(core) Switch excel import parsing from messytables+xlrd to openpyxl, and ignore empty rows

Summary:
Use openpyxl instead of messytables (which used xlrd internally) in import_xls.py.

Skip empty rows since excel files can easily contain huge numbers of them.

Drop support for xls files (which openpyxl doesn't support) in favour of the newer xlsx format.

Fix some details relating to python virtualenvs and dependencies, as Jenkins was failing to find new Python dependencies.

Test Plan: Mostly relying on existing tests. Updated various tests which referred to xls files instead of xlsx. Added a Python test for skipping empty rows.

Reviewers: georgegevoian

Reviewed By: georgegevoian

Differential Revision: https://phab.getgrist.com/D3406

app/client/lib/uploads.ts

@@ -32,7 +32,7 @@ export interface SelectFileOptions extends UploadOptions {
                           // e.g. [".jpg", ".png"]
 }
 
-export const IMPORTABLE_EXTENSIONS = [".grist", ".csv", ".tsv", ".txt", ".xls", ".xlsx", ".xlsm"];
+export const IMPORTABLE_EXTENSIONS = [".grist", ".csv", ".tsv", ".txt", ".xlsx", ".xlsm"];
 
 /**
  * Shows the file-picker dialog with the given options, and uploads the selected files. If under

app/server/lib/ActiveDocImport.ts

@@ -271,8 +271,8 @@ export class ActiveDocImport {
   /**
    * Imports the data stored at tmpPath.
    *
-   * Currently it starts a python parser (that relies on the messytables library) as a child process
-   * outside the sandbox, and supports xls(x), csv, txt, and perhaps some other formats. It may
+   * Currently it starts a python parser as a child process
+   * outside the sandbox, and supports xlsx, csv, and perhaps some other formats. It may
    * result in the import of multiple tables, in case of e.g. Excel formats.
    * @param {OptDocSession} docSession: Session instance to use for importing.
    * @param {String} tmpPath: The path from of the original file.

app/server/lib/DocPluginManager.ts

@@ -136,7 +136,6 @@ export class DocPluginManager {
     if (messages.length) {
       const extToType: Record<string, string> = {
         '.xlsx' : 'Excel',
-        '.xls' : 'Excel',
         '.json' : 'JSON',
         '.csv' : 'CSV',
       };

app/server/lib/NSandbox.ts

@@ -559,7 +559,8 @@ function gvisor(options: ISandboxOptions): SandboxProcess {
   // if checkpoints are in use.
   const venv = path.join(process.cwd(),
                          pythonVersion === '2' ? 'venv' : 'sandbox_venv3');
-  if (fs.existsSync(venv) && !process.env.GRIST_CHECKPOINT) {
+  const useCheckpoint = process.env.GRIST_CHECKPOINT && !paths.importDir;
+  if (fs.existsSync(venv) && !useCheckpoint) {
     wrapperArgs.addMount(venv);
     wrapperArgs.push('-s', path.join(venv, 'bin', 'python'));
   }
@@ -570,17 +571,16 @@ function gvisor(options: ISandboxOptions): SandboxProcess {
   // between the checkpoint and how it gets used later).
   // If a sandbox is being used for import, it will have a special mount we can't
   // deal with easily right now. Should be possible to do in future if desired.
-  if (options.useGristEntrypoint && pythonVersion === '3' && !paths.importDir &&
-      process.env.GRIST_CHECKPOINT) {
+  if (options.useGristEntrypoint && pythonVersion === '3' && useCheckpoint) {
     if (process.env.GRIST_CHECKPOINT_MAKE) {
       const child =
-        spawn(command, [...wrapperArgs.get(), '--checkpoint', process.env.GRIST_CHECKPOINT,
+        spawn(command, [...wrapperArgs.get(), '--checkpoint', process.env.GRIST_CHECKPOINT!,
                         `python${pythonVersion}`, '--', ...pythonArgs]);
       // We don't want process control for this.
       return {child, control: new NoProcessControl(child)};
     }
     wrapperArgs.push('--restore');
-    wrapperArgs.push(process.env.GRIST_CHECKPOINT);
+    wrapperArgs.push(process.env.GRIST_CHECKPOINT!);
   }
   const child = spawn(command, [...wrapperArgs.get(), `python${pythonVersion}`, '--', ...pythonArgs]);
   // For gvisor under ptrace, main work is done by a traced process identifiable as

app/server/lib/guessExt.ts

@@ -21,7 +21,7 @@ export async function guessExt(filePath: string, fileName: string, mimeType: str
     return mimeExt;
   }
 
-  if (origExt === ".csv" || origExt === ".xls") {
+  if (origExt === ".csv") {
     // File type detection doesn't work for these, and mime type can't be trusted. E.g. Windows
     // may report "application/vnd.ms-excel" for .csv files. See
     // https://github.com/ManifoldScholar/manifold/issues/2409#issuecomment-545152220

plugins/core/manifest.yml

@@ -5,11 +5,11 @@ components:
   safePython: sandbox/main.py
 contributions:
   fileParsers:
-    - fileExtensions: ["csv"]
+    - fileExtensions: ["csv", "tsv", "txt"]
       parseFile:
         component: safePython
         name: csv_parser
-    - fileExtensions: ["xls", "xlsx", "tsv", "txt", "xlsm"]
+    - fileExtensions: ["xlsx", "xlsm"]
       parseFile:
         component: safePython
         name: xls_parser

sandbox/grist/imports/__init__.py

@@ -0,0 +1,16 @@
+import warnings
+
+import six
+
+original_formatwarning = warnings.formatwarning
+
+
+def formatwarning(*args, **kwargs):
+  """
+  Fixes an error on Jenkins where byte strings (instead of unicode)
+  were being written to stderr due to a warning from an internal library.
+  """
+  return six.ensure_text(original_formatwarning(*args, **kwargs))
+
+
+warnings.formatwarning = formatwarning

sandbox/grist/imports/import_csv.py

@@ -150,8 +150,11 @@ def _parse_open_file(file_obj, parse_options=None):
       headers = [''] * len(headers)
 
     row_set.register_processor(messytables.offset_processor(data_offset))
-
-    table_data_with_types = parse_data.get_table_data(row_set, len(headers), num_rows)
+    rows = [
+      [cell.value for cell in row]
+      for row in row_set
+    ]
+    table_data_with_types = parse_data.get_table_data(rows, len(headers), num_rows)
 
     # Identify and remove empty columns, and populate separate metadata and data lists.
     column_metadata = []

sandbox/grist/imports/import_xls.py

@@ -2,13 +2,10 @@
 This module reads a file path that is passed in using ActiveDoc.importFile()
 and returns a object formatted so that it can be used by grist for a bulk add records action
 """
-import csv
 import logging
-import os
 
-import chardet
 import messytables
-import messytables.excel
+import openpyxl
 import six
 from six.moves import zip
 
@@ -18,76 +15,58 @@ from imports import import_utils
 log = logging.getLogger(__name__)
 
 
-def import_file(file_source, parse_options):
+def import_file(file_source):
   path = import_utils.get_path(file_source["path"])
-  orig_name = file_source["origName"]
-  parse_options, tables = parse_file(path, orig_name, parse_options)
+  parse_options, tables = parse_file(path)
   return {"parseOptions": parse_options, "tables": tables}
 
-# messytable is painfully un-extensible, so we have to jump through dumb hoops to override any
-# behavior.
-orig_dialect = messytables.CSVRowSet._dialect
-def override_dialect(self):
-  if self.delimiter == '\t':
-    return csv.excel_tab
-  return orig_dialect.fget(self)
-messytables.CSVRowSet._dialect = property(override_dialect)
-
-def parse_file(file_path, orig_name, parse_options=None, table_name_hint=None, num_rows=None):
-  # pylint: disable=unused-argument
+
+def parse_file(file_path):
   with open(file_path, "rb") as f:
-    try:
-      return parse_open_file(f, orig_name, table_name_hint=table_name_hint)
-    except Exception as e:
-      # Log the full error, but simplify the thrown error to omit the unhelpful extra args.
-      log.warn("import_xls parse_file failed: %s", e)
-      if six.PY2 and e.args and isinstance(e.args[0], six.string_types):
-        raise Exception(e.args[0])
-      raise
-
-
-def parse_open_file(file_obj, orig_name, table_name_hint=None):
-  file_root, file_ext = os.path.splitext(orig_name)
-  table_set = messytables.any.any_tableset(file_obj, extension=file_ext, auto_detect=False)
-
-  # Messytable's encoding detection uses too small a sample, so we override it here.
-  if isinstance(table_set, messytables.CSVTableSet):
-    sample = file_obj.read(100000)
-    table_set.encoding = chardet.detect(sample)['encoding']
-    # In addition, always prefer UTF8 over ASCII.
-    if table_set.encoding == 'ascii':
-      table_set.encoding = 'utf8'
+    return parse_open_file(f)
+
+
+def parse_open_file(file_obj):
+  workbook = openpyxl.load_workbook(
+    file_obj,
+    read_only=True,
+    keep_vba=False,
+    data_only=True,
+    keep_links=False,
+  )
 
   skipped_tables = 0
   export_list = []
   # A table set is a collection of tables:
-  for row_set in table_set.tables:
-    table_name = row_set.name
-
-    if isinstance(row_set, messytables.CSVRowSet):
-      # For csv files, we can do better for table_name by using the filename.
-      table_name = import_utils.capitalize(table_name_hint or
-                                           os.path.basename(file_root.decode('utf8')))
-
-      # Messytables doesn't guess whether headers are present, so we need to step in.
-      data_offset, headers = import_utils.headers_guess(list(row_set.sample))
-    else:
-      # Let messytables guess header names and the offset of the header.
-      offset, headers = messytables.headers_guess(row_set.sample)
-      data_offset = offset + 1    # Add the header line
+  for sheet in workbook:
+    table_name = sheet.title
+    rows = [
+      row
+      for row in sheet.iter_rows(values_only=True)
+      # Exclude empty rows, i.e. rows with only empty values.
+      # `if not any(row)` would be slightly faster, but would count `0` as empty.
+      if not set(row) <= {None, ""}
+    ]
+    sample = [
+      # Create messytables.Cells for the sake of messytables.headers_guess
+      [messytables.Cell(cell) for cell in row]
+      for row in rows[:1000]
+    ]
+    offset, headers = messytables.headers_guess(sample)
+    data_offset = offset + 1  # Add the header line
+    rows = rows[data_offset:]
 
     # Make sure all header values are strings.
     for i, header in enumerate(headers):
-      if not isinstance(header, six.string_types):
+      if header is None:
+        headers[i] = u''
+      elif not isinstance(header, six.string_types):
         headers[i] = six.text_type(header)
 
     log.debug("Guessed data_offset as %s", data_offset)
     log.debug("Guessed headers as: %s", headers)
 
-    row_set.register_processor(messytables.offset_processor(data_offset))
-
-
-    table_data_with_types = parse_data.get_table_data(row_set, len(headers))
+    table_data_with_types = parse_data.get_table_data(rows, len(headers))
 
     # Identify and remove empty columns, and populate separate metadata and data lists.
     column_metadata = []
@@ -121,36 +100,3 @@ def parse_open_file(file_obj, orig_name, table_name_hint=None):
 
   parse_options = {}
   return parse_options, export_list
-
-
-# This change was initially introduced in https://phab.getgrist.com/D2145
-# Monkey-patching done in https://phab.getgrist.com/D2965
-# to move towards normal dependency management
-@staticmethod
-def from_xlrdcell(xlrd_cell, sheet, col, row):
-  from messytables.excel import (
-    XLS_TYPES, StringType, DateType, InvalidDateError, xlrd, time, datetime, XLSCell
-  )
-  value = xlrd_cell.value
-  cell_type = XLS_TYPES.get(xlrd_cell.ctype, StringType())
-  if cell_type == DateType(None):
-    # Try-catch added by Dmitry, to avoid failing even if we see a date we can't handle.
-    try:
-      if value == 0:
-        raise InvalidDateError
-      year, month, day, hour, minute, second = \
-        xlrd.xldate_as_tuple(value, sheet.book.datemode)
-      if (year, month, day) == (0, 0, 0):
-        value = time(hour, minute, second)
-      else:
-        value = datetime(year, month, day, hour, minute, second)
-    except Exception:
-      # Keep going, and we'll just interpret the date as a number.
-      pass
-  messy_cell = XLSCell(value, type=cell_type)
-  messy_cell.sheet = sheet
-  messy_cell.xlrd_cell = xlrd_cell
-  messy_cell.xlrd_pos = (row, col)  # necessary for properties, note not (x,y)
-  return messy_cell
-
-messytables.excel.XLSCell.from_xlrdcell = from_xlrdcell

sandbox/grist/imports/register.py

@@ -6,8 +6,9 @@ def register_import_parsers(sandbox):
   sandbox.register("csv_parser.parseFile", parse_csv)
 
   def parse_excel(file_source, parse_options):
+    # pylint: disable=unused-argument
     from imports.import_xls import import_file
-    return import_file(file_source, parse_options)
+    return import_file(file_source)
 
   sandbox.register("xls_parser.parseFile", parse_excel)
 

sandbox/grist/imports/test_import_xls.py

@@ -8,7 +8,7 @@ import unittest
 from imports import import_xls
 
 def _get_fixture(filename):
-  return [os.path.join(os.path.dirname(__file__), "fixtures", filename), filename]
+  return [os.path.join(os.path.dirname(__file__), "fixtures", filename)]
 
 
 class TestImportXLS(unittest.TestCase):
@@ -42,7 +42,7 @@ class TestImportXLS(unittest.TestCase):
                      {"type": "Any", "id": "corner-cases"})
     self.assertEqual(parsed_file[1][0]["table_data"][3],
       # The type is detected as text, so all values should be text.
-      [u'=function()', u'3.0', u'two spaces after  ',
+      [u'=function()', u'3', u'two spaces after  ',
         u'  two spaces before', u'!@#$', u'€€€', u'√∫abc$$', u'line\nbreak'])
 
     # check that multiple tables are created when there are multiple sheets in a document
@@ -53,12 +53,12 @@ class TestImportXLS(unittest.TestCase):
   def test_excel_types(self):
     parsed_file = import_xls.parse_file(*_get_fixture('test_excel_types.xlsx'))
     sheet = parsed_file[1][0]
-    self._check_col(sheet, 0, "int1", "Numeric", [-1234123, '', ''])
-    self._check_col(sheet, 1, "int2", "Numeric", [5, '', ''])
+    self._check_col(sheet, 0, "int1", "Numeric", [-1234123, None, None])
+    self._check_col(sheet, 1, "int2", "Numeric", [5, None, None])
     self._check_col(sheet, 2, "textint", "Any", ["12345678902345689", '', ''])
     self._check_col(sheet, 3, "bigint", "Any", ["320150170634561830", '', ''])
-    self._check_col(sheet, 4, "num2", "Numeric", [123456789.123456, '', ''])
-    self._check_col(sheet, 5, "bignum", "Numeric", [math.exp(200), '', ''])
+    self._check_col(sheet, 4, "num2", "Numeric", [123456789.123456, None, None])
+    self._check_col(sheet, 5, "bignum", "Numeric", [math.exp(200), None, None])
     self._check_col(sheet, 6, "date1", "DateTime",
              [calendar.timegm(datetime.datetime(2015, 12, 22, 11, 59, 00).timetuple()), None, None])
     self._check_col(sheet, 7, "date2", "Date",
@@ -78,7 +78,7 @@ class TestImportXLS(unittest.TestCase):
                      None, 1452038400.0, 1451549340.0, 1483214940.0, None,
                      1454544000.0, 1199577600.0, 1451692800.0, 1451549340.0, 1483214940.0])
     self._check_col(sheet, 1, "float_not_int", "Numeric",
-                    [1,2,3,4,5,"",6,7,8,9,10,10.25,11,12,13,14,15,16,17,18])
+                    [1,2,3,4,5,None,6,7,8,9,10,10.25,11,12,13,14,15,16,17,18])
     self._check_col(sheet, 2, "int_not_bool", "Any",
                     [0, 0, 1, 0, 1, 0, 0, 1, 0, 2, 0, 0, 1, 0, 0, 1, 0, 0, 1, 0])
     self._check_col(sheet, 3, "float_not_bool", "Any",
@@ -91,12 +91,11 @@ class TestImportXLS(unittest.TestCase):
                     [4.0, 6.0, 4.0, 4.0, 6.0, 4.0, '--', 6.0, 4.0, 4.0, 4.0, 4.0, 4.0, 6.0, 6.0,
                      4.0, 6.0, '3-4', 4.0, 6.5])
     self._check_col(sheet, 7, "float_not_text", "Numeric",
-                    [-10.25, -8.00, -5.75, -3.50, "n/a", '  1.  ', "   ???   ", 5.50, "", "-",
-                     12.25, 0.00, "", 0.00, "--", 23.50, "NA", 28.00, 30.25, 32.50])
+                    [-10.25, -8.00, -5.75, -3.50, "n/a", '  1.  ', "   ???   ", 5.50, None, "-",
+                     12.25, 0.00, None, 0.00, "--", 23.50, "NA", 28.00, 30.25, 32.50])
 
   def test_excel_single_merged_cell(self):
-    # An older version of xlrd had a bug where a single cell marked as 'merged' would cause an
-    # exception.
+    # An older version had a bug where a single cell marked as 'merged' would cause an exception.
     parsed_file = import_xls.parse_file(*_get_fixture('test_single_merged_cell.xlsx'))
     tables = parsed_file[1]
     self.assertEqual(tables, [{
@@ -110,21 +109,19 @@ class TestImportXLS(unittest.TestCase):
       ],
       'table_data': [
         [u'SINGLE MERGED', u'The End'],
-        [1637384.52, u''],
-        [2444344.06, u''],
-        [2444344.06, u''],
+        [1637384.52, None],
+        [2444344.06, None],
+        [2444344.06, None],
         [u'', u''],
       ],
     }])
 
   def test_excel_strange_dates(self):
-    # TODO fails with xlrd.xldate.XLDateAmbiguous: 4.180902777777778
     # Check that we don't fail when encountering unusual dates and times (e.g. 0 or 38:00:00).
     parsed_file = import_xls.parse_file(*_get_fixture('strange_dates.xlsx'))
     tables = parsed_file[1]
     # We test non-failure, but the result is not really what we want. E.g. "1:10" and "100:20:30"
-    # would be best left as text, but here become "01:10:00" (after xlrd parses the first as
-    # datetime.time), and as 4.18... (after xlrd fails and we resort to the numerical value).
+    # would be best left as text.
     self.assertEqual(tables, [{
       'table_name': u'Sheet1',
       'column_metadata': [
@@ -132,22 +129,41 @@ class TestImportXLS(unittest.TestCase):
         {'id': 'b', 'type': 'Date'},
         {'id': 'c', 'type': 'Any'},
         {'id': 'd', 'type': 'Any'},
-        {'id': 'e', 'type': 'Numeric'},
-        {'id': 'f', 'type': 'Numeric'},
+        {'id': 'e', 'type': 'DateTime'},
+        {'id': 'f', 'type': 'Date'},
         {'id': 'g', 'type': 'Any'},
         {'id': 'h', 'type': 'Date'},
-        {'id': 'i', 'type': 'Numeric'},
+        {'id': 'i', 'type': 'Date'},
       ],
       'table_data': [
         [u'21:14:00'],
         [1568851200.0],
         [u'01:10:00'],
         [u'10:20:30'],
-        [4.180902777777778],
-        [20],
+        [-2208713970.0],
+        [-2207347200.0],
         [u'7/4/1776'],
         [205286400.0],
-        [0],
+        [-2209161600.0],
+      ],
+    }])
+
+  def test_empty_rows(self):
+    # Check that empty rows aren't imported,
+    # and that files with lots of empty rows are imported quickly.
+    # The fixture file is mostly empty but has data in the last row,
+    # with over a million empty rows in between.
+    parsed_file = import_xls.parse_file(*_get_fixture('test_empty_rows.xlsx'))
+    tables = parsed_file[1]
+    self.assertEqual(tables, [{
+      'table_name': u'Sheet1',
+      'column_metadata': [
+        {'id': 'a', 'type': 'Numeric'},
+        {'id': 'b', 'type': 'Numeric'},
+      ],
+      'table_data': [
+        [0, None, 1],
+        [None, 0, 2],
       ],
     }])
 

sandbox/grist/parse_data.py

@@ -10,7 +10,6 @@ of values. All "data" lists will have the same length.
 import datetime
 import logging
 import re
-import messytables
 import moment # TODO grist internal libraries might not be available to plugins in the future.
 import six
 from six.moves import zip, xrange
@@ -59,12 +58,16 @@ class BaseConverter(object):
     raise NotImplementedError()
 
 
+numeric_types = six.integer_types + (float, complex, type(None))
+
 class NumericConverter(BaseConverter):
   """Handles the Grist Numeric type"""
 
   @classmethod
   def convert(cls, value):
-    if type(value) in six.integer_types + (float, complex):
+    if type(value) is bool:
+      return int(value)
+    elif type(value) in numeric_types:
       return value
     raise ValueError()
 
@@ -80,7 +83,7 @@ class SimpleDateTimeConverter(BaseConverter):
   def convert(cls, value):
     if type(value) is datetime.datetime:
       return value
-    elif value == "":
+    elif not value:
       return None
     raise ValueError()
 
@@ -103,6 +106,8 @@ class AnyConverter(BaseConverter):
   """
   @classmethod
   def convert(cls, value):
+    if value is None:
+      return u''
     return six.text_type(value)
 
   @classmethod
@@ -156,7 +161,7 @@ def _guess_basic_types(rows, num_columns):
   column_detectors = [ColumnDetector() for i in xrange(num_columns)]
   for row in rows:
     for cell, detector in zip(row, column_detectors):
-      detector.add_value(cell.value)
+      detector.add_value(cell)
 
   return [detector.get_converter() for detector in column_detectors]
 
@@ -194,10 +199,10 @@ class ColumnConverter(object):
     return {"type": grist_type, "data": self._all_col_values}
 
 
-def get_table_data(row_set, num_columns, num_rows=0):
-  converters = _guess_basic_types(row_set.sample, num_columns)
+def get_table_data(rows, num_columns, num_rows=0):
+  converters = _guess_basic_types(rows[:1000], num_columns)
   col_converters = [ColumnConverter(c) for c in converters]
-  for num, row in enumerate(row_set):
+  for num, row in enumerate(rows):
     if num_rows and num == num_rows:
       break
 
@@ -207,9 +212,9 @@ def get_table_data(row_set, num_columns, num_rows=0):
     # Make sure we have a value for every column.
     missing_values = len(converters) - len(row)
     if missing_values > 0:
-      row.extend([messytables.Cell("")] * missing_values)
+      row.extend([""] * missing_values)
 
     for cell, conv in zip(row, col_converters):
-      conv.convert_and_add(cell.value)
+      conv.convert_and_add(cell)
 
   return [conv.get_grist_column() for conv in col_converters]

sandbox/requirements.txt

@@ -3,12 +3,15 @@ asttokens==2.0.5
 backports.functools-lru-cache==1.6.4
 chardet==4.0.0
 enum34==1.1.10
+et-xmlfile==1.0.1
 html5lib==0.999999999
 iso8601==0.1.12
+jdcal==1.4.1
 json_table_schema==0.2.1
 lazy_object_proxy==1.6.0
 lxml==4.6.3                # used in csv plugin only?
 messytables==0.15.2
+openpyxl==2.6.4
 python_dateutil==2.6.0
 python_magic==0.4.12
 roman==2.0.0

sandbox/requirements3.txt

@@ -3,12 +3,15 @@ asttokens==2.0.5
 backports.functools-lru-cache==1.6.4
 chardet==4.0.0
 enum34==1.1.10
+et-xmlfile==1.0.1
 html5lib==0.999999999
 iso8601==0.1.12
+jdcal==1.4.1
 json_table_schema==0.2.1
 lazy_object_proxy==1.6.0
 lxml==4.6.3                # used in csv plugin only?
 messytables==0.15.2
+openpyxl==2.6.4
 python_dateutil==2.6.0
 python_magic==0.4.12
 roman==2.0.0