nilmdb/nilmdb/layout.pyx

# cython: profile=False

import time
import sys
import inspect
import cStringIO
import numpy as np

cdef enum:
    max_value_count = 64

cimport cython
cimport libc.stdlib
cimport libc.stdio
cimport libc.string

class ParserError(Exception):
    def __init__(self, line, message):
        self.message = "line " + str(line) + ": " + message
        Exception.__init__(self, self.message)

class FormatterError(Exception):
    pass

class Layout:
    """Represents a NILM database layout"""

    def __init__(self, typestring):
        """Initialize this Layout object to handle the specified
        type string"""
        try:
            [ datatype, count ] = typestring.split("_")
        except:
            raise KeyError("invalid layout string")

        try:
            self.count = int(count)
        except ValueError:
            raise KeyError("invalid count")
        if self.count < 1 or self.count > max_value_count:
            raise KeyError("invalid count")

        if datatype == 'uint16':
            self.parse = self.parse_uint16
            self.format = self.format_uint16
        elif datatype == 'float32' or datatype == 'float64':
            self.parse = self.parse_float64
            self.format = self.format_float64
        else:
            raise KeyError("invalid type")

        self.datatype = datatype

    # Parsers
    def parse_float64(self, char *text):
        cdef int n
        cdef double ts
        # Return doubles even in float32 case, since they're going into
        # a Python array which would upconvert to double anyway.
        result = []
        cdef char *end
        ts = libc.stdlib.strtod(text, &end)
        if end == text:
            raise ValueError("bad timestamp")
        result.append(ts)
        for n in range(self.count):
            text = end
            result.append(libc.stdlib.strtod(text, &end))
            if end == text:
                raise ValueError("wrong number of values")
        n = 0
        while end[n] == ' ':
            n += 1
        if end[n] != '\n' and end[n] != '#' and end[n] != '\0':
            raise ValueError("extra data on line")
        return (ts, result)

    def parse_uint16(self, char *text):
        cdef int n
        cdef double ts
        cdef int v
        result = []
        cdef char *end
        ts = libc.stdlib.strtod(text, &end)
        if end == text:
            raise ValueError("bad timestamp")
        result.append(ts)
        for n in range(self.count):
            text = end
            v = libc.stdlib.strtol(text, &end, 10)
            if v < 0 or v > 65535:
                raise ValueError("value out of range")
            result.append(v)
            if end == text:
                raise ValueError("wrong number of values")
        n = 0
        while end[n] == ' ':
            n += 1
        if end[n] != '\n' and end[n] != '#' and end[n] != '\0':
            raise ValueError("extra data on line")
        return (ts, result)

    # Formatters
    def format_float64(self, d):
        n = len(d) - 1
        if n != self.count:
            raise ValueError("wrong number of values for layout type: "
                             "got %d, wanted %d" % (n, self.count))
        s = "%.6f" % d[0]
        for i in range(n):
            s += " %f" % d[i+1]
        return s + "\n"

    def format_uint16(self, d):
        n = len(d) - 1
        if n != self.count:
            raise ValueError("wrong number of values for layout type: "
                             "got %d, wanted %d" % (n, self.count))
        s = "%.6f" % d[0]
        for i in range(n):
            s += " %d" % d[i+1]
        return s + "\n"

# Get a layout by name
def get_named(typestring):
    try:
        return Layout(typestring)
    except KeyError:
        compat = { "PrepData": "float32_8",
                   "RawData": "uint16_6",
                   "RawNotchedData": "uint16_9" }
        return Layout(compat[typestring])

class Parser(object):
    """Object that parses and stores ASCII data for inclusion into the
    database"""

    def __init__(self, layout):
        if issubclass(layout.__class__, Layout):
            self.layout = layout
        else:
            try:
                self.layout = get_named(layout)
            except KeyError:
                raise TypeError("unknown layout")

        self.data = []
        self.min_timestamp = None
        self.max_timestamp = None

    def parse(self, textdata):
        """
        Parse the data, provided as lines of text, using the current
        layout, into an internal data structure suitable for a
        pytables 'table.append(parser.data)'.
        """
        cdef double last_ts = 0, ts
        cdef int n = 0, i
        cdef char *line

        indata = cStringIO.StringIO(textdata)
        # Assume any parsing error is a real error.
        # In the future we might want to skip completely empty lines,
        # or partial lines right before EOF?
        try:
            self.data = []
            for pyline in indata:
                line = pyline
                n += 1
                if line[0] == '\#':
                    continue
                (ts, row) = self.layout.parse(line)
                if ts < last_ts:
                    raise ValueError("timestamp is not "
                                     "monotonically increasing")
                last_ts = ts
                self.data.append(row)
        except (ValueError, IndexError, TypeError) as e:
            raise ParserError(n, "error: " + e.message)

        # Mark timestamp ranges
        if len(self.data):
            self.min_timestamp = self.data[0][0]
            self.max_timestamp = self.data[-1][0]

class Formatter(object):
    """Object that formats database data into ASCII"""

    def __init__(self, layout):
        if issubclass(layout.__class__, Layout):
            self.layout = layout
        else:
            try:
                self.layout = get_named(layout)
            except KeyError:
                raise TypeError("unknown layout")

    def format(self, data):
        """
        Format raw data from the database, using the current layout,
        as lines of ACSII text.
        """
        text = cStringIO.StringIO()
        try:
            for row in data:
                text.write(self.layout.format(row))
        except (ValueError, IndexError, TypeError) as e:
            raise FormatterError("formatting error: " + e.message)
        return text.getvalue()