1 | .\"------------------------------------------------------------
|
---|
2 | .\" Id - set Rv,revision, and Dt, Date using rcs-Id tag.
|
---|
3 | .de Id
|
---|
4 | .ds Rv \\$3
|
---|
5 | .ds Dt \\$4
|
---|
6 | ..
|
---|
7 | .Id $Id: mg_passes.1 16583 2008-07-29 10:20:36Z davidb $
|
---|
8 | .\"------------------------------------------------------------
|
---|
9 | .TH mg_passes 1 \*(Dt CITRI
|
---|
10 | .SH NAME
|
---|
11 | mg_passes \- builds mg databases
|
---|
12 | .SH SYNOPSIS
|
---|
13 | .B mg_passes
|
---|
14 | [
|
---|
15 | .B \-h
|
---|
16 | ]
|
---|
17 | [
|
---|
18 | .B \-G
|
---|
19 | ]
|
---|
20 | [
|
---|
21 | .B \-S
|
---|
22 | ]
|
---|
23 | [
|
---|
24 | .B \-D
|
---|
25 | ]
|
---|
26 | [
|
---|
27 | .B \-W
|
---|
28 | ]
|
---|
29 | .if n .ti +9n
|
---|
30 | [
|
---|
31 | .BR \-1 " |"
|
---|
32 | .BR \-2 " |"
|
---|
33 | .B \-3
|
---|
34 | ]
|
---|
35 | [
|
---|
36 | .BI \-C " compstatpoint"
|
---|
37 | ]
|
---|
38 | .if n .ti +9n
|
---|
39 | [
|
---|
40 | .BI \-n " tracename"
|
---|
41 | ]
|
---|
42 | .if t .ti +.5i
|
---|
43 | [
|
---|
44 | .BI \-b " bufsize"
|
---|
45 | ]
|
---|
46 | [
|
---|
47 | .BI \-m " memlimit"
|
---|
48 | ]
|
---|
49 | .if n .ti +9n
|
---|
50 | [
|
---|
51 | .BI \-c " numchunks"
|
---|
52 | ]
|
---|
53 | [
|
---|
54 | .BI \-a " stemmer"
|
---|
55 | ]
|
---|
56 | [
|
---|
57 | .BI \-s " stemmethod"
|
---|
58 | ]
|
---|
59 | [
|
---|
60 | .BI \-t " tracepos"
|
---|
61 | ]
|
---|
62 | .if n .ti +9n
|
---|
63 | [
|
---|
64 | .B \-T1
|
---|
65 | ]
|
---|
66 | [
|
---|
67 | .B \-T2
|
---|
68 | ]
|
---|
69 | .if t .ti +.5i
|
---|
70 | [
|
---|
71 | .B \-I1
|
---|
72 | ]
|
---|
73 | [
|
---|
74 | .B \-I2
|
---|
75 | ]
|
---|
76 | .if n .ti +9n
|
---|
77 | [
|
---|
78 | .BI \-d " directory"
|
---|
79 | ]
|
---|
80 | .BI \-f " name"
|
---|
81 | [
|
---|
82 | .I filename(s)
|
---|
83 | ]
|
---|
84 | .SH DESCRIPTION
|
---|
85 | .B mg_passes
|
---|
86 | is the program that does most of the work when building
|
---|
87 | .BR mg (1)
|
---|
88 | database systems. The input documents can come from either
|
---|
89 | .I stdin
|
---|
90 | or from a list of files on the command line. Individual documents
|
---|
91 | must be separated with control-B characters. In general,
|
---|
92 | .B mg_passes
|
---|
93 | must be run twice to build a database, first with the
|
---|
94 | .B \-T1
|
---|
95 | and
|
---|
96 | .B \-I1
|
---|
97 | options, and second with the
|
---|
98 | .B \-T2
|
---|
99 | and
|
---|
100 | .B \-I2
|
---|
101 | options. Several other programs must be run in order to get an
|
---|
102 | .BR mg (1)
|
---|
103 | database that is ready for the
|
---|
104 | .BR mgquery (1)
|
---|
105 | program. The
|
---|
106 | .SB EXAMPLE
|
---|
107 | section below gives an example of how to build a complete
|
---|
108 | .BR mg (1)
|
---|
109 | database.
|
---|
110 | .SH OPTIONS
|
---|
111 | Options may appear in any order, but the
|
---|
112 | .IR filename(s) ,
|
---|
113 | if specified, must be last.
|
---|
114 | .TP "\w'\fB\-C\fP \fIcompstatpoint\fP'u+2n"
|
---|
115 | .B \-h
|
---|
116 | This displays a usage line on
|
---|
117 | .IR stderr .
|
---|
118 | .TP
|
---|
119 | .B \-G
|
---|
120 | Treat SGML tags as non-words when building the inverted file. An SGML
|
---|
121 | tag is anything between angle brackets, i.e., `<' and `>'.
|
---|
122 | .TP
|
---|
123 | .B \-S
|
---|
124 | This option causes a special pass to be executed. It is up to a user
|
---|
125 | to modify
|
---|
126 | .I mg.special.c
|
---|
127 | in the source code to do something with the documents it is given.
|
---|
128 | .TP
|
---|
129 | .B \-D
|
---|
130 | If
|
---|
131 | .B mg_passes
|
---|
132 | fails, then print the document that caused the failure to the trace
|
---|
133 | file if tracing is active, or to
|
---|
134 | .I stderr
|
---|
135 | if it is not.
|
---|
136 | .TP
|
---|
137 | .B \-W
|
---|
138 | This option enables the generation of the weights file when
|
---|
139 | .B \-I2
|
---|
140 | is specified. It causes
|
---|
141 | .B \-I2
|
---|
142 | to use a little more memory and CPU.
|
---|
143 | .TP
|
---|
144 | .B \-1
|
---|
145 | Produce a level-1 inverted file. This option is only useful when
|
---|
146 | specified with
|
---|
147 | .BR "\-I1 ".
|
---|
148 | A level-1 inverted file makes it possible for
|
---|
149 | .BR mgquery (1)
|
---|
150 | to do Boolean queries. Ranked queries can still be done,
|
---|
151 | although the quality of the ranking is abysmal.
|
---|
152 | .TP
|
---|
153 | .B \-2
|
---|
154 | Produce a level-2 inverted file. This option is only useful when
|
---|
155 | specified with
|
---|
156 | .BR "\-I1 ".
|
---|
157 | This is the default when neither
|
---|
158 | .BR \-1 ", " "\-2 " "or " \-3
|
---|
159 | is specified.
|
---|
160 | A level-2 inverted file makes it possible for
|
---|
161 | .BR mgquery (1)
|
---|
162 | to do Boolean queries and cosine-ranked queries.
|
---|
163 | .TP
|
---|
164 | .B \-3
|
---|
165 | Produce a level-3 inverted file. This option is only useful when
|
---|
166 | specified with
|
---|
167 | .BR "\-I1 ".
|
---|
168 | This has been implemented to enable paragraph-level inversion.
|
---|
169 | Paragraphs are delimited by control-C characters in the source text.
|
---|
170 | .TP
|
---|
171 | .BI \-C " compstatpoint"
|
---|
172 | This option causes statistics on the compression performance to be
|
---|
173 | output to a file called
|
---|
174 | .IR *.compression.stats .
|
---|
175 | .I compstatpoint
|
---|
176 | specifies the interval between outputting each line of statistics. The
|
---|
177 | units of
|
---|
178 | .I compstatpoint
|
---|
179 | are kilobytes of source text. E.g., if
|
---|
180 | .I compstatpoint
|
---|
181 | is 10, then a line is output to the file every 10 KB of input
|
---|
182 | source. Each line of the file consists of 4 numbers The first number
|
---|
183 | is the amount of input text, in bytes, processed so far. The second
|
---|
184 | number is the amount of input text, in bytes, processed since the
|
---|
185 | last line was output to the file. The third number is the number of
|
---|
186 | output bytes generated since the last line was output to the file, and
|
---|
187 | the fourth number is the compression achieved since the last line was
|
---|
188 | output, i.e., the third number divided by the second number.
|
---|
189 | .TP
|
---|
190 | .BI \-n " tracename"
|
---|
191 | This specifies the filename to use for the trace log, if tracing is
|
---|
192 | enabled using the
|
---|
193 | .B \-t
|
---|
194 | option. If
|
---|
195 | .BI \-n " tracename"
|
---|
196 | is not given and tracing is enabled, a default trace filename will be
|
---|
197 | used.
|
---|
198 | .TP
|
---|
199 | .BI \-s " stemmethod"
|
---|
200 | This specifies the method to use to \*(lqstem\*(rq the words in the
|
---|
201 | inverted file dictionary. This is a bit mask specifying the
|
---|
202 | operations to do on words as they are parsed out of the text, where
|
---|
203 | bit number 0 is the low-order (rightmost) bit. Bit 0 does case
|
---|
204 | folding, and bit 1 does simple stemming, so the value 3 for
|
---|
205 | .I stemmethod
|
---|
206 | does both case folding and stemming.
|
---|
207 | .TP
|
---|
208 | .BI \-a " stemmer"
|
---|
209 | This specifies the stemmer to use when stemming words. This
|
---|
210 | is a description of the language the stemmer is intended for
|
---|
211 | or a description of the stemmer. Valid options include:
|
---|
212 | english, lovin, french, and simplefrench.
|
---|
213 | .TP
|
---|
214 | .BI \-b " bufsize"
|
---|
215 | Specify the size of the document buffer in kilobytes. If any document
|
---|
216 | is larger than
|
---|
217 | .IR bufsize ,
|
---|
218 | the program will abort with an error message. This should probably be
|
---|
219 | replaced with some system which automatically increases the buffer
|
---|
220 | size as required. The default size is 3072 KB (3 MB).
|
---|
221 | .TP
|
---|
222 | .BI \-m " memlimit"
|
---|
223 | Maximum amount of memory to use for the pass-2 file inversion in
|
---|
224 | megabytes. This option is only useful when used in conjunction with
|
---|
225 | the option
|
---|
226 | .BR \-I1 .
|
---|
227 | The larger this value, the faster the pass-2 inversion will proceed.
|
---|
228 | The default value is 5 MB.
|
---|
229 | .TP
|
---|
230 | .BI \-c " numchunks"
|
---|
231 | The maximum number of inversion chunks to write to disk. Each chunk
|
---|
232 | will be approximately as large as
|
---|
233 | .IR memlimit .
|
---|
234 | This option is only useful when used in conjunction with the option
|
---|
235 | .BR \-I2 .
|
---|
236 | The larger this value, the faster the pass-2 inversion will proceed.
|
---|
237 | The default value is 5 MB.
|
---|
238 | .TP
|
---|
239 | .BI \-t " tracepos"
|
---|
240 | This option activates tracing. A line will be generated in the
|
---|
241 | trace file for every
|
---|
242 | .I tracepos
|
---|
243 | input bytes processed. The default name for the trace file can be
|
---|
244 | overridden using the
|
---|
245 | .BI \-n " tracename"
|
---|
246 | option.
|
---|
247 | .TP
|
---|
248 | .B \-T1
|
---|
249 | Generate the
|
---|
250 | .I *.text.stats
|
---|
251 | file.
|
---|
252 | .TP
|
---|
253 | .B \-T2
|
---|
254 | Generate the
|
---|
255 | .IR *.text ,
|
---|
256 | .IR *.text.idx ,
|
---|
257 | and possibly the
|
---|
258 | .I *.text.dict.aux
|
---|
259 | files. Using this option requires that the
|
---|
260 | .I *.text.dict
|
---|
261 | file be present.
|
---|
262 | .TP
|
---|
263 | .B \-I1
|
---|
264 | Generate the
|
---|
265 | .IR *.invf.dict ,
|
---|
266 | .IR *.invf.chunk ,
|
---|
267 | and
|
---|
268 | .I *.invf.chunk.trans
|
---|
269 | files.
|
---|
270 | .TP
|
---|
271 | .B \-I2
|
---|
272 | Generate the
|
---|
273 | .I *.invf
|
---|
274 | and
|
---|
275 | .I *.invf.idx
|
---|
276 | files. Using this option requires
|
---|
277 | that the
|
---|
278 | .IR *.invf.dict.hash ,
|
---|
279 | .IR *.invf.chunk ,
|
---|
280 | and
|
---|
281 | .I *.invf.chunk.trans
|
---|
282 | files
|
---|
283 | be present. The
|
---|
284 | .I *.invf.dict.hash
|
---|
285 | file is generated by
|
---|
286 | .BR mg_perf_hash_build (1)
|
---|
287 | from the
|
---|
288 | .I *.invf.dict.build
|
---|
289 | file. If the
|
---|
290 | .B \-W
|
---|
291 | option is specified, the
|
---|
292 | .I *.weight
|
---|
293 | file will also be generated.
|
---|
294 | .TP
|
---|
295 | .BI \-d " directory"
|
---|
296 | This specifies the directory where the document collection is to be
|
---|
297 | written.
|
---|
298 | .TP
|
---|
299 | .BI \-f " name"
|
---|
300 | This specifies the base name of the document collection that will be
|
---|
301 | created.
|
---|
302 | .TP
|
---|
303 | .I filename(s)
|
---|
304 | This specifies the source text. If this is not specified, then the
|
---|
305 | program expects the source text from
|
---|
306 | .IR stdin .
|
---|
307 | .SH EXAMPLE
|
---|
308 | What follows is a UNIX
|
---|
309 | .BR csh (1)
|
---|
310 | script as an example of how to build an
|
---|
311 | .BR mg (1)
|
---|
312 | document collection.
|
---|
313 | .LP
|
---|
314 | .nf
|
---|
315 | .DT
|
---|
316 | .ft B
|
---|
317 | .I #! /bin/csh
|
---|
318 | .I
|
---|
319 | # The first argument on the command line specifies the
|
---|
320 | .I
|
---|
321 | # source of the text
|
---|
322 | set source = ($1)
|
---|
323 | .PP
|
---|
324 | .I
|
---|
325 | # The second argument is the name of the collection
|
---|
326 | set text = ($2)
|
---|
327 | .PP
|
---|
328 | .I
|
---|
329 | # Create *.text.stats, *.invf.dict.build,
|
---|
330 | .I
|
---|
331 | # *.invf.chunk and *.invf.chunks.trans
|
---|
332 | ${source} | mg_passes -T1 -I1 -m 1 -t 1 -f ${text}
|
---|
333 | .PP
|
---|
334 | .I
|
---|
335 | # Create *.text.dict
|
---|
336 | mg_compression_dict -f ${text}
|
---|
337 | .PP
|
---|
338 | .I
|
---|
339 | # Create *.invf.dict.hash
|
---|
340 | mg_perf_hash_build -f ${text}
|
---|
341 | .PP
|
---|
342 | .I
|
---|
343 | # Create *.text, *.text.idx,
|
---|
344 | .I
|
---|
345 | # *.invf and *.invf.idx
|
---|
346 | ${source} | mg_passes -T2 -I2 -c 2 -t 1 -f ${text}
|
---|
347 | .PP
|
---|
348 | .I
|
---|
349 | # Create *.text.idx.wgt and *.weight.approx
|
---|
350 | mg_weights_build -f ${text} -b 8
|
---|
351 | .PP
|
---|
352 | .I
|
---|
353 | # Create *.invf.dict
|
---|
354 | mg_invf_dict -f ${text} -b 4096
|
---|
355 | .PP
|
---|
356 | .I
|
---|
357 | # Create *.text.dict
|
---|
358 | mg_fast_comp_dict -f ${text}
|
---|
359 | .ft R
|
---|
360 | .fi
|
---|
361 | .SH ENVIRONMENT
|
---|
362 | .TP "\w'\fBMGDATA\fP'u+2n"
|
---|
363 | .SB MGDATA
|
---|
364 | If this environment variable exists, then its value is used as the
|
---|
365 | default directory where the
|
---|
366 | .BR mg (1)
|
---|
367 | collection files are. If this variable does not exist, then the
|
---|
368 | directory \*(lq\fB.\fP\*(rq is used by default. The command line
|
---|
369 | option
|
---|
370 | .BI \-d " directory"
|
---|
371 | overrides the directory in
|
---|
372 | .BR MGDATA .
|
---|
373 | .SH FILES
|
---|
374 | .TP 20
|
---|
375 | .B *.invf
|
---|
376 | Inverted file.
|
---|
377 | .TP
|
---|
378 | .B *.invf.chunk
|
---|
379 | Inverted file chunk descriptor file. When the inverted file is
|
---|
380 | created it is created in chunks that use no more than a set amount of
|
---|
381 | memory. This file describes those chunks.
|
---|
382 | .TP
|
---|
383 | .B *.invf.chunk.trans
|
---|
384 | Word-occurrence-order to lexical-order translation file. The
|
---|
385 | .B *.invf.chunk
|
---|
386 | file is written in word-occurrence order but is required by
|
---|
387 | .B \-I2
|
---|
388 | to be in lexical order.
|
---|
389 | .TP
|
---|
390 | .B *.invf.dict.build
|
---|
391 | Compressed stemmed dictionary.
|
---|
392 | .TP
|
---|
393 | .B *.invf.dict.hash
|
---|
394 | Data for an order-preserving perfect hash function.
|
---|
395 | .TP
|
---|
396 | .B *.invf.idx
|
---|
397 | The index into the inverted file.
|
---|
398 | .TP
|
---|
399 | .B *.weight
|
---|
400 | The exact weights file.
|
---|
401 | .TP
|
---|
402 | .B *.text
|
---|
403 | Compressed documents.
|
---|
404 | .TP
|
---|
405 | .B *.text.stats
|
---|
406 | Statistics about the text.
|
---|
407 | .TP
|
---|
408 | .B *.text.dict
|
---|
409 | Compressed compression dictionary.
|
---|
410 | .TP
|
---|
411 | .B *.text.idx
|
---|
412 | Index into the compressed documents.
|
---|
413 | .TP
|
---|
414 | .B *.trace
|
---|
415 | The default trace file.
|
---|
416 | .TP
|
---|
417 | .B *.compression.stats
|
---|
418 | Statistics about the compression of the text.
|
---|
419 | .SH "SEE ALSO"
|
---|
420 | .na
|
---|
421 | .BR mg (1),
|
---|
422 | .BR mg_compression_dict (1),
|
---|
423 | .BR mg_fast_comp_dict (1),
|
---|
424 | .BR mg_get (1),
|
---|
425 | .BR mg_invf_dict (1),
|
---|
426 | .BR mg_invf_dump (1),
|
---|
427 | .BR mg_invf_rebuild (1),
|
---|
428 | .BR mg_perf_hash_build (1),
|
---|
429 | .BR mg_text_estimate (1),
|
---|
430 | .BR mg_weights_build (1),
|
---|
431 | .BR mgbilevel (1),
|
---|
432 | .BR mgbuild (1),
|
---|
433 | .BR mgdictlist (1),
|
---|
434 | .BR mgfelics (1),
|
---|
435 | .BR mgquery (1),
|
---|
436 | .BR mgstat (1),
|
---|
437 | .BR mgtic (1),
|
---|
438 | .BR mgticbuild (1),
|
---|
439 | .BR mgticdump (1),
|
---|
440 | .BR mgticprune (1),
|
---|
441 | .BR mgticstat (1).
|
---|