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