Merge pull request #1475 from burnash/feature/find_table

[feature] Add 'expand_table' feature

docs/user-guide.rst

@@ -350,7 +350,73 @@ Check out the api docs for `DataValidationRule`_ and `CondtionType`_ for more de
 
 .. _CondtionType: https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets/other#ConditionType
 
-.. _DataValidationRule: https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets/cells#DataValidationRule 
+.. _DataValidationRule: https://developers.google.com/sheets/api/reference/rest/v4/spreadsheets/cells#DataValidationRule
+
+Extract table
+~~~~~~~~~~~~~
+
+Gspread provides a function to extract a data table.
+A data table is defined as a rectangular table that stops either on the **first empty** cell or
+the **enge of the sheet**.
+
+You can extract table from any address by providing the top left corner of the desired table.
+
+Gspread provides 3 directions for searching the end of the table:
+
+   * :attr:`~gspread.utils.TableDirection.right`: extract a single row searching on the right of the starting cell
+   * :attr:`~gspread.utils.TableDirection.down`: extract a single column searching on the bottom of the starting cell
+   * :attr:`~gspread.utils.TableDirection.table`: extract a rectangular table by first searching right from starting cell,
+     then searching down from starting cell.
+
+      .. note::
+
+        Gspread will not look for empty cell inside the table. it only look at the top row and first column.
+
+Example extracting a table from the below sample sheet:
+
+.. list-table:: Find table
+   :header-rows: 1
+
+   * - ID
+     - Name
+     - Universe
+     - Super power
+   * - 1
+     - Batman
+     - DC
+     - Very rich
+   * - 2
+     - DeadPool
+     - Marvel
+     - self healing
+   * - 3
+     - Superman
+     - DC
+     - super human
+   * -
+     - \-
+     - \-
+     - \-
+   * - 5
+     - Lavigne958
+     -
+     - maintains Gspread
+   * - 6
+     - Alifee
+     -
+     - maintains Gspread
+
+Using the below code will result in rows 2 to 4:
+
+.. code:: python
+
+   worksheet.expand("A2")
+
+   [
+      ["Batman", "DC", "Very rich"],
+      ["DeadPool", "Marvel", "self healing"],
+      ["Superman", "DC", "super human"],
+   ]
 
 
 

gspread/utils.py

@@ -168,6 +168,12 @@ class ValidationConditionType(StrEnum):
     filter_expression = "FILTER_EXPRESSION"
 
 
+class TableDirection(StrEnum):
+    table = "TABLE"
+    down = "DOWN"
+    right = "RIGHT"
+
+
 def convert_credentials(credentials: Credentials) -> Credentials:
     module = credentials.__module__
     cls = credentials.__class__.__name__
@@ -979,6 +985,129 @@ def to_records(
     return [dict(zip(headers, row)) for row in values]
 
 
+def _expand_right(values: List[List[str]], start: int, end: int, row: int) -> int:
+    """This is a private function, returning the column index of the last non empty cell
+    on the given row.
+
+    Search starts from ``start`` index column.
+    Search ends on ``end`` index column.
+    Searches only in the row pointed by ``row``.
+    """
+    try:
+        return values[row].index("", start, end) - 1
+    except ValueError:
+        return end
+
+
+def _expand_bottom(values: List[List[str]], start: int, end: int, col: int) -> int:
+    """This is a private function, returning the row index of the last non empty cell
+    on the given column.
+
+    Search starts from ``start`` index row.
+    Search ends on ``end`` index row.
+    Searches only in the column pointed by ``col``.
+    """
+    for rows in range(start, end):
+        # in case we try to look further than last row
+        if rows >= len(values):
+            return len(values) - 1
+
+        # check if cell is empty (or the row => empty cell)
+        if col >= len(values[rows]) or values[rows][col] == "":
+            return rows - 1
+
+    return end - 1
+
+
+def find_table(
+    values: List[List[str]],
+    start_range: str,
+    direction: TableDirection = TableDirection.table,
+) -> List[List[str]]:
+    """Expands a list of values based on non-null adjacent cells.
+
+    Expand can be done in 3 directions defined in :class:`~gspread.utils.TableDirection`
+
+        * ``TableDirection.right``: expands right until the first empty cell
+        * ``TableDirection.down``: expands down until the first empty cell
+        * ``TableDirection.table``: expands right until the first empty cell and down until first empty cell
+
+    In case of empty result an empty list is restuned.
+
+    When the given ``start_range`` is outside the given matrix of values the exception
+    :class:`~gspread.exceptions.InvalidInputValue` is raised.
+
+    Example::
+
+        values = [
+            ['', '',   '',   '', ''  ],
+            ['', 'B2', 'C2', '', 'E2'],
+            ['', 'B3', 'C3', '', 'E3'],
+            ['', ''  , ''  , '', 'E4'],
+        ]
+        >>> utils.find_table(TableDirection.table, 'B2')
+        [
+            ['B2', 'C2'],
+            ['B3', 'C3'],
+        ]
+
+
+    .. note::
+
+       the ``TableDirection.table`` will look right from starting cell then look down from starting cell.
+       It will not check cells located inside the table. This could lead to
+       potential empty values located in the middle of the table.
+
+    .. warning::
+
+       Given values must be padded with `''` empty values.
+
+    :param list[list] values: values where to find the table.
+    :param gspread.utils.TableDirection direction: the expand direction.
+    :param str start_range: the starting cell range.
+    :rtype list(list): the resulting matrix
+    """
+    row, col = a1_to_rowcol(start_range)
+
+    # a1_to_rowcol returns coordinates starting form 1
+    row -= 1
+    col -= 1
+
+    if row >= len(values):
+        raise InvalidInputValue(
+            "given row for start_range is outside given values: start range row ({}) >= rows in values {}".format(
+                row, len(values)
+            )
+        )
+
+    if col >= len(values[row]):
+        raise InvalidInputValue(
+            "given column for start_range is outside given values: start range column ({}) >= columns in values {}".format(
+                col, len(values[row])
+            )
+        )
+
+    if direction == TableDirection.down:
+        rightMost = col
+        bottomMost = _expand_bottom(values, row, len(values), col)
+
+    if direction == TableDirection.right:
+        bottomMost = row
+        rightMost = _expand_right(values, col, len(values[row]), row)
+
+    if direction == TableDirection.table:
+        rightMost = _expand_right(values, col, len(values[row]), row)
+        bottomMost = _expand_bottom(values, row, len(values), col)
+
+    result = []
+
+    # build resulting array
+    for rows in values[row : bottomMost + 1]:
+        result.append(rows[col : rightMost + 1])
+
+    return result
+
+
 # SHOULD NOT BE NEEDED UNTIL NEXT MAJOR VERSION
 # DEPRECATION_WARNING_TEMPLATE = (
 #     "[Deprecated][in version {v_deprecated}]: {msg_deprecated}"

gspread/worksheet.py

@@ -41,6 +41,7 @@
     PasteOrientation,
     PasteType,
     T,
+    TableDirection,
     ValidationConditionType,
     ValueInputOption,
     ValueRenderOption,
@@ -53,6 +54,7 @@
     convert_colors_to_hex_value,
     convert_hex_to_colors_dict,
     fill_gaps,
+    find_table,
     finditem,
     get_a1_from_absolute_range,
     is_full_a1_notation,
@@ -3336,3 +3338,56 @@ def add_validation(
         }
 
         return self.client.batch_update(self.spreadsheet_id, body)
+
+    def expand(
+        self,
+        top_left_range_name: str = "A1",
+        direction: TableDirection = TableDirection.table,
+    ) -> List[List[str]]:
+        """Expands a cell range based on non-null adjacent cells.
+
+        Expand can be done in 3 directions defined in :class:`~gspread.utils.TableDirection`
+
+        * ``TableDirection.right``: expands right until the first empty cell
+        * ``TableDirection.down``: expands down until the first empty cell
+        * ``TableDirection.table``: expands right until the first empty cell and down until the first empty cell
+
+        In case of empty result an empty list is restuned.
+
+        When the given ``start_range`` is outside the given matrix of values the exception
+        :class:`~gspread.exceptions.InvalidInputValue` is raised.
+
+        Example::
+
+            values = [
+                ['', '',   '',   '', ''  ],
+                ['', 'B2', 'C2', '', 'E2'],
+                ['', 'B3', 'C3', '', 'E3'],
+                ['', ''  , ''  , '', 'E4'],
+            ]
+            >>> utils.find_table(TableDirection.table, 'B2')
+            [
+                ['B2', 'C2'],
+                ['B3', 'C3'],
+            ]
+
+
+        .. note::
+
+            the ``TableDirection.table`` will look right from starting cell then look down from starting cell.
+            It will not check cells located inside the table. This could lead to
+            potential empty values located in the middle of the table.
+
+        .. note::
+
+            when it is necessary to use non-default options for :meth:`~gspread.worksheet.Worksheet.get`,
+            please get the data first using desired options then use the function
+            :func:`gspread.utils.find_table` to extract the desired table.
+
+        :param str top_left_range_name: the top left corner of the table to expand.
+        :param gspread.utils.TableDirection direction: the expand direction
+        :rtype list(list): the resulting matrix
+        """
+
+        values = self.get(pad_values=True)
+        return find_table(values, top_left_range_name, direction)