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 4334 aaronmk
def map_names(func, params):
36
    return [(func(name), value) for name, value in params]
37
38
def variadic_args(node):
39
    args = map_names(float, xml_dom.NodeEntryIter(node))
40
    args.sort()
41
    return [value for name, value in args]
42
43 1992 aaronmk
def map_items(func, items):
44
    return [(name, func(value)) for name, value in items]
45
46
def cast(type_, val):
47
    '''Throws FormatException if can't cast'''
48
    try: return type_(val)
49
    except ValueError, e: raise FormatException(e)
50
51
def conv_items(type_, items):
52
    return map_items(lambda val: cast(type_, val),
53
        xml_dom.TextEntryOnlyIter(items))
54
55
def pop_value(items, name='value'):
56
    '''@param name Name of value param, or None to accept any name'''
57
    try: last = items.pop() # last entry contains value
58
    except IndexError: return None # input is empty and no actions
59
    if name != None and last[0] != name: return None # input is empty
60
    return last[1]
61
62 3335 aaronmk
def merge_tagged(root):
63
    '''Merges siblings in root that are marked as mergeable.
64
    Used to recombine pieces of nodes that were split apart in the mappings.
65
    '''
66
    for name in set((c.tagName for c in xpath.get(root, '*[@merge=1]'))):
67
        xml_dom.merge_by_name(root, name)
68
69
    # Recurse
70
    for child in xml_dom.NodeElemIter(root): merge_tagged(child)
71
72 995 aaronmk
funcs = {}
73 4236 aaronmk
simplifying_funcs = {}
74 995 aaronmk
75 1992 aaronmk
##### Public functions
76
77 4239 aaronmk
var_name_prefix = '$'
78
79
def is_var_name(str_): return str_.startswith(var_name_prefix)
80
81
def is_var(node):
82
    return xml_dom.is_text_node(node) and is_var_name(xml_dom.value(node))
83
84 2112 aaronmk
def is_func_name(name):
85
    return name.startswith('_') and name != '_' # '_' is default root node name
86
87
def is_func(node): return is_func_name(node.tagName)
88
89
def is_xml_func_name(name): return is_func_name(name) and name in funcs
90
91
def is_xml_func(node): return is_xml_func_name(node.tagName)
92
93 4300 aaronmk
def passthru(node):
94 4302 aaronmk
    '''Passes through single child node. First prunes the node.'''
95 4322 aaronmk
    xml_dom.prune(node)
96 4300 aaronmk
    children = list(xml_dom.NodeEntryIter(node))
97
    if len(children) == 1: xml_dom.replace(node, children[0][1])
98
99 4041 aaronmk
def simplify(node):
100 4305 aaronmk
    '''Simplifies an XML tree.
101 4041 aaronmk
    * Merges nodes tagged as mergable
102 4236 aaronmk
    * Runs simplifying functions
103 4041 aaronmk
    '''
104
    for child in xml_dom.NodeElemIter(node): simplify(child)
105
    merge_tagged(node)
106
107 4227 aaronmk
    name = node.tagName
108 4041 aaronmk
109 4078 aaronmk
    # Pass-through optimizations
110 4228 aaronmk
    if is_func_name(name):
111 4236 aaronmk
        try: func = simplifying_funcs[name]
112 4305 aaronmk
        except KeyError: pass
113 4236 aaronmk
        else: func(node)
114 4229 aaronmk
    # Pruning optimizations
115
    else: # these should not run on functions because they would remove args
116 4319 aaronmk
        xml_dom.prune_children(node)
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 4303 aaronmk
#### Merging
200
201
simplifying_funcs['_alt'] = passthru
202 4326 aaronmk
simplifying_funcs['_join'] = passthru
203 4303 aaronmk
simplifying_funcs['_merge'] = passthru
204 4304 aaronmk
simplifying_funcs['_name'] = passthru
205 4303 aaronmk
206 1469 aaronmk
##### XML functions
207 995 aaronmk
208
# Function names must start with _ to avoid collisions with real tags
209 4144 aaronmk
# Functions take arguments (items, node)
210 995 aaronmk
211 1469 aaronmk
#### Transforming values
212
213 1602 aaronmk
def repl(repls, value):
214 1537 aaronmk
    '''Raises error if value not in map and no special '*' entry
215 1602 aaronmk
    @param repls dict repl:with
216
        repl "*" means all other input values
217
        with "*" means keep input value the same
218
        with "" means ignore input value
219 1537 aaronmk
    '''
220 1602 aaronmk
    try: new_value = repls[value]
221 1304 aaronmk
    except KeyError, e:
222 1537 aaronmk
        # Save traceback right away in case another exception raised
223 2984 aaronmk
        fe = FormatException(e)
224 1602 aaronmk
        try: new_value = repls['*']
225 1609 aaronmk
        except KeyError: raise fe
226 1537 aaronmk
    if new_value == '*': new_value = value # '*' means keep input value the same
227 1607 aaronmk
    return new_value
228 1602 aaronmk
229 2016 aaronmk
def _map(items, node):
230 1602 aaronmk
    '''See repl()
231
    @param items
232
        <last_entry> Value
233
        <other_entries> name=value Mappings. Special values: See repl() repls.
234
    '''
235
    value = pop_value(items)
236
    if value == None: return None # input is empty
237 1607 aaronmk
    return util.none_if(repl(dict(items), value), u'') # empty value means None
238 1219 aaronmk
funcs['_map'] = _map
239
240 2016 aaronmk
def _replace(items, node):
241 1562 aaronmk
    items = conv_items(strings.ustr, items) # get *once* from iter, check types
242 1581 aaronmk
    value = pop_value(items)
243
    if value == None: return None # input is empty
244 1219 aaronmk
    try:
245
        for repl, with_ in items:
246
            if re.match(r'^\w+$', repl):
247
                repl = r'(?<![^\W_])'+repl+r'(?![^\W_])' # match whole word
248
            value = re.sub(repl, with_, value)
249 1612 aaronmk
    except sre_constants.error, e: raise SyntaxError(e)
250 1624 aaronmk
    return util.none_if(value.strip(), u'') # empty strings always mean None
251 1219 aaronmk
funcs['_replace'] = _replace
252
253 1469 aaronmk
#### Quantities
254
255 2016 aaronmk
def _units(items, node):
256 1581 aaronmk
    value = pop_value(items)
257
    if value == None: return None # input is empty
258 1471 aaronmk
259 1581 aaronmk
    quantity = units.str2quantity(value)
260 1471 aaronmk
    try:
261
        for action, units_ in items:
262
            units_ = util.none_if(units_, u'')
263
            if action == 'default': units.set_default_units(quantity, units_)
264 1567 aaronmk
            elif action == 'to':
265
                try: quantity = units.convert(quantity, units_)
266 1609 aaronmk
                except ValueError, e: raise FormatException(e)
267 1612 aaronmk
            else: raise SyntaxError(ValueError('Invalid action: '+action))
268 1609 aaronmk
    except units.MissingUnitsException, e: raise FormatException(e)
269 1471 aaronmk
    return units.quantity2str(quantity)
270 1225 aaronmk
funcs['_units'] = _units
271
272 1399 aaronmk
def parse_range(str_, range_sep='-'):
273
    default = (str_, None)
274
    start, sep, end = str_.partition(range_sep)
275
    if sep == '': return default # not a range
276 1427 aaronmk
    if start == '' and range_sep == '-': return default # negative number
277 1399 aaronmk
    return tuple(d.strip() for d in (start, end))
278
279 2016 aaronmk
def _rangeStart(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)[0]
284
funcs['_rangeStart'] = _rangeStart
285
286 2016 aaronmk
def _rangeEnd(items, node):
287 1562 aaronmk
    items = dict(conv_items(strings.ustr, items))
288 1399 aaronmk
    try: value = items['value']
289 1406 aaronmk
    except KeyError: return None # input is empty
290 1399 aaronmk
    return parse_range(value)[1]
291
funcs['_rangeEnd'] = _rangeEnd
292
293 2016 aaronmk
def _range(items, node):
294 1472 aaronmk
    items = dict(conv_items(float, items))
295
    from_ = items.get('from', None)
296
    to = items.get('to', None)
297
    if from_ == None or to == None: return None
298
    return str(to - from_)
299
funcs['_range'] = _range
300
301 2016 aaronmk
def _avg(items, node):
302 86 aaronmk
    count = 0
303
    sum_ = 0.
304 278 aaronmk
    for name, value in conv_items(float, items):
305 86 aaronmk
        count += 1
306
        sum_ += value
307 1472 aaronmk
    if count == 0: return None # input is empty
308
    else: return str(sum_/count)
309 995 aaronmk
funcs['_avg'] = _avg
310 86 aaronmk
311 968 aaronmk
class CvException(Exception):
312
    def __init__(self):
313
        Exception.__init__(self, 'CV (coefficient of variation) values are only'
314
            ' allowed for ratio scale data '
315
            '(see <http://en.wikipedia.org/wiki/Coefficient_of_variation>)')
316
317 2016 aaronmk
def _noCV(items, node):
318 3631 aaronmk
    items = list(conv_items(strings.ustr, items))
319 3046 aaronmk
    try: name, value = items.pop() # last entry contains value
320
    except IndexError: return None # input is empty
321 1609 aaronmk
    if re.match('^(?i)CV *\d+$', value): raise FormatException(CvException())
322 968 aaronmk
    return value
323 995 aaronmk
funcs['_noCV'] = _noCV
324 968 aaronmk
325 1469 aaronmk
#### Names
326
327 328 aaronmk
_name_parts_slices_items = [
328
    ('first', slice(None, 1)),
329
    ('middle', slice(1, -1)),
330
    ('last', slice(-1, None)),
331
]
332
name_parts_slices = dict(_name_parts_slices_items)
333
name_parts = [name for name, slice_ in _name_parts_slices_items]
334
335 2016 aaronmk
def _name(items, node):
336 3631 aaronmk
    items = dict(list(conv_items(strings.ustr, items)))
337 102 aaronmk
    parts = []
338 328 aaronmk
    for part in name_parts:
339
        if part in items: parts.append(items[part])
340 4035 aaronmk
    if not parts: return None # pass None values through; handle no name parts
341 102 aaronmk
    return ' '.join(parts)
342 995 aaronmk
funcs['_name'] = _name
343 102 aaronmk
344 1607 aaronmk
#### Angles
345
346 2016 aaronmk
def _compass(items, node):
347 1607 aaronmk
    '''Converts a compass direction (N, NE, NNE, etc.) into a degree heading'''
348
    items = dict(conv_items(strings.ustr, items))
349
    try: value = items['value']
350
    except KeyError: return None # input is empty
351
352
    if not value.isupper(): return value # pass through other coordinate formats
353
    try: return util.cast(str, angles.compass2heading(value)) # ignore None
354
    except KeyError, e: raise FormatException(e)
355
funcs['_compass'] = _compass