mirror of
https://github.com/gsi-upm/sitc
synced 2024-11-25 07:52:27 +00:00
130 lines
3.9 KiB
Python
130 lines
3.9 KiB
Python
'''
|
|
Helper functions and ipython magic for the SPARQL exercises.
|
|
|
|
The tests in the notebooks rely on the `LAST_QUERY` variable, which is updated by the `%%sparql` magic after every query.
|
|
This variable contains the full query used (`LAST_QUERY["query"]`), the endpoint it was sent to (`LAST_QUERY["endpoint"]`), and a dictionary with the response of the endpoint (`LAST_QUERY["results"]`).
|
|
For convenience, the results are also given as tuples (`LAST_QUERY["tuples"]`), and as a dictionary of of `{column:[values]}` (`LAST_QUERY["columns"]`).
|
|
'''
|
|
from IPython.core.magic import (register_line_magic, register_cell_magic,
|
|
register_line_cell_magic)
|
|
from IPython.display import HTML, display, Image, display_javascript
|
|
from urllib.request import Request, urlopen
|
|
from urllib.parse import quote_plus, urlencode
|
|
from urllib.error import HTTPError
|
|
|
|
import json
|
|
import sys
|
|
|
|
js = "IPython.CodeCell.options_default.highlight_modes['magic_sparql'] = {'reg':[/^%%sparql/]};"
|
|
display_javascript(js, raw=True)
|
|
|
|
|
|
def send_query(query, endpoint):
|
|
FORMATS = ",".join(["application/sparql-results+json", "text/javascript", "application/json"])
|
|
|
|
data = {'query': query}
|
|
# b = quote_plus(query)
|
|
|
|
r = Request(endpoint,
|
|
data=urlencode(data).encode('utf-8'),
|
|
headers={'content-type': 'application/x-www-form-urlencoded',
|
|
'accept': FORMATS},
|
|
method='POST')
|
|
res = urlopen(r)
|
|
data = res.read().decode('utf-8')
|
|
if res.getcode() == 200:
|
|
return json.loads(data)
|
|
raise Exception('Error getting results: {}'.format(data))
|
|
|
|
|
|
def tabulate(tuples, header=None):
|
|
if not header:
|
|
header, tuples = tuples[0], tuples[1:]
|
|
header = '<tr>{}<tr>'.format(''.join('<th>{}</th>'.format(h) for h in header))
|
|
rows = []
|
|
for row in tuples:
|
|
inner = ''.join('<td>{}</td>'.format(c) for c in row)
|
|
rows.append('<tr>{}</tr>'.format(inner))
|
|
body = ''.join(rows)
|
|
table = '<table><thead>{header}</thead><tbody>{body}</tbody></table>'.format(body=body,
|
|
header=header)
|
|
return table
|
|
|
|
|
|
LAST_QUERY = {}
|
|
|
|
def solution():
|
|
return LAST_QUERY
|
|
|
|
|
|
def query(query, endpoint=None, print_table=False):
|
|
global LAST_QUERY
|
|
|
|
endpoint = endpoint or "http://fuseki.cluster.gsi.dit.upm.es/sitc/"
|
|
results = send_query(query, endpoint)
|
|
tuples = to_table(results)
|
|
|
|
|
|
columns = {}
|
|
header, values = tuples[0], tuples[1:]
|
|
|
|
for ix, h in enumerate(header):
|
|
columns[h] = []
|
|
for row in values:
|
|
columns[h].append(row[ix])
|
|
|
|
LAST_QUERY.update({
|
|
'query': query,
|
|
'endpoint': query,
|
|
'results': results,
|
|
'tuples': values,
|
|
'columns': columns
|
|
})
|
|
|
|
if not print_table:
|
|
return tuples
|
|
return HTML(tabulate(tuples))
|
|
|
|
|
|
def to_table(results):
|
|
table = []
|
|
header = results['head']['vars']
|
|
table.append(header)
|
|
for result in results["results"]["bindings"]:
|
|
table.append(tuple(result.get(h, {}).get('value', "") for h in header))
|
|
return table
|
|
|
|
|
|
@register_cell_magic
|
|
def sparql(line, cell):
|
|
'''
|
|
Sparql magic command for ipython. It can be used in a cell like this:
|
|
|
|
```
|
|
%%sparql
|
|
|
|
... Your SPARQL query ...
|
|
|
|
```
|
|
|
|
by default, it will use the DBpedia endpoint, but you can use a different endpoint like this:
|
|
|
|
```
|
|
%%sparql http://my-sparql-endpoint...
|
|
|
|
... Your SPARQL query ...
|
|
```
|
|
'''
|
|
try:
|
|
return query(cell, endpoint=line, print_table=True)
|
|
except HTTPError as ex:
|
|
error_message = ex.read().decode('utf-8')
|
|
print('Error {}. Reason: {}'.format(ex.status, ex.reason))
|
|
print(error_message, file=sys.stderr)
|
|
|
|
|
|
def show_photos(values):
|
|
for value in values:
|
|
if 'http://' in value:
|
|
display(Image(url=value))
|