Project

General

Profile

1 996 aaronmk
# XML "function" nodes that transform their contents
2 86 aaronmk
3 111 aaronmk
import datetime
4 5190 aaronmk
import operator
5 968 aaronmk
import re
6 1219 aaronmk
import sre_constants
7 2017 aaronmk
import warnings
8 111 aaronmk
9 1607 aaronmk
import angles
10 818 aaronmk
import dates
11 300 aaronmk
import exc
12 1580 aaronmk
import format
13 917 aaronmk
import maps
14 3688 aaronmk
import sql
15 3077 aaronmk
import sql_io
16 1234 aaronmk
import strings
17 827 aaronmk
import term
18 1468 aaronmk
import units
19 1047 aaronmk
import util
20 86 aaronmk
import xml_dom
21 1321 aaronmk
import xpath
22 86 aaronmk
23 995 aaronmk
##### Exceptions
24
25 1612 aaronmk
class SyntaxError(exc.ExceptionWithCause):
26 797 aaronmk
    def __init__(self, cause):
27 1611 aaronmk
        exc.ExceptionWithCause.__init__(self, 'Invalid XML function syntax',
28
            cause)
29 278 aaronmk
30 1613 aaronmk
class FormatException(exc.ExceptionWithCause):
31
    def __init__(self, cause):
32
        exc.ExceptionWithCause.__init__(self, 'Invalid input value', cause)
33 843 aaronmk
34 1992 aaronmk
##### Helper functions
35 995 aaronmk
36 4334 aaronmk
def map_names(func, params):
37
    return [(func(name), value) for name, value in params]
38
39
def variadic_args(node):
40
    args = map_names(float, xml_dom.NodeEntryIter(node))
41
    args.sort()
42
    return [value for name, value in args]
43
44 1992 aaronmk
def map_items(func, items):
45
    return [(name, func(value)) for name, value in items]
46
47
def cast(type_, val):
48
    '''Throws FormatException if can't cast'''
49
    try: return type_(val)
50
    except ValueError, e: raise FormatException(e)
51
52
def conv_items(type_, items):
53
    return map_items(lambda val: cast(type_, val),
54
        xml_dom.TextEntryOnlyIter(items))
55
56
def pop_value(items, name='value'):
57
    '''@param name Name of value param, or None to accept any name'''
58
    try: last = items.pop() # last entry contains value
59
    except IndexError: return None # input is empty and no actions
60
    if name != None and last[0] != name: return None # input is empty
61
    return last[1]
62
63 3335 aaronmk
def merge_tagged(root):
64
    '''Merges siblings in root that are marked as mergeable.
65
    Used to recombine pieces of nodes that were split apart in the mappings.
66
    '''
67
    for name in set((c.tagName for c in xpath.get(root, '*[@merge=1]'))):
68
        xml_dom.merge_by_name(root, name)
69
70
    # Recurse
71
    for child in xml_dom.NodeElemIter(root): merge_tagged(child)
72
73 995 aaronmk
funcs = {}
74 4236 aaronmk
simplifying_funcs = {}
75 995 aaronmk
76 1992 aaronmk
##### Public functions
77
78 4239 aaronmk
var_name_prefix = '$'
79
80
def is_var_name(str_): return str_.startswith(var_name_prefix)
81
82
def is_var(node):
83
    return xml_dom.is_text_node(node) and is_var_name(xml_dom.value(node))
84
85 2112 aaronmk
def is_func_name(name):
86
    return name.startswith('_') and name != '_' # '_' is default root node name
87
88
def is_func(node): return is_func_name(node.tagName)
89
90
def is_xml_func_name(name): return is_func_name(name) and name in funcs
91
92
def is_xml_func(node): return is_xml_func_name(node.tagName)
93
94 4300 aaronmk
def passthru(node):
95 4302 aaronmk
    '''Passes through single child node. First prunes the node.'''
96 4322 aaronmk
    xml_dom.prune(node)
97 4300 aaronmk
    children = list(xml_dom.NodeEntryIter(node))
98
    if len(children) == 1: xml_dom.replace(node, children[0][1])
99
100 4041 aaronmk
def simplify(node):
101 4305 aaronmk
    '''Simplifies an XML tree.
102 4041 aaronmk
    * Merges nodes tagged as mergable
103 4236 aaronmk
    * Runs simplifying functions
104 4041 aaronmk
    '''
105
    for child in xml_dom.NodeElemIter(node): simplify(child)
106
    merge_tagged(node)
107
108 4227 aaronmk
    name = node.tagName
109 4041 aaronmk
110 4078 aaronmk
    # Pass-through optimizations
111 4228 aaronmk
    if is_func_name(name):
112 4236 aaronmk
        try: func = simplifying_funcs[name]
113 4756 aaronmk
        except KeyError: xml_dom.prune_empty(node)
114 4236 aaronmk
        else: func(node)
115 4229 aaronmk
    # Pruning optimizations
116
    else: # these should not run on functions because they would remove args
117 4319 aaronmk
        xml_dom.prune_children(node)
118 4041 aaronmk
119 3660 aaronmk
def process(node, on_error=exc.reraise, is_rel_func=None, db=None):
120 2597 aaronmk
    '''Evaluates the XML functions in an XML tree.
121 3424 aaronmk
    @param is_rel_func None|f(str) Tests if a name is a relational function.
122 2602 aaronmk
        * If != None: Non-relational functions are removed, or relational
123
          functions are treated specially, depending on the db param (below).
124
    @param db
125
        * If None: Non-relational functions other than structural functions are
126
          replaced with their last parameter (usually the value), not evaluated.
127
          This is used in column-based mode to remove XML-only functions.
128
        * If != None: Relational functions are evaluated directly. This is used
129
          in row-based mode to combine relational and XML functions.
130 2597 aaronmk
    '''
131 3424 aaronmk
    has_rel_funcs = is_rel_func != None
132 2602 aaronmk
    assert db == None or has_rel_funcs # rel_funcs required if db set
133 2597 aaronmk
134 3333 aaronmk
    for child in xml_dom.NodeElemIter(node):
135 3424 aaronmk
        process(child, on_error, is_rel_func, db)
136 3335 aaronmk
    merge_tagged(node)
137 3333 aaronmk
138 995 aaronmk
    name = node.tagName
139 3227 aaronmk
    if not is_func_name(name): return node # not any kind of function
140 2602 aaronmk
141
    row_mode = has_rel_funcs and db != None
142
    column_mode = has_rel_funcs and db == None
143 3629 aaronmk
    func = funcs.get(name, None)
144 3028 aaronmk
    items = list(xml_dom.NodeTextEntryIter(node))
145 2602 aaronmk
146 3029 aaronmk
    # Parse function
147
    if len(items) == 1 and items[0][0].isdigit(): # has single numeric param
148
        # pass-through optimization for aggregating functions with one arg
149
        value = items[0][1] # pass through first arg
150 3629 aaronmk
    elif row_mode and (is_rel_func(name) or func == None): # row-based mode
151 5190 aaronmk
        if reduce(operator.or_, (xml_dom.is_node(v) for n, v in items)):
152
            return # preserve complex funcs
153 3688 aaronmk
        try: value = sql_io.put(db, name, dict(items)) # evaluate using DB
154
        except sql.DoesNotExistException: return # preserve unknown funcs
155
            # possibly a built-in function of db_xml.put()
156 4024 aaronmk
    elif column_mode or func == None:
157 3640 aaronmk
        # local XML function can't be used or does not exist
158
        if column_mode and is_rel_func(name): return # preserve relational funcs
159
        # otherwise XML-only in column mode, or DB-only in XML output mode
160
        value = pop_value(items, None) # just replace with last param
161 2602 aaronmk
    else: # local XML function
162 3629 aaronmk
        try: value = func(items, node)
163 1613 aaronmk
        except Exception, e: # also catch non-wrapped exceptions (XML func bugs)
164 1371 aaronmk
            # Save in case another exception raised, overwriting sys.exc_info()
165
            exc.add_traceback(e)
166 1562 aaronmk
            str_ = strings.ustr(node)
167 995 aaronmk
            exc.add_msg(e, 'function:\n'+str_)
168 1810 aaronmk
            xml_dom.replace(node, xml_dom.mk_comment(node.ownerDocument,
169
                '\n'+term.emph_multiline(str_)))
170
171 995 aaronmk
            on_error(e)
172 2602 aaronmk
            return # in case on_error() returns
173 3227 aaronmk
174 3333 aaronmk
    xml_dom.replace_with_text(node, value)
175 995 aaronmk
176 4236 aaronmk
##### Simplifying functions
177
178
# Function names must start with _ to avoid collisions with real tags
179
# Functions take params (node) and have no return value
180
181 4237 aaronmk
#### Logic
182
183
def _exists(node):
184
    '''Returns whether its node is non-empty'''
185
    xml_dom.replace_with_text(node, not xml_dom.is_empty(node))
186
simplifying_funcs['_exists'] = _exists
187
188 4240 aaronmk
def _if(node):
189 4477 aaronmk
    '''
190
    *Must* be run to remove conditions that functions._if() can't handle.
191
    Note: Can add `@name` attr to distinguish separate _if statements.
192
    '''
193 4240 aaronmk
    params = dict(xml_dom.NodeEntryIter(node))
194
    then = params.get('then', None)
195
    cond = params.get('cond', None)
196
    else_ = params.get('else', None)
197
198
    if cond == None: xml_dom.replace(node, else_) # always False
199
    elif then == else_: xml_dom.replace(node, then) # always same value
200
    elif is_var(cond): pass # can't simplify variable conditions
201
    elif xml_dom.is_text_node(cond) and bool(xml_dom.value(cond)): # always True
202
        xml_dom.replace(node, then)
203
simplifying_funcs['_if'] = _if
204
205 4303 aaronmk
#### Merging
206
207
simplifying_funcs['_alt'] = passthru
208 4326 aaronmk
simplifying_funcs['_join'] = passthru
209 5011 aaronmk
simplifying_funcs['_join_words'] = passthru
210 4303 aaronmk
simplifying_funcs['_merge'] = passthru
211 5409 aaronmk
simplifying_funcs['_min'] = passthru
212
simplifying_funcs['_max'] = passthru
213 4303 aaronmk
214 4335 aaronmk
def _first(node):
215
    '''Chooses the first param (after sorting by numeric param name)'''
216
    args = variadic_args(node)
217
    try: first = args[0]
218
    except IndexError: first = None
219
    xml_dom.replace(node, first)
220
simplifying_funcs['_first'] = _first
221
222 1469 aaronmk
##### XML functions
223 995 aaronmk
224
# Function names must start with _ to avoid collisions with real tags
225 4144 aaronmk
# Functions take arguments (items, node)
226 995 aaronmk
227 1469 aaronmk
#### Transforming values
228
229 1602 aaronmk
def repl(repls, value):
230 1537 aaronmk
    '''Raises error if value not in map and no special '*' entry
231 1602 aaronmk
    @param repls dict repl:with
232
        repl "*" means all other input values
233
        with "*" means keep input value the same
234
        with "" means ignore input value
235 1537 aaronmk
    '''
236 1602 aaronmk
    try: new_value = repls[value]
237 1304 aaronmk
    except KeyError, e:
238 1537 aaronmk
        # Save traceback right away in case another exception raised
239 2984 aaronmk
        fe = FormatException(e)
240 1602 aaronmk
        try: new_value = repls['*']
241 1609 aaronmk
        except KeyError: raise fe
242 1537 aaronmk
    if new_value == '*': new_value = value # '*' means keep input value the same
243 1607 aaronmk
    return new_value
244 1602 aaronmk
245 2016 aaronmk
def _map(items, node):
246 1602 aaronmk
    '''See repl()
247
    @param items
248
        <last_entry> Value
249
        <other_entries> name=value Mappings. Special values: See repl() repls.
250
    '''
251
    value = pop_value(items)
252
    if value == None: return None # input is empty
253 1607 aaronmk
    return util.none_if(repl(dict(items), value), u'') # empty value means None
254 1219 aaronmk
funcs['_map'] = _map
255
256 2016 aaronmk
def _replace(items, node):
257 1562 aaronmk
    items = conv_items(strings.ustr, items) # get *once* from iter, check types
258 1581 aaronmk
    value = pop_value(items)
259
    if value == None: return None # input is empty
260 1219 aaronmk
    try:
261
        for repl, with_ in items:
262
            if re.match(r'^\w+$', repl):
263
                repl = r'(?<![^\W_])'+repl+r'(?![^\W_])' # match whole word
264
            value = re.sub(repl, with_, value)
265 1612 aaronmk
    except sre_constants.error, e: raise SyntaxError(e)
266 1624 aaronmk
    return util.none_if(value.strip(), u'') # empty strings always mean None
267 1219 aaronmk
funcs['_replace'] = _replace
268
269 1469 aaronmk
#### Quantities
270
271 2016 aaronmk
def _units(items, node):
272 1581 aaronmk
    value = pop_value(items)
273
    if value == None: return None # input is empty
274 1471 aaronmk
275 1581 aaronmk
    quantity = units.str2quantity(value)
276 1471 aaronmk
    try:
277
        for action, units_ in items:
278
            units_ = util.none_if(units_, u'')
279
            if action == 'default': units.set_default_units(quantity, units_)
280 1567 aaronmk
            elif action == 'to':
281
                try: quantity = units.convert(quantity, units_)
282 1609 aaronmk
                except ValueError, e: raise FormatException(e)
283 1612 aaronmk
            else: raise SyntaxError(ValueError('Invalid action: '+action))
284 1609 aaronmk
    except units.MissingUnitsException, e: raise FormatException(e)
285 1471 aaronmk
    return units.quantity2str(quantity)
286 1225 aaronmk
funcs['_units'] = _units
287
288 1399 aaronmk
def parse_range(str_, range_sep='-'):
289
    default = (str_, None)
290
    start, sep, end = str_.partition(range_sep)
291
    if sep == '': return default # not a range
292 1427 aaronmk
    if start == '' and range_sep == '-': return default # negative number
293 1399 aaronmk
    return tuple(d.strip() for d in (start, end))
294
295 2016 aaronmk
def _rangeStart(items, node):
296 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
297 1399 aaronmk
    try: value = items['value']
298 1406 aaronmk
    except KeyError: return None # input is empty
299 1399 aaronmk
    return parse_range(value)[0]
300
funcs['_rangeStart'] = _rangeStart
301
302 2016 aaronmk
def _rangeEnd(items, node):
303 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
304 1399 aaronmk
    try: value = items['value']
305 1406 aaronmk
    except KeyError: return None # input is empty
306 1399 aaronmk
    return parse_range(value)[1]
307
funcs['_rangeEnd'] = _rangeEnd
308
309 2016 aaronmk
def _range(items, node):
310 1472 aaronmk
    items = dict(conv_items(float, items))
311
    from_ = items.get('from', None)
312
    to = items.get('to', None)
313
    if from_ == None or to == None: return None
314
    return str(to - from_)
315
funcs['_range'] = _range
316
317 2016 aaronmk
def _avg(items, node):
318 86 aaronmk
    count = 0
319
    sum_ = 0.
320 278 aaronmk
    for name, value in conv_items(float, items):
321 86 aaronmk
        count += 1
322
        sum_ += value
323 1472 aaronmk
    if count == 0: return None # input is empty
324
    else: return str(sum_/count)
325 995 aaronmk
funcs['_avg'] = _avg
326 86 aaronmk
327 968 aaronmk
class CvException(Exception):
328
    def __init__(self):
329
        Exception.__init__(self, 'CV (coefficient of variation) values are only'
330
            ' allowed for ratio scale data '
331
            '(see <http://en.wikipedia.org/wiki/Coefficient_of_variation>)')
332
333 2016 aaronmk
def _noCV(items, node):
334 3631 aaronmk
    items = list(conv_items(strings.ustr, items))
335 3046 aaronmk
    try: name, value = items.pop() # last entry contains value
336
    except IndexError: return None # input is empty
337 1609 aaronmk
    if re.match('^(?i)CV *\d+$', value): raise FormatException(CvException())
338 968 aaronmk
    return value
339 995 aaronmk
funcs['_noCV'] = _noCV
340 968 aaronmk
341 1607 aaronmk
#### Angles
342
343 2016 aaronmk
def _compass(items, node):
344 1607 aaronmk
    '''Converts a compass direction (N, NE, NNE, etc.) into a degree heading'''
345
    items = dict(conv_items(strings.ustr, items))
346
    try: value = items['value']
347
    except KeyError: return None # input is empty
348
349
    if not value.isupper(): return value # pass through other coordinate formats
350
    try: return util.cast(str, angles.compass2heading(value)) # ignore None
351
    except KeyError, e: raise FormatException(e)
352
funcs['_compass'] = _compass