def fromsarray(cls, array: np.ndarray, dtype: Optional[Union[str, type, np.ndarray.dtype]] = None, headerpos: Optional[Union[Sequence[int], np.ndarray]] = None) -> Table:
_r = re.compile('#<::([<>|]?[biufcmMOSUV]\\d*)::>')
_findt = lambda x: (lambda v: v[0] if len(v) > 0 else '')(_r.findall(x))
if missing(headerpos):
mtab = np.vectorize(_findt)(array[:100,:100])
dpos = np.c_[np.where(mtab != '')]
if dpos.shape[0] >= 2: raise ValueError('string array has multiple headers')
if dpos.shape[0] == 0: raise ValueError('string array has no header in the first 100 rows / cols')
headerpos = dpos[0]
rids, cids = headerpos
if missing(dtype):
dtype = _findt(array[rids,cids])
if dtype == '': raise ValueError('unknown array data type')
ridx = StructuredArray.fromsarray(array[rids:,:cids].T) if cids > 0 else None
cidx = StructuredArray.fromsarray(array[:rids,cids:]) if rids > 0 else None
rnam = array[rids+1:,cids]
if np.all(rnam == smap(range(rnam.shape[0]), lambda x: f'[{x}]')): rnam = None
cnam = array[rids,cids+1:]
if np.all(cnam == smap(range(cnam.shape[0]), lambda x: f'[{x}]')): cnam = None
dmtx = array[rids+1:,cids+1:]
return Table(dmtx, dtype = dtype, rownames = rnam, colnames = cnam, rowindex = ridx, colindex = cidx)
def _fmt(mtx, rnam, cnam, ridx, cidx):
nr, nc = mtx.shape
if missing(rnam): rnam = smap(range(nr), lambda x: f'[{x}]')
if missing(cnam): cnam = smap(range(nc), lambda x: f'[{x}]')
_sln = lambda x,sr,hd,tl,rp: (smap(x[:hd],str) + [rp] + smap(x[tl:],str)) if sr else smap(x, str)
_scol = lambda x: _sln(x, nc > strinkcols, 3, -1, ' ... ')
_srow = lambda x: _sln(x, nr > strinkrows, 5, -3, '')
slns = [_scol(cnam)] + \
([_scol(ln) for ln in mtx] if nr <= strinkrows else
([_scol(ln) for ln in mtx[:5]] + [_scol([' ... ... '] + [''] * (nc-1))] + [_scol(ln) for ln in mtx[-3:]]))
slns = [['#'] + slns[0]] + [[n] + ln for n,ln in zip(_srow(rnam), slns[1:])]
nri = ridx.size if available(ridx) else 0
nci = cidx.size if available(cidx) else 0
if nci > 0: slns = [[f'<{k}>'] + _scol(cidx[k]) for k in cidx.names] + slns
if nri > 0:
sidx = [[''] * nci + [f'<{k}>'] + _srow(ridx[k]) for k in ridx.names]
slns = [list(ix) + ln for ix,ln in zip(zip(*sidx), slns)]
def _sfmt(lns, pos):
size = max(collapse(smap(lns, lambda x: smap(x[pos], lambda v: len(v) if v not in (' ... ', ' ... ... ') else 0)))) + 1
for ln in lns: ln[pos] = smap(ln[pos], lambda x: '{0:>{1}s}'.format(x, size) if x != ' ... ' else x)
return lns
if nri > 0: slns = _sfmt(slns, slice(None,nri))
slns = _sfmt(slns, slice(nri,nri+1))
slns = _sfmt(slns, slice(nri+1,None))
return smap(slns, lambda ln: paste(ln, sep = delimiter))
def _exec(self, pms):
params, stdin, timeout = pms # for multiproc
exlst = [self._bin] + ([] if missing(params) else smap(
params, lambda x: str(x).strip()))
if self._shell:
exlst = paste(smap(exlst, lambda x: x.replace(' ', r'\ ')),
sep=' ')
procs = Popen(exlst,
stdin=PIPE,
stdout=PIPE,
stderr=PIPE,
shell=self._shell)
try:
rvals = procs.communicate(input=stdin, timeout=timeout)
rstrs = smap(
rvals, lambda x: ''
if x is None else x.decode('utf-8').strip())
rcode = procs.returncode
except TimeoutExpired:
procs.kill()
rstrs = ['subprocess terminated as timeout expired', '']
rcode = 124
prstr = paste(rstrs, sep=' | ')
if rcode in self._ncode:
logging.log((logging.DEBUG if self._mute else logging.INFO), prstr)
else:
raise RuntimeError(f'execution failed [{rcode}]:\n{prstr}')
return rcode, rstrs
def __str__(self):
if len(self._arrs) == 0: return '[ ]'
nlen = max(smap(self._arrs.keys(), len))
olns = [
(('{' + f':{nlen}s' + '} : ').format(k) if i == 0 else
(' ' * (nlen + 3))) + ln
for k, v in zip(self._arrs.keys(), smap(self._arrs.values(), str))
for i, ln in enumerate(v.split('\n'))
]
return paste(olns, sep='\n')
def _parsevals(self, value):
if isinstance(value, StructuredArray):
return [value._arrs[n].copy() for n in self._arrs.keys()]
if not iterable(value): return value
value = ll(value)
if not iterable(value[0]): return np.asarray(value)
value = smap(value, lambda x: np.asarray(ll(x)))
if not len(set(smap(value, len))) == 1:
raise ValueError('input arrays not in the same size')
return value
def tosarray(self, withindex: bool = True) -> np.ndarray:
rnam = np.asarray(self._rnames) if available(self._rnames) else np.array(smap(range(self.nrow), lambda x: f'[{x}]'))
cnam = np.asarray(self._cnames) if available(self._cnames) else np.array(smap(range(self.ncol), lambda x: f'[{x}]'))
smtx = np.vstack([np.hstack([f'#<::{self.dtype.str}::>', cnam]), np.hstack([rnam.reshape((-1,1)), np.asarray(self._dmatx, dtype = str)])])
if not withindex: return smtx
if available(self._rindex):
ridx = self._rindex.tosarray()
smtx = np.hstack([ridx.T, smtx])
if available(self._cindex):
cidx = self._cindex.tosarray()
if available(self._rindex): cidx = np.hstack([np.tile([''], (self._cindex.size, self._rindex.size)), cidx])
smtx = np.vstack([cidx, smtx])
return smtx
def loadlns(fname: Union[str, Path],
mode: str = 'r',
strip: bool = True) -> List[str]:
checkInputFile(fname)
with open(fname, mode) as f:
lns = f.readlines()
lns = smap(lns, lambda x: x.rstrip('\n'))
if strip: lns = drop(lns, lambda x: x.strip() == '')
return lns
def fromsarray(cls, array: np.ndarray) -> StructuredArray:
_r = re.compile('<(.*)::([<>|]?[biufcmMOSUV]\\d*)>')
nams, vals = array[:, 0], array[:, 1:]
nams, vdts = np.vectorize(lambda x: (lambda v: v[0] if len(v) > 0 else
'')(_r.findall(x)))(nams)
vals = smap(
zip(vals, vdts),
unpack(lambda v, d: np.asarray(v).astype(d)
if d != '|b1' else v == 'True'))
return StructuredArray(zip(nams, vals))
def asMatrix(val: Iterable[Iterable],
nrow: Optional[int] = None,
ncol: Optional[int] = None,
rownames: Optional[Iterable] = None,
colnames: Optional[Iterable] = None) -> robj.Matrix:
if not (isinstance(val, np.ndarray) and val.ndim == 2):
val = np.asarray(smap(val, ll))
if missing(nrow) and missing(ncol): nrow, ncol = val.shape
matx = robj.r.matrix(val, nrow=nrow, ncol=ncol)
if available(rownames):
matx.rownames = robj.StrVector(np.asarray(ll(rownames), dtype=str))
if available(colnames):
matx.colnames = robj.StrVector(np.asarray(ll(colnames), dtype=str))
return matx
def __init__(self, X: Iterable[Iterable], *, dtype: Optional[Union[str, type, np.ndarray.dtype]] = None,
rownames: Optional[Union[Iterable[str], NamedIndex]] = None, rowindex: Optional[Union[Iterable, Mapping, np.ndarray, StructuredArray]] = None,
colnames: Optional[Union[Iterable[str], NamedIndex]] = None, colindex: Optional[Union[Iterable, Mapping, np.ndarray, StructuredArray]] = None,
metadata: Optional[Union[Sequence, Mapping]] = None, memmap: Optional[Union[str, Path]] = None):
if not isinstance(X, np.ndarray): X = smap(X, ll)
self._dmatx = np.array(X, dtype = dtype) # make a copy
if self._dmatx.ndim != 2: raise ValueError('input data is not a 2-dimensional matrix')
self._memmap = None
if available(memmap): self.offload(memmap)
self._rnames = self._cnames = None
self.rows_ = rownames
self.cols_ = colnames
self._rindex = self._cindex = None
self.ridx_ = rowindex
self.cidx_ = colindex
self._metas = Metadata(optional(metadata, ()))
def _parseids(self, idx, axis=None, mapslice=True):
if missing(axis):
sids, aids = (idx, slice(None)) if not isinstance(idx, tuple) else \
(idx[0], slice(None)) if len(idx) == 1 else idx
else:
if isinstance(idx, tuple):
raise IndexError('too many dimensions for array')
if axis not in (0, 1): raise IndexError('invalid axis value')
sids, aids = (idx, slice(None)) if axis == 0 else (slice(None),
idx)
def _wrap(ids):
if ids is None: return slice(None)
if isinstance(ids, slice): return ids
if not listable(ids): return [ids]
return ids
sids, aids = smap((sids, aids), _wrap)
if (isinstance(sids, slice) and mapslice) or (
listable(sids) and checkany(sids, lambda x: not isstring(x))):
sids = self.names[sids]
return sids, aids
def _parsevals(self, value):
if isinstance(value, Table): return value._dmatx.astype(self.dtype)
if isinstance(value, np.ndarray): return value.astype(self.dtype)
if not iterable(value): return value
value = ll(value)
if not iterable(value[0]): return np.asarray(value, dtype = self.dtype)
value = np.asarray(value if isinstance(value, np.ndarray) and value.ndim == 2 else smap(value, ll), dtype = self.dtype)
return value
def _sfmt(lns, pos):
size = max(collapse(smap(lns, lambda x: smap(x[pos], lambda v: len(v) if v not in (' ... ', ' ... ... ') else 0)))) + 1
for ln in lns: ln[pos] = smap(ln[pos], lambda x: '{0:>{1}s}'.format(x, size) if x != ' ... ' else x)
return lns
def createTable(
self, tableName: str,
columns: Iterable[Iterable[str]] = ()) -> SQLiteWrapper:
tcols = paste(smap(columns, lambda x: paste(x, sep=' ')), sep=', ')
self.execute(f"CREATE TABLE '{tableName}'({tcols})")
return self