paul@44 | 1 | #!/usr/bin/env python |
paul@44 | 2 | |
paul@44 | 3 | """ |
paul@44 | 4 | Specific classes for storing document information. |
paul@44 | 5 | |
paul@89 | 6 | Copyright (C) 2009, 2010, 2011 Paul Boddie <paul@boddie.org.uk> |
paul@44 | 7 | |
paul@44 | 8 | This program is free software; you can redistribute it and/or modify it under |
paul@44 | 9 | the terms of the GNU General Public License as published by the Free Software |
paul@44 | 10 | Foundation; either version 3 of the License, or (at your option) any later |
paul@44 | 11 | version. |
paul@44 | 12 | |
paul@44 | 13 | This program is distributed in the hope that it will be useful, but WITHOUT ANY |
paul@44 | 14 | WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A |
paul@44 | 15 | PARTICULAR PURPOSE. See the GNU General Public License for more details. |
paul@44 | 16 | |
paul@44 | 17 | You should have received a copy of the GNU General Public License along |
paul@44 | 18 | with this program. If not, see <http://www.gnu.org/licenses/>. |
paul@44 | 19 | """ |
paul@44 | 20 | |
paul@89 | 21 | from iixr.data import * |
paul@44 | 22 | from iixr.files import * |
paul@44 | 23 | from bisect import bisect_right # to find terms in the dictionary index |
paul@44 | 24 | |
paul@67 | 25 | DOCUMENT_CACHE_LIMIT = 10000 |
paul@67 | 26 | |
paul@44 | 27 | class FieldWriter(FileWriter): |
paul@44 | 28 | |
paul@44 | 29 | "Writing field data to files." |
paul@44 | 30 | |
paul@44 | 31 | def reset(self): |
paul@90 | 32 | self.end_record() |
paul@74 | 33 | self.last_docnum = None |
paul@89 | 34 | self.subtractor = None |
paul@44 | 35 | |
paul@44 | 36 | def write_fields(self, docnum, fields): |
paul@44 | 37 | |
paul@44 | 38 | """ |
paul@44 | 39 | Write for the given 'docnum', a list of 'fields' (integer, string pairs |
paul@44 | 40 | representing field identifiers and values respectively). |
paul@44 | 41 | """ |
paul@44 | 42 | |
paul@74 | 43 | # Find the size of document number values. |
paul@74 | 44 | |
paul@89 | 45 | if self.last_docnum is not None: |
paul@89 | 46 | docnum_seq = self.subtractor(docnum, self.last_docnum) |
paul@89 | 47 | else: |
paul@89 | 48 | self.subtractor = get_subtractor(docnum) |
paul@89 | 49 | docnum_seq = docnum |
paul@74 | 50 | |
paul@89 | 51 | # Write the document number. |
paul@89 | 52 | |
paul@89 | 53 | self.write_sequence_value(docnum_seq) |
paul@44 | 54 | |
paul@44 | 55 | # Write the number of fields. |
paul@44 | 56 | |
paul@88 | 57 | self.write_number(len(fields)) |
paul@44 | 58 | |
paul@44 | 59 | # Write the fields themselves. |
paul@44 | 60 | |
paul@44 | 61 | for i, field in fields: |
paul@44 | 62 | self.write_number(i) |
paul@44 | 63 | self.write_string(field, 1) # compress |
paul@44 | 64 | |
paul@89 | 65 | self.last_docnum = docnum |
paul@89 | 66 | |
paul@44 | 67 | class FieldReader(FileReader): |
paul@44 | 68 | |
paul@44 | 69 | "Reading field data from files." |
paul@44 | 70 | |
paul@44 | 71 | def reset(self): |
paul@74 | 72 | self.last_docnum = None |
paul@89 | 73 | self.adder = None |
paul@90 | 74 | self.begin_record() |
paul@44 | 75 | |
paul@44 | 76 | def read_fields(self): |
paul@44 | 77 | |
paul@44 | 78 | """ |
paul@44 | 79 | Read fields from the file, returning a tuple containing the document |
paul@44 | 80 | number and a list of field (identifier, value) pairs. |
paul@44 | 81 | """ |
paul@44 | 82 | |
paul@89 | 83 | # Read the document number. |
paul@89 | 84 | |
paul@89 | 85 | docnum = self.read_sequence_value() |
paul@44 | 86 | |
paul@89 | 87 | if self.last_docnum is not None: |
paul@89 | 88 | self.last_docnum = self.adder(docnum, self.last_docnum) |
paul@89 | 89 | else: |
paul@89 | 90 | self.adder = get_adder(docnum) |
paul@89 | 91 | self.last_docnum = docnum |
paul@44 | 92 | |
paul@44 | 93 | # Read the number of fields. |
paul@44 | 94 | |
paul@44 | 95 | nfields = self.read_number() |
paul@44 | 96 | |
paul@44 | 97 | # Collect the fields. |
paul@44 | 98 | |
paul@44 | 99 | fields = [] |
paul@44 | 100 | i = 0 |
paul@44 | 101 | |
paul@44 | 102 | while i < nfields: |
paul@44 | 103 | identifier = self.read_number() |
paul@44 | 104 | value = self.read_string(1) # decompress |
paul@44 | 105 | fields.append((identifier, value)) |
paul@44 | 106 | i += 1 |
paul@44 | 107 | |
paul@44 | 108 | return self.last_docnum, fields |
paul@44 | 109 | |
paul@44 | 110 | def read_document_fields(self, docnum, offset): |
paul@44 | 111 | |
paul@44 | 112 | """ |
paul@44 | 113 | Read fields for 'docnum' at the given 'offset'. This permits the |
paul@44 | 114 | retrieval of details for the specified document, as well as scanning for |
paul@44 | 115 | later documents. |
paul@44 | 116 | """ |
paul@44 | 117 | |
paul@69 | 118 | self.seek(offset) |
paul@44 | 119 | bad_docnum, fields = self.read_fields() |
paul@44 | 120 | self.last_docnum = docnum |
paul@44 | 121 | return docnum, fields |
paul@44 | 122 | |
paul@44 | 123 | class FieldIndexWriter(FileWriter): |
paul@44 | 124 | |
paul@44 | 125 | "Writing field index details to files." |
paul@44 | 126 | |
paul@44 | 127 | def reset(self): |
paul@90 | 128 | self.end_record() |
paul@74 | 129 | self.last_docnum = None |
paul@89 | 130 | self.subtractor = None |
paul@44 | 131 | self.last_offset = 0 |
paul@44 | 132 | |
paul@44 | 133 | def write_document(self, docnum, offset): |
paul@44 | 134 | |
paul@44 | 135 | """ |
paul@44 | 136 | Write for the given 'docnum', the 'offset' at which the fields for the |
paul@44 | 137 | document are stored in the fields file. |
paul@44 | 138 | """ |
paul@44 | 139 | |
paul@74 | 140 | # Find the size of document number values. |
paul@74 | 141 | |
paul@89 | 142 | if self.last_docnum is not None: |
paul@89 | 143 | docnum_seq = self.subtractor(docnum, self.last_docnum) |
paul@89 | 144 | else: |
paul@89 | 145 | self.subtractor = get_subtractor(docnum) |
paul@89 | 146 | docnum_seq = docnum |
paul@74 | 147 | |
paul@89 | 148 | # Write the document number. |
paul@89 | 149 | |
paul@89 | 150 | self.write_sequence_value(docnum_seq) |
paul@74 | 151 | |
paul@74 | 152 | # Write the offset delta. |
paul@74 | 153 | |
paul@44 | 154 | self.write_number(offset - self.last_offset) |
paul@89 | 155 | |
paul@89 | 156 | self.last_docnum = docnum |
paul@44 | 157 | self.last_offset = offset |
paul@44 | 158 | |
paul@44 | 159 | class FieldIndexReader(FileReader): |
paul@44 | 160 | |
paul@44 | 161 | "Reading field index details from files." |
paul@44 | 162 | |
paul@44 | 163 | def reset(self): |
paul@74 | 164 | self.last_docnum = None |
paul@89 | 165 | self.adder = None |
paul@44 | 166 | self.last_offset = 0 |
paul@90 | 167 | self.begin_record() |
paul@44 | 168 | |
paul@44 | 169 | def read_document(self): |
paul@44 | 170 | |
paul@44 | 171 | "Read a document number and field file offset." |
paul@44 | 172 | |
paul@89 | 173 | # Read the document number. |
paul@89 | 174 | |
paul@89 | 175 | docnum = self.read_sequence_value() |
paul@74 | 176 | |
paul@89 | 177 | if self.last_docnum is not None: |
paul@89 | 178 | self.last_docnum = self.adder(docnum, self.last_docnum) |
paul@89 | 179 | else: |
paul@89 | 180 | self.adder = get_adder(docnum) |
paul@89 | 181 | self.last_docnum = docnum |
paul@74 | 182 | |
paul@74 | 183 | # Read the offset. |
paul@74 | 184 | |
paul@44 | 185 | self.last_offset += self.read_number() |
paul@44 | 186 | |
paul@44 | 187 | return self.last_docnum, self.last_offset |
paul@44 | 188 | |
paul@44 | 189 | class FieldDictionaryWriter: |
paul@44 | 190 | |
paul@44 | 191 | "Writing field dictionary details." |
paul@44 | 192 | |
paul@44 | 193 | def __init__(self, field_writer, field_index_writer, interval): |
paul@44 | 194 | self.field_writer = field_writer |
paul@44 | 195 | self.field_index_writer = field_index_writer |
paul@44 | 196 | self.interval = interval |
paul@44 | 197 | self.entry = 0 |
paul@44 | 198 | |
paul@44 | 199 | def write_fields(self, docnum, fields): |
paul@44 | 200 | |
paul@44 | 201 | "Write details of the document with the given 'docnum' and 'fields'." |
paul@44 | 202 | |
paul@44 | 203 | if self.entry % self.interval == 0: |
paul@90 | 204 | self.field_writer.reset() |
paul@88 | 205 | offset = self.field_writer.tell() |
paul@55 | 206 | self.field_writer.write_fields(docnum, fields) |
paul@44 | 207 | self.field_index_writer.write_document(docnum, offset) |
paul@55 | 208 | else: |
paul@55 | 209 | self.field_writer.write_fields(docnum, fields) |
paul@44 | 210 | |
paul@44 | 211 | self.entry += 1 |
paul@44 | 212 | |
paul@44 | 213 | def close(self): |
paul@44 | 214 | self.field_writer.close() |
paul@44 | 215 | self.field_index_writer.close() |
paul@44 | 216 | |
paul@44 | 217 | class FieldDictionaryReader: |
paul@44 | 218 | |
paul@44 | 219 | "Reading field dictionary details." |
paul@44 | 220 | |
paul@44 | 221 | def __init__(self, field_reader, field_index_reader): |
paul@44 | 222 | self.field_reader = field_reader |
paul@44 | 223 | self.field_index_reader = field_index_reader |
paul@90 | 224 | self.entry = 0 |
paul@44 | 225 | |
paul@67 | 226 | self.cache = {} |
paul@44 | 227 | self.docs = [] |
paul@44 | 228 | try: |
paul@44 | 229 | while 1: |
paul@44 | 230 | self.docs.append(self.field_index_reader.read_document()) |
paul@44 | 231 | except EOFError: |
paul@44 | 232 | pass |
paul@44 | 233 | |
paul@44 | 234 | # Large numbers for ordering purposes. |
paul@44 | 235 | |
paul@44 | 236 | if self.docs: |
paul@44 | 237 | self.max_offset = self.docs[-1][1] |
paul@44 | 238 | else: |
paul@44 | 239 | self.max_offset = None |
paul@44 | 240 | |
paul@44 | 241 | # Iterator convenience methods. |
paul@44 | 242 | |
paul@44 | 243 | def __iter__(self): |
paul@44 | 244 | self.rewind() |
paul@44 | 245 | return self |
paul@44 | 246 | |
paul@44 | 247 | def next(self): |
paul@44 | 248 | try: |
paul@44 | 249 | return self.read_fields() |
paul@44 | 250 | except EOFError: |
paul@44 | 251 | raise StopIteration |
paul@44 | 252 | |
paul@44 | 253 | # Sequential access methods. |
paul@44 | 254 | |
paul@44 | 255 | def rewind(self): |
paul@44 | 256 | self.field_reader.rewind() |
paul@44 | 257 | |
paul@44 | 258 | def read_fields(self): |
paul@44 | 259 | |
paul@44 | 260 | "Return the next document number and fields." |
paul@44 | 261 | |
paul@90 | 262 | try: |
paul@90 | 263 | return self.field_reader.read_fields() |
paul@90 | 264 | except EOFError: |
paul@90 | 265 | self.entry += 1 |
paul@90 | 266 | try: |
paul@90 | 267 | found_docnum, offset = self.docs[self.entry] |
paul@90 | 268 | except IndexError: |
paul@90 | 269 | raise EOFError |
paul@90 | 270 | else: |
paul@90 | 271 | self.field_reader.reset() |
paul@90 | 272 | return self.field_reader.read_fields() |
paul@44 | 273 | |
paul@44 | 274 | # Random access methods. |
paul@44 | 275 | |
paul@44 | 276 | def get_fields(self, docnum): |
paul@44 | 277 | |
paul@44 | 278 | "Read the fields of the document with the given 'docnum'." |
paul@44 | 279 | |
paul@67 | 280 | if self.cache.has_key(docnum): |
paul@67 | 281 | return self.cache[docnum] |
paul@67 | 282 | |
paul@44 | 283 | i = bisect_right(self.docs, (docnum, self.max_offset)) - 1 |
paul@44 | 284 | |
paul@44 | 285 | # Get the entry position providing the term or one preceding it. |
paul@44 | 286 | |
paul@44 | 287 | if i == -1: |
paul@44 | 288 | return None |
paul@44 | 289 | |
paul@44 | 290 | found_docnum, offset = self.docs[i] |
paul@44 | 291 | |
paul@44 | 292 | # Read from the fields file. |
paul@44 | 293 | |
paul@44 | 294 | found_docnum, fields = self.field_reader.read_document_fields(found_docnum, offset) |
paul@44 | 295 | |
paul@44 | 296 | # Scan for the document, if necessary. |
paul@44 | 297 | |
paul@44 | 298 | try: |
paul@44 | 299 | while docnum > found_docnum: |
paul@44 | 300 | found_docnum, fields = self.field_reader.read_fields() |
paul@44 | 301 | except EOFError: |
paul@44 | 302 | pass |
paul@44 | 303 | |
paul@44 | 304 | # If the document is found, return the fields. |
paul@44 | 305 | |
paul@44 | 306 | if docnum == found_docnum: |
paul@67 | 307 | |
paul@67 | 308 | # Store the fields in the cache, removing entries if the limit has |
paul@67 | 309 | # been reached. |
paul@67 | 310 | |
paul@67 | 311 | keys = self.cache.keys() |
paul@67 | 312 | |
paul@67 | 313 | if len(keys) == DOCUMENT_CACHE_LIMIT: |
paul@67 | 314 | del self.cache[keys[0]] |
paul@67 | 315 | |
paul@67 | 316 | self.cache[docnum] = fields |
paul@44 | 317 | return fields |
paul@44 | 318 | else: |
paul@44 | 319 | return None |
paul@44 | 320 | |
paul@44 | 321 | def close(self): |
paul@44 | 322 | self.field_reader.close() |
paul@44 | 323 | self.field_index_reader.close() |
paul@44 | 324 | |
paul@44 | 325 | # vim: tabstop=4 expandtab shiftwidth=4 |