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 1369 aaronmk
import sys
7 111 aaronmk
8 818 aaronmk
import dates
9 300 aaronmk
import exc
10 917 aaronmk
import maps
11 1234 aaronmk
import strings
12 827 aaronmk
import term
13 1047 aaronmk
import util
14 86 aaronmk
import xml_dom
15 1321 aaronmk
import xpath
16 86 aaronmk
17 995 aaronmk
##### Exceptions
18
19 962 aaronmk
class SyntaxException(Exception):
20 797 aaronmk
    def __init__(self, cause):
21 962 aaronmk
        Exception.__init__(self, 'Invalid XML function syntax: '
22
            +exc.str_(cause))
23 278 aaronmk
24 843 aaronmk
class FormatException(SyntaxException): pass
25
26 995 aaronmk
##### Functions
27
28
funcs = {}
29
30
def process(node, on_error=exc.raise_):
31
    for child in xml_dom.NodeElemIter(node): process(child, on_error)
32
    name = node.tagName
33
    if name.startswith('_') and name in funcs:
34 1369 aaronmk
        try:
35
            value = funcs[name](xml_dom.NodeTextEntryIter(node))
36
            xml_dom.replace_with_text(node, value)
37
        except Exception, e: # also catch XML func internal errors
38 1371 aaronmk
            # Save in case another exception raised, overwriting sys.exc_info()
39
            exc.add_traceback(e)
40 995 aaronmk
            str_ = str(node)
41
            exc.add_msg(e, 'function:\n'+str_)
42
            xml_dom.replace(node, node.ownerDocument.createComment(
43 1234 aaronmk
                '\n'+term.emph_multiline(str_).replace('--','-')))
44
                # comments can't contain '--'
45 995 aaronmk
            on_error(e)
46
47 86 aaronmk
def map_items(func, items):
48
    return [(name, func(value)) for name, value in items]
49
50 1234 aaronmk
def cast(type_, val):
51
    '''Throws SyntaxException if can't cast'''
52
    try: return type_(val)
53
    except ValueError, e: raise SyntaxException(e)
54
55 278 aaronmk
def conv_items(type_, items):
56 1234 aaronmk
    return map_items(lambda val: cast(type_, val),
57
        xml_dom.TextEntryOnlyIter(items))
58 278 aaronmk
59 995 aaronmk
#### XML functions
60
61
# Function names must start with _ to avoid collisions with real tags
62
# Functions take arguments (items)
63
64
def _ignore(items):
65 994 aaronmk
    '''Used to "comment out" an XML subtree'''
66
    return None
67 995 aaronmk
funcs['_ignore'] = _ignore
68 994 aaronmk
69 1234 aaronmk
def _eq(items):
70
    items = dict(items)
71
    try:
72
        left = items['left']
73
        right = items['right']
74
    except KeyError: return '' # a value was None
75
    return util.bool2str(left == right)
76
funcs['_eq'] = _eq
77
78
def _if(items):
79
    items = dict(items)
80
    try:
81
        cond = items['cond']
82
        then = items['then']
83
    except KeyError, e: raise SyntaxException(e)
84
    else_ = items.get('else', None)
85
    cond = bool(cast(str, cond))
86
    if cond: return then
87
    else: return else_
88
funcs['_if'] = _if
89
90 995 aaronmk
def _alt(items):
91 113 aaronmk
    items = list(items)
92
    items.sort()
93 1186 aaronmk
    try: return items[0][1] # value of lowest-numbered item
94 1187 aaronmk
    except IndexError: return None # input got removed by e.g. SyntaxException
95 995 aaronmk
funcs['_alt'] = _alt
96 113 aaronmk
97 995 aaronmk
def _merge(items):
98 1234 aaronmk
    items = list(conv_items(strings.ustr, items))
99
        # get *once* from iter and check types
100 917 aaronmk
    items.sort()
101
    return maps.merge_values(*[v for k, v in items])
102 995 aaronmk
funcs['_merge'] = _merge
103 917 aaronmk
104 995 aaronmk
def _label(items):
105 917 aaronmk
    items = dict(conv_items(str, items)) # get *once* from iter and check types
106
    try:
107
        label = items['label']
108
        value = items['value']
109
    except KeyError, e: raise SyntaxException(e)
110
    return label+': '+value
111 995 aaronmk
funcs['_label'] = _label
112 917 aaronmk
113 1047 aaronmk
def _nullIf(items):
114
    items = dict(conv_items(str, items))
115
    try:
116
        null = items['null']
117
        value = items['value']
118
    except KeyError, e: raise SyntaxException(e)
119 1219 aaronmk
    type_str = items.get('type', None)
120
    type_ = str
121
    if type_str == 'float': type_ = float
122
    return util.none_if(value, type_(null))
123 1047 aaronmk
funcs['_nullIf'] = _nullIf
124
125 1219 aaronmk
def _map(items):
126
    items = conv_items(str, items) # get *once* from iter and check types
127
    try: value = items.pop()[1] # value is last entry's value
128
    except IndexError, e: raise SyntaxException(e)
129
    map_ = dict(items)
130 1304 aaronmk
    closed = bool(map_.pop('_closed', False))
131 1219 aaronmk
    try: return map_[value]
132 1304 aaronmk
    except KeyError, e:
133
        if closed: raise SyntaxException(e)
134
        else: return value
135 1219 aaronmk
funcs['_map'] = _map
136
137
def _replace(items):
138
    items = conv_items(str, items) # get *once* from iter and check types
139
    try: value = items.pop() # value is last entry
140
    except IndexError, e: raise SyntaxException(e)
141
    try:
142
        for repl, with_ in items:
143
            if re.match(r'^\w+$', repl):
144
                repl = r'(?<![^\W_])'+repl+r'(?![^\W_])' # match whole word
145
            value = re.sub(repl, with_, value)
146
    except sre_constants.error, e: raise SyntaxException(e)
147
    return value
148
funcs['_replace'] = _replace
149
150 1225 aaronmk
def _units(items):
151
    items = dict(conv_items(str, items))
152
    try:
153
        units = items['units']
154
        value = items['value']
155
    except KeyError, e: raise SyntaxException(e)
156
    return value#+' '+units # don't add yet because unit conversion isn't ready
157
funcs['_units'] = _units
158
159 995 aaronmk
def _range(items):
160 278 aaronmk
    items = dict(conv_items(float, items))
161 965 aaronmk
    from_ = items.get('from', None)
162
    to = items.get('to', None)
163
    if from_ == None or to == None: return None
164 326 aaronmk
    return str(to - from_)
165 995 aaronmk
funcs['_range'] = _range
166 86 aaronmk
167 1399 aaronmk
def parse_range(str_, range_sep='-'):
168
    default = (str_, None)
169
    start, sep, end = str_.partition(range_sep)
170
    if sep == '': return default # not a range
171
    return tuple(d.strip() for d in (start, end))
172
173
def _rangeStart(items):
174
    items = dict(conv_items(str, items))
175
    try: value = items['value']
176 1406 aaronmk
    except KeyError: return None # input is empty
177 1399 aaronmk
    return parse_range(value)[0]
178
funcs['_rangeStart'] = _rangeStart
179
180
def _rangeEnd(items):
181
    items = dict(conv_items(str, items))
182
    try: value = items['value']
183 1406 aaronmk
    except KeyError: return None # input is empty
184 1399 aaronmk
    return parse_range(value)[1]
185
funcs['_rangeEnd'] = _rangeEnd
186
187 995 aaronmk
def _avg(items):
188 86 aaronmk
    count = 0
189
    sum_ = 0.
190 278 aaronmk
    for name, value in conv_items(float, items):
191 86 aaronmk
        count += 1
192
        sum_ += value
193
    return str(sum_/count)
194 995 aaronmk
funcs['_avg'] = _avg
195 86 aaronmk
196 968 aaronmk
class CvException(Exception):
197
    def __init__(self):
198
        Exception.__init__(self, 'CV (coefficient of variation) values are only'
199
            ' allowed for ratio scale data '
200
            '(see <http://en.wikipedia.org/wiki/Coefficient_of_variation>)')
201
202 995 aaronmk
def _noCV(items):
203 968 aaronmk
    try: name, value = items.next()
204
    except StopIteration: return None
205
    if re.match('^(?i)CV *\d+$', value): raise SyntaxException(CvException())
206
    return value
207 995 aaronmk
funcs['_noCV'] = _noCV
208 968 aaronmk
209 995 aaronmk
def _date(items):
210 917 aaronmk
    items = conv_items(str, items) # get *once* from iter and check types
211 786 aaronmk
    try: str_ = dict(items)['date']
212
    except KeyError:
213 1308 aaronmk
        items = dict(conv_items(int, items))
214 1292 aaronmk
        try: items['year'] # year is required
215 1309 aaronmk
        except KeyError, e:
216
            if items == {}: return None # entire date is empty
217
            else: raise SyntaxException(e)
218 786 aaronmk
        items.setdefault('month', 1)
219
        items.setdefault('day', 1)
220
        try: date = datetime.date(**items)
221
        except ValueError, e: raise SyntaxException(e)
222
    else:
223 324 aaronmk
        try: year = float(str_)
224
        except ValueError:
225 1264 aaronmk
            try: date = dates.strtotime(str_)
226 324 aaronmk
            except ImportError: return str_
227
            except ValueError, e: raise SyntaxException(e)
228
        else: date = (datetime.date(int(year), 1, 1) +
229
            datetime.timedelta(round((year % 1.)*365)))
230 818 aaronmk
    try: return dates.strftime('%Y-%m-%d', date)
231 843 aaronmk
    except ValueError, e: raise FormatException(e)
232 995 aaronmk
funcs['_date'] = _date
233 86 aaronmk
234 1366 aaronmk
def _dateRangeStart(items):
235
    items = dict(conv_items(str, items))
236
    try: value = items['value']
237 1406 aaronmk
    except KeyError: return None # input is empty
238 1366 aaronmk
    return dates.parse_date_range(value)[0]
239
funcs['_dateRangeStart'] = _dateRangeStart
240 1311 aaronmk
241 1366 aaronmk
def _dateRangeEnd(items):
242 1311 aaronmk
    items = dict(conv_items(str, items))
243 1366 aaronmk
    try: value = items['value']
244 1406 aaronmk
    except KeyError: return None # input is empty
245 1366 aaronmk
    return dates.parse_date_range(value)[1]
246
funcs['_dateRangeEnd'] = _dateRangeEnd
247 1311 aaronmk
248 328 aaronmk
_name_parts_slices_items = [
249
    ('first', slice(None, 1)),
250
    ('middle', slice(1, -1)),
251
    ('last', slice(-1, None)),
252
]
253
name_parts_slices = dict(_name_parts_slices_items)
254
name_parts = [name for name, slice_ in _name_parts_slices_items]
255
256 995 aaronmk
def _name(items):
257 89 aaronmk
    items = dict(items)
258 102 aaronmk
    parts = []
259 328 aaronmk
    for part in name_parts:
260
        if part in items: parts.append(items[part])
261 102 aaronmk
    return ' '.join(parts)
262 995 aaronmk
funcs['_name'] = _name
263 102 aaronmk
264 995 aaronmk
def _namePart(items):
265 328 aaronmk
    out_items = []
266
    for part, value in items:
267
        try: slice_ = name_parts_slices[part]
268
        except KeyError, e: raise SyntaxException(e)
269 1219 aaronmk
        out_items.append((part, ' '.join(value.split(' ')[slice_])))
270 995 aaronmk
    return _name(out_items)
271
funcs['_namePart'] = _namePart
272 1321 aaronmk
273
def _simplifyPath(items):
274
    items = dict(items)
275
    try:
276
        next = cast(str, items['next'])
277
        require = cast(str, items['require'])
278
        root = items['path']
279
    except KeyError, e: raise SyntaxException(e)
280
281
    node = root
282
    while node != None:
283
        new_node = xpath.get_1(node, next, allow_rooted=False)
284
        if xpath.get_1(node, require, allow_rooted=False) == None: # empty elem
285
            xml_dom.replace(node, new_node) # remove current elem
286
            if node is root: root = new_node # also update root
287
        node = new_node
288
    return root
289
funcs['_simplifyPath'] = _simplifyPath