Project

General

Profile

1 996 aaronmk
# XML "function" nodes that transform their contents
2 86 aaronmk
3 111 aaronmk
import datetime
4 968 aaronmk
import re
5 1219 aaronmk
import sre_constants
6 111 aaronmk
7 818 aaronmk
import dates
8 300 aaronmk
import exc
9 917 aaronmk
import maps
10 1234 aaronmk
import strings
11 827 aaronmk
import term
12 1468 aaronmk
import units
13 1047 aaronmk
import util
14 86 aaronmk
import xml_dom
15 1321 aaronmk
import xpath
16 86 aaronmk
17 995 aaronmk
##### Exceptions
18
19 962 aaronmk
class SyntaxException(Exception):
20 797 aaronmk
    def __init__(self, cause):
21 962 aaronmk
        Exception.__init__(self, 'Invalid XML function syntax: '
22
            +exc.str_(cause))
23 278 aaronmk
24 843 aaronmk
class FormatException(SyntaxException): pass
25
26 995 aaronmk
##### Functions
27
28
funcs = {}
29
30
def process(node, on_error=exc.raise_):
31
    for child in xml_dom.NodeElemIter(node): process(child, on_error)
32
    name = node.tagName
33
    if name.startswith('_') and name in funcs:
34 1369 aaronmk
        try:
35
            value = funcs[name](xml_dom.NodeTextEntryIter(node))
36
            xml_dom.replace_with_text(node, value)
37
        except Exception, e: # also catch XML func internal errors
38 1371 aaronmk
            # Save in case another exception raised, overwriting sys.exc_info()
39
            exc.add_traceback(e)
40 995 aaronmk
            str_ = str(node)
41
            exc.add_msg(e, 'function:\n'+str_)
42
            xml_dom.replace(node, node.ownerDocument.createComment(
43 1234 aaronmk
                '\n'+term.emph_multiline(str_).replace('--','-')))
44
                # comments can't contain '--'
45 995 aaronmk
            on_error(e)
46
47 86 aaronmk
def map_items(func, items):
48
    return [(name, func(value)) for name, value in items]
49
50 1234 aaronmk
def cast(type_, val):
51
    '''Throws SyntaxException if can't cast'''
52
    try: return type_(val)
53
    except ValueError, e: raise SyntaxException(e)
54
55 278 aaronmk
def conv_items(type_, items):
56 1234 aaronmk
    return map_items(lambda val: cast(type_, val),
57
        xml_dom.TextEntryOnlyIter(items))
58 278 aaronmk
59 1469 aaronmk
##### XML functions
60 995 aaronmk
61
# Function names must start with _ to avoid collisions with real tags
62
# Functions take arguments (items)
63
64 1469 aaronmk
#### General
65
66 995 aaronmk
def _ignore(items):
67 994 aaronmk
    '''Used to "comment out" an XML subtree'''
68
    return None
69 995 aaronmk
funcs['_ignore'] = _ignore
70 994 aaronmk
71 1469 aaronmk
#### Conditionals
72
73 1234 aaronmk
def _eq(items):
74
    items = dict(items)
75
    try:
76
        left = items['left']
77
        right = items['right']
78
    except KeyError: return '' # a value was None
79
    return util.bool2str(left == right)
80
funcs['_eq'] = _eq
81
82
def _if(items):
83
    items = dict(items)
84
    try:
85
        cond = items['cond']
86
        then = items['then']
87
    except KeyError, e: raise SyntaxException(e)
88
    else_ = items.get('else', None)
89
    cond = bool(cast(str, cond))
90
    if cond: return then
91
    else: return else_
92
funcs['_if'] = _if
93
94 1469 aaronmk
#### Combining values
95
96 995 aaronmk
def _alt(items):
97 113 aaronmk
    items = list(items)
98
    items.sort()
99 1186 aaronmk
    try: return items[0][1] # value of lowest-numbered item
100 1187 aaronmk
    except IndexError: return None # input got removed by e.g. SyntaxException
101 995 aaronmk
funcs['_alt'] = _alt
102 113 aaronmk
103 995 aaronmk
def _merge(items):
104 1234 aaronmk
    items = list(conv_items(strings.ustr, items))
105
        # get *once* from iter and check types
106 917 aaronmk
    items.sort()
107
    return maps.merge_values(*[v for k, v in items])
108 995 aaronmk
funcs['_merge'] = _merge
109 917 aaronmk
110 995 aaronmk
def _label(items):
111 1412 aaronmk
    items = dict(conv_items(strings.ustr, items))
112
        # get *once* from iter and check types
113 917 aaronmk
    try:
114
        label = items['label']
115
        value = items['value']
116
    except KeyError, e: raise SyntaxException(e)
117
    return label+': '+value
118 995 aaronmk
funcs['_label'] = _label
119 917 aaronmk
120 1469 aaronmk
#### Transforming values
121
122 1047 aaronmk
def _nullIf(items):
123
    items = dict(conv_items(str, items))
124
    try:
125
        null = items['null']
126
        value = items['value']
127
    except KeyError, e: raise SyntaxException(e)
128 1219 aaronmk
    type_str = items.get('type', None)
129
    type_ = str
130
    if type_str == 'float': type_ = float
131
    return util.none_if(value, type_(null))
132 1047 aaronmk
funcs['_nullIf'] = _nullIf
133
134 1219 aaronmk
def _map(items):
135
    items = conv_items(str, items) # get *once* from iter and check types
136
    try: value = items.pop()[1] # value is last entry's value
137
    except IndexError, e: raise SyntaxException(e)
138
    map_ = dict(items)
139 1304 aaronmk
    closed = bool(map_.pop('_closed', False))
140 1219 aaronmk
    try: return map_[value]
141 1304 aaronmk
    except KeyError, e:
142
        if closed: raise SyntaxException(e)
143
        else: return value
144 1219 aaronmk
funcs['_map'] = _map
145
146
def _replace(items):
147
    items = conv_items(str, items) # get *once* from iter and check types
148 1424 aaronmk
    try: value = items.pop()[1] # last entry contains value
149 1219 aaronmk
    except IndexError, e: raise SyntaxException(e)
150
    try:
151
        for repl, with_ in items:
152
            if re.match(r'^\w+$', repl):
153
                repl = r'(?<![^\W_])'+repl+r'(?![^\W_])' # match whole word
154
            value = re.sub(repl, with_, value)
155
    except sre_constants.error, e: raise SyntaxException(e)
156 1427 aaronmk
    return util.none_if(value, u'') # empty strings always mean None
157 1219 aaronmk
funcs['_replace'] = _replace
158
159 1469 aaronmk
#### Quantities
160
161 1225 aaronmk
def _units(items):
162
    items = dict(conv_items(str, items))
163 1463 aaronmk
    try: value = items['value']
164 1464 aaronmk
    except KeyError: return None # input is empty
165 1463 aaronmk
    default_units = items.get('units', None)
166
    # DB unit conversion isn't ready yet, so just return number
167 1468 aaronmk
    try: return units.cleanup_units(value, default_units).split(' ')[0]
168
    except units.MissingUnitsException, e: raise SyntaxException(e)
169 1225 aaronmk
funcs['_units'] = _units
170
171 995 aaronmk
def _range(items):
172 278 aaronmk
    items = dict(conv_items(float, items))
173 965 aaronmk
    from_ = items.get('from', None)
174
    to = items.get('to', None)
175
    if from_ == None or to == None: return None
176 326 aaronmk
    return str(to - from_)
177 995 aaronmk
funcs['_range'] = _range
178 86 aaronmk
179 1399 aaronmk
def parse_range(str_, range_sep='-'):
180
    default = (str_, None)
181
    start, sep, end = str_.partition(range_sep)
182
    if sep == '': return default # not a range
183 1427 aaronmk
    if start == '' and range_sep == '-': return default # negative number
184 1399 aaronmk
    return tuple(d.strip() for d in (start, end))
185
186
def _rangeStart(items):
187
    items = dict(conv_items(str, items))
188
    try: value = items['value']
189 1406 aaronmk
    except KeyError: return None # input is empty
190 1399 aaronmk
    return parse_range(value)[0]
191
funcs['_rangeStart'] = _rangeStart
192
193
def _rangeEnd(items):
194
    items = dict(conv_items(str, items))
195
    try: value = items['value']
196 1406 aaronmk
    except KeyError: return None # input is empty
197 1399 aaronmk
    return parse_range(value)[1]
198
funcs['_rangeEnd'] = _rangeEnd
199
200 995 aaronmk
def _avg(items):
201 86 aaronmk
    count = 0
202
    sum_ = 0.
203 278 aaronmk
    for name, value in conv_items(float, items):
204 86 aaronmk
        count += 1
205
        sum_ += value
206
    return str(sum_/count)
207 995 aaronmk
funcs['_avg'] = _avg
208 86 aaronmk
209 968 aaronmk
class CvException(Exception):
210
    def __init__(self):
211
        Exception.__init__(self, 'CV (coefficient of variation) values are only'
212
            ' allowed for ratio scale data '
213
            '(see <http://en.wikipedia.org/wiki/Coefficient_of_variation>)')
214
215 995 aaronmk
def _noCV(items):
216 968 aaronmk
    try: name, value = items.next()
217
    except StopIteration: return None
218
    if re.match('^(?i)CV *\d+$', value): raise SyntaxException(CvException())
219
    return value
220 995 aaronmk
funcs['_noCV'] = _noCV
221 968 aaronmk
222 1469 aaronmk
#### Dates
223
224 995 aaronmk
def _date(items):
225 917 aaronmk
    items = conv_items(str, items) # get *once* from iter and check types
226 786 aaronmk
    try: str_ = dict(items)['date']
227
    except KeyError:
228 1308 aaronmk
        items = dict(conv_items(int, items))
229 1292 aaronmk
        try: items['year'] # year is required
230 1309 aaronmk
        except KeyError, e:
231
            if items == {}: return None # entire date is empty
232
            else: raise SyntaxException(e)
233 786 aaronmk
        items.setdefault('month', 1)
234
        items.setdefault('day', 1)
235
        try: date = datetime.date(**items)
236
        except ValueError, e: raise SyntaxException(e)
237
    else:
238 324 aaronmk
        try: year = float(str_)
239
        except ValueError:
240 1264 aaronmk
            try: date = dates.strtotime(str_)
241 324 aaronmk
            except ImportError: return str_
242
            except ValueError, e: raise SyntaxException(e)
243
        else: date = (datetime.date(int(year), 1, 1) +
244
            datetime.timedelta(round((year % 1.)*365)))
245 818 aaronmk
    try: return dates.strftime('%Y-%m-%d', date)
246 843 aaronmk
    except ValueError, e: raise FormatException(e)
247 995 aaronmk
funcs['_date'] = _date
248 86 aaronmk
249 1366 aaronmk
def _dateRangeStart(items):
250
    items = dict(conv_items(str, items))
251
    try: value = items['value']
252 1406 aaronmk
    except KeyError: return None # input is empty
253 1366 aaronmk
    return dates.parse_date_range(value)[0]
254
funcs['_dateRangeStart'] = _dateRangeStart
255 1311 aaronmk
256 1366 aaronmk
def _dateRangeEnd(items):
257 1311 aaronmk
    items = dict(conv_items(str, items))
258 1366 aaronmk
    try: value = items['value']
259 1406 aaronmk
    except KeyError: return None # input is empty
260 1366 aaronmk
    return dates.parse_date_range(value)[1]
261
funcs['_dateRangeEnd'] = _dateRangeEnd
262 1311 aaronmk
263 1469 aaronmk
#### Names
264
265 328 aaronmk
_name_parts_slices_items = [
266
    ('first', slice(None, 1)),
267
    ('middle', slice(1, -1)),
268
    ('last', slice(-1, None)),
269
]
270
name_parts_slices = dict(_name_parts_slices_items)
271
name_parts = [name for name, slice_ in _name_parts_slices_items]
272
273 995 aaronmk
def _name(items):
274 89 aaronmk
    items = dict(items)
275 102 aaronmk
    parts = []
276 328 aaronmk
    for part in name_parts:
277
        if part in items: parts.append(items[part])
278 102 aaronmk
    return ' '.join(parts)
279 995 aaronmk
funcs['_name'] = _name
280 102 aaronmk
281 995 aaronmk
def _namePart(items):
282 328 aaronmk
    out_items = []
283
    for part, value in items:
284
        try: slice_ = name_parts_slices[part]
285
        except KeyError, e: raise SyntaxException(e)
286 1219 aaronmk
        out_items.append((part, ' '.join(value.split(' ')[slice_])))
287 995 aaronmk
    return _name(out_items)
288
funcs['_namePart'] = _namePart
289 1321 aaronmk
290 1469 aaronmk
#### Paths
291
292 1321 aaronmk
def _simplifyPath(items):
293
    items = dict(items)
294
    try:
295
        next = cast(str, items['next'])
296
        require = cast(str, items['require'])
297
        root = items['path']
298
    except KeyError, e: raise SyntaxException(e)
299
300
    node = root
301
    while node != None:
302
        new_node = xpath.get_1(node, next, allow_rooted=False)
303
        if xpath.get_1(node, require, allow_rooted=False) == None: # empty elem
304
            xml_dom.replace(node, new_node) # remove current elem
305
            if node is root: root = new_node # also update root
306
        node = new_node
307
    return root
308
funcs['_simplifyPath'] = _simplifyPath