Revision 141
Added by Aaron Marcuse-Kubitza almost 13 years ago
| scripts/test/accepted_output/SALVIAS_db.organisms.VegX.xml | ||
|---|---|---|
| 30 | 30 |
</simpleUserdefined> |
| 31 | 31 |
<simpleUserdefined> |
| 32 | 32 |
<name>individualCode</name> |
| 33 |
<value>1</value>
|
|
| 33 |
<value>2</value>
|
|
| 34 | 34 |
</simpleUserdefined> |
| 35 | 35 |
<simpleUserdefined> |
| 36 | 36 |
<name>habit</name> |
| ... | ... | |
| 55 | 55 |
<voucher>1</voucher> |
| 56 | 56 |
</taxonNameUsageConcept> |
| 57 | 57 |
<taxonNameUsageConcept id="1"> |
| 58 |
<voucher>1</voucher>
|
|
| 58 |
<voucher>2</voucher>
|
|
| 59 | 59 |
</taxonNameUsageConcept> |
| 60 | 60 |
</taxonNameUsageConcepts> |
| 61 | 61 |
<plotObservations> |
| ... | ... | |
| 198 | 198 |
</taxonConcept> |
| 199 | 199 |
<taxonConcept id="5"> |
| 200 | 200 |
<Rank code="fam"/> |
| 201 |
<Name>Arecaceae</Name>
|
|
| 201 |
<Name>Clusiaceae</Name>
|
|
| 202 | 202 |
</taxonConcept> |
| 203 | 203 |
<taxonConcept id="6"> |
| 204 | 204 |
<Rank code="gen"/> |
| 205 |
<Name>Geonoma</Name>
|
|
| 205 |
<Name>Tovomitopsis</Name>
|
|
| 206 | 206 |
</taxonConcept> |
| 207 | 207 |
<taxonConcept id="7"> |
| 208 | 208 |
<Rank code="sp"/> |
| 209 |
<Name>seleri</Name>
|
|
| 209 |
<Name>allenii</Name>
|
|
| 210 | 210 |
<AccordingTo> |
| 211 |
<Simple>Burret</Simple>
|
|
| 211 |
<Simple>Maguire</Simple>
|
|
| 212 | 212 |
</AccordingTo> |
| 213 | 213 |
</taxonConcept> |
| 214 | 214 |
<taxonConcept id="8"> |
| 215 | 215 |
<Rank code="fam"/> |
| 216 |
<Name>Arecaceae</Name>
|
|
| 216 |
<Name>Clusiaceae</Name>
|
|
| 217 | 217 |
</taxonConcept> |
| 218 | 218 |
<taxonConcept id="9"> |
| 219 | 219 |
<Rank code="sp"/> |
| 220 |
<Name>seleri</Name>
|
|
| 220 |
<Name>allenii</Name>
|
|
| 221 | 221 |
</taxonConcept> |
| 222 | 222 |
</taxonConcepts> |
| 223 | 223 |
</VegX> |
| scripts/test/accepted_output/SALVIAS_db.organisms.VegBank.xml | ||
|---|---|---|
| 289 | 289 |
</reference_ID> |
| 290 | 290 |
<plantName_ID> |
| 291 | 291 |
<plantName> |
| 292 |
<plantName>seleri</plantName>
|
|
| 292 |
<plantName>allenii</plantName>
|
|
| 293 | 293 |
</plantName> |
| 294 | 294 |
</plantName_ID> |
| 295 | 295 |
</plantConcept> |
| ... | ... | |
| 297 | 297 |
<party_ID> |
| 298 | 298 |
<party/> |
| 299 | 299 |
</party_ID> |
| 300 |
<museumAccessionNumber>1</museumAccessionNumber>
|
|
| 300 |
<museumAccessionNumber>2</museumAccessionNumber>
|
|
| 301 | 301 |
<definedValue fkey="tableRecord_ID"> |
| 302 | 302 |
<userDefined_ID> |
| 303 | 303 |
<userDefined> |
| ... | ... | |
| 316 | 316 |
<userDefinedType>varchar</userDefinedType> |
| 317 | 317 |
</userDefined> |
| 318 | 318 |
</userDefined_ID> |
| 319 |
<definedValue>1</definedValue>
|
|
| 319 |
<definedValue>2</definedValue>
|
|
| 320 | 320 |
</definedValue> |
| 321 | 321 |
<definedValue fkey="tableRecord_ID"> |
| 322 | 322 |
<userDefined_ID> |
| ... | ... | |
| 360 | 360 |
</reference_ID> |
| 361 | 361 |
<plantName_ID> |
| 362 | 362 |
<plantName> |
| 363 |
<plantName>Arecaceae</plantName>
|
|
| 363 |
<plantName>Clusiaceae</plantName>
|
|
| 364 | 364 |
</plantName> |
| 365 | 365 |
</plantName_ID> |
| 366 | 366 |
</plantConcept> |
| ... | ... | |
| 394 | 394 |
</reference_ID> |
| 395 | 395 |
<plantName_ID> |
| 396 | 396 |
<plantName> |
| 397 |
<plantName>Geonoma</plantName>
|
|
| 397 |
<plantName>Tovomitopsis</plantName>
|
|
| 398 | 398 |
</plantName> |
| 399 | 399 |
</plantName_ID> |
| 400 | 400 |
</plantConcept> |
| ... | ... | |
| 428 | 428 |
</reference_ID> |
| 429 | 429 |
<plantName_ID> |
| 430 | 430 |
<plantName> |
| 431 |
<plantName>seleri</plantName>
|
|
| 431 |
<plantName>allenii</plantName>
|
|
| 432 | 432 |
<reference_ID> |
| 433 | 433 |
<reference> |
| 434 |
<shortName>Burret</shortName>
|
|
| 434 |
<shortName>Maguire</shortName>
|
|
| 435 | 435 |
</reference> |
| 436 | 436 |
</reference_ID> |
| 437 | 437 |
</plantName> |
| ... | ... | |
| 472 | 472 |
</reference_ID> |
| 473 | 473 |
<plantName_ID> |
| 474 | 474 |
<plantName> |
| 475 |
<plantName>Arecaceae</plantName>
|
|
| 475 |
<plantName>Clusiaceae</plantName>
|
|
| 476 | 476 |
</plantName> |
| 477 | 477 |
</plantName_ID> |
| 478 | 478 |
</plantConcept> |
| ... | ... | |
| 506 | 506 |
</reference_ID> |
| 507 | 507 |
<plantName_ID> |
| 508 | 508 |
<plantName> |
| 509 |
<plantName>seleri</plantName>
|
|
| 509 |
<plantName>allenii</plantName>
|
|
| 510 | 510 |
</plantName> |
| 511 | 511 |
</plantName_ID> |
| 512 | 512 |
</plantConcept> |
| scripts/lib/xpath.py | ||
|---|---|---|
| 18 | 18 |
|
| 19 | 19 |
def __repr__(self): |
| 20 | 20 |
str_ = '' |
| 21 |
if self.is_positive: str_ += '!' |
|
| 21 |
if not self.is_positive: str_ += '!'
|
|
| 22 | 22 |
if self.is_attr: str_ += '@' |
| 23 | 23 |
str_ += self.name |
| 24 | 24 |
if self.keys != []: str_ += repr(self.keys) |
| ... | ... | |
| 135 | 135 |
|
| 136 | 136 |
def is_instance(elem): return elem.keys != [] and is_id(elem.keys[0]) |
| 137 | 137 |
|
| 138 |
def get(doc, xpath, create=False, last_only=None, parent=None): |
|
| 139 |
# Warning: The last_only optimization may put data that should be together |
|
| 140 |
# into separate nodes |
|
| 141 |
if parent == None: parent = doc.documentElement |
|
| 138 |
def get(parent, xpath, create=False, last_only=None): |
|
| 139 |
'''Warning: The last_only optimization may put data that should be together |
|
| 140 |
into separate nodes''' |
|
| 142 | 141 |
if last_only == None: last_only = create |
| 143 | 142 |
|
| 144 | 143 |
if create and not is_positive(xpath): return None |
| 145 | 144 |
doc = parent.ownerDocument |
| 145 |
root = doc.documentElement |
|
| 146 | 146 |
for elem_idx, elem in enumerate(xpath): |
| 147 | 147 |
# Find possible matches |
| 148 | 148 |
children = [] |
| ... | ... | |
| 160 | 160 |
is_match = elem.value == None or xml_dom.value(child) == elem.value |
| 161 | 161 |
for attr in elem.keys: |
| 162 | 162 |
if not is_match: break |
| 163 |
is_match = (get(doc, attr, False, last_only, child) != None)\
|
|
| 163 |
is_match = (get(child, attr, False, last_only) != None)\
|
|
| 164 | 164 |
== is_positive(attr) |
| 165 | 165 |
if is_match: node = child; break |
| 166 | 166 |
|
| ... | ... | |
| 174 | 174 |
if elem.value != None: xml_dom.set_value(doc, node, elem.value) |
| 175 | 175 |
if create: |
| 176 | 176 |
for attr in elem.keys + elem.attrs: |
| 177 |
get(doc, attr, create, last_only, node)
|
|
| 177 |
get(node, attr, create, last_only)
|
|
| 178 | 178 |
|
| 179 | 179 |
for branch in elem.other_branches: |
| 180 | 180 |
branch = copy.deepcopy(branch) |
| 181 | 181 |
set_value(branch, value(xpath)) |
| 182 |
get(doc, branch, create, last_only, node)
|
|
| 182 |
get(node, branch, create, last_only)
|
|
| 183 | 183 |
|
| 184 | 184 |
# Follow pointer |
| 185 | 185 |
if elem.is_ptr: |
| ... | ... | |
| 191 | 191 |
else: # forward (parent-to-child) pointer |
| 192 | 192 |
id_ = xml_dom.value(node) |
| 193 | 193 |
obj_xpath = obj(xpath) # target object |
| 194 |
if id_ == None or get(doc, obj_xpath, False, True) == None:
|
|
| 194 |
if id_ == None or get(root, obj_xpath, False, True) == None:
|
|
| 195 | 195 |
# no target or target keys don't match |
| 196 | 196 |
if not create: return None |
| 197 | 197 |
|
| 198 | 198 |
# Use last target object's ID + 1 |
| 199 | 199 |
obj_xpath[-1].keys = [] # just get by tag name |
| 200 |
last = get(doc, obj_xpath, False, True)
|
|
| 200 |
last = get(root, obj_xpath, False, True)
|
|
| 201 | 201 |
if last != None: id_ = str(int(xml_dom.get_id(last)) + 1) |
| 202 | 202 |
else: id_ = '0' |
| 203 | 203 |
|
| ... | ... | |
| 206 | 206 |
xml_dom.set_value(doc, node, id_) |
| 207 | 207 |
else: last_only = False |
| 208 | 208 |
set_id(xpath, id_) |
| 209 |
return get(doc, xpath, create, last_only)
|
|
| 209 |
return get(root, xpath, create, last_only)
|
|
| 210 | 210 |
|
| 211 | 211 |
parent = node |
| 212 | 212 |
return parent |
| 213 | 213 |
|
| 214 |
def put_obj(doc, xpath, id_, has_types, value=None):
|
|
| 214 |
def put_obj(root, xpath, id_, has_types, value=None):
|
|
| 215 | 215 |
xpath = copy.deepcopy(xpath) # don't modify input! |
| 216 | 216 |
set_id(xpath, id_, has_types) |
| 217 | 217 |
if value != None: set_value(xpath, value) |
| 218 |
get(doc, xpath, True)
|
|
| 218 |
get(root, xpath, True)
|
|
| 219 | 219 |
|
| 220 | 220 |
def path2xml(xpath, first_branch=True): |
| 221 | 221 |
root = xml_dom.create_doc().documentElement |
| 222 |
get(root.ownerDocument, xpath, True) |
|
| 223 |
if first_branch: root = root.firstChild |
|
| 222 |
get(root, xpath, True) |
|
| 224 | 223 |
return root |
| 225 | 224 |
|
| 226 | 225 |
def str2xml(xpath): return path2xml(parse(xpath)) |
| 226 |
|
|
| 227 |
def xml_set_id(root, id_): xml_dom.set_id(root.firstChild, id_) |
|
| scripts/lib/db_xml.py | ||
|---|---|---|
| 37 | 37 |
if pkeys == None: pkeys = {}
|
| 38 | 38 |
def pkey(table): return sql.pkey(db, pkeys, table) |
| 39 | 39 |
|
| 40 |
node = node.firstChild |
|
| 40 | 41 |
table = name_of(node) |
| 41 | 42 |
pkey_ = pkey(table) |
| 42 | 43 |
|
| scripts/map | ||
|---|---|---|
| 79 | 79 |
if in_is_xml: doc0 = xml.dom.minidom.parse(sys.stdin) |
| 80 | 80 |
if map_path != None: |
| 81 | 81 |
doc1 = xml_dom.create_doc(dest) |
| 82 |
root = doc1.documentElement |
|
| 82 | 83 |
if in_is_db: |
| 83 | 84 |
assert in_is_xpaths |
| 84 | 85 |
|
| ... | ... | |
| 97 | 98 |
row_id = str(row_id) |
| 98 | 99 |
|
| 99 | 100 |
def put_col(path, value): |
| 100 |
xpath.put_obj(doc1, path, row_id, has_types, value)
|
|
| 101 |
xpath.put_obj(root, path, row_id, has_types, value)
|
|
| 101 | 102 |
for value, out in metadata: put_col(out, value) |
| 102 | 103 |
for in_, out in mappings: |
| 103 |
root = xpath.get(in_.ownerDocument, src_root) |
|
| 104 |
xml_dom.replace(root, root.cloneNode(False)) |
|
| 105 |
xml_dom.set_id(root, row_id) |
|
| 106 |
|
|
| 104 |
in_ = in_.cloneNode(True) # don't modify orig value! |
|
| 105 |
xml_dom.set_id(xpath.get(in_, src_root), row_id) |
|
| 107 | 106 |
value = sql.value_or_none(db_xml.get(in_db, in_, in_pkeys)) |
| 108 | 107 |
if value != None: put_col(out, str(value)) |
| 109 | 108 |
in_db.close() |
| ... | ... | |
| 117 | 116 |
row_id = str(row_idx) |
| 118 | 117 |
|
| 119 | 118 |
def put_col(path, value): |
| 120 |
xpath.put_obj(doc1, path, row_id, has_types, value)
|
|
| 119 |
xpath.put_obj(root, path, row_id, has_types, value)
|
|
| 121 | 120 |
for value, out in metadata: put_col(out, value) |
| 122 | 121 |
for i, col in enumerate(cols): |
| 123 | 122 |
if row[i] != '' and col in map_: put_col(map_[col], row[i]) |
Also available in: Unified diff
xpath.py: Refactored to avoid needing a doc parameter for the XML document