self.elements = ['choice', 'text', 'textarea', 'repeat']

        # Open files and load xml
        try:
            f1 = None
            f2 = None
            log.debug('Old: %s'% filename_old)
            f1 = open(filename_old, "r")
            log.debug('New: %s'% filename_new)
            f2 = open(filename_new, "r")
            self.old_xml_doc = etree.parse(f1)
            self.new_xml_doc = etree.parse(f2)
        except:
            log.exception('Error while opening file')
        finally:
            if f1 is not None:
                f1.close()
            if f2 is not None:
                f2.close()

    def perform(self, elements=None, check_attributes=False):
        log.info('Only checking %s elements' % ",".join(elements))
        for element in self.elements:
            if elements is not None:
                if element not in elements: continue 
            self.elements_old[element] = []
            self.elements_new[element] = []
            for e in self.old_xml_doc.findall("//%s" % element):
                self.elements_old[element].append(e)
            for e in self.new_xml_doc.findall("//%s" % element):
                self.elements_new[element].append(e)
            self.diff_element(element, check_attributes)

    def diff_element(self, element, check_attributes):
        log.debug('Checking %s elements' % element)
        out = []

        # Search for old field in new 
        for o in self.elements_old.get(element):
            error = []
            name = o.attrib.get('name')
            found = False
            for n in self.elements_new.get(element):
                if name == n.attrib.get('name'):
                    found = True
                    try:
                        if check_attributes:
                            self.diff_attributes(o, n)
                    except AttributeChangeException, e:
                        error.append("Attributes changed for '%s': %s" % (name, e.message))
                    if element == 'choice':
                        error.extend(self.diff_choicelist(o, n, check_attributes))
                    break
            if not found:
                error.append("Deleted")

            if error:
                out.append("Field: %s (%s)" % (name, element))
                out.append('---')
                out.append('Changes:')
                for num, err in enumerate(error):
                    out.append('%s. %s' % (num, err))
                out.append('===')

        # Searching for new fields in old
        for o in self.elements_new.get(element):
            error = []
            name = o.attrib.get('name')
            # Search for element in new
            found = False
            for n in self.elements_old.get(element):
                if name == n.attrib.get('name'):
                    found = True
                    break
            if not found:
                    error.append("New")

            if error:
                out.append("Field: %s (%s)" % (name, element))
                out.append('---')
                out.append('Changes:')
                for num, err in enumerate(error):
                    out.append('%s. %s' % (num, err))
                out.append('===')

        for o in out:
            print unicode(o).encode("utf-8")
    
    def diff_choicelist(self, o, n, check_attributes):
        old = {}
        new = {}
        errors = []
        for c in o.getchildren():
            old[c.attrib.get('value')] = c
        for c in n.getchildren():
            new[c.attrib.get('value')] = c

        for key in old.keys():
            if new.has_key(key):
                try:
                    if check_attributes:
                        self.diff_attributes(old.get(key), new.get(key))
                except AttributeChangeException, e:
                    errors.append("Attributes changed for '%s': %s " % (key, e.message) )
            else:
                errors.append("Deleted option: '%s' value: '%s'" % (old.get(key).attrib.get('description'), key))
        for key in new.keys():
            if not old.has_key(key):
                errors.append("New option: '%s' value: '%s'" % (new.get(key).attrib.get('description'), key))

        return errors

    def diff_attributes(self, o, n):
        ok = False
        new = []
        changed = []
        deleted = []
        for key, value in o.attrib.iteritems():
            if n.attrib.has_key(key):
                if n.get(key) == value:
                    pass #Ok
                else:
                    changed.append(key)
            else:
                deleted.append(key)
        for key, value in n.attrib.iteritems():
            if not o.attrib.has_key(key):
                new.append(key)

        if len(new) > 0 or len(changed) > 0 or len(deleted) > 0:
            raise AttributeChangeException(new, changed, deleted)

def usage(code, msg=''):
    print >> sys.stderr, __doc__
    if msg:
        print >> sys.stderr, msg
    sys.exit(code)

def main():
    try:
        opts, args = getopt.getopt(sys.argv[1:], 'hvVe:',
        ['elements=', 'check-attributes'])
   
    except getopt.error, msg:
        usage(1, msg)
    
    elements = None
    check_attributes = False
    for opt, arg in opts:
        if opt in ('-v'):
            log.setLevel(logging.INFO)
        if opt in ('-V'):
            log.setLevel(logging.DEBUG)
        if opt in ('-h', '--help'):
            usage(0)
        if opt in ('-e', '--elements'):
            elements = arg.split(',')
            log.debug('Elements: %s' % elements)
        if opt in ('--check-attributes'):
            check_attributes = True
    if len(args) < 2:
        usage(1)

    ins = Inspector(args[0], args[1])
    ins.perform(elements, check_attributes)

if __name__ == '__main__':
    main()
1	torsten	393	# -- coding: utf-8 --
2			#!/usr/bin/python
3			"""\n
4			Diffs to given formedtrees. Note that currently no checks on external choice
5			lists are done. Furter checks on Repeatgroups are very simple for now.
6			Warning: If a field with the same name is moved from master_tbl into a RG, it will not be
7			noticed, as we do not diff strutural changes.
8
9			Usage: diff_formed.py old new
10
11			Options:
12			-h / --help Print this message and exit.
13			-e / --elements comma-seperated list of elements to check
14			--check-attributes Check changes in attributes (changed descriptions, target...)
15
16			"""
17			# 20100810 Torsten Irlaender <[email protected]>
18			#
19			# This program is free software under the GNU GPL (>=v2)
20			# Read the file COPYING coming with the software for details.
21
22
23			import sys
24			import getopt
25			import logging
26
27			from lxml import etree
28
29			logging.basicConfig(level=logging.ERROR)
30			log = logging.getLogger('diff_formed')
31
32			class AttributeChangeException(Exception):
33			def __init__(self, n, c, d):
34			self.new = n
35			self.changed = c
36			self.deleted = d
37
38			msg = []
39			if self.new:
40			msg.append('New: %s' % ",".join(self.new))
41			if self.changed:
42			msg.append('Changed: %s' % ",".join(self.changed))
43			if self.deleted:
44			msg.append('Deleted: %s' % ",".join(self.deleted))
45			self.message = " \| ".join(msg)
46
47			class Inspector():
48
49			def __init__(self, filename_old, filename_new):
50
51			self.old_xml_doc = None
52			self.new_xml_doc = None
53			self.elements_old = {}
54			self.elements_new = {}
55			self.elements = ['choice', 'text', 'textarea', 'repeat']
56
57			# Open files and load xml
58			try:
59			f1 = None
60			f2 = None
61			log.debug('Old: %s'% filename_old)
62			f1 = open(filename_old, "r")
63			log.debug('New: %s'% filename_new)
64			f2 = open(filename_new, "r")
65			self.old_xml_doc = etree.parse(f1)
66			self.new_xml_doc = etree.parse(f2)
67			except:
68			log.exception('Error while opening file')
69			finally:
70			if f1 is not None:
71			f1.close()
72			if f2 is not None:
73			f2.close()
74
75			def perform(self, elements=None, check_attributes=False):
76			log.info('Only checking %s elements' % ",".join(elements))
77			for element in self.elements:
78			if elements is not None:
79			if element not in elements: continue
80			self.elements_old[element] = []
81			self.elements_new[element] = []
82			for e in self.old_xml_doc.findall("//%s" % element):
83			self.elements_old[element].append(e)
84			for e in self.new_xml_doc.findall("//%s" % element):
85			self.elements_new[element].append(e)
86			self.diff_element(element, check_attributes)
87
88			def diff_element(self, element, check_attributes):
89			log.debug('Checking %s elements' % element)
90			out = []
91
92			# Search for old field in new
93			for o in self.elements_old.get(element):
94			error = []
95			name = o.attrib.get('name')
96			found = False
97			for n in self.elements_new.get(element):
98			if name == n.attrib.get('name'):
99			found = True
100			try:
101			if check_attributes:
102			self.diff_attributes(o, n)
103			except AttributeChangeException, e:
104			error.append("Attributes changed for '%s': %s" % (name, e.message))
105			if element == 'choice':
106			error.extend(self.diff_choicelist(o, n, check_attributes))
107			break
108			if not found:
109			error.append("Deleted")
110
111			if error:
112			out.append("Field: %s (%s)" % (name, element))
113			out.append('---')
114			out.append('Changes:')
115			for num, err in enumerate(error):
116			out.append('%s. %s' % (num, err))
117			out.append('===')
118
119			# Searching for new fields in old
120			for o in self.elements_new.get(element):
121			error = []
122			name = o.attrib.get('name')
123			# Search for element in new
124			found = False
125			for n in self.elements_old.get(element):
126			if name == n.attrib.get('name'):
127			found = True
128			break
129			if not found:
130			error.append("New")
131
132			if error:
133			out.append("Field: %s (%s)" % (name, element))
134			out.append('---')
135			out.append('Changes:')
136			for num, err in enumerate(error):
137			out.append('%s. %s' % (num, err))
138			out.append('===')
139
140			for o in out:
141			print unicode(o).encode("utf-8")
142
143			def diff_choicelist(self, o, n, check_attributes):
144			old = {}
145			new = {}
146			errors = []
147			for c in o.getchildren():
148			old[c.attrib.get('value')] = c
149			for c in n.getchildren():
150			new[c.attrib.get('value')] = c
151
152			for key in old.keys():
153			if new.has_key(key):
154			try:
155			if check_attributes:
156			self.diff_attributes(old.get(key), new.get(key))
157			except AttributeChangeException, e:
158			errors.append("Attributes changed for '%s': %s " % (key, e.message) )
159			else:
160			errors.append("Deleted option: '%s' value: '%s'" % (old.get(key).attrib.get('description'), key))
161			for key in new.keys():
162			if not old.has_key(key):
163			errors.append("New option: '%s' value: '%s'" % (new.get(key).attrib.get('description'), key))
164
165			return errors
166
167			def diff_attributes(self, o, n):
168			ok = False
169			new = []
170			changed = []
171			deleted = []
172			for key, value in o.attrib.iteritems():
173			if n.attrib.has_key(key):
174			if n.get(key) == value:
175			pass #Ok
176			else:
177			changed.append(key)
178			else:
179			deleted.append(key)
180			for key, value in n.attrib.iteritems():
181			if not o.attrib.has_key(key):
182			new.append(key)
183
184			if len(new) > 0 or len(changed) > 0 or len(deleted) > 0:
185			raise AttributeChangeException(new, changed, deleted)
186
187			def usage(code, msg=''):
188			print >> sys.stderr, __doc__
189			if msg:
190			print >> sys.stderr, msg
191			sys.exit(code)
192
193			def main():
194			try:
195			opts, args = getopt.getopt(sys.argv[1:], 'hvVe:',
196			['elements=', 'check-attributes'])
197
198			except getopt.error, msg:
199			usage(1, msg)
200
201			elements = None
202			check_attributes = False
203			for opt, arg in opts:
204			if opt in ('-v'):
205			log.setLevel(logging.INFO)
206			if opt in ('-V'):
207			log.setLevel(logging.DEBUG)
208			if opt in ('-h', '--help'):
209			usage(0)
210			if opt in ('-e', '--elements'):
211			elements = arg.split(',')
212			log.debug('Elements: %s' % elements)
213			if opt in ('--check-attributes'):
214			check_attributes = True
215			if len(args) < 2:
216			usage(1)
217
218			ins = Inspector(args[0], args[1])
219			ins.perform(elements, check_attributes)
220
221			if __name__ == '__main__':
222			main()