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 2017 aaronmk
import warnings
7 111 aaronmk
8 1607 aaronmk
import angles
9 818 aaronmk
import dates
10 300 aaronmk
import exc
11 1580 aaronmk
import format
12 917 aaronmk
import maps
13 3077 aaronmk
import sql_io
14 1234 aaronmk
import strings
15 827 aaronmk
import term
16 1468 aaronmk
import units
17 1047 aaronmk
import util
18 86 aaronmk
import xml_dom
19 1321 aaronmk
import xpath
20 86 aaronmk
21 995 aaronmk
##### Exceptions
22
23 1612 aaronmk
class SyntaxError(exc.ExceptionWithCause):
24 797 aaronmk
    def __init__(self, cause):
25 1611 aaronmk
        exc.ExceptionWithCause.__init__(self, 'Invalid XML function syntax',
26
            cause)
27 278 aaronmk
28 1613 aaronmk
class FormatException(exc.ExceptionWithCause):
29
    def __init__(self, cause):
30
        exc.ExceptionWithCause.__init__(self, 'Invalid input value', cause)
31 843 aaronmk
32 1992 aaronmk
##### Helper functions
33 995 aaronmk
34 1992 aaronmk
def map_items(func, items):
35
    return [(name, func(value)) for name, value in items]
36
37
def cast(type_, val):
38
    '''Throws FormatException if can't cast'''
39
    try: return type_(val)
40
    except ValueError, e: raise FormatException(e)
41
42
def conv_items(type_, items):
43
    return map_items(lambda val: cast(type_, val),
44
        xml_dom.TextEntryOnlyIter(items))
45
46
def pop_value(items, name='value'):
47
    '''@param name Name of value param, or None to accept any name'''
48
    try: last = items.pop() # last entry contains value
49
    except IndexError: return None # input is empty and no actions
50
    if name != None and last[0] != name: return None # input is empty
51
    return last[1]
52
53 3335 aaronmk
def merge_tagged(root):
54
    '''Merges siblings in root that are marked as mergeable.
55
    Used to recombine pieces of nodes that were split apart in the mappings.
56
    '''
57
    for name in set((c.tagName for c in xpath.get(root, '*[@merge=1]'))):
58
        xml_dom.merge_by_name(root, name)
59
60
    # Recurse
61
    for child in xml_dom.NodeElemIter(root): merge_tagged(child)
62
63 995 aaronmk
funcs = {}
64
65 2557 aaronmk
structural_funcs = set()
66
67 1992 aaronmk
##### Public functions
68
69 2112 aaronmk
def is_func_name(name):
70
    return name.startswith('_') and name != '_' # '_' is default root node name
71
72
def is_func(node): return is_func_name(node.tagName)
73
74
def is_xml_func_name(name): return is_func_name(name) and name in funcs
75
76
def is_xml_func(node): return is_xml_func_name(node.tagName)
77
78 3424 aaronmk
def process(node, on_error=exc.raise_, is_rel_func=None, db=None):
79 2597 aaronmk
    '''Evaluates the XML functions in an XML tree.
80 3424 aaronmk
    @param is_rel_func None|f(str) Tests if a name is a relational function.
81 2602 aaronmk
        * If != None: Non-relational functions are removed, or relational
82
          functions are treated specially, depending on the db param (below).
83
    @param db
84
        * If None: Non-relational functions other than structural functions are
85
          replaced with their last parameter (usually the value), not evaluated.
86
          This is used in column-based mode to remove XML-only functions.
87
        * If != None: Relational functions are evaluated directly. This is used
88
          in row-based mode to combine relational and XML functions.
89 2597 aaronmk
    '''
90 3424 aaronmk
    has_rel_funcs = is_rel_func != None
91 2602 aaronmk
    assert db == None or has_rel_funcs # rel_funcs required if db set
92 2597 aaronmk
93 3333 aaronmk
    for child in xml_dom.NodeElemIter(node):
94 3424 aaronmk
        process(child, on_error, is_rel_func, db)
95 3335 aaronmk
    merge_tagged(node)
96 3333 aaronmk
97 995 aaronmk
    name = node.tagName
98 3227 aaronmk
    if not is_func_name(name): return node # not any kind of function
99 2602 aaronmk
100
    row_mode = has_rel_funcs and db != None
101
    column_mode = has_rel_funcs and db == None
102 3629 aaronmk
    func = funcs.get(name, None)
103 3028 aaronmk
    items = list(xml_dom.NodeTextEntryIter(node))
104 2602 aaronmk
105 3029 aaronmk
    # Parse function
106
    if len(items) == 1 and items[0][0].isdigit(): # has single numeric param
107
        # pass-through optimization for aggregating functions with one arg
108
        value = items[0][1] # pass through first arg
109 3629 aaronmk
    elif row_mode and (is_rel_func(name) or func == None): # row-based mode
110
        value = sql_io.put(db, name, dict(items)) # evaluate using DB
111 3640 aaronmk
    elif (column_mode and not name in structural_funcs) or func == None:
112
        # local XML function can't be used or does not exist
113
        if column_mode and is_rel_func(name): return # preserve relational funcs
114
        # otherwise XML-only in column mode, or DB-only in XML output mode
115
        value = pop_value(items, None) # just replace with last param
116 2602 aaronmk
    else: # local XML function
117 3629 aaronmk
        try: value = func(items, node)
118 1613 aaronmk
        except Exception, e: # also catch non-wrapped exceptions (XML func bugs)
119 1371 aaronmk
            # Save in case another exception raised, overwriting sys.exc_info()
120
            exc.add_traceback(e)
121 1562 aaronmk
            str_ = strings.ustr(node)
122 995 aaronmk
            exc.add_msg(e, 'function:\n'+str_)
123 1810 aaronmk
            xml_dom.replace(node, xml_dom.mk_comment(node.ownerDocument,
124
                '\n'+term.emph_multiline(str_)))
125
126 995 aaronmk
            on_error(e)
127 2602 aaronmk
            return # in case on_error() returns
128 3227 aaronmk
129 3333 aaronmk
    xml_dom.replace_with_text(node, value)
130 995 aaronmk
131 1469 aaronmk
##### XML functions
132 995 aaronmk
133
# Function names must start with _ to avoid collisions with real tags
134
# Functions take arguments (items)
135
136 2557 aaronmk
#### Structural
137 1469 aaronmk
138 2017 aaronmk
def _ignore(items, node):
139 994 aaronmk
    '''Used to "comment out" an XML subtree'''
140
    return None
141 995 aaronmk
funcs['_ignore'] = _ignore
142 2557 aaronmk
structural_funcs.add('_ignore')
143 994 aaronmk
144 2017 aaronmk
def _ref(items, node):
145
    '''Used to retrieve a value from another XML node
146
    @param items
147
        addr=<path> XPath to value, relative to the XML func's parent node
148
    '''
149
    items = dict(items)
150
    try: addr = items['addr']
151
    except KeyError, e: raise SyntaxError(e)
152
153
    value = xpath.get_value(node.parentNode, addr)
154
    if value == None:
155
        warnings.warn(UserWarning('_ref: XPath reference target missing: '
156
            +str(addr)))
157
    return value
158
funcs['_ref'] = _ref
159 2557 aaronmk
structural_funcs.add('_ref')
160 2017 aaronmk
161 1469 aaronmk
#### Conditionals
162
163 2016 aaronmk
def _eq(items, node):
164 1234 aaronmk
    items = dict(items)
165
    try:
166
        left = items['left']
167
        right = items['right']
168
    except KeyError: return '' # a value was None
169
    return util.bool2str(left == right)
170
funcs['_eq'] = _eq
171
172 2016 aaronmk
def _if(items, node):
173 1234 aaronmk
    items = dict(items)
174
    try:
175
        cond = items['cond']
176
        then = items['then']
177 1612 aaronmk
    except KeyError, e: raise SyntaxError(e)
178 1234 aaronmk
    else_ = items.get('else', None)
179 1562 aaronmk
    cond = bool(cast(strings.ustr, cond))
180 1234 aaronmk
    if cond: return then
181
    else: return else_
182
funcs['_if'] = _if
183
184 1469 aaronmk
#### Combining values
185
186 2016 aaronmk
def _alt(items, node):
187 113 aaronmk
    items = list(items)
188
    items.sort()
189 1186 aaronmk
    try: return items[0][1] # value of lowest-numbered item
190 1609 aaronmk
    except IndexError: return None # input got removed by e.g. FormatException
191 995 aaronmk
funcs['_alt'] = _alt
192 113 aaronmk
193 2016 aaronmk
def _merge(items, node):
194 1234 aaronmk
    items = list(conv_items(strings.ustr, items))
195 1562 aaronmk
        # get *once* from iter, check types
196 917 aaronmk
    items.sort()
197
    return maps.merge_values(*[v for k, v in items])
198 995 aaronmk
funcs['_merge'] = _merge
199 917 aaronmk
200 2016 aaronmk
def _label(items, node):
201 1412 aaronmk
    items = dict(conv_items(strings.ustr, items))
202 1562 aaronmk
        # get *once* from iter, check types
203 2014 aaronmk
    value = items.get('value', None)
204
    if value == None: return None # input is empty
205
    try: label = items['label']
206 1612 aaronmk
    except KeyError, e: raise SyntaxError(e)
207 917 aaronmk
    return label+': '+value
208 995 aaronmk
funcs['_label'] = _label
209 917 aaronmk
210 1469 aaronmk
#### Transforming values
211
212 2016 aaronmk
def _collapse(items, node):
213 2012 aaronmk
    '''Collapses a subtree if the "value" element in it is NULL'''
214
    items = dict(items)
215
    try: require = cast(strings.ustr, items['require'])
216
    except KeyError, e: raise SyntaxError(e)
217
    value = items.get('value', None)
218
219 2558 aaronmk
    if xpath.get_value(value, require, allow_rooted=False) == None: return None
220 2012 aaronmk
    else: return value
221
funcs['_collapse'] = _collapse
222
223 1478 aaronmk
types_by_name = {None: strings.ustr, 'str': strings.ustr, 'float': float}
224 1477 aaronmk
225 2016 aaronmk
def _nullIf(items, node):
226 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
227 1477 aaronmk
    try: null = items['null']
228 1612 aaronmk
    except KeyError, e: raise SyntaxError(e)
229 1477 aaronmk
    value = items.get('value', None)
230 1219 aaronmk
    type_str = items.get('type', None)
231 1477 aaronmk
232
    try: type_ = types_by_name[type_str]
233 1612 aaronmk
    except KeyError, e: raise SyntaxError(e)
234 1477 aaronmk
    null = type_(null)
235
236
    try: return util.none_if(value, null)
237
    except ValueError: return value # value not convertible, so can't equal null
238 1047 aaronmk
funcs['_nullIf'] = _nullIf
239
240 1602 aaronmk
def repl(repls, value):
241 1537 aaronmk
    '''Raises error if value not in map and no special '*' entry
242 1602 aaronmk
    @param repls dict repl:with
243
        repl "*" means all other input values
244
        with "*" means keep input value the same
245
        with "" means ignore input value
246 1537 aaronmk
    '''
247 1602 aaronmk
    try: new_value = repls[value]
248 1304 aaronmk
    except KeyError, e:
249 1537 aaronmk
        # Save traceback right away in case another exception raised
250 2984 aaronmk
        fe = FormatException(e)
251 1602 aaronmk
        try: new_value = repls['*']
252 1609 aaronmk
        except KeyError: raise fe
253 1537 aaronmk
    if new_value == '*': new_value = value # '*' means keep input value the same
254 1607 aaronmk
    return new_value
255 1602 aaronmk
256 2016 aaronmk
def _map(items, node):
257 1602 aaronmk
    '''See repl()
258
    @param items
259
        <last_entry> Value
260
        <other_entries> name=value Mappings. Special values: See repl() repls.
261
    '''
262
    items = conv_items(strings.ustr, items) # get *once* from iter, check types
263
    value = pop_value(items)
264
    if value == None: return None # input is empty
265 1607 aaronmk
    return util.none_if(repl(dict(items), value), u'') # empty value means None
266 1219 aaronmk
funcs['_map'] = _map
267
268 2016 aaronmk
def _replace(items, node):
269 1562 aaronmk
    items = conv_items(strings.ustr, items) # get *once* from iter, check types
270 1581 aaronmk
    value = pop_value(items)
271
    if value == None: return None # input is empty
272 1219 aaronmk
    try:
273
        for repl, with_ in items:
274
            if re.match(r'^\w+$', repl):
275
                repl = r'(?<![^\W_])'+repl+r'(?![^\W_])' # match whole word
276
            value = re.sub(repl, with_, value)
277 1612 aaronmk
    except sre_constants.error, e: raise SyntaxError(e)
278 1624 aaronmk
    return util.none_if(value.strip(), u'') # empty strings always mean None
279 1219 aaronmk
funcs['_replace'] = _replace
280
281 1469 aaronmk
#### Quantities
282
283 2016 aaronmk
def _units(items, node):
284 1562 aaronmk
    items = conv_items(strings.ustr, items) # get *once* from iter, check types
285 1581 aaronmk
    value = pop_value(items)
286
    if value == None: return None # input is empty
287 1471 aaronmk
288 1581 aaronmk
    quantity = units.str2quantity(value)
289 1471 aaronmk
    try:
290
        for action, units_ in items:
291
            units_ = util.none_if(units_, u'')
292
            if action == 'default': units.set_default_units(quantity, units_)
293 1567 aaronmk
            elif action == 'to':
294
                try: quantity = units.convert(quantity, units_)
295 1609 aaronmk
                except ValueError, e: raise FormatException(e)
296 1612 aaronmk
            else: raise SyntaxError(ValueError('Invalid action: '+action))
297 1609 aaronmk
    except units.MissingUnitsException, e: raise FormatException(e)
298 1471 aaronmk
    return units.quantity2str(quantity)
299 1225 aaronmk
funcs['_units'] = _units
300
301 1399 aaronmk
def parse_range(str_, range_sep='-'):
302
    default = (str_, None)
303
    start, sep, end = str_.partition(range_sep)
304
    if sep == '': return default # not a range
305 1427 aaronmk
    if start == '' and range_sep == '-': return default # negative number
306 1399 aaronmk
    return tuple(d.strip() for d in (start, end))
307
308 2016 aaronmk
def _rangeStart(items, node):
309 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
310 1399 aaronmk
    try: value = items['value']
311 1406 aaronmk
    except KeyError: return None # input is empty
312 1399 aaronmk
    return parse_range(value)[0]
313
funcs['_rangeStart'] = _rangeStart
314
315 2016 aaronmk
def _rangeEnd(items, node):
316 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
317 1399 aaronmk
    try: value = items['value']
318 1406 aaronmk
    except KeyError: return None # input is empty
319 1399 aaronmk
    return parse_range(value)[1]
320
funcs['_rangeEnd'] = _rangeEnd
321
322 2016 aaronmk
def _range(items, node):
323 1472 aaronmk
    items = dict(conv_items(float, items))
324
    from_ = items.get('from', None)
325
    to = items.get('to', None)
326
    if from_ == None or to == None: return None
327
    return str(to - from_)
328
funcs['_range'] = _range
329
330 2016 aaronmk
def _avg(items, node):
331 86 aaronmk
    count = 0
332
    sum_ = 0.
333 278 aaronmk
    for name, value in conv_items(float, items):
334 86 aaronmk
        count += 1
335
        sum_ += value
336 1472 aaronmk
    if count == 0: return None # input is empty
337
    else: return str(sum_/count)
338 995 aaronmk
funcs['_avg'] = _avg
339 86 aaronmk
340 968 aaronmk
class CvException(Exception):
341
    def __init__(self):
342
        Exception.__init__(self, 'CV (coefficient of variation) values are only'
343
            ' allowed for ratio scale data '
344
            '(see <http://en.wikipedia.org/wiki/Coefficient_of_variation>)')
345
346 2016 aaronmk
def _noCV(items, node):
347 3631 aaronmk
    items = list(conv_items(strings.ustr, items))
348 3046 aaronmk
    try: name, value = items.pop() # last entry contains value
349
    except IndexError: return None # input is empty
350 1609 aaronmk
    if re.match('^(?i)CV *\d+$', value): raise FormatException(CvException())
351 968 aaronmk
    return value
352 995 aaronmk
funcs['_noCV'] = _noCV
353 968 aaronmk
354 1469 aaronmk
#### Dates
355
356 2016 aaronmk
def _date(items, node):
357 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
358
        # get *once* from iter, check types
359 1514 aaronmk
    try: str_ = items['date']
360 786 aaronmk
    except KeyError:
361 1515 aaronmk
        # Year is required
362
        try: items['year']
363 1309 aaronmk
        except KeyError, e:
364
            if items == {}: return None # entire date is empty
365 1609 aaronmk
            else: raise FormatException(e)
366 1515 aaronmk
367
        # Convert month name to number
368
        try: month = items['month']
369
        except KeyError: pass
370
        else:
371
            if not month.isdigit(): # month is name
372 1582 aaronmk
                try: items['month'] = str(dates.strtotime(month).month)
373 1609 aaronmk
                except ValueError, e: raise FormatException(e)
374 1515 aaronmk
375 1580 aaronmk
        items = dict(conv_items(format.str2int, items.iteritems()))
376 786 aaronmk
        items.setdefault('month', 1)
377
        items.setdefault('day', 1)
378 1535 aaronmk
379
        for try_num in xrange(2):
380
            try:
381
                date = datetime.date(**items)
382
                break
383
            except ValueError, e:
384 1609 aaronmk
                if try_num > 0: raise FormatException(e)
385 1536 aaronmk
                    # exception still raised after retry
386 1562 aaronmk
                msg = strings.ustr(e)
387 1535 aaronmk
                if msg == 'month must be in 1..12': # try swapping month and day
388
                    items['month'], items['day'] = items['day'], items['month']
389 1609 aaronmk
                else: raise FormatException(e)
390 786 aaronmk
    else:
391 324 aaronmk
        try: year = float(str_)
392
        except ValueError:
393 1264 aaronmk
            try: date = dates.strtotime(str_)
394 324 aaronmk
            except ImportError: return str_
395 1609 aaronmk
            except ValueError, e: raise FormatException(e)
396 324 aaronmk
        else: date = (datetime.date(int(year), 1, 1) +
397
            datetime.timedelta(round((year % 1.)*365)))
398 818 aaronmk
    try: return dates.strftime('%Y-%m-%d', date)
399 843 aaronmk
    except ValueError, e: raise FormatException(e)
400 995 aaronmk
funcs['_date'] = _date
401 86 aaronmk
402 2016 aaronmk
def _dateRangeStart(items, node):
403 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
404 1366 aaronmk
    try: value = items['value']
405 1406 aaronmk
    except KeyError: return None # input is empty
406 1366 aaronmk
    return dates.parse_date_range(value)[0]
407
funcs['_dateRangeStart'] = _dateRangeStart
408 1311 aaronmk
409 2016 aaronmk
def _dateRangeEnd(items, node):
410 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
411 1366 aaronmk
    try: value = items['value']
412 1406 aaronmk
    except KeyError: return None # input is empty
413 1366 aaronmk
    return dates.parse_date_range(value)[1]
414
funcs['_dateRangeEnd'] = _dateRangeEnd
415 1311 aaronmk
416 1469 aaronmk
#### Names
417
418 328 aaronmk
_name_parts_slices_items = [
419
    ('first', slice(None, 1)),
420
    ('middle', slice(1, -1)),
421
    ('last', slice(-1, None)),
422
]
423
name_parts_slices = dict(_name_parts_slices_items)
424
name_parts = [name for name, slice_ in _name_parts_slices_items]
425
426 2016 aaronmk
def _name(items, node):
427 3631 aaronmk
    items = dict(list(conv_items(strings.ustr, items)))
428 102 aaronmk
    parts = []
429 328 aaronmk
    for part in name_parts:
430
        if part in items: parts.append(items[part])
431 102 aaronmk
    return ' '.join(parts)
432 995 aaronmk
funcs['_name'] = _name
433 102 aaronmk
434 2016 aaronmk
def _namePart(items, node):
435 3631 aaronmk
    items = list(conv_items(strings.ustr, items))
436 328 aaronmk
    out_items = []
437
    for part, value in items:
438
        try: slice_ = name_parts_slices[part]
439 1612 aaronmk
        except KeyError, e: raise SyntaxError(e)
440 1219 aaronmk
        out_items.append((part, ' '.join(value.split(' ')[slice_])))
441 2016 aaronmk
    return _name(out_items, node)
442 995 aaronmk
funcs['_namePart'] = _namePart
443 1321 aaronmk
444 1607 aaronmk
#### Angles
445
446 2016 aaronmk
def _compass(items, node):
447 1607 aaronmk
    '''Converts a compass direction (N, NE, NNE, etc.) into a degree heading'''
448
    items = dict(conv_items(strings.ustr, items))
449
    try: value = items['value']
450
    except KeyError: return None # input is empty
451
452
    if not value.isupper(): return value # pass through other coordinate formats
453
    try: return util.cast(str, angles.compass2heading(value)) # ignore None
454
    except KeyError, e: raise FormatException(e)
455
funcs['_compass'] = _compass
456
457 1469 aaronmk
#### Paths
458
459 2016 aaronmk
def _simplifyPath(items, node):
460 1321 aaronmk
    items = dict(items)
461
    try:
462 1562 aaronmk
        next = cast(strings.ustr, items['next'])
463
        require = cast(strings.ustr, items['require'])
464 1321 aaronmk
        root = items['path']
465 1612 aaronmk
    except KeyError, e: raise SyntaxError(e)
466 1321 aaronmk
467
    node = root
468
    while node != None:
469
        new_node = xpath.get_1(node, next, allow_rooted=False)
470 2558 aaronmk
        if xpath.get_value(node, require, allow_rooted=False) == None: # empty
471 1321 aaronmk
            xml_dom.replace(node, new_node) # remove current elem
472
            if node is root: root = new_node # also update root
473
        node = new_node
474
    return root
475
funcs['_simplifyPath'] = _simplifyPath