def merge_colspan_headers_in_place(table) -> None:
"""
Turn tuple colnames into strings.
Pandas `read_html()` returns tuples for column names when scraping tables
with colspan. Collapse duplicate entries and reformats to be human
readable. E.g. ('year', 'year') -> 'year' and
('year', 'month') -> 'year - month'
Alter the table in place, no return value.
"""
newcols = []
for c in table.columns:
if isinstance(c, tuple):
# collapse all runs of duplicate values:
# 'a','a','b','c','c','c' -> 'a','b','c'
vals = list(c)
idx = 0
while idx < len(vals) - 1:
if vals[idx] == vals[idx + 1]:
vals.pop(idx)
else:
idx += 1
# put dashes between all remaining header values
newcols.append(" - ".join(vals))
elif isinstance(c, int):
# If first row isn't header and there's no <thead>, table.columns
# will be an integer index.
newcols.append("Column %d" % (c + 1))
else:
newcols.append(c)
# newcols can contain duplicates. Rename them.
table.columns = [c.name for c in gen_unique_clean_colnames(newcols)]
def test_gen_truncate_during_conflict_consider_unicode():
assert gen_unique_clean_colnames(["aéé"] * 10,
settings=MockSettings(5)) == [
UniqueCleanColname("aéé"),
UniqueCleanColname("aé 2",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 3",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 4",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 5",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 6",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 7",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 8",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("aé 9",
is_numbered=True,
is_truncated=True),
UniqueCleanColname("a 10",
is_numbered=True,
is_truncated=True),
]
def test_gen_truncate_during_conflict():
assert gen_unique_clean_colnames(
[
"abcd",
"abcd",
"abcd",
"abcd",
"abcd",
"abcd",
"abcd",
"abcd",
"abcd",
"abcd",
"a 100",
],
settings=MockSettings(4),
) == [
UniqueCleanColname("abcd"),
UniqueCleanColname("ab 2", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 3", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 4", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 5", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 6", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 7", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 8", is_numbered=True, is_truncated=True),
UniqueCleanColname("ab 9", is_numbered=True, is_truncated=True),
UniqueCleanColname("a 11", is_numbered=True, is_truncated=True),
UniqueCleanColname("a 10", is_truncated=True), # was "a 100"
]
def test_gen_name_default_columns_without_conflict():
assert gen_unique_clean_colnames(["Column 2", "", ""]) == [
UniqueCleanColname("Column 2"),
UniqueCleanColname("Column 4", is_default=True, is_numbered=True),
UniqueCleanColname("Column 3",
is_default=True), # this 3 is "reserved"
]
def test_gen_calls_clean():
assert gen_unique_clean_colnames(["ab\n\ud800cd"],
settings=MockSettings(6)) == [
UniqueCleanColname(
"ab�c",
is_ascii_cleaned=True,
is_unicode_fixed=True,
is_truncated=True)
]
def test_gen_avoid_existing_names():
assert gen_unique_clean_colnames(["", "foo"],
existing_names=["Column 3", "foo"]) == [
UniqueCleanColname("Column 4",
is_default=True,
is_numbered=True),
UniqueCleanColname("foo 2",
is_numbered=True),
]
def parse_xls_file(path: Path, *, output_path: Path, has_header: bool,
autoconvert_types: bool) -> RenderResult:
"""
Build a RenderResult from xls bytes or raise parse error.
Peculiarities:
* Error can be xlrd.XLRDError or pandas error
* We read the entire file contents into memory before parsing
"""
# Use xlrd.open_workbook(): if we call pandas.read_excel(bytesio) it
# will read the entire file into RAM.
# dtype='category' crashes as of 2018-09-11
try:
workbook = xlrd.open_workbook(path.as_posix())
table = pd.read_excel(workbook,
engine="xlrd",
dtype=object,
header=(0 if has_header else None))
except xlrd.XLRDError as err:
return RenderResult(errors=[
RenderError(
I18nMessage.TODO_i18n(f"Error reading Excel file: %s" %
str(err)))
])
if has_header:
# pd.read_excel() _badly_ uniquifies column names: it adds ".1", ".2",
# etc. This is hard to fix. We'd need to stop using pd.read_excel().
# [2019-12-09, adamhooper] Not today.
#
# In the meantime, ensure valid colnames so at least the user sees
# _something_. Ignore all warnings.
table.columns = [
cn.name for cn in gen_unique_clean_colnames(
[str(c) for c in table.columns], settings=settings)
]
else:
table.columns = [f"Column {i + 1}" for i in range(len(table.columns))]
autocast_dtypes_in_place(table)
return ProcessResult(table).to_arrow(output_path)
def render(table, params, *, fetch_result):
if not fetch_result:
return table
if fetch_result.status == "error":
return fetch_result
table = fetch_result.dataframe
has_header: bool = params["first_row_is_header"]
if has_header and len(table) >= 1: # if len == 0, no-op
# TODO inform user of column-rename warnings
table.columns = [
uccn.name for uccn in gen_unique_clean_colnames(
[str(c) for c in table.iloc[0, :]])
]
table.drop(index=0, inplace=True)
table.reset_index(drop=True, inplace=True)
autocast_dtypes_in_place(table)
if fetch_result.errors:
return (table, fetch_result.errors)
else:
return table
def test_gen_name_default_columns():
assert gen_unique_clean_colnames(["", ""]) == [
UniqueCleanColname("Column 1", is_default=True),
UniqueCleanColname("Column 2", is_default=True),
]
def test_gen_add_number_that_does_not_overwrite_existing_number():
assert gen_unique_clean_colnames(["A", "A", "A 2"]) == [
UniqueCleanColname("A"),
UniqueCleanColname("A 3", is_numbered=True),
UniqueCleanColname("A 2"),
]
def test_gen_add_number():
assert gen_unique_clean_colnames(["A", "A", "A"]) == [
UniqueCleanColname("A"),
UniqueCleanColname("A 2", is_numbered=True),
UniqueCleanColname("A 3", is_numbered=True),
]
def test_gen_number_1_is_unique():
assert gen_unique_clean_colnames(["A", "A 1", "A 2"]) == [
UniqueCleanColname("A"),
UniqueCleanColname("A 1"),
UniqueCleanColname("A 2"),
]
def test_gen_whitespace_only_key():
# issue #174927345: handle empty-string key
assert gen_unique_clean_colnames([" 1", " 1"]) == [
UniqueCleanColname(" 1"),
UniqueCleanColname(" 2", is_numbered=True),
]
def test_gen_do_not_number_name_without_key():
# issue #174927345: handle empty-string key
assert gen_unique_clean_colnames([" 1", " 1"]) == [
UniqueCleanColname(" 1"),
UniqueCleanColname(" 1 2", is_numbered=True),
]
def _postprocess_name_columns(
table: pyarrow.Table,
has_header: bool) -> Tuple[pyarrow.Table, List[ParseCsvWarning]]:
"""
Return `table`, with final column names but still String values.
"""
warnings = []
if has_header and table.num_rows > 0:
n_ascii_cleaned = 0
first_ascii_cleaned = None
n_truncated = 0
first_truncated = None
n_numbered = 0
first_numbered = None
names = []
for colname in gen_unique_clean_colnames(
list(("" if c[0] is pyarrow.NULL else c[0].as_py())
for c in table.columns),
settings=settings,
):
names.append(colname.name)
if colname.is_ascii_cleaned:
if n_ascii_cleaned == 0:
first_ascii_cleaned = colname.name
n_ascii_cleaned += 1
if colname.is_truncated:
if n_truncated == 0:
first_truncated = colname.name
n_truncated += 1
if colname.is_numbered:
if n_numbered == 0:
first_numbered = colname.name
n_numbered += 1
# Unicode can't be fixed, because we assume valid UTF-8 input
assert not colname.is_unicode_fixed
# Stay silent if colname.is_default. Users expect us to
# auto-generate default column names.
if n_ascii_cleaned:
warnings.append(
ParseCsvWarning.CleanedAsciiColumnNames(
n_ascii_cleaned, first_ascii_cleaned))
if n_truncated:
warnings.append(
ParseCsvWarning.TruncatedColumnNames(n_truncated,
first_truncated))
if n_numbered:
warnings.append(
ParseCsvWarning.NumberedColumnNames(n_numbered,
first_numbered))
# Remove header (zero-copy: builds new pa.Table with same backing data)
table = table.slice(1)
else:
names = [f"Column {i + 1}" for i in range(len(table.columns))]
return (
pyarrow.table({name: table.column(i)
for i, name in enumerate(names)}),
warnings,
)
