Thuban/Model/table.py

# Copyright (c) 2001, 2002, 2003 by Intevation GmbH
# Authors:
# Bernhard Herzog <[email protected]>
# Jan-Oliver Wagner <[email protected]>
#
# This program is free software under the GPL (>=v2)
# Read the file COPYING coming with Thuban for details.

"""
Classes for handling tables of data.
"""

__version__ = "$Revision$"

import inspect
import warnings

import dbflib

# the field types supported by a Table instance.
FIELDTYPE_INT = "int"
FIELDTYPE_STRING = "string"
FIELDTYPE_DOUBLE = "double"


# map the dbflib constants for the field types to our constants
dbflib_fieldtypes = {dbflib.FTString: FIELDTYPE_STRING,
                     dbflib.FTInteger: FIELDTYPE_INT,
                     dbflib.FTDouble: FIELDTYPE_DOUBLE}


class OldTableInterfaceMixin:

    """Mixin to implement the old table interface using the new one"""

    def __deprecation_warning(self):
        """Issue a DeprecationWarning for code hat uses the old interface"""
        callername = inspect.currentframe().f_back.f_code.co_name
        warnings.warn("The %s method of the old table interface"
                      " is deprecated" % callername,
                      DeprecationWarning, stacklevel = 3)

    def record_count(self):
        self.__deprecation_warning()
        return self.NumRows()

    def field_count(self):
        self.__deprecation_warning()
        return self.NumColumns()

    def field_info(self, field):
        """Return a tuple (type, name, width, prec) for the field no. field

        type is the data type of the field, name the name, width the
        field width in characters and prec the decimal precision. width
        and prec will be zero if the information returned by the Column
        method doesn't provide values for them.
        """
        self.__deprecation_warning()
        col = self.Column(field)
        return (col.type, col.name,
               getattr(col, "width", 0), getattr(col, "prec", 0))

    def field_info_by_name(self, col):
        self.__deprecation_warning()
        try:
            return self.field_info(col)
        except KeyError:
            # FIXME: It may be that field_info raises other exceptions
            # when the name is not a valid column name.
            return None

    def field_range(self, fieldName):
        self.__deprecation_warning()
        min, max = self.ValueRange(fieldName)
        return ((min, None), (max, None))

    def GetUniqueValues(self, field):
        self.__deprecation_warning()
        return self.UniqueValues(field)

    def read_record(self, r):
        self.__deprecation_warning()
        return self.ReadRowAsDict(r)


class DBFColumn:

    """Description of a column in a DBFTable

    Instances have the following public attributes:

    name -- Name of the column
    type -- Type of the column (one of FIELDTYPE_STRING, FIELDTYPE_INT or\
            FIELDTYPE_DOUBLE)
    index -- The index of the column
    width -- the width of the data in the column
    prec -- The precision of the data (only valid for type == FIELDTYPE_DOUBLE)
    """

    def __init__(self, name, type, width, prec, index):
        self.name = name
        self.type = type
        self.width = width
        self.prec = prec
        self.index = index


class DBFTable(OldTableInterfaceMixin):

    """
    Table interface for the data in a DBF file
    """

    # Implementation strategy regarding writing to a DBF file:
    #
    # Most of the time Thuban only needs to read from a table and it is
    # important that Thuban can work with read-only files. Therefore the
    # DBF file is opened only for reading initially. Only when
    # write_record is called we try to open the DBF file for writing as
    # well. If that succeeds the read/write DBF file will be used for
    # all IO afterwards.
    #
    # It's important to use the same DBF file object for both reading
    # and writing to make sure that reading a records after writing
    # returns the new values. With two separate objects this wouldn't
    # work because a DBF file object buffers some data

    def __init__(self, filename):
        self.filename = filename
        self.dbf = dbflib.DBFFile(filename)

        # If true, self.dbf is open for writing.
        self._writable = 0

        # Create the column information objects
        self.columns = []
        self.column_map = {}
        for i in range(self.NumColumns()):
            ftype, name, width, prec = self.dbf.field_info(i)
            ftype = dbflib_fieldtypes[ftype]
            index = len(self.columns)
            col = DBFColumn(name, ftype, width, prec, index)
            self.columns.append(col)
            self.column_map[name] = col
            self.column_map[index] = col

    def NumRows(self):
        """Return the number of rows in the table"""
        return self.dbf.record_count()

    def NumColumns(self):
        """Return the number of columns in the table"""
        return self.dbf.field_count()

    def Columns(self):
        """Return the table's colum definitions

        The return value is a sequence of DBFColumn instances, one for
        each column.
        """
        return self.columns

    def Column(self, col):
        """Return information about the column given by its name or index

        The returned object is an instance of DBFColumn
        """
        return self.column_map[col]

    def HasColumn(self, col):
        """Return whether the table has a column with the given name or index
        """
        return self.column_map.has_key(col)

    def ReadRowAsDict(self, row):
        """Return the entire row as a dictionary with column names as keys"""
        return self.dbf.read_record(row)

    def ReadValue(self, row, col):
        """Return the value of the specified row and column

        The col parameter may be the index of the column or its name.
        """
        return self.dbf.read_record(row)[self.column_map[col].name]

    def ValueRange(self, col):
        """Return the minimum and maximum values of the values in the column

        The return value is a tuple (min, max) unless the table is empty
        in which case the return value is None.
        """
        count = self.NumRows()

        if count == 0:
            return None

        min = max = self.ReadValue(0, col)
        for i in range(1, count):
            value = self.ReadValue(i, col)
            if value < min:
                min = value
            elif value > max:
                max = value

        return (min, max)

    def UniqueValues(self, col):
        """Return a sorted list of all unique values in the column col"""
        dict = {}

        for i in range(self.NumRows()):
            value = self.ReadValue(i, col)
            dict[value] = 0

        values = dict.keys()
        values.sort()
        return values


    # DBF specific interface parts.

    def Destroy(self):
        self.dbf.close()
        self.dbf = None

    def write_record(self, record, values):
        """Write the values into the record

        The values parameter may either be a dictionary or a sequence.

        If it's a dictionary the keys must be the names of the fields
        and their value must have a suitable type. Only the fields
        actually contained in the dictionary are written. Fields for
        which there's no item in the dict are not modified.

        If it's a sequence, all fields must be present in the right
        order.
        """
        if not self._writable:
            new_dbf = dbflib.DBFFile(self.filename, "r+b")
            self.dbf.close()
            self.dbf = new_dbf
            self._writable = 1
        self.dbf.write_record(record, values)
        self.dbf.commit()


class MemoryColumn:

    def __init__(self, name, type, index):
        self.name = name
        self.type = type
        self.index = index

class MemoryTable(OldTableInterfaceMixin):

    """Very simple table implementation that operates on a list of tuples"""

    def __init__(self, fields, data):
        """Initialize the MemoryTable

        Parameters:
        fields -- List of (name, field_type) pairs
        data -- List of tuples, one for each row of data
        """
        self.data = data

        # Create the column information objects
        self.columns = []
        self.column_map = {}
        for name, ftype in fields:
            index = len(self.columns)
            col = MemoryColumn(name, ftype, index)
            self.columns.append(col)
            self.column_map[name] = col
            self.column_map[index] = col

    def NumColumns(self):
        """Return the number of columns in the table"""
        return len(self.columns)

    def Column(self, col):
        """Return information about the column given by its name or index

        The returned object is an instance of MemoryColumn.
        """
        return self.column_map[col]

    def Columns(self):
        """Return the table's colum definitions

        The return value is a sequence of MemoryColumn instances, one
        for each column.
        """
        return self.columns

    def HasColumn(self, col):
        """Return whether the table has a column with the given name or index
        """
        return self.column_map.has_key(col)

    def NumRows(self):
        """Return the number of rows in the table"""
        return len(self.data)

    def ReadValue(self, row, col):
        """Return the value of the specified row and column

        The col parameter may be the index of the column or its name.
        """
        return self.data[row][self.column_map[col].index]

    def ReadRowAsDict(self, index):
        """Return the entire row as a dictionary with column names as keys"""
        return dict([(col.name, self.data[index][col.index])
                      for col in self.columns])

    def ValueRange(self, col):
        """Return the minimum and maximum values of the values in the column

        The return value is a tuple (min, max) unless the table is empty
        in which case the return value is None.
        """

        index = self.column_map[col].index
        values = [row[index] for row in self.data]
        if not values:
            return None

        return min(values), max(values)

    def UniqueValues(self, col):
        """Return a sorted list of all unique values in the column col"""
        dict = {}

        for i in range(self.NumRows()):
            value = self.ReadValue(i, col)
            dict[value] = 0

        values = dict.keys()
        values.sort()
        return values


    def write_record(self, record, values):
        # TODO: Check for correct lenght and perhaps also
        # for correct types in case values is a tuple. How to report problems?
        # TODO: Allow values to be a dictionary and write the single
        # fields that are specified.
        self.data[record] = values
1	bh	590	# Copyright (c) 2001, 2002, 2003 by Intevation GmbH
2	bh	6	# Authors:
3			# Bernhard Herzog <[email protected]>
4	jan	806	# Jan-Oliver Wagner <[email protected]>
5	bh	6	#
6			# This program is free software under the GPL (>=v2)
7			# Read the file COPYING coming with Thuban for details.
8
9			"""
10			Classes for handling tables of data.
11			"""
12
13			__version__ = "$Revision$"
14
15	bh	839	import inspect
16			import warnings
17
18	bh	6	import dbflib
19
20			# the field types supported by a Table instance.
21	jonathan	474	FIELDTYPE_INT = "int"
22			FIELDTYPE_STRING = "string"
23			FIELDTYPE_DOUBLE = "double"
24	bh	6
25
26			# map the dbflib constants for the field types to our constants
27			dbflib_fieldtypes = {dbflib.FTString: FIELDTYPE_STRING,
28			dbflib.FTInteger: FIELDTYPE_INT,
29			dbflib.FTDouble: FIELDTYPE_DOUBLE}
30
31	jan	806
32	bh	818	class OldTableInterfaceMixin:
33	jan	806
34	bh	818	"""Mixin to implement the old table interface using the new one"""
35	jan	806
36	bh	839	def __deprecation_warning(self):
37			"""Issue a DeprecationWarning for code hat uses the old interface"""
38			callername = inspect.currentframe().f_back.f_code.co_name
39			warnings.warn("The %s method of the old table interface"
40			" is deprecated" % callername,
41			DeprecationWarning, stacklevel = 3)
42
43	bh	818	def record_count(self):
44	bh	839	self.__deprecation_warning()
45	bh	818	return self.NumRows()
46	jan	806
47			def field_count(self):
48	bh	839	self.__deprecation_warning()
49	bh	818	return self.NumColumns()
50	jan	806
51	bh	818	def field_info(self, field):
52			"""Return a tuple (type, name, width, prec) for the field no. field
53	jan	806
54	bh	818	type is the data type of the field, name the name, width the
55			field width in characters and prec the decimal precision. width
56			and prec will be zero if the information returned by the Column
57			method doesn't provide values for them.
58			"""
59	bh	839	self.__deprecation_warning()
60	bh	818	col = self.Column(field)
61			return (col.type, col.name,
62			getattr(col, "width", 0), getattr(col, "prec", 0))
63	jan	806
64	bh	818	def field_info_by_name(self, col):
65	bh	839	self.__deprecation_warning()
66	bh	818	try:
67			return self.field_info(col)
68			except KeyError:
69			# FIXME: It may be that field_info raises other exceptions
70			# when the name is not a valid column name.
71			return None
72	jan	806
73	bh	818	def field_range(self, fieldName):
74	bh	839	self.__deprecation_warning()
75	bh	818	min, max = self.ValueRange(fieldName)
76			return ((min, None), (max, None))
77	jan	806
78	bh	818	def GetUniqueValues(self, field):
79	bh	839	self.__deprecation_warning()
80	bh	818	return self.UniqueValues(field)
81	jan	806
82	bh	818	def read_record(self, r):
83	bh	839	self.__deprecation_warning()
84	bh	818	return self.ReadRowAsDict(r)
85	bh	6
86	bh	818
87
88			class DBFColumn:
89
90			"""Description of a column in a DBFTable
91
92			Instances have the following public attributes:
93
94			name -- Name of the column
95			type -- Type of the column (one of FIELDTYPE_STRING, FIELDTYPE_INT or\
96			FIELDTYPE_DOUBLE)
97			index -- The index of the column
98			width -- the width of the data in the column
99			prec -- The precision of the data (only valid for type == FIELDTYPE_DOUBLE)
100	bh	6	"""
101	bh	818
102			def __init__(self, name, type, width, prec, index):
103			self.name = name
104			self.type = type
105			self.width = width
106			self.prec = prec
107			self.index = index
108
109
110			class DBFTable(OldTableInterfaceMixin):
111
112			"""
113	bh	765	Table interface for the data in a DBF file
114	bh	6	"""
115
116	bh	286	# Implementation strategy regarding writing to a DBF file:
117			#
118			# Most of the time Thuban only needs to read from a table and it is
119			# important that Thuban can work with read-only files. Therefore the
120			# DBF file is opened only for reading initially. Only when
121			# write_record is called we try to open the DBF file for writing as
122	bh	590	# well. If that succeeds the read/write DBF file will be used for
123			# all IO afterwards.
124	bh	286	#
125			# It's important to use the same DBF file object for both reading
126			# and writing to make sure that reading a records after writing
127			# returns the new values. With two separate objects this wouldn't
128			# work because a DBF file object buffers some data
129
130	bh	6	def __init__(self, filename):
131			self.filename = filename
132	bh	284	self.dbf = dbflib.DBFFile(filename)
133	bh	6
134	bh	286	# If true, self.dbf is open for writing.
135			self._writable = 0
136
137	bh	818	# Create the column information objects
138			self.columns = []
139			self.column_map = {}
140			for i in range(self.NumColumns()):
141			ftype, name, width, prec = self.dbf.field_info(i)
142			ftype = dbflib_fieldtypes[ftype]
143			index = len(self.columns)
144			col = DBFColumn(name, ftype, width, prec, index)
145			self.columns.append(col)
146			self.column_map[name] = col
147			self.column_map[index] = col
148	bh	257
149	bh	818	def NumRows(self):
150			"""Return the number of rows in the table"""
151	bh	6	return self.dbf.record_count()
152
153	bh	818	def NumColumns(self):
154			"""Return the number of columns in the table"""
155	bh	6	return self.dbf.field_count()
156
157	bh	818	def Columns(self):
158			"""Return the table's colum definitions
159	bh	6
160	bh	818	The return value is a sequence of DBFColumn instances, one for
161			each column.
162	bh	6	"""
163	bh	818	return self.columns
164	bh	6
165	bh	818	def Column(self, col):
166			"""Return information about the column given by its name or index
167	jonathan	467
168	bh	818	The returned object is an instance of DBFColumn
169			"""
170			return self.column_map[col]
171	jonathan	467
172	bh	839	def HasColumn(self, col):
173			"""Return whether the table has a column with the given name or index
174			"""
175			return self.column_map.has_key(col)
176
177	bh	818	def ReadRowAsDict(self, row):
178			"""Return the entire row as a dictionary with column names as keys"""
179			return self.dbf.read_record(row)
180	jonathan	467
181	bh	818	def ReadValue(self, row, col):
182			"""Return the value of the specified row and column
183	jonathan	628
184	bh	818	The col parameter may be the index of the column or its name.
185			"""
186			return self.dbf.read_record(row)[self.column_map[col].name]
187	jonathan	628
188	bh	818	def ValueRange(self, col):
189			"""Return the minimum and maximum values of the values in the column
190	jonathan	628
191	bh	818	The return value is a tuple (min, max) unless the table is empty
192			in which case the return value is None.
193	jonathan	628	"""
194	bh	818	count = self.NumRows()
195	jonathan	628
196			if count == 0:
197			return None
198
199	bh	818	min = max = self.ReadValue(0, col)
200	jonathan	628	for i in range(1, count):
201	bh	818	value = self.ReadValue(i, col)
202			if value < min:
203			min = value
204			elif value > max:
205			max = value
206	jonathan	628
207	bh	818	return (min, max)
208	jonathan	628
209	bh	818	def UniqueValues(self, col):
210			"""Return a sorted list of all unique values in the column col"""
211			dict = {}
212	jonathan	628
213	bh	818	for i in range(self.NumRows()):
214			value = self.ReadValue(i, col)
215			dict[value] = 0
216	jonathan	628
217	bh	818	values = dict.keys()
218			values.sort()
219			return values
220	jonathan	628
221
222	bh	818	# DBF specific interface parts.
223	jonathan	628
224	bh	818	def Destroy(self):
225			self.dbf.close()
226			self.dbf = None
227	jonathan	628
228	bh	274	def write_record(self, record, values):
229			"""Write the values into the record
230
231			The values parameter may either be a dictionary or a sequence.
232
233			If it's a dictionary the keys must be the names of the fields
234			and their value must have a suitable type. Only the fields
235			actually contained in the dictionary are written. Fields for
236			which there's no item in the dict are not modified.
237
238			If it's a sequence, all fields must be present in the right
239			order.
240			"""
241	bh	286	if not self._writable:
242			new_dbf = dbflib.DBFFile(self.filename, "r+b")
243			self.dbf.close()
244			self.dbf = new_dbf
245			self._writable = 1
246			self.dbf.write_record(record, values)
247			self.dbf.commit()
248	jonathan	467
249	bh	765
250
251	bh	818	class MemoryColumn:
252
253			def __init__(self, name, type, index):
254			self.name = name
255			self.type = type
256			self.index = index
257
258			class MemoryTable(OldTableInterfaceMixin):
259
260			"""Very simple table implementation that operates on a list of tuples"""
261
262			def __init__(self, fields, data):
263			"""Initialize the MemoryTable
264
265			Parameters:
266			fields -- List of (name, field_type) pairs
267			data -- List of tuples, one for each row of data
268			"""
269			self.data = data
270
271			# Create the column information objects
272			self.columns = []
273			self.column_map = {}
274			for name, ftype in fields:
275			index = len(self.columns)
276			col = MemoryColumn(name, ftype, index)
277			self.columns.append(col)
278			self.column_map[name] = col
279			self.column_map[index] = col
280
281			def NumColumns(self):
282			"""Return the number of columns in the table"""
283			return len(self.columns)
284
285			def Column(self, col):
286			"""Return information about the column given by its name or index
287
288			The returned object is an instance of MemoryColumn.
289			"""
290			return self.column_map[col]
291
292			def Columns(self):
293			"""Return the table's colum definitions
294
295			The return value is a sequence of MemoryColumn instances, one
296			for each column.
297			"""
298			return self.columns
299
300	bh	839	def HasColumn(self, col):
301			"""Return whether the table has a column with the given name or index
302			"""
303			return self.column_map.has_key(col)
304
305	bh	818	def NumRows(self):
306			"""Return the number of rows in the table"""
307			return len(self.data)
308
309			def ReadValue(self, row, col):
310			"""Return the value of the specified row and column
311
312			The col parameter may be the index of the column or its name.
313			"""
314			return self.data[row][self.column_map[col].index]
315
316			def ReadRowAsDict(self, index):
317			"""Return the entire row as a dictionary with column names as keys"""
318			return dict([(col.name, self.data[index][col.index])
319			for col in self.columns])
320
321			def ValueRange(self, col):
322			"""Return the minimum and maximum values of the values in the column
323
324			The return value is a tuple (min, max) unless the table is empty
325			in which case the return value is None.
326			"""
327
328			index = self.column_map[col].index
329			values = [row[index] for row in self.data]
330			if not values:
331			return None
332
333			return min(values), max(values)
334
335			def UniqueValues(self, col):
336			"""Return a sorted list of all unique values in the column col"""
337			dict = {}
338
339			for i in range(self.NumRows()):
340			value = self.ReadValue(i, col)
341			dict[value] = 0
342
343			values = dict.keys()
344			values.sort()
345			return values
346
347
348			def write_record(self, record, values):
349			# TODO: Check for correct lenght and perhaps also
350			# for correct types in case values is a tuple. How to report problems?
351			# TODO: Allow values to be a dictionary and write the single
352			# fields that are specified.
353			self.data[record] = values
Name	Value
svn:eol-style	native
svn:keywords	Author Date Id Revision