support public published spreadsheets

gspread/__init__.py

@@ -18,7 +18,7 @@
     from urllib.parse import urlencode
 
 
-from .client import Client, authorize
+from .client import Client, authorize, public
 from .models import Spreadsheet, Worksheet, Cell
 from .exceptions import (GSpreadException, AuthenticationError,
                          SpreadsheetNotFound, NoValidUrlKeyFound,

gspread/client.py

@@ -25,7 +25,7 @@
     DRIVE_FILES_API_V2_URL,
     DRIVE_FILES_UPLOAD_API_V2_URL
 )
-from .utils import finditem, extract_id_from_url
+from .utils import finditem, extract_id_from_url, is_public_url
 from .exceptions import (SpreadsheetNotFound, UpdateCellError)
 
 
@@ -35,6 +35,9 @@ class Client(object):
 
     :param auth: An OAuth2 credential object. Credential objects are those created by the
                  oauth2client library. https://github.com/google/oauth2client
+                 To access public, published spreadsheets, set auth to None.  Be careful,
+                 it is insufficient for the spreadsheet to simply be public, it must
+                 also be published (in the sense of File->Publish to the web->Entire Document->Publish)
     :param http_session: (optional) A session object capable of making HTTP requests while persisting headers.
                                     Defaults to :class:`~gspread.httpsession.HTTPSession`.
 
@@ -102,13 +105,17 @@ def open_by_key(self, key):
         >>> c.open_by_key('0BmgG6nO_6dprdS1MN3d3MkdPa142WFRrdnRRUWl1UFE')
 
         """
-        feed = self.get_spreadsheets_feed()
+        feed = self.get_spreadsheets_feed(hint=key)
         for elem in feed.findall(_ns('entry')):
             alter_link = finditem(lambda x: x.get('rel') == 'alternate',
                                   elem.findall(_ns('link')))
             spreadsheet_id = extract_id_from_url(alter_link.get('href'))
             if spreadsheet_id == key:
                 return Spreadsheet(self, elem)
+
+            if is_public_url(alter_link.get('href')):
+                return Spreadsheet(self, elem)
+
         else:
             raise SpreadsheetNotFound
 
@@ -146,9 +153,32 @@ def openall(self, title=None):
                     continue
             result.append(Spreadsheet(self, elem))
 
+        if self.auth is None:
+            raise SpreadsheetNotFound
+
         return result
 
-    def get_spreadsheets_feed(self, visibility='private', projection='full'):
+    def get_spreadsheets_feed(self, visibility='private', projection='full', hint=None):
+        if not self.auth:
+            # If we have a spreadsheet key, we can try public route
+            if not hint:
+                # No joy, back out
+                return ElementTree.Element('feed')
+            url = construct_url('worksheets', spreadsheet_id=hint,
+                                visibility='public', projection='full')
+            r = self.session.get(url)
+            # Construct a single-entry feed in the expected format
+            feed = ElementTree.Element('feed')
+            try:
+                entry = ElementTree.Element(_ns('entry'))
+                for elem in ElementTree.fromstring(r.content):
+                    entry.append(elem)
+                feed.append(entry)
+            except ElementTree.ParseError:
+                # we get a html error page if sheet is public-but-not-published
+                return ElementTree.Element('feed')
+            return feed
+
         url = construct_url('spreadsheets',
                             visibility=visibility, projection=projection)
 
@@ -157,6 +187,9 @@ def get_spreadsheets_feed(self, visibility='private', projection='full'):
 
     def get_worksheets_feed(self, spreadsheet,
                             visibility='private', projection='full'):
+        if not self.auth:
+            # fall back to public
+            visibility = 'public'
         url = construct_url('worksheets', spreadsheet,
                             visibility=visibility, projection=projection)
 
@@ -166,6 +199,9 @@ def get_worksheets_feed(self, spreadsheet,
     def get_cells_feed(self, worksheet,
                        visibility='private', projection='full', params=None):
 
+        if not self.auth:
+            # fall back to public
+            visibility = 'public'
         url = construct_url('cells', worksheet,
                             visibility=visibility, projection=projection)
 
@@ -401,3 +437,15 @@ def authorize(credentials):
     client = Client(auth=credentials)
     client.login()
     return client
+
+
+def public():
+    """Prepare to access public, published spreadsheets.
+
+    No private spreadsheets will be accessible , for that you
+    need to authorize or login instead.
+
+    :returns: :class:`Client` instance.
+
+    """
+    return Client(auth=None)

gspread/models.py

@@ -82,6 +82,9 @@ def __init__(self, client, feed_entry):
         self._sheet_list = []
         self._feed_entry = feed_entry
         self._id = feed_entry.find(_ns('id')).text.split('/')[-1]
+        if self._id == 'full':
+            # public sheet case, id is elsewhere
+            self._id = feed_entry.find(_ns('id')).text.split('/')[-3]
         self._title = feed_entry.find(_ns('title')).text
         self._updated = feed_entry.find(_ns('updated')).text
 

gspread/utils.py

@@ -18,6 +18,7 @@
 
 URL_KEY_V1_RE = re.compile(r'key=([^&#]+)')
 URL_KEY_V2_RE = re.compile(r'/spreadsheets/d/([a-zA-Z0-9-_]+)')
+URL_KEY_V2_PUBLIC_RE = re.compile(r'spreadsheets/d/([a-zA-Z0-9-_]+)/pubhtml')
 
 
 def finditem(func, seq):
@@ -190,6 +191,11 @@ def wid_to_gid(wid):
     return str(int(str(widval), 36) ^ xorval)
 
 
+def is_public_url(url):
+    m = URL_KEY_V2_PUBLIC_RE.search(url)
+    return bool(m)
+
+
 if __name__ == '__main__':
     import doctest
     doctest.testmod()

tests/test.py

@@ -120,6 +120,21 @@ def setUp(self):
         self.assertTrue(isinstance(self.gc, gspread.Client))
 
 
+class PublicGspreadTest(unittest.TestCase):
+
+    @classmethod
+    def setUpClass(cls):
+        try:
+            cls.config = read_config(CONFIG_FILENAME)
+            cls.gc = gspread.public()
+        except IOError as e:
+            msg = "Can't find %s for reading test configuration. "
+            raise Exception(msg % e.filename)
+
+    def setUp(self):
+        self.assertTrue(isinstance(self.gc, gspread.Client))
+
+
 class ClientTest(GspreadTest):
 
     """Test for gspread.client."""
@@ -156,6 +171,31 @@ def test_create(self):
         self.assertTrue(isinstance(new_spreadsheet, gspread.Spreadsheet))
 
 
+class PublicClientTest(PublicGspreadTest):
+
+    """Test for gspread.client without authorization."""
+
+    def test_open(self):
+        title = self.config.get('PublicSpreadsheet', 'title')
+        self.assertRaises(gspread.SpreadsheetNotFound,
+                          self.gc.open,
+                          title)
+
+    def test_open_by_key(self):
+        key = self.config.get('PublicSpreadsheet', 'key')
+        spreadsheet = self.gc.open_by_key(key)
+        self.assertTrue(isinstance(spreadsheet, gspread.Spreadsheet))
+
+    def test_open_by_url(self):
+        url = self.config.get('PublicSpreadsheet', 'url')
+        spreadsheet = self.gc.open_by_url(url)
+        self.assertTrue(isinstance(spreadsheet, gspread.Spreadsheet))
+
+    def test_openall(self):
+        self.assertRaises(gspread.SpreadsheetNotFound,
+                          self.gc.openall)
+
+
 class SpreadsheetTest(GspreadTest):
 
     """Test for gspread.Spreadsheet."""
@@ -189,6 +229,37 @@ def test_worksheet_iteration(self):
                          [sheet for sheet in self.spreadsheet])
 
 
+class PublicSpreadsheetTest(PublicGspreadTest):
+
+    """Test for gspread.Spreadsheet without authorization."""
+
+    def setUp(self):
+        super(PublicSpreadsheetTest, self).setUp()
+        key = self.config.get('PublicSpreadsheet', 'key')
+        self.spreadsheet = self.gc.open_by_key(key)
+
+    def test_properties(self):
+        self.assertEqual(self.config.get('PublicSpreadsheet', 'title'),
+                         self.spreadsheet.title)
+
+    def test_sheet1(self):
+        sheet1 = self.spreadsheet.sheet1
+        self.assertTrue(isinstance(sheet1, gspread.Worksheet))
+
+    def test_get_worksheet(self):
+        sheet1 = self.spreadsheet.get_worksheet(0)
+        self.assertTrue(isinstance(sheet1, gspread.Worksheet))
+
+    def test_worksheet(self):
+        sheet_title = self.config.get('Spreadsheet', 'sheet1_title')
+        sheet = self.spreadsheet.worksheet(sheet_title)
+        self.assertTrue(isinstance(sheet, gspread.Worksheet))
+
+    def test_worksheet_iteration(self):
+        self.assertEqual(self.spreadsheet.worksheets(),
+            [sheet for sheet in self.spreadsheet])
+
+
 class WorksheetTest(GspreadTest):
 
     """Test for gspread.Worksheet."""
@@ -531,6 +602,21 @@ def test_export(self):
         self.assertEqual(exported_values, value_list)
 
 
+class PublicWorksheetTest(PublicGspreadTest):
+
+    """Test for gspread.Worksheet without authorization."""
+
+    def setUp(self):
+        super(PublicWorksheetTest, self).setUp()
+        key = self.config.get('PublicSpreadsheet', 'key')
+        self.spreadsheet = self.gc.open_by_key(key)
+        self.sheet = self.spreadsheet.sheet1
+
+    def test_get_all_values(self):
+        read_data = self.sheet.get_all_values()
+        self.assertTrue(len(read_data) > 0)
+
+
 class WorksheetDeleteTest(GspreadTest):
 
     def setUp(self):

tests/tests.config.example

@@ -10,6 +10,12 @@ url: full url of this spreadsheet
 sheet1_title: title of the first worksheet
 new_spreadsheet_title: could be anything you like
 
+[PublicSpreadsheet]
+title: gspread public test 2 (public and published)
+key: 16i5mfs194m44guNIBUylgZrvNslQlkp7F9JNbQar7E4
+url: https://docs.google.com/spreadsheets/d/16i5mfs194m44guNIBUylgZrvNslQlkp7F9JNbQar7E4/edit?usp=sharing
+sheet1_title: Sheet1
+
 [Worksheet]
 id: string following "#gid="
 title: title of this worksheet