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):
119 if self.result is not None:
120 self.result = [r['geocoding'] for r in self.result]
122 def parse_html(self):
123 content, errors = tidy_document(self.page,
124 options={'char-encoding' : 'utf8'})
125 #eq_(len(errors), 0 , "Errors found in HTML document:\n%s" % errors)
128 b = content.find('nominatim_results =')
129 e = content.find('</script>')
130 if b >= 0 and e >= 0:
131 content = content[b:e]
133 b = content.find('[')
134 e = content.rfind(']')
135 if b >= 0 and e >= 0:
136 self.result = json.JSONDecoder(object_pairs_hook=OrderedDict)\
137 .decode(content[b:e+1])
140 et = ET.fromstring(self.page)
142 self.header = dict(et.attrib)
145 assert_equal(child.tag, "place")
146 self.result.append(dict(child.attrib))
150 if sub.tag == 'extratags':
151 self.result[-1]['extratags'] = {}
153 self.result[-1]['extratags'][tag.attrib['key']] = tag.attrib['value']
154 elif sub.tag == 'namedetails':
155 self.result[-1]['namedetails'] = {}
157 self.result[-1]['namedetails'][tag.attrib['desc']] = tag.text
158 elif sub.tag in ('geokml'):
159 self.result[-1][sub.tag] = True
161 address[sub.tag] = sub.text
164 self.result[-1]['address'] = address
167 class ReverseResponse(GenericResponse):
169 def __init__(self, page, fmt='json', errorcode=200):
172 self.errorcode = errorcode
177 getattr(self, 'parse_' + fmt)()
179 def parse_html(self):
180 content, errors = tidy_document(self.page,
181 options={'char-encoding' : 'utf8'})
182 #eq_(len(errors), 0 , "Errors found in HTML document:\n%s" % errors)
184 b = content.find('nominatim_results =')
185 e = content.find('</script>')
186 content = content[b:e]
187 b = content.find('[')
188 e = content.rfind(']')
190 self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(content[b:e+1])
192 def parse_json(self):
193 m = re.fullmatch(r'([\w$][^(]*)\((.*)\)', self.page)
198 self.header['json_func'] = m.group(1)
199 self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)]
201 def parse_geojson(self):
203 if 'error' in self.result:
205 self.result = geojson_results_to_json_results(self.result[0])
207 def parse_geocodejson(self):
209 if self.result is not None:
210 self.result = [r['geocoding'] for r in self.result]
213 et = ET.fromstring(self.page)
215 self.header = dict(et.attrib)
219 if child.tag == 'result':
220 eq_(0, len(self.result), "More than one result in reverse result")
221 self.result.append(dict(child.attrib))
222 elif child.tag == 'addressparts':
225 address[sub.tag] = sub.text
226 self.result[0]['address'] = address
227 elif child.tag == 'extratags':
228 self.result[0]['extratags'] = {}
230 self.result[0]['extratags'][tag.attrib['key']] = tag.attrib['value']
231 elif child.tag == 'namedetails':
232 self.result[0]['namedetails'] = {}
234 self.result[0]['namedetails'][tag.attrib['desc']] = tag.text
235 elif child.tag in ('geokml'):
236 self.result[0][child.tag] = True
238 assert child.tag == 'error', \
239 "Unknown XML tag %s on page: %s" % (child.tag, self.page)
242 class DetailsResponse(GenericResponse):
244 def __init__(self, page, fmt='json', errorcode=200):
247 self.errorcode = errorcode
252 getattr(self, 'parse_' + fmt)()
254 def parse_html(self):
255 content, errors = tidy_document(self.page,
256 options={'char-encoding' : 'utf8'})
259 def parse_json(self):
260 self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
263 class StatusResponse(GenericResponse):
265 def __init__(self, page, fmt='text', errorcode=200):
268 self.errorcode = errorcode
270 if errorcode == 200 and fmt != 'text':
271 getattr(self, 'parse_' + fmt)()
273 def parse_json(self):
274 self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
277 def geojson_result_to_json_result(geojson_result):
278 result = geojson_result['properties']
279 result['geojson'] = geojson_result['geometry']
280 if 'bbox' in geojson_result:
281 # bbox is minlon, minlat, maxlon, maxlat
282 # boundingbox is minlat, maxlat, minlon, maxlon
283 result['boundingbox'] = [
284 geojson_result['bbox'][1],
285 geojson_result['bbox'][3],
286 geojson_result['bbox'][0],
287 geojson_result['bbox'][2]
292 def geojson_results_to_json_results(geojson_results):
293 if 'error' in geojson_results:
295 return list(map(geojson_result_to_json_result, geojson_results['features']))
298 @when(u'searching for "(?P<query>.*)"(?P<dups> with dups)?')
299 def query_cmd(context, query, dups):
300 """ Query directly via PHP script.
302 cmd = ['/usr/bin/env', 'php']
303 cmd.append(os.path.join(context.nominatim.build_dir, 'utils', 'query.php'))
305 cmd.extend(['--search', query])
306 # add more parameters in table form
308 for h in context.table.headings:
309 value = context.table[0][h].strip()
311 cmd.extend(('--' + h, value))
314 cmd.extend(('--dedupe', '0'))
316 proc = subprocess.Popen(cmd, cwd=context.nominatim.build_dir,
317 stdout=subprocess.PIPE, stderr=subprocess.PIPE)
318 (outp, err) = proc.communicate()
320 assert_equals (0, proc.returncode, "query.php failed with message: %s\noutput: %s" % (err, outp))
322 context.response = SearchResponse(outp.decode('utf-8'), 'json')
324 def send_api_query(endpoint, params, fmt, context):
326 params['format'] = fmt.strip()
328 if context.table.headings[0] == 'param':
329 for line in context.table:
330 params[line['param']] = line['value']
332 for h in context.table.headings:
333 params[h] = context.table[0][h]
335 env = dict(BASE_SERVER_ENV)
336 env['QUERY_STRING'] = urlencode(params)
338 env['SCRIPT_NAME'] = '/%s.php' % endpoint
339 env['REQUEST_URI'] = '%s?%s' % (env['SCRIPT_NAME'], env['QUERY_STRING'])
340 env['CONTEXT_DOCUMENT_ROOT'] = os.path.join(context.nominatim.build_dir, 'website')
341 env['SCRIPT_FILENAME'] = os.path.join(env['CONTEXT_DOCUMENT_ROOT'],
343 env['NOMINATIM_SETTINGS'] = context.nominatim.local_settings_file
345 logger.debug("Environment:" + json.dumps(env, sort_keys=True, indent=2))
347 if hasattr(context, 'http_headers'):
348 env.update(context.http_headers)
350 cmd = ['/usr/bin/env', 'php-cgi', '-f']
351 if context.nominatim.code_coverage_path:
352 env['COV_SCRIPT_FILENAME'] = env['SCRIPT_FILENAME']
353 env['COV_PHP_DIR'] = os.path.join(context.nominatim.src_dir, "lib")
354 env['COV_TEST_NAME'] = '%s:%s' % (context.scenario.filename, context.scenario.line)
355 env['SCRIPT_FILENAME'] = \
356 os.path.join(os.path.split(__file__)[0], 'cgi-with-coverage.php')
357 cmd.append(env['SCRIPT_FILENAME'])
358 env['PHP_CODE_COVERAGE_FILE'] = context.nominatim.next_code_coverage_file()
360 cmd.append(env['SCRIPT_FILENAME'])
362 for k,v in params.items():
363 cmd.append("%s=%s" % (k, v))
365 proc = subprocess.Popen(cmd, cwd=context.nominatim.build_dir, env=env,
366 stdout=subprocess.PIPE, stderr=subprocess.PIPE)
368 (outp, err) = proc.communicate()
369 outp = outp.decode('utf-8')
370 err = err.decode("utf-8")
372 logger.debug("Result: \n===============================\n"
373 + outp + "\n===============================\n")
375 assert_equals(0, proc.returncode,
376 "%s failed with message: %s" % (
377 os.path.basename(env['SCRIPT_FILENAME']),
380 assert_equals(0, len(err), "Unexpected PHP error: %s" % (err))
382 if outp.startswith('Status: '):
383 status = int(outp[8:11])
387 content_start = outp.find('\r\n\r\n')
389 return outp[content_start + 4:], status
391 @given(u'the HTTP header')
392 def add_http_header(context):
393 if not hasattr(context, 'http_headers'):
394 context.http_headers = {}
396 for h in context.table.headings:
397 envvar = 'HTTP_' + h.upper().replace('-', '_')
398 context.http_headers[envvar] = context.table[0][h]
401 @when(u'sending (?P<fmt>\S+ )?search query "(?P<query>.*)"(?P<addr> with address)?')
402 def website_search_request(context, fmt, query, addr):
407 params['addressdetails'] = '1'
409 outp, status = send_api_query('search', params, fmt, context)
413 elif fmt == 'jsonv2 ':
418 context.response = SearchResponse(outp, outfmt, status)
420 @when(u'sending (?P<fmt>\S+ )?reverse coordinates (?P<lat>.+)?,(?P<lon>.+)?')
421 def website_reverse_request(context, fmt, lat, lon):
428 outp, status = send_api_query('reverse', params, fmt, context)
432 elif fmt == 'jsonv2 ':
437 context.response = ReverseResponse(outp, outfmt, status)
439 @when(u'sending (?P<fmt>\S+ )?details query for (?P<query>.*)')
440 def website_details_request(context, fmt, query):
442 if query[0] in 'NWR':
443 params['osmtype'] = query[0]
444 params['osmid'] = query[1:]
446 params['place_id'] = query
447 outp, status = send_api_query('details', params, fmt, context)
454 context.response = DetailsResponse(outp, outfmt, status)
456 @when(u'sending (?P<fmt>\S+ )?lookup query for (?P<query>.*)')
457 def website_lookup_request(context, fmt, query):
458 params = { 'osm_ids' : query }
459 outp, status = send_api_query('lookup', params, fmt, context)
463 elif fmt == 'jsonv2 ':
465 elif fmt == 'geojson ':
467 elif fmt == 'geocodejson ':
468 outfmt = 'geocodejson'
472 context.response = SearchResponse(outp, outfmt, status)
474 @when(u'sending (?P<fmt>\S+ )?status query')
475 def website_status_request(context, fmt):
477 outp, status = send_api_query('status', params, fmt, context)
484 context.response = StatusResponse(outp, outfmt, status)
486 @step(u'(?P<operator>less than|more than|exactly|at least|at most) (?P<number>\d+) results? (?:is|are) returned')
487 def validate_result_number(context, operator, number):
488 eq_(context.response.errorcode, 200)
489 numres = len(context.response.result)
490 ok_(compare(operator, numres, int(number)),
491 "Bad number of results: expected %s %s, got %d." % (operator, number, numres))
493 @then(u'a HTTP (?P<status>\d+) is returned')
494 def check_http_return_status(context, status):
495 eq_(context.response.errorcode, int(status))
497 @then(u'the page contents equals "(?P<text>.+)"')
498 def check_page_content_equals(context, text):
499 eq_(context.response.page, text)
501 @then(u'the result is valid (?P<fmt>\w+)')
502 def step_impl(context, fmt):
503 context.execute_steps("Then a HTTP 200 is returned")
504 eq_(context.response.format, fmt)
506 @then(u'a (?P<fmt>\w+) user error is returned')
507 def check_page_error(context, fmt):
508 context.execute_steps("Then a HTTP 400 is returned")
509 eq_(context.response.format, fmt)
512 assert_is_not_none(re.search(r'<html( |>).+</html>', context.response.page, re.DOTALL))
514 assert_is_not_none(re.search(r'<error>.+</error>', context.response.page, re.DOTALL))
516 assert_is_not_none(re.search(r'({"error":)', context.response.page, re.DOTALL))
518 @then(u'result header contains')
519 def check_header_attr(context):
520 for line in context.table:
521 assert_is_not_none(re.fullmatch(line['value'], context.response.header[line['attr']]),
522 "attribute '%s': expected: '%s', got '%s'"
523 % (line['attr'], line['value'],
524 context.response.header[line['attr']]))
526 @then(u'result header has (?P<neg>not )?attributes (?P<attrs>.*)')
527 def check_header_no_attr(context, neg, attrs):
528 for attr in attrs.split(','):
530 assert_not_in(attr, context.response.header)
532 assert_in(attr, context.response.header)
534 @then(u'results contain')
535 def step_impl(context):
536 context.execute_steps("then at least 1 result is returned")
538 for line in context.table:
539 context.response.match_row(line)
541 @then(u'result (?P<lid>\d+ )?has (?P<neg>not )?attributes (?P<attrs>.*)')
542 def validate_attributes(context, lid, neg, attrs):
544 idx = range(len(context.response.result))
545 context.execute_steps("then at least 1 result is returned")
547 idx = [int(lid.strip())]
548 context.execute_steps("then more than %sresults are returned" % lid)
551 for attr in attrs.split(','):
553 assert_not_in(attr, context.response.result[i])
555 assert_in(attr, context.response.result[i])
557 @then(u'result addresses contain')
558 def step_impl(context):
559 context.execute_steps("then at least 1 result is returned")
561 if 'ID' not in context.table.headings:
562 addr_parts = context.response.property_list('address')
564 for line in context.table:
565 if 'ID' in context.table.headings:
566 addr_parts = [dict(context.response.result[int(line['ID'])]['address'])]
568 for h in context.table.headings:
572 assert_equal(p[h], line[h], "Bad address value for %s" % h)
574 @then(u'address of result (?P<lid>\d+) has(?P<neg> no)? types (?P<attrs>.*)')
575 def check_address(context, lid, neg, attrs):
576 context.execute_steps("then more than %s results are returned" % lid)
578 addr_parts = context.response.result[int(lid)]['address']
580 for attr in attrs.split(','):
582 assert_not_in(attr, addr_parts)
584 assert_in(attr, addr_parts)
586 @then(u'address of result (?P<lid>\d+) is')
587 def check_address(context, lid):
588 context.execute_steps("then more than %s results are returned" % lid)
590 addr_parts = dict(context.response.result[int(lid)]['address'])
592 for line in context.table:
593 assert_in(line['type'], addr_parts)
594 assert_equal(addr_parts[line['type']], line['value'],
595 "Bad address value for %s" % line['type'])
596 del addr_parts[line['type']]
598 eq_(0, len(addr_parts), "Additional address parts found: %s" % str(addr_parts))
600 @then(u'result (?P<lid>\d+ )?has bounding box in (?P<coords>[\d,.-]+)')
601 def step_impl(context, lid, coords):
603 context.execute_steps("then at least 1 result is returned")
604 bboxes = context.response.property_list('boundingbox')
606 context.execute_steps("then more than %sresults are returned" % lid)
607 bboxes = [ context.response.result[int(lid)]['boundingbox']]
609 coord = [ float(x) for x in coords.split(',') ]
612 if isinstance(bbox, str):
613 bbox = bbox.split(',')
614 bbox = [ float(x) for x in bbox ]
616 assert_greater_equal(bbox[0], coord[0])
617 assert_less_equal(bbox[1], coord[1])
618 assert_greater_equal(bbox[2], coord[2])
619 assert_less_equal(bbox[3], coord[3])
621 @then(u'result (?P<lid>\d+ )?has centroid in (?P<coords>[\d,.-]+)')
622 def step_impl(context, lid, coords):
624 context.execute_steps("then at least 1 result is returned")
625 bboxes = zip(context.response.property_list('lat'),
626 context.response.property_list('lon'))
628 context.execute_steps("then more than %sresults are returned" % lid)
629 res = context.response.result[int(lid)]
630 bboxes = [ (res['lat'], res['lon']) ]
632 coord = [ float(x) for x in coords.split(',') ]
634 for lat, lon in bboxes:
637 assert_greater_equal(lat, coord[0])
638 assert_less_equal(lat, coord[1])
639 assert_greater_equal(lon, coord[2])
640 assert_less_equal(lon, coord[3])
642 @then(u'there are(?P<neg> no)? duplicates')
643 def check_for_duplicates(context, neg):
644 context.execute_steps("then at least 1 result is returned")
649 for res in context.response.result:
650 dup = (res['osm_type'], res['class'], res['type'], res['display_name'])
657 assert not has_dupe, "Found duplicate for %s" % (dup, )
659 assert has_dupe, "No duplicates found"