Revision 316
Added by Aaron Marcuse-Kubitza about 13 years ago
| map | ||
|---|---|---|
| 75 | 75 |
|
| 76 | 76 |
if in_is_xml: doc0 = minidom.parse(sys.stdin) |
| 77 | 77 |
|
| 78 |
def process_input(use_row):
|
|
| 78 |
def process_input(root, process_row):
|
|
| 79 | 79 |
'''Inputs datasource to XML tree, mapping if needed''' |
| 80 |
doc1 = xml_dom.create_doc(out_label) |
|
| 81 |
root = doc1.documentElement |
|
| 82 |
|
|
| 83 | 80 |
def process_rows(get_value, rows): |
| 84 | 81 |
'''Processes input values |
| 85 | 82 |
@param get_value f(in_, row):str |
| 86 | 83 |
''' |
| 87 | 84 |
for i, row in enumerate(rows): |
| 88 |
try: |
|
| 89 |
if not (limit == None or i < limit): break |
|
| 90 |
row_id = str(i) |
|
| 91 |
for in_, out in mappings: |
|
| 92 |
value = metadata_value(in_) |
|
| 93 |
if value == None: value = get_value(in_, row) |
|
| 94 |
if value != None: |
|
| 95 |
xpath.put_obj(root, out, row_id, has_types, value) |
|
| 96 |
except Exception: traceback.print_exc() |
|
| 85 |
if not (limit == None or i < limit): break |
|
| 86 |
row_id = str(i) |
|
| 87 |
for in_, out in mappings: |
|
| 88 |
value = metadata_value(in_) |
|
| 89 |
if value == None: value = get_value(in_, row) |
|
| 90 |
if value != None: |
|
| 91 |
xpath.put_obj(root, out, row_id, has_types, value) |
|
| 92 |
process_row() |
|
| 97 | 93 |
|
| 98 | 94 |
if map_path == None: |
| 99 | 95 |
iter_ = xml_dom.NodeElemIter(doc0.documentElement) |
| 100 | 96 |
util.skip(iter_, xml_dom.is_text) # skip metadata |
| 101 |
#map(use_row, iter_)
|
|
| 97 |
map(process_row, iter_)
|
|
| 102 | 98 |
return doc0 |
| 103 | 99 |
elif in_is_db: |
| 104 | 100 |
assert in_is_xpaths |
| ... | ... | |
| 146 | 142 |
if value != '': return value |
| 147 | 143 |
else: return None |
| 148 | 144 |
process_rows(get_value, reader) |
| 149 |
xml_func.process(root) |
|
| 150 |
return doc1 |
|
| 151 | 145 |
|
| 152 | 146 |
# Output XML tree |
| 147 |
doc = xml_dom.create_doc(out_label) |
|
| 148 |
root = doc.documentElement |
|
| 153 | 149 |
if out_is_db: |
| 154 | 150 |
from psycopg2.extensions import ISOLATION_LEVEL_SERIALIZABLE |
| 155 | 151 |
import db_xml |
| ... | ... | |
| 159 | 155 |
out_pkeys = {}
|
| 160 | 156 |
try: |
| 161 | 157 |
row_ct_ref = [0] |
| 162 |
def use_row(root): |
|
| 163 |
try: |
|
| 164 |
db_xml.put(out_db, root, False, row_ct_ref, out_pkeys) |
|
| 165 |
if commit: db.commit() |
|
| 166 |
except Exception: |
|
| 167 |
out_db.rollback() |
|
| 168 |
traceback.print_exc() |
|
| 169 |
map(use_row, process_input(use_row).firstChild) |
|
| 158 |
def process_row(): |
|
| 159 |
try: xml_func.process(root) |
|
| 160 |
except xml_func.SyntaxException: traceback.print_exc() |
|
| 161 |
else: |
|
| 162 |
assert xml_dom.has_one_child(root) |
|
| 163 |
child = root.firstChild |
|
| 164 |
try: |
|
| 165 |
db_xml.put(out_db, child, False, row_ct_ref, out_pkeys) |
|
| 166 |
if commit: db.commit() |
|
| 167 |
except Exception: |
|
| 168 |
out_db.rollback() |
|
| 169 |
traceback.print_exc() |
|
| 170 |
root.clear() |
|
| 171 |
process_input(root, process_row) |
|
| 170 | 172 |
print 'Inserted '+str(row_ct_ref[0])+' rows' |
| 171 | 173 |
finally: |
| 172 | 174 |
out_db.rollback() |
| 173 | 175 |
out_db.close() |
| 174 | 176 |
else: # output is XML |
| 175 |
output = sys.stdout |
|
| 176 |
config = xml_dom.prettyxml_config |
|
| 177 |
doc = xml_dom.create_doc(out_label) |
|
| 178 |
doc.write_opening(output, **config) |
|
| 179 |
def use_row(root): root.writexml(output, indent=1, **config) |
|
| 180 |
map(use_row, process_input(use_row).firstChild) |
|
| 181 |
doc.write_closing(output, **config) |
|
| 177 |
def process_row(): pass |
|
| 178 |
process_input(root, process_row) |
|
| 179 |
xml_func.process(root) |
|
| 180 |
doc.writexml(sys.stdout, **xml_dom.prettyxml_config) |
|
| 182 | 181 |
|
| 183 | 182 |
try: main() |
| 184 | 183 |
except Parser.SyntaxException, e: raise SystemExit(str(e)) |
Also available in: Unified diff
bin/map: Refactored to process each row separately when out_is_db