1 """ Steps that run search queries.
3 Queries may either be run directly via PHP using the query script
4 or via the HTTP interface.
12 from tidylib import tidy_document
13 import xml.etree.ElementTree as ET
15 from urllib.parse import urlencode
16 from collections import OrderedDict
17 from nose.tools import * # for assert functions
19 logger = logging.getLogger(__name__)
22 'HTTP_HOST' : 'localhost',
23 'HTTP_USER_AGENT' : 'Mozilla/5.0 (X11; Linux x86_64; rv:51.0) Gecko/20100101 Firefox/51.0',
24 'HTTP_ACCEPT' : 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
25 'HTTP_ACCEPT_ENCODING' : 'gzip, deflate',
26 'HTTP_CONNECTION' : 'keep-alive',
27 'SERVER_SIGNATURE' : '<address>Nominatim BDD Tests</address>',
28 'SERVER_SOFTWARE' : 'Nominatim test',
29 'SERVER_NAME' : 'localhost',
30 'SERVER_ADDR' : '127.0.1.1',
32 'REMOTE_ADDR' : '127.0.0.1',
33 'DOCUMENT_ROOT' : '/var/www',
34 'REQUEST_SCHEME' : 'http',
35 'CONTEXT_PREFIX' : '/',
36 'SERVER_ADMIN' : 'webmaster@localhost',
37 'REMOTE_PORT' : '49319',
38 'GATEWAY_INTERFACE' : 'CGI/1.1',
39 'SERVER_PROTOCOL' : 'HTTP/1.1',
40 'REQUEST_METHOD' : 'GET',
41 'REDIRECT_STATUS' : 'CGI'
45 def compare(operator, op1, op2):
46 if operator == 'less than':
48 elif operator == 'more than':
50 elif operator == 'exactly':
52 elif operator == 'at least':
54 elif operator == 'at most':
57 raise Exception("unknown operator '%s'" % operator)
59 class GenericResponse(object):
61 def match_row(self, row):
62 if 'ID' in row.headings:
63 todo = [int(row['ID'])]
65 todo = range(len(self.result))
69 for h in row.headings:
73 assert_equal(res['osm_type'], row[h][0])
74 assert_equal(res['osm_id'], int(row[h][1:]))
76 x, y = row[h].split(' ')
77 assert_almost_equal(float(y), float(res['lat']))
78 assert_almost_equal(float(x), float(res['lon']))
79 elif row[h].startswith("^"):
81 assert_is_not_none(re.fullmatch(row[h], res[h]),
82 "attribute '%s': expected: '%s', got '%s'"
83 % (h, row[h], res[h]))
86 assert_equal(str(res[h]), str(row[h]))
88 def property_list(self, prop):
89 return [ x[prop] for x in self.result ]
92 class SearchResponse(GenericResponse):
94 def __init__(self, page, fmt='json', errorcode=200):
97 self.errorcode = errorcode
102 getattr(self, 'parse_' + fmt)()
104 def parse_json(self):
105 m = re.fullmatch(r'([\w$][^(]*)\((.*)\)', self.page)
110 self.header['json_func'] = m.group(1)
111 self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)
113 def parse_geojson(self):
115 self.result = geojson_results_to_json_results(self.result)
117 def parse_geocodejson(self):
118 return self.parse_geojson()
120 def parse_html(self):
121 content, errors = tidy_document(self.page,
122 options={'char-encoding' : 'utf8'})
123 #eq_(len(errors), 0 , "Errors found in HTML document:\n%s" % errors)
126 b = content.find('nominatim_results =')
127 e = content.find('</script>')
128 if b >= 0 and e >= 0:
129 content = content[b:e]
131 b = content.find('[')
132 e = content.rfind(']')
133 if b >= 0 and e >= 0:
134 self.result = json.JSONDecoder(object_pairs_hook=OrderedDict)\
135 .decode(content[b:e+1])
138 et = ET.fromstring(self.page)
140 self.header = dict(et.attrib)
143 assert_equal(child.tag, "place")
144 self.result.append(dict(child.attrib))
148 if sub.tag == 'extratags':
149 self.result[-1]['extratags'] = {}
151 self.result[-1]['extratags'][tag.attrib['key']] = tag.attrib['value']
152 elif sub.tag == 'namedetails':
153 self.result[-1]['namedetails'] = {}
155 self.result[-1]['namedetails'][tag.attrib['desc']] = tag.text
156 elif sub.tag in ('geokml'):
157 self.result[-1][sub.tag] = True
159 address[sub.tag] = sub.text
162 self.result[-1]['address'] = address
165 class ReverseResponse(GenericResponse):
167 def __init__(self, page, fmt='json', errorcode=200):
170 self.errorcode = errorcode
175 getattr(self, 'parse_' + fmt)()
177 def parse_html(self):
178 content, errors = tidy_document(self.page,
179 options={'char-encoding' : 'utf8'})
180 #eq_(len(errors), 0 , "Errors found in HTML document:\n%s" % errors)
182 b = content.find('nominatim_results =')
183 e = content.find('</script>')
184 content = content[b:e]
185 b = content.find('[')
186 e = content.rfind(']')
188 self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(content[b:e+1])
190 def parse_json(self):
191 m = re.fullmatch(r'([\w$][^(]*)\((.*)\)', self.page)
196 self.header['json_func'] = m.group(1)
197 self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)]
199 def parse_geojson(self):
201 if 'error' in self.result:
203 self.result = geojson_results_to_json_results(self.result[0])
205 def parse_geocodejson(self):
206 return self.parse_geojson()
209 et = ET.fromstring(self.page)
211 self.header = dict(et.attrib)
215 if child.tag == 'result':
216 eq_(0, len(self.result), "More than one result in reverse result")
217 self.result.append(dict(child.attrib))
218 elif child.tag == 'addressparts':
221 address[sub.tag] = sub.text
222 self.result[0]['address'] = address
223 elif child.tag == 'extratags':
224 self.result[0]['extratags'] = {}
226 self.result[0]['extratags'][tag.attrib['key']] = tag.attrib['value']
227 elif child.tag == 'namedetails':
228 self.result[0]['namedetails'] = {}
230 self.result[0]['namedetails'][tag.attrib['desc']] = tag.text
231 elif child.tag in ('geokml'):
232 self.result[0][child.tag] = True
234 assert child.tag == 'error', \
235 "Unknown XML tag %s on page: %s" % (child.tag, self.page)
238 class DetailsResponse(GenericResponse):
240 def __init__(self, page, fmt='json', errorcode=200):
243 self.errorcode = errorcode
248 getattr(self, 'parse_' + fmt)()
250 def parse_html(self):
251 content, errors = tidy_document(self.page,
252 options={'char-encoding' : 'utf8'})
255 def parse_json(self):
256 self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
259 class StatusResponse(GenericResponse):
261 def __init__(self, page, fmt='text', errorcode=200):
264 self.errorcode = errorcode
266 if errorcode == 200 and fmt != 'text':
267 getattr(self, 'parse_' + fmt)()
269 def parse_json(self):
270 self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
273 def geojson_result_to_json_result(geojson_result):
274 result = geojson_result['properties']
275 result['geojson'] = geojson_result['geometry']
276 if 'bbox' in geojson_result:
277 # bbox is minlon, minlat, maxlon, maxlat
278 # boundingbox is minlat, maxlat, minlon, maxlon
279 result['boundingbox'] = [
280 geojson_result['bbox'][1],
281 geojson_result['bbox'][3],
282 geojson_result['bbox'][0],
283 geojson_result['bbox'][2]
288 def geojson_results_to_json_results(geojson_results):
289 if 'error' in geojson_results:
291 return list(map(geojson_result_to_json_result, geojson_results['features']))
294 @when(u'searching for "(?P<query>.*)"(?P<dups> with dups)?')
295 def query_cmd(context, query, dups):
296 """ Query directly via PHP script.
298 cmd = ['/usr/bin/env', 'php']
299 cmd.append(os.path.join(context.nominatim.build_dir, 'utils', 'query.php'))
300 cmd.extend(['--search', query])
301 # add more parameters in table form
303 for h in context.table.headings:
304 value = context.table[0][h].strip()
306 cmd.extend(('--' + h, value))
309 cmd.extend(('--dedupe', '0'))
311 proc = subprocess.Popen(cmd, cwd=context.nominatim.build_dir,
312 stdout=subprocess.PIPE, stderr=subprocess.PIPE)
313 (outp, err) = proc.communicate()
315 assert_equals (0, proc.returncode, "query.php failed with message: %s\noutput: %s" % (err, outp))
317 context.response = SearchResponse(outp.decode('utf-8'), 'json')
319 def send_api_query(endpoint, params, fmt, context):
321 params['format'] = fmt.strip()
323 if context.table.headings[0] == 'param':
324 for line in context.table:
325 params[line['param']] = line['value']
327 for h in context.table.headings:
328 params[h] = context.table[0][h]
330 env = dict(BASE_SERVER_ENV)
331 env['QUERY_STRING'] = urlencode(params)
333 env['SCRIPT_NAME'] = '/%s.php' % endpoint
334 env['REQUEST_URI'] = '%s?%s' % (env['SCRIPT_NAME'], env['QUERY_STRING'])
335 env['CONTEXT_DOCUMENT_ROOT'] = os.path.join(context.nominatim.build_dir, 'website')
336 env['SCRIPT_FILENAME'] = os.path.join(env['CONTEXT_DOCUMENT_ROOT'],
338 env['NOMINATIM_SETTINGS'] = context.nominatim.local_settings_file
340 logger.debug("Environment:" + json.dumps(env, sort_keys=True, indent=2))
342 if hasattr(context, 'http_headers'):
343 env.update(context.http_headers)
345 cmd = ['/usr/bin/env', 'php-cgi', '-f']
346 if context.nominatim.code_coverage_path:
347 env['COV_SCRIPT_FILENAME'] = env['SCRIPT_FILENAME']
348 env['COV_PHP_DIR'] = os.path.join(context.nominatim.src_dir, "lib")
349 env['COV_TEST_NAME'] = '%s:%s' % (context.scenario.filename, context.scenario.line)
350 env['SCRIPT_FILENAME'] = \
351 os.path.join(os.path.split(__file__)[0], 'cgi-with-coverage.php')
352 cmd.append(env['SCRIPT_FILENAME'])
353 env['PHP_CODE_COVERAGE_FILE'] = context.nominatim.next_code_coverage_file()
355 cmd.append(env['SCRIPT_FILENAME'])
357 for k,v in params.items():
358 cmd.append("%s=%s" % (k, v))
360 proc = subprocess.Popen(cmd, cwd=context.nominatim.build_dir, env=env,
361 stdout=subprocess.PIPE, stderr=subprocess.PIPE)
363 (outp, err) = proc.communicate()
364 outp = outp.decode('utf-8')
365 err = err.decode("utf-8")
367 logger.debug("Result: \n===============================\n"
368 + outp + "\n===============================\n")
370 assert_equals(0, proc.returncode,
371 "%s failed with message: %s" % (
372 os.path.basename(env['SCRIPT_FILENAME']),
375 assert_equals(0, len(err), "Unexpected PHP error: %s" % (err))
377 if outp.startswith('Status: '):
378 status = int(outp[8:11])
382 content_start = outp.find('\r\n\r\n')
384 return outp[content_start + 4:], status
386 @given(u'the HTTP header')
387 def add_http_header(context):
388 if not hasattr(context, 'http_headers'):
389 context.http_headers = {}
391 for h in context.table.headings:
392 envvar = 'HTTP_' + h.upper().replace('-', '_')
393 context.http_headers[envvar] = context.table[0][h]
396 @when(u'sending (?P<fmt>\S+ )?search query "(?P<query>.*)"(?P<addr> with address)?')
397 def website_search_request(context, fmt, query, addr):
402 params['addressdetails'] = '1'
404 outp, status = send_api_query('search', params, fmt, context)
408 elif fmt == 'jsonv2 ':
413 context.response = SearchResponse(outp, outfmt, status)
415 @when(u'sending (?P<fmt>\S+ )?reverse coordinates (?P<lat>.+)?,(?P<lon>.+)?')
416 def website_reverse_request(context, fmt, lat, lon):
423 outp, status = send_api_query('reverse', params, fmt, context)
427 elif fmt == 'jsonv2 ':
432 context.response = ReverseResponse(outp, outfmt, status)
434 @when(u'sending (?P<fmt>\S+ )?details query for (?P<query>.*)')
435 def website_details_request(context, fmt, query):
437 if query[0] in 'NWR':
438 params['osmtype'] = query[0]
439 params['osmid'] = query[1:]
441 params['place_id'] = query
442 outp, status = send_api_query('details', params, fmt, context)
449 context.response = DetailsResponse(outp, outfmt, status)
451 @when(u'sending (?P<fmt>\S+ )?lookup query for (?P<query>.*)')
452 def website_lookup_request(context, fmt, query):
453 params = { 'osm_ids' : query }
454 outp, status = send_api_query('lookup', params, fmt, context)
458 elif fmt == 'geojson ':
463 context.response = SearchResponse(outp, outfmt, status)
465 @when(u'sending (?P<fmt>\S+ )?status query')
466 def website_status_request(context, fmt):
468 outp, status = send_api_query('status', params, fmt, context)
475 context.response = StatusResponse(outp, outfmt, status)
477 @step(u'(?P<operator>less than|more than|exactly|at least|at most) (?P<number>\d+) results? (?:is|are) returned')
478 def validate_result_number(context, operator, number):
479 eq_(context.response.errorcode, 200)
480 numres = len(context.response.result)
481 ok_(compare(operator, numres, int(number)),
482 "Bad number of results: expected %s %s, got %d." % (operator, number, numres))
484 @then(u'a HTTP (?P<status>\d+) is returned')
485 def check_http_return_status(context, status):
486 eq_(context.response.errorcode, int(status))
488 @then(u'the page contents equals "(?P<text>.+)"')
489 def check_page_content_equals(context, text):
490 eq_(context.response.page, text)
492 @then(u'the result is valid (?P<fmt>\w+)')
493 def step_impl(context, fmt):
494 context.execute_steps("Then a HTTP 200 is returned")
495 eq_(context.response.format, fmt)
497 @then(u'a (?P<fmt>\w+) user error is returned')
498 def check_page_error(context, fmt):
499 context.execute_steps("Then a HTTP 400 is returned")
500 eq_(context.response.format, fmt)
503 assert_is_not_none(re.search(r'<html( |>).+</html>', context.response.page, re.DOTALL))
505 assert_is_not_none(re.search(r'<error>.+</error>', context.response.page, re.DOTALL))
507 assert_is_not_none(re.search(r'({"error":)', context.response.page, re.DOTALL))
509 @then(u'result header contains')
510 def check_header_attr(context):
511 for line in context.table:
512 assert_is_not_none(re.fullmatch(line['value'], context.response.header[line['attr']]),
513 "attribute '%s': expected: '%s', got '%s'"
514 % (line['attr'], line['value'],
515 context.response.header[line['attr']]))
517 @then(u'result header has (?P<neg>not )?attributes (?P<attrs>.*)')
518 def check_header_no_attr(context, neg, attrs):
519 for attr in attrs.split(','):
521 assert_not_in(attr, context.response.header)
523 assert_in(attr, context.response.header)
525 @then(u'results contain')
526 def step_impl(context):
527 context.execute_steps("then at least 1 result is returned")
529 for line in context.table:
530 context.response.match_row(line)
532 @then(u'result (?P<lid>\d+ )?has (?P<neg>not )?attributes (?P<attrs>.*)')
533 def validate_attributes(context, lid, neg, attrs):
535 idx = range(len(context.response.result))
536 context.execute_steps("then at least 1 result is returned")
538 idx = [int(lid.strip())]
539 context.execute_steps("then more than %sresults are returned" % lid)
542 for attr in attrs.split(','):
544 assert_not_in(attr, context.response.result[i])
546 assert_in(attr, context.response.result[i])
548 @then(u'result addresses contain')
549 def step_impl(context):
550 context.execute_steps("then at least 1 result is returned")
552 if 'ID' not in context.table.headings:
553 addr_parts = context.response.property_list('address')
555 for line in context.table:
556 if 'ID' in context.table.headings:
557 addr_parts = [dict(context.response.result[int(line['ID'])]['address'])]
559 for h in context.table.headings:
563 assert_equal(p[h], line[h], "Bad address value for %s" % h)
565 @then(u'address of result (?P<lid>\d+) has(?P<neg> no)? types (?P<attrs>.*)')
566 def check_address(context, lid, neg, attrs):
567 context.execute_steps("then more than %s results are returned" % lid)
569 addr_parts = context.response.result[int(lid)]['address']
571 for attr in attrs.split(','):
573 assert_not_in(attr, addr_parts)
575 assert_in(attr, addr_parts)
577 @then(u'address of result (?P<lid>\d+) is')
578 def check_address(context, lid):
579 context.execute_steps("then more than %s results are returned" % lid)
581 addr_parts = dict(context.response.result[int(lid)]['address'])
583 for line in context.table:
584 assert_in(line['type'], addr_parts)
585 assert_equal(addr_parts[line['type']], line['value'],
586 "Bad address value for %s" % line['type'])
587 del addr_parts[line['type']]
589 eq_(0, len(addr_parts), "Additional address parts found: %s" % str(addr_parts))
591 @then(u'result (?P<lid>\d+ )?has bounding box in (?P<coords>[\d,.-]+)')
592 def step_impl(context, lid, coords):
594 context.execute_steps("then at least 1 result is returned")
595 bboxes = context.response.property_list('boundingbox')
597 context.execute_steps("then more than %sresults are returned" % lid)
598 bboxes = [ context.response.result[int(lid)]['boundingbox']]
600 coord = [ float(x) for x in coords.split(',') ]
603 if isinstance(bbox, str):
604 bbox = bbox.split(',')
605 bbox = [ float(x) for x in bbox ]
607 assert_greater_equal(bbox[0], coord[0])
608 assert_less_equal(bbox[1], coord[1])
609 assert_greater_equal(bbox[2], coord[2])
610 assert_less_equal(bbox[3], coord[3])
612 @then(u'result (?P<lid>\d+ )?has centroid in (?P<coords>[\d,.-]+)')
613 def step_impl(context, lid, coords):
615 context.execute_steps("then at least 1 result is returned")
616 bboxes = zip(context.response.property_list('lat'),
617 context.response.property_list('lon'))
619 context.execute_steps("then more than %sresults are returned" % lid)
620 res = context.response.result[int(lid)]
621 bboxes = [ (res['lat'], res['lon']) ]
623 coord = [ float(x) for x in coords.split(',') ]
625 for lat, lon in bboxes:
628 assert_greater_equal(lat, coord[0])
629 assert_less_equal(lat, coord[1])
630 assert_greater_equal(lon, coord[2])
631 assert_less_equal(lon, coord[3])
633 @then(u'there are(?P<neg> no)? duplicates')
634 def check_for_duplicates(context, neg):
635 context.execute_steps("then at least 1 result is returned")
640 for res in context.response.result:
641 dup = (res['osm_type'], res['class'], res['type'], res['display_name'])
648 assert not has_dupe, "Found duplicate for %s" % (dup, )
650 assert has_dupe, "No duplicates found"