Merge pull request #90 from ni1o1/0.5.3

0.5.3

docs/source/conf.py

@@ -26,8 +26,8 @@
 author = 'Qing Yu'
 
 # The full version, including alpha/beta/rc tags
-release = '0.5.2'
-version = '0.5.2'
+release = '0.5.3'
+version = '0.5.3'
 html_logo = "_static/logo-wordmark-light.png"
 html_favicon = '_static/logo2.ico'
 # -- General configuration ---------------------------------------------------

setup.py

@@ -5,7 +5,7 @@
 
 setuptools.setup(
     name="transbigdata",
-    version="0.5.2",
+    version="0.5.3",
     author="Qing Yu",
     author_email="[email protected]",
     description="A Python package developed for transportation spatio-temporal big data processing and analysis.",

src/transbigdata/__init__.py

@@ -33,7 +33,7 @@
 OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 """
 
-__version__ = '0.5.2'
+__version__ = '0.5.3'
 __author__ = 'Qing Yu <[email protected]>'
 
 # module level doc-string

src/transbigdata/crawler.py

@@ -48,6 +48,9 @@
     wgs84togcj02
 )
 
+headers={
+  'User-Agent': 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_12_0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3497.100 Safari/537.36',
+  }
 
 def getadmin(keyword, ak, jscode='', subdistricts=False, timeout=20):
     '''
@@ -96,7 +99,7 @@ def getadmin(keyword, ak, jscode='', subdistricts=False, timeout=20):
         'sdkversion': '1.4.10'
     }
     # 发送请求
-    response = requests.get(url,params = dict1,timeout=timeout)
+    response = requests.get(url,params = dict1,timeout=timeout,headers = headers)
     result = json.loads(response.text)
     if result['info'] == 'INVALID_USER_SCODE':
         raise ValueError('缺少jscode，请将高德开放平台Key中的安全密钥以jscode参数的形式传入该方法')   # pragma: no cover
@@ -181,7 +184,8 @@ def getbusdata(city, keywords, accurate=True, timeout=20):
     def getlineuid(keyword, c, acc=True):
         url = 'http://map.baidu.com/?qt=s&wd=' + \
             urllib.parse.quote(keyword)+'&c='+c+'&from=webmap'
-        response = requests.get(url)
+        response = requests.get(url, timeout=timeout,headers = headers)
+
         searchinfo = json.loads(response.text)
         try:
             res = pd.DataFrame(searchinfo['content'])
@@ -197,13 +201,13 @@ def getlineuid(keyword, c, acc=True):
 
     def getcitycode(c):
         url = 'http://map.baidu.com/?qt=s&wd='+urllib.parse.quote(c)
-        response1 = requests.get(url, timeout=timeout)
+        response1 = requests.get(url, timeout=timeout,headers = headers)
         searchinfo = json.loads(response1.text)
         return str(searchinfo['content']['code'])
 
     def getlinegeo(uid, c):
         url = 'http://map.baidu.com/?qt=bsl&uid='+uid+'&c='+c+"&auth=1"
-        response1 = requests.get(url, timeout=timeout)
+        response1 = requests.get(url, timeout=timeout,headers = headers)
         searchinfo = json.loads(response1.text)
         linename = searchinfo['content'][0]['name']
         stations = searchinfo['content'][0]['stations']
@@ -247,8 +251,9 @@ def coodconvert(coo):
     if type(keywords) != list:
         keywords = [str(keywords)]   # pragma: no cover
     for keyword in keywords:
-        print(keyword)
+        print('Get bus data: '+str(keyword))
         for uid in getlineuid(keyword, c, accurate):
+
             if uid not in uids:
                 try:
                     linename, coo, stationnames, stationgeo = getlinegeo(
@@ -334,7 +339,7 @@ def get_isochrone_amap(lon, lat, reachtime, ak, jscode='', mode=2, timeout=20):
         'extensions': 'all',
         'strategy': str(strategy)
     }
-    response = requests.get(url,params = dict1,timeout=timeout)
+    response = requests.get(url,params = dict1,timeout=timeout,headers = headers)
     result = json.loads(response.text)
 
     P_all = []
@@ -389,7 +394,7 @@ def get_isochrone_mapbox(lon, lat, reachtime, access_token='auto',
     url = 'https://api.mapbox.com/isochrone/v1/mapbox/'+mode+'/' +\
         str(lon)+','+str(lat)+'?contours_minutes='+str(reachtime) +\
         '&polygons=true&access_token='+access_token
-    response = requests.get(url,timeout = timeout)
+    response = requests.get(url,timeout = timeout,headers = headers)
     result = json.loads(response.text)
     isochrone = gpd.GeoDataFrame.from_features(result)
     isochrone['lon'] = lon

src/transbigdata/quality.py

@@ -116,7 +116,7 @@ def data_summary(data, col=['Vehicleid', 'Time'], show_sample_duration=False,
         fig = plt.figure(1,(8,3),dpi=300)
         ax = plt.subplot(111)
         plt.subplots_adjust(left=0.19,right=0.98,top=0.9,bottom=0.19)
-        sns.kdeplot(sd[sd['duration']<sd['duration'].quantile(0.95)],ax=ax,legend=False)
+        sns.kdeplot(sd[sd['duration']<sd['duration'].quantile(0.95)]['duration'],ax=ax,legend=False)
         from matplotlib import ticker
         formatter = ticker.ScalarFormatter(useMathText=True)
         formatter.set_scientific(True)

src/transbigdata/traj.py

@@ -633,6 +633,7 @@ def traj_sparsify(data, col=['Vehicleid', 'Time', 'Lng', 'Lat'], timegap=15,
             [Vehicleid+'_new', 'utctime', 'utctime_new'], axis=1)
     return data1
 
+
 def traj_stay_move(data, params,
                      col=['ID', 'dataTime', 'longitude', 'latitude'],
                      activitytime=1800):
@@ -670,9 +671,12 @@ def traj_stay_move(data, params,
     stay['status_id'] = ((stay['LONCOL'] != stay['LONCOL'].shift()) |
                          (stay['LATCOL'] != stay['LATCOL'].shift()) |
                          (stay[uid] != stay[uid].shift())).astype(int)
+    stay.loc[stay[uid] != stay[uid].shift(-1),'status_id'] = 1
+
     stay['status_id'] = stay.groupby([uid])['status_id'].cumsum()
     stay = stay.drop_duplicates(
         subset=[uid, 'status_id'], keep='first').copy()
+
     stay['etime'] = stay['stime'].shift(-1)
     stay = stay[stay[uid] == stay[uid].shift(-1)].copy()
     # Remove the duration shorter than given activitytime
@@ -681,7 +685,7 @@ def traj_stay_move(data, params,
     stay = stay[stay['duration'] >= activitytime].copy()
     stay = stay[[uid, 'stime', 'LONCOL', 'LATCOL',
                  'etime', 'lon', 'lat', 'duration']]
-    
+
     # Add the first and last two data points for each ID in the Stay dataset before conducting move detection, so that the movement patterns of individuals at the beginning and end of the study period can also be identified.
     first_data = data.drop_duplicates(subset=[uid],keep='first').copy()
     last_data = data.drop_duplicates(subset=[uid],keep='last').copy()
@@ -730,6 +734,7 @@ def traj_stay_move(data, params,
 
 
 
+
 def traj_to_linestring(traj_points, col=['Lng', 'Lat', 'ID'], timecol=None):
     '''
     Input trajectory, generate GeoDataFrame
@@ -785,3 +790,11 @@ def traj_to_linestring(traj_points, col=['Lng', 'Lat', 'ID'], timecol=None):
         traj['geometry'] = geometry
         traj = gpd.GeoDataFrame(traj)
     return traj
+
+''' old namespace '''
+
+import warnings
+
+def points_to_traj(*args, **kwargs):
+    warnings.warn("The 'points_to_traj' function is deprecated. Use 'traj_to_linestring' instead.", DeprecationWarning)
+    return traj_to_linestring(*args, **kwargs)