bdd: rename and clean up osm_data steps

[nominatim.git] / test / bdd / steps / queries.py

diff --git a/test/bdd/steps/queries.py b/test/bdd/steps/queries.py
index 5779c7637b301b188a443fb6e85835345d7cf89d..c7fdd6dba7fdb90aa6dd9345b0649b1fa6041d18 100644 (file)
--- a/test/bdd/steps/queries.py
+++ b/test/bdd/steps/queries.py
@@ -9,12 +9,12 @@ import os
 import io
 import re
 import logging
 import io
 import re
 import logging

-from tidylib import tidy_document

 import xml.etree.ElementTree as ET
 import xml.etree.ElementTree as ET

-import subprocess

 from urllib.parse import urlencode
 from collections import OrderedDict
 from urllib.parse import urlencode
 from collections import OrderedDict

-from nose.tools import * # for assert functions
+
+from check_functions import Almost
+from utils import run_script

 
 logger = logging.getLogger(__name__)
 
 
 logger = logging.getLogger(__name__)
 


@@ -70,20 +70,19 @@ class GenericResponse(object):
                 if h == 'ID':
                     pass
                 elif h == 'osm':
                 if h == 'ID':
                     pass
                 elif h == 'osm':

-                    assert_equal(res['osm_type'], row[h][0])
-                    assert_equal(res['osm_id'], row[h][1:])
+                    assert res['osm_type'] == row[h][0]
+                    assert res['osm_id'] == int(row[h][1:])

                 elif h == 'centroid':
                     x, y = row[h].split(' ')
                 elif h == 'centroid':
                     x, y = row[h].split(' ')

-                    assert_almost_equal(float(y), float(res['lat']))
-                    assert_almost_equal(float(x), float(res['lon']))
+                    assert Almost(float(y)) == float(res['lat'])
+                    assert Almost(float(x)) == float(res['lon'])

                 elif row[h].startswith("^"):
                 elif row[h].startswith("^"):

-                    assert_in(h, res)
-                    assert_is_not_none(re.fullmatch(row[h], res[h]),
-                                       "attribute '%s': expected: '%s', got '%s'"
-                                          % (h, row[h], res[h]))
+                    assert h in res
+                    assert re.fullmatch(row[h], res[h]) is not None, \
+                           "attribute '%s': expected: '%s', got '%s'" % (h, row[h], res[h])

                 else:
                 else:

-                    assert_in(h, res)
-                    assert_equal(str(res[h]), str(row[h]))
+                    assert h in res
+                    assert str(res[h]) == str(row[h])

 
     def property_list(self, prop):
         return [ x[prop] for x in self.result ]
 
     def property_list(self, prop):
         return [ x[prop] for x in self.result ]


@@ -110,18 +109,14 @@ class SearchResponse(GenericResponse):
             self.header['json_func'] = m.group(1)
         self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)
 
             self.header['json_func'] = m.group(1)
         self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)
 

-    def parse_html(self):
-        content, errors = tidy_document(self.page,
-                                        options={'char-encoding' : 'utf8'})
-        #eq_(len(errors), 0 , "Errors found in HTML document:\n%s" % errors)
-
-        b = content.find('nominatim_results =')
-        e = content.find('</script>')
-        content = content[b:e]
-        b = content.find('[')
-        e = content.rfind(']')
+    def parse_geojson(self):
+        self.parse_json()
+        self.result = geojson_results_to_json_results(self.result)

 
 

-        self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(content[b:e+1])
+    def parse_geocodejson(self):
+        self.parse_geojson()
+        if self.result is not None:
+            self.result = [r['geocoding'] for r in self.result]

 
     def parse_xml(self):
         et = ET.fromstring(self.page)
 
     def parse_xml(self):
         et = ET.fromstring(self.page)


@@ -129,7 +124,7 @@ class SearchResponse(GenericResponse):
         self.header = dict(et.attrib)
 
         for child in et:
         self.header = dict(et.attrib)
 
         for child in et:

-            assert_equal(child.tag, "place")
+            assert child.tag == "place"

             self.result.append(dict(child.attrib))
 
             address = {}
             self.result.append(dict(child.attrib))
 
             address = {}


@@ -163,19 +158,6 @@ class ReverseResponse(GenericResponse):
         if errorcode == 200:
             getattr(self, 'parse_' + fmt)()
 
         if errorcode == 200:
             getattr(self, 'parse_' + fmt)()
 

-    def parse_html(self):
-        content, errors = tidy_document(self.page,
-                                        options={'char-encoding' : 'utf8'})
-        #eq_(len(errors), 0 , "Errors found in HTML document:\n%s" % errors)
-
-        b = content.find('nominatim_results =')
-        e = content.find('</script>')
-        content = content[b:e]
-        b = content.find('[')
-        e = content.rfind(']')
-
-        self.result = json.JSONDecoder(object_pairs_hook=OrderedDict).decode(content[b:e+1])
-

     def parse_json(self):
         m = re.fullmatch(r'([\w$][^(]*)\((.*)\)', self.page)
         if m is None:
     def parse_json(self):
         m = re.fullmatch(r'([\w$][^(]*)\((.*)\)', self.page)
         if m is None:


@@ -185,6 +167,17 @@ class ReverseResponse(GenericResponse):
             self.header['json_func'] = m.group(1)
         self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)]
 
             self.header['json_func'] = m.group(1)
         self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(code)]
 

+    def parse_geojson(self):
+        self.parse_json()
+        if 'error' in self.result:
+            return
+        self.result = geojson_results_to_json_results(self.result[0])
+
+    def parse_geocodejson(self):
+        self.parse_geojson()
+        if self.result is not None:
+            self.result = [r['geocoding'] for r in self.result]
+

     def parse_xml(self):
         et = ET.fromstring(self.page)
 
     def parse_xml(self):
         et = ET.fromstring(self.page)
 


@@ -193,7 +186,7 @@ class ReverseResponse(GenericResponse):
 
         for child in et:
             if child.tag == 'result':
 
         for child in et:
             if child.tag == 'result':

-                eq_(0, len(self.result), "More than one result in reverse result")
+                assert len(self.result) == 0, "More than one result in reverse result"

                 self.result.append(dict(child.attrib))
             elif child.tag == 'addressparts':
                 address = {}
                 self.result.append(dict(child.attrib))
             elif child.tag == 'addressparts':
                 address = {}


@@ -227,22 +220,53 @@ class DetailsResponse(GenericResponse):
         if errorcode == 200:
             getattr(self, 'parse_' + fmt)()
 
         if errorcode == 200:
             getattr(self, 'parse_' + fmt)()
 

-    def parse_html(self):
-        content, errors = tidy_document(self.page,
-                                        options={'char-encoding' : 'utf8'})
-        self.result = {}
+    def parse_json(self):
+        self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
+
+
+class StatusResponse(GenericResponse):
+
+    def __init__(self, page, fmt='text', errorcode=200):
+        self.page = page
+        self.format = fmt
+        self.errorcode = errorcode
+
+        if errorcode == 200 and fmt != 'text':
+            getattr(self, 'parse_' + fmt)()

 
     def parse_json(self):
         self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
 
 
 
     def parse_json(self):
         self.result = [json.JSONDecoder(object_pairs_hook=OrderedDict).decode(self.page)]
 
 

+def geojson_result_to_json_result(geojson_result):
+    result = geojson_result['properties']
+    result['geojson'] = geojson_result['geometry']
+    if 'bbox' in geojson_result:
+        # bbox is  minlon, minlat, maxlon, maxlat
+        # boundingbox is minlat, maxlat, minlon, maxlon
+        result['boundingbox'] = [
+                                    geojson_result['bbox'][1],
+                                    geojson_result['bbox'][3],
+                                    geojson_result['bbox'][0],
+                                    geojson_result['bbox'][2]
+                                ]
+    return result
+
+
+def geojson_results_to_json_results(geojson_results):
+    if 'error' in geojson_results:
+        return
+    return list(map(geojson_result_to_json_result, geojson_results['features']))
+
+

 @when(u'searching for "(?P<query>.*)"(?P<dups> with dups)?')
 def query_cmd(context, query, dups):
     """ Query directly via PHP script.
     """
     cmd = ['/usr/bin/env', 'php']
     cmd.append(os.path.join(context.nominatim.build_dir, 'utils', 'query.php'))
 @when(u'searching for "(?P<query>.*)"(?P<dups> with dups)?')
 def query_cmd(context, query, dups):
     """ Query directly via PHP script.
     """
     cmd = ['/usr/bin/env', 'php']
     cmd.append(os.path.join(context.nominatim.build_dir, 'utils', 'query.php'))

-    cmd.extend(['--search', query])
+    if query:
+        cmd.extend(['--search', query])

     # add more parameters in table form
     if context.table:
         for h in context.table.headings:
     # add more parameters in table form
     if context.table:
         for h in context.table.headings:


@@ -253,13 +277,9 @@ def query_cmd(context, query, dups):
     if dups:
         cmd.extend(('--dedupe', '0'))
 
     if dups:
         cmd.extend(('--dedupe', '0'))
 

-    proc = subprocess.Popen(cmd, cwd=context.nominatim.build_dir,
-                            stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-    (outp, err) = proc.communicate()
-
-    assert_equals (0, proc.returncode, "query.php failed with message: %s\noutput: %s" % (err, outp))
+    outp, err = run_script(cmd, cwd=context.nominatim.build_dir)

 
 

-    context.response = SearchResponse(outp.decode('utf-8'), 'json')
+    context.response = SearchResponse(outp, 'json')

 
 def send_api_query(endpoint, params, fmt, context):
     if fmt is not None:
 
 def send_api_query(endpoint, params, fmt, context):
     if fmt is not None:


@@ -277,10 +297,9 @@ def send_api_query(endpoint, params, fmt, context):
 
     env['SCRIPT_NAME'] = '/%s.php' % endpoint
     env['REQUEST_URI'] = '%s?%s' % (env['SCRIPT_NAME'], env['QUERY_STRING'])
 
     env['SCRIPT_NAME'] = '/%s.php' % endpoint
     env['REQUEST_URI'] = '%s?%s' % (env['SCRIPT_NAME'], env['QUERY_STRING'])

-    env['CONTEXT_DOCUMENT_ROOT'] = os.path.join(context.nominatim.build_dir, 'website')
+    env['CONTEXT_DOCUMENT_ROOT'] = os.path.join(context.nominatim.website_dir.name, 'website')

     env['SCRIPT_FILENAME'] = os.path.join(env['CONTEXT_DOCUMENT_ROOT'],
                                           '%s.php' % endpoint)
     env['SCRIPT_FILENAME'] = os.path.join(env['CONTEXT_DOCUMENT_ROOT'],
                                           '%s.php' % endpoint)

-    env['NOMINATIM_SETTINGS'] = context.nominatim.local_settings_file

 
     logger.debug("Environment:" + json.dumps(env, sort_keys=True, indent=2))
 
 
     logger.debug("Environment:" + json.dumps(env, sort_keys=True, indent=2))
 


@@ -302,19 +321,9 @@ def send_api_query(endpoint, params, fmt, context):
     for k,v in params.items():
         cmd.append("%s=%s" % (k, v))
 
     for k,v in params.items():
         cmd.append("%s=%s" % (k, v))
 

-    proc = subprocess.Popen(cmd, cwd=context.nominatim.build_dir, env=env,
-                            stdout=subprocess.PIPE, stderr=subprocess.PIPE)
-
-    (outp, err) = proc.communicate()
-    outp = outp.decode('utf-8')
-
-    logger.debug("Result: \n===============================\n"
-                 + outp + "\n===============================\n")
-
-    assert_equals(0, proc.returncode,
-                  "%s failed with message: %s\noutput: %s" % (env['SCRIPT_FILENAME'], err, outp))
+    outp, err = run_script(cmd, cwd=context.nominatim.website_dir.name, env=env)

 
 

-    assert_equals(0, len(err), "Unexpected PHP error: %s" % (err))
+    assert len(err) == 0, "Unexpected PHP error: %s" % (err)

 
     if outp.startswith('Status: '):
         status = int(outp[8:11])
 
     if outp.startswith('Status: '):
         status = int(outp[8:11])


@@ -345,9 +354,7 @@ def website_search_request(context, fmt, query, addr):
 
     outp, status = send_api_query('search', params, fmt, context)
 
 
     outp, status = send_api_query('search', params, fmt, context)
 

-    if fmt is None:
-        outfmt = 'html'
-    elif fmt == 'jsonv2 ':
+    if fmt is None or fmt == 'jsonv2 ':

         outfmt = 'json'
     else:
         outfmt = fmt.strip()
         outfmt = 'json'
     else:
         outfmt = fmt.strip()


@@ -384,7 +391,7 @@ def website_details_request(context, fmt, query):
     outp, status = send_api_query('details', params, fmt, context)
 
     if fmt is None:
     outp, status = send_api_query('details', params, fmt, context)
 
     if fmt is None:

-        outfmt = 'html'
+        outfmt = 'json'

     else:
         outfmt = fmt.strip()
 
     else:
         outfmt = fmt.strip()
 


@@ -397,42 +404,74 @@ def website_lookup_request(context, fmt, query):
 
     if fmt == 'json ':
         outfmt = 'json'
 
     if fmt == 'json ':
         outfmt = 'json'

+    elif fmt == 'jsonv2 ':
+        outfmt = 'json'
+    elif fmt == 'geojson ':
+        outfmt = 'geojson'
+    elif fmt == 'geocodejson ':
+        outfmt = 'geocodejson'

     else:
         outfmt = 'xml'
 
     context.response = SearchResponse(outp, outfmt, status)
 
     else:
         outfmt = 'xml'
 
     context.response = SearchResponse(outp, outfmt, status)
 

+@when(u'sending (?P<fmt>\S+ )?status query')
+def website_status_request(context, fmt):
+    params = {}
+    outp, status = send_api_query('status', params, fmt, context)
+
+    if fmt is None:
+        outfmt = 'text'
+    else:
+        outfmt = fmt.strip()
+
+    context.response = StatusResponse(outp, outfmt, status)
+

 @step(u'(?P<operator>less than|more than|exactly|at least|at most) (?P<number>\d+) results? (?:is|are) returned')
 def validate_result_number(context, operator, number):
 @step(u'(?P<operator>less than|more than|exactly|at least|at most) (?P<number>\d+) results? (?:is|are) returned')
 def validate_result_number(context, operator, number):

-    eq_(context.response.errorcode, 200)
+    assert context.response.errorcode == 200

     numres = len(context.response.result)
     numres = len(context.response.result)

-    ok_(compare(operator, numres, int(number)),
-        "Bad number of results: expected %s %s, got %d." % (operator, number, numres))
+    assert compare(operator, numres, int(number)), \
+        "Bad number of results: expected %s %s, got %d." % (operator, number, numres)

 
 @then(u'a HTTP (?P<status>\d+) is returned')
 def check_http_return_status(context, status):
 
 @then(u'a HTTP (?P<status>\d+) is returned')
 def check_http_return_status(context, status):

-    eq_(context.response.errorcode, int(status))
+    assert context.response.errorcode == int(status)
+
+@then(u'the page contents equals "(?P<text>.+)"')
+def check_page_content_equals(context, text):
+    assert context.response.page == text

 
 @then(u'the result is valid (?P<fmt>\w+)')
 def step_impl(context, fmt):
     context.execute_steps("Then a HTTP 200 is returned")
 
 @then(u'the result is valid (?P<fmt>\w+)')
 def step_impl(context, fmt):
     context.execute_steps("Then a HTTP 200 is returned")

-    eq_(context.response.format, fmt)
+    assert context.response.format == fmt
+
+@then(u'a (?P<fmt>\w+) user error is returned')
+def check_page_error(context, fmt):
+    context.execute_steps("Then a HTTP 400 is returned")
+    assert context.response.format == fmt
+
+    if fmt == 'xml':
+        assert re.search(r'<error>.+</error>', context.response.page, re.DOTALL) is not None
+    else:
+        assert re.search(r'({"error":)', context.response.page, re.DOTALL) is not None

 
 @then(u'result header contains')
 def check_header_attr(context):
     for line in context.table:
 
 @then(u'result header contains')
 def check_header_attr(context):
     for line in context.table:

-        assert_is_not_none(re.fullmatch(line['value'], context.response.header[line['attr']]),
-                     "attribute '%s': expected: '%s', got '%s'"
-                       % (line['attr'], line['value'],
-                          context.response.header[line['attr']]))
+        assert re.fullmatch(line['value'], context.response.header[line['attr']]) is not None, \
+               "attribute '%s': expected: '%s', got '%s'" % (
+                    line['attr'], line['value'],
+                    context.response.header[line['attr']])

 
 @then(u'result header has (?P<neg>not )?attributes (?P<attrs>.*)')
 def check_header_no_attr(context, neg, attrs):
     for attr in attrs.split(','):
         if neg:
 
 @then(u'result header has (?P<neg>not )?attributes (?P<attrs>.*)')
 def check_header_no_attr(context, neg, attrs):
     for attr in attrs.split(','):
         if neg:

-            assert_not_in(attr, context.response.header)
+            assert attr not in context.response.header

         else:
         else:

-            assert_in(attr, context.response.header)
+            assert attr in context.response.header

 
 @then(u'results contain')
 def step_impl(context):
 
 @then(u'results contain')
 def step_impl(context):


@@ -453,9 +492,9 @@ def validate_attributes(context, lid, neg, attrs):
     for i in idx:
         for attr in attrs.split(','):
             if neg:
     for i in idx:
         for attr in attrs.split(','):
             if neg:

-                assert_not_in(attr, context.response.result[i])
+                assert attr not in context.response.result[i]

             else:
             else:

-                assert_in(attr, context.response.result[i])
+                assert attr in context.response.result[i]

 
 @then(u'result addresses contain')
 def step_impl(context):
 
 @then(u'result addresses contain')
 def step_impl(context):


@@ -471,8 +510,8 @@ def step_impl(context):
         for h in context.table.headings:
             if h != 'ID':
                 for p in addr_parts:
         for h in context.table.headings:
             if h != 'ID':
                 for p in addr_parts:

-                    assert_in(h, p)
-                    assert_equal(p[h], line[h], "Bad address value for %s" % h)
+                    assert h in p
+                    assert p[h] == line[h], "Bad address value for %s" % h

 
 @then(u'address of result (?P<lid>\d+) has(?P<neg> no)? types (?P<attrs>.*)')
 def check_address(context, lid, neg, attrs):
 
 @then(u'address of result (?P<lid>\d+) has(?P<neg> no)? types (?P<attrs>.*)')
 def check_address(context, lid, neg, attrs):


@@ -482,23 +521,24 @@ def check_address(context, lid, neg, attrs):
 
     for attr in attrs.split(','):
         if neg:
 
     for attr in attrs.split(','):
         if neg:

-            assert_not_in(attr, addr_parts)
+            assert attr not in addr_parts

         else:
         else:

-            assert_in(attr, addr_parts)
+            assert attr in addr_parts

 
 

-@then(u'address of result (?P<lid>\d+) is')
-def check_address(context, lid):
+@then(u'address of result (?P<lid>\d+) (?P<complete>is|contains)')
+def check_address(context, lid, complete):

     context.execute_steps("then more than %s results are returned" % lid)
 
     addr_parts = dict(context.response.result[int(lid)]['address'])
 
     for line in context.table:
     context.execute_steps("then more than %s results are returned" % lid)
 
     addr_parts = dict(context.response.result[int(lid)]['address'])
 
     for line in context.table:

-        assert_in(line['type'], addr_parts)
-        assert_equal(addr_parts[line['type']], line['value'],
-                     "Bad address value for %s" % line['type'])
+        assert line['type'] in addr_parts
+        assert addr_parts[line['type']] == line['value'], \
+                     "Bad address value for %s" % line['type']

         del addr_parts[line['type']]
 
         del addr_parts[line['type']]
 

-    eq_(0, len(addr_parts), "Additional address parts found: %s" % str(addr_parts))
+    if complete == 'is':
+        assert len(addr_parts) == 0, "Additional address parts found: %s" % str(addr_parts)

 
 @then(u'result (?P<lid>\d+ )?has bounding box in (?P<coords>[\d,.-]+)')
 def step_impl(context, lid, coords):
 
 @then(u'result (?P<lid>\d+ )?has bounding box in (?P<coords>[\d,.-]+)')
 def step_impl(context, lid, coords):


@@ -516,10 +556,31 @@ def step_impl(context, lid, coords):
             bbox = bbox.split(',')
         bbox = [ float(x) for x in bbox ]
 
             bbox = bbox.split(',')
         bbox = [ float(x) for x in bbox ]
 

-        assert_greater_equal(bbox[0], coord[0])
-        assert_less_equal(bbox[1], coord[1])
-        assert_greater_equal(bbox[2], coord[2])
-        assert_less_equal(bbox[3], coord[3])
+        assert bbox[0] >= coord[0]
+        assert bbox[1] <= coord[1]
+        assert bbox[2] >= coord[2]
+        assert bbox[3] <= coord[3]
+
+@then(u'result (?P<lid>\d+ )?has centroid in (?P<coords>[\d,.-]+)')
+def step_impl(context, lid, coords):
+    if lid is None:
+        context.execute_steps("then at least 1 result is returned")
+        bboxes = zip(context.response.property_list('lat'),
+                     context.response.property_list('lon'))
+    else:
+        context.execute_steps("then more than %sresults are returned" % lid)
+        res = context.response.result[int(lid)]
+        bboxes = [ (res['lat'], res['lon']) ]
+
+    coord = [ float(x) for x in coords.split(',') ]
+
+    for lat, lon in bboxes:
+        lat = float(lat)
+        lon = float(lon)
+        assert lat >= coord[0]
+        assert lat <= coord[1]
+        assert lon >= coord[2]
+        assert lon <= coord[3]

 
 @then(u'there are(?P<neg> no)? duplicates')
 def check_for_duplicates(context, neg):
 
 @then(u'there are(?P<neg> no)? duplicates')
 def check_for_duplicates(context, neg):