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 3688 aaronmk
import sql
14 3077 aaronmk
import sql_io
15 1234 aaronmk
import strings
16 827 aaronmk
import term
17 1468 aaronmk
import units
18 1047 aaronmk
import util
19 86 aaronmk
import xml_dom
20 1321 aaronmk
import xpath
21 86 aaronmk
22 995 aaronmk
##### Exceptions
23
24 1612 aaronmk
class SyntaxError(exc.ExceptionWithCause):
25 797 aaronmk
    def __init__(self, cause):
26 1611 aaronmk
        exc.ExceptionWithCause.__init__(self, 'Invalid XML function syntax',
27
            cause)
28 278 aaronmk
29 1613 aaronmk
class FormatException(exc.ExceptionWithCause):
30
    def __init__(self, cause):
31
        exc.ExceptionWithCause.__init__(self, 'Invalid input value', cause)
32 843 aaronmk
33 1992 aaronmk
##### Helper functions
34 995 aaronmk
35 1992 aaronmk
def map_items(func, items):
36
    return [(name, func(value)) for name, value in items]
37
38
def cast(type_, val):
39
    '''Throws FormatException if can't cast'''
40
    try: return type_(val)
41
    except ValueError, e: raise FormatException(e)
42
43
def conv_items(type_, items):
44
    return map_items(lambda val: cast(type_, val),
45
        xml_dom.TextEntryOnlyIter(items))
46
47
def pop_value(items, name='value'):
48
    '''@param name Name of value param, or None to accept any name'''
49
    try: last = items.pop() # last entry contains value
50
    except IndexError: return None # input is empty and no actions
51
    if name != None and last[0] != name: return None # input is empty
52
    return last[1]
53
54 3335 aaronmk
def merge_tagged(root):
55
    '''Merges siblings in root that are marked as mergeable.
56
    Used to recombine pieces of nodes that were split apart in the mappings.
57
    '''
58
    for name in set((c.tagName for c in xpath.get(root, '*[@merge=1]'))):
59
        xml_dom.merge_by_name(root, name)
60
61
    # Recurse
62
    for child in xml_dom.NodeElemIter(root): merge_tagged(child)
63
64 995 aaronmk
funcs = {}
65 4236 aaronmk
simplifying_funcs = {}
66 995 aaronmk
67 1992 aaronmk
##### Public functions
68
69 4239 aaronmk
var_name_prefix = '$'
70
71
def is_var_name(str_): return str_.startswith(var_name_prefix)
72
73
def is_var(node):
74
    return xml_dom.is_text_node(node) and is_var_name(xml_dom.value(node))
75
76 2112 aaronmk
def is_func_name(name):
77
    return name.startswith('_') and name != '_' # '_' is default root node name
78
79
def is_func(node): return is_func_name(node.tagName)
80
81
def is_xml_func_name(name): return is_func_name(name) and name in funcs
82
83
def is_xml_func(node): return is_xml_func_name(node.tagName)
84
85 4298 aaronmk
def prune(node):
86
    '''Removes empty children'''
87
    for child in xml_dom.NodeElemIter(node):
88
        if xml_dom.is_empty(child): xml_dom.remove(child)
89
90 4300 aaronmk
def passthru(node):
91 4302 aaronmk
    '''Passes through single child node. First prunes the node.'''
92
    prune(node)
93 4300 aaronmk
    children = list(xml_dom.NodeEntryIter(node))
94
    if len(children) == 1: xml_dom.replace(node, children[0][1])
95
96 4041 aaronmk
def simplify(node):
97
    '''Simplifies the XML functions in an XML tree.
98
    * Merges nodes tagged as mergable
99 4236 aaronmk
    * Runs simplifying functions
100 4078 aaronmk
    * Applies pass-through optimizations for XML functions
101 4041 aaronmk
    '''
102
    for child in xml_dom.NodeElemIter(node): simplify(child)
103
    merge_tagged(node)
104
105 4227 aaronmk
    name = node.tagName
106 4041 aaronmk
107 4078 aaronmk
    # Pass-through optimizations
108 4228 aaronmk
    if is_func_name(name):
109 4236 aaronmk
        try: func = simplifying_funcs[name]
110
        except KeyError:
111
            children = list(xml_dom.NodeElemIter(node))
112
            if len(children) == 1: # single arg function
113
                func = name
114
                param_node = children[0]
115
                param = param_node.tagName
116
                value = param_node.firstChild
117
118
                is_agg_func = param.isdigit() or func == '_name'
119 4301 aaronmk
                if is_agg_func: passthru(node)
120 4236 aaronmk
        else: func(node)
121 4229 aaronmk
    # Pruning optimizations
122
    else: # these should not run on functions because they would remove args
123 4299 aaronmk
        prune(node)
124 4041 aaronmk
125 3660 aaronmk
def process(node, on_error=exc.reraise, is_rel_func=None, db=None):
126 2597 aaronmk
    '''Evaluates the XML functions in an XML tree.
127 3424 aaronmk
    @param is_rel_func None|f(str) Tests if a name is a relational function.
128 2602 aaronmk
        * If != None: Non-relational functions are removed, or relational
129
          functions are treated specially, depending on the db param (below).
130
    @param db
131
        * If None: Non-relational functions other than structural functions are
132
          replaced with their last parameter (usually the value), not evaluated.
133
          This is used in column-based mode to remove XML-only functions.
134
        * If != None: Relational functions are evaluated directly. This is used
135
          in row-based mode to combine relational and XML functions.
136 2597 aaronmk
    '''
137 3424 aaronmk
    has_rel_funcs = is_rel_func != None
138 2602 aaronmk
    assert db == None or has_rel_funcs # rel_funcs required if db set
139 2597 aaronmk
140 3333 aaronmk
    for child in xml_dom.NodeElemIter(node):
141 3424 aaronmk
        process(child, on_error, is_rel_func, db)
142 3335 aaronmk
    merge_tagged(node)
143 3333 aaronmk
144 995 aaronmk
    name = node.tagName
145 3227 aaronmk
    if not is_func_name(name): return node # not any kind of function
146 2602 aaronmk
147
    row_mode = has_rel_funcs and db != None
148
    column_mode = has_rel_funcs and db == None
149 3629 aaronmk
    func = funcs.get(name, None)
150 3028 aaronmk
    items = list(xml_dom.NodeTextEntryIter(node))
151 2602 aaronmk
152 3029 aaronmk
    # Parse function
153
    if len(items) == 1 and items[0][0].isdigit(): # has single numeric param
154
        # pass-through optimization for aggregating functions with one arg
155
        value = items[0][1] # pass through first arg
156 3629 aaronmk
    elif row_mode and (is_rel_func(name) or func == None): # row-based mode
157 3688 aaronmk
        try: value = sql_io.put(db, name, dict(items)) # evaluate using DB
158
        except sql.DoesNotExistException: return # preserve unknown funcs
159
            # possibly a built-in function of db_xml.put()
160 4024 aaronmk
    elif column_mode or func == None:
161 3640 aaronmk
        # local XML function can't be used or does not exist
162
        if column_mode and is_rel_func(name): return # preserve relational funcs
163
        # otherwise XML-only in column mode, or DB-only in XML output mode
164
        value = pop_value(items, None) # just replace with last param
165 2602 aaronmk
    else: # local XML function
166 3629 aaronmk
        try: value = func(items, node)
167 1613 aaronmk
        except Exception, e: # also catch non-wrapped exceptions (XML func bugs)
168 1371 aaronmk
            # Save in case another exception raised, overwriting sys.exc_info()
169
            exc.add_traceback(e)
170 1562 aaronmk
            str_ = strings.ustr(node)
171 995 aaronmk
            exc.add_msg(e, 'function:\n'+str_)
172 1810 aaronmk
            xml_dom.replace(node, xml_dom.mk_comment(node.ownerDocument,
173
                '\n'+term.emph_multiline(str_)))
174
175 995 aaronmk
            on_error(e)
176 2602 aaronmk
            return # in case on_error() returns
177 3227 aaronmk
178 3333 aaronmk
    xml_dom.replace_with_text(node, value)
179 995 aaronmk
180 4236 aaronmk
##### Simplifying functions
181
182
# Function names must start with _ to avoid collisions with real tags
183
# Functions take params (node) and have no return value
184
185 4237 aaronmk
#### Logic
186
187
def _exists(node):
188
    '''Returns whether its node is non-empty'''
189
    xml_dom.replace_with_text(node, not xml_dom.is_empty(node))
190
simplifying_funcs['_exists'] = _exists
191
192 4240 aaronmk
def _if(node):
193
    '''Can add `@name` attr to distinguish separate _if statements'''
194
    params = dict(xml_dom.NodeEntryIter(node))
195
    then = params.get('then', None)
196
    cond = params.get('cond', None)
197
    else_ = params.get('else', None)
198
199
    if cond == None: xml_dom.replace(node, else_) # always False
200
    elif then == else_: xml_dom.replace(node, then) # always same value
201
    elif is_var(cond): pass # can't simplify variable conditions
202
    elif xml_dom.is_text_node(cond) and bool(xml_dom.value(cond)): # always True
203
        xml_dom.replace(node, then)
204
simplifying_funcs['_if'] = _if
205
206 4303 aaronmk
#### Merging
207
208
simplifying_funcs['_alt'] = passthru
209
simplifying_funcs['_merge'] = passthru
210
211 1469 aaronmk
##### XML functions
212 995 aaronmk
213
# Function names must start with _ to avoid collisions with real tags
214 4144 aaronmk
# Functions take arguments (items, node)
215 995 aaronmk
216 1469 aaronmk
#### Transforming values
217
218 1602 aaronmk
def repl(repls, value):
219 1537 aaronmk
    '''Raises error if value not in map and no special '*' entry
220 1602 aaronmk
    @param repls dict repl:with
221
        repl "*" means all other input values
222
        with "*" means keep input value the same
223
        with "" means ignore input value
224 1537 aaronmk
    '''
225 1602 aaronmk
    try: new_value = repls[value]
226 1304 aaronmk
    except KeyError, e:
227 1537 aaronmk
        # Save traceback right away in case another exception raised
228 2984 aaronmk
        fe = FormatException(e)
229 1602 aaronmk
        try: new_value = repls['*']
230 1609 aaronmk
        except KeyError: raise fe
231 1537 aaronmk
    if new_value == '*': new_value = value # '*' means keep input value the same
232 1607 aaronmk
    return new_value
233 1602 aaronmk
234 2016 aaronmk
def _map(items, node):
235 1602 aaronmk
    '''See repl()
236
    @param items
237
        <last_entry> Value
238
        <other_entries> name=value Mappings. Special values: See repl() repls.
239
    '''
240
    value = pop_value(items)
241
    if value == None: return None # input is empty
242 1607 aaronmk
    return util.none_if(repl(dict(items), value), u'') # empty value means None
243 1219 aaronmk
funcs['_map'] = _map
244
245 2016 aaronmk
def _replace(items, node):
246 1562 aaronmk
    items = conv_items(strings.ustr, items) # get *once* from iter, check types
247 1581 aaronmk
    value = pop_value(items)
248
    if value == None: return None # input is empty
249 1219 aaronmk
    try:
250
        for repl, with_ in items:
251
            if re.match(r'^\w+$', repl):
252
                repl = r'(?<![^\W_])'+repl+r'(?![^\W_])' # match whole word
253
            value = re.sub(repl, with_, value)
254 1612 aaronmk
    except sre_constants.error, e: raise SyntaxError(e)
255 1624 aaronmk
    return util.none_if(value.strip(), u'') # empty strings always mean None
256 1219 aaronmk
funcs['_replace'] = _replace
257
258 1469 aaronmk
#### Quantities
259
260 2016 aaronmk
def _units(items, node):
261 1581 aaronmk
    value = pop_value(items)
262
    if value == None: return None # input is empty
263 1471 aaronmk
264 1581 aaronmk
    quantity = units.str2quantity(value)
265 1471 aaronmk
    try:
266
        for action, units_ in items:
267
            units_ = util.none_if(units_, u'')
268
            if action == 'default': units.set_default_units(quantity, units_)
269 1567 aaronmk
            elif action == 'to':
270
                try: quantity = units.convert(quantity, units_)
271 1609 aaronmk
                except ValueError, e: raise FormatException(e)
272 1612 aaronmk
            else: raise SyntaxError(ValueError('Invalid action: '+action))
273 1609 aaronmk
    except units.MissingUnitsException, e: raise FormatException(e)
274 1471 aaronmk
    return units.quantity2str(quantity)
275 1225 aaronmk
funcs['_units'] = _units
276
277 1399 aaronmk
def parse_range(str_, range_sep='-'):
278
    default = (str_, None)
279
    start, sep, end = str_.partition(range_sep)
280
    if sep == '': return default # not a range
281 1427 aaronmk
    if start == '' and range_sep == '-': return default # negative number
282 1399 aaronmk
    return tuple(d.strip() for d in (start, end))
283
284 2016 aaronmk
def _rangeStart(items, node):
285 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
286 1399 aaronmk
    try: value = items['value']
287 1406 aaronmk
    except KeyError: return None # input is empty
288 1399 aaronmk
    return parse_range(value)[0]
289
funcs['_rangeStart'] = _rangeStart
290
291 2016 aaronmk
def _rangeEnd(items, node):
292 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
293 1399 aaronmk
    try: value = items['value']
294 1406 aaronmk
    except KeyError: return None # input is empty
295 1399 aaronmk
    return parse_range(value)[1]
296
funcs['_rangeEnd'] = _rangeEnd
297
298 2016 aaronmk
def _range(items, node):
299 1472 aaronmk
    items = dict(conv_items(float, items))
300
    from_ = items.get('from', None)
301
    to = items.get('to', None)
302
    if from_ == None or to == None: return None
303
    return str(to - from_)
304
funcs['_range'] = _range
305
306 2016 aaronmk
def _avg(items, node):
307 86 aaronmk
    count = 0
308
    sum_ = 0.
309 278 aaronmk
    for name, value in conv_items(float, items):
310 86 aaronmk
        count += 1
311
        sum_ += value
312 1472 aaronmk
    if count == 0: return None # input is empty
313
    else: return str(sum_/count)
314 995 aaronmk
funcs['_avg'] = _avg
315 86 aaronmk
316 968 aaronmk
class CvException(Exception):
317
    def __init__(self):
318
        Exception.__init__(self, 'CV (coefficient of variation) values are only'
319
            ' allowed for ratio scale data '
320
            '(see <http://en.wikipedia.org/wiki/Coefficient_of_variation>)')
321
322 2016 aaronmk
def _noCV(items, node):
323 3631 aaronmk
    items = list(conv_items(strings.ustr, items))
324 3046 aaronmk
    try: name, value = items.pop() # last entry contains value
325
    except IndexError: return None # input is empty
326 1609 aaronmk
    if re.match('^(?i)CV *\d+$', value): raise FormatException(CvException())
327 968 aaronmk
    return value
328 995 aaronmk
funcs['_noCV'] = _noCV
329 968 aaronmk
330 1469 aaronmk
#### Names
331
332 328 aaronmk
_name_parts_slices_items = [
333
    ('first', slice(None, 1)),
334
    ('middle', slice(1, -1)),
335
    ('last', slice(-1, None)),
336
]
337
name_parts_slices = dict(_name_parts_slices_items)
338
name_parts = [name for name, slice_ in _name_parts_slices_items]
339
340 2016 aaronmk
def _name(items, node):
341 3631 aaronmk
    items = dict(list(conv_items(strings.ustr, items)))
342 102 aaronmk
    parts = []
343 328 aaronmk
    for part in name_parts:
344
        if part in items: parts.append(items[part])
345 4035 aaronmk
    if not parts: return None # pass None values through; handle no name parts
346 102 aaronmk
    return ' '.join(parts)
347 995 aaronmk
funcs['_name'] = _name
348 102 aaronmk
349 1607 aaronmk
#### Angles
350
351 2016 aaronmk
def _compass(items, node):
352 1607 aaronmk
    '''Converts a compass direction (N, NE, NNE, etc.) into a degree heading'''
353
    items = dict(conv_items(strings.ustr, items))
354
    try: value = items['value']
355
    except KeyError: return None # input is empty
356
357
    if not value.isupper(): return value # pass through other coordinate formats
358
    try: return util.cast(str, angles.compass2heading(value)) # ignore None
359
    except KeyError, e: raise FormatException(e)
360
funcs['_compass'] = _compass