Context Navigation

Back to Ticket #186

Ticket #186: patch.2

File patch.2, 29.1 KB (added by Richard Boulton, 16 years ago)
Updated patch for subclassing Stem

Line
1	Index: queryparser/termgenerator_internal.h
2	===================================================================
3	--- queryparser/termgenerator_internal.h (revision 10133)
4	+++ queryparser/termgenerator_internal.h (working copy)
5	@@ -33,7 +33,7 @@
6
7	class TermGenerator::Internal : public Xapian::Internal::RefCntBase {
8	friend class TermGenerator;
9	- Stem stemmer;
10	+ Xapian::Internal::RefCntPtr<Xapian::BaseStem> stemmer;
11	const Stopper * stopper;
12	Document doc;
13	termcount termpos;
14	Index: queryparser/termgenerator.cc
15	===================================================================
16	--- queryparser/termgenerator.cc (revision 10133)
17	+++ queryparser/termgenerator.cc (working copy)
18	@@ -46,7 +46,7 @@
19	void
20	TermGenerator::set_stemmer(const Xapian::Stem & stemmer)
21	{
22	- internal->stemmer = stemmer;
23	+ internal->stemmer = stemmer.internal;
24	}
25
26	void
27	@@ -122,7 +122,7 @@
28	string s("Xapian::TermGenerator(");
29	if (internal.get()) {
30	s += "stem=";
31	- s += internal->stemmer.get_description();
32	+ s += internal->stemmer->get_description();
33	if (internal->stopper) {
34	s += ", stopper set";
35	}
36	Index: queryparser/queryparser.lemony
37	===================================================================
38	--- queryparser/queryparser.lemony (revision 10133)
39	+++ queryparser/queryparser.lemony (working copy)
40	@@ -182,7 +182,7 @@
41	: qpi(qpi_), error(NULL), flags(flags_) { }
42
43	string stem_term(const string &term) {
44	- return qpi->stemmer(term);
45	+ return (*qpi->stemmer)(term);
46	}
47
48	void add_to_stoplist(const Term * term) {
49	@@ -907,7 +907,7 @@
50	// when used with positional operators".
51	stem_strategy stem_term = stem_action;
52	if (stem_term != STEM_NONE) {
53	- if (!stemmer.internal.get()) {
54	+ if (!stemmer.get()) {
55	// No stemmer is set.
56	stem_term = STEM_NONE;
57	} else if (stem_term == STEM_SOME) {
58	Index: queryparser/queryparser_internal.h
59	===================================================================
60	--- queryparser/queryparser_internal.h (revision 10133)
61	+++ queryparser/queryparser_internal.h (working copy)
62	@@ -57,7 +57,7 @@
63	class QueryParser::Internal : public Xapian::Internal::RefCntBase {
64	friend class QueryParser;
65	friend class ::State;
66	- Stem stemmer;
67	+ Xapian::Internal::RefCntPtr<Xapian::BaseStem> stemmer;
68	stem_strategy stem_action;
69	const Stopper * stopper;
70	Query::op default_op;
71	Index: queryparser/queryparser.cc
72	===================================================================
73	--- queryparser/queryparser.cc (revision 10133)
74	+++ queryparser/queryparser.cc (working copy)
75	@@ -74,7 +74,7 @@
76	void
77	QueryParser::set_stemmer(const Xapian::Stem & stemmer)
78	{
79	- internal->stemmer = stemmer;
80	+ internal->stemmer = stemmer.internal;
81	}
82
83	void
84	Index: queryparser/termgenerator_internal.cc
85	===================================================================
86	--- queryparser/termgenerator_internal.cc (revision 10133)
87	+++ queryparser/termgenerator_internal.cc (working copy)
88	@@ -206,7 +206,7 @@
89	}
90	if ((flags & FLAG_SPELLING) && prefix.empty()) db.add_spelling(term);
91
92	- if (!stemmer.internal.get()) continue;
93	+ if (!stemmer.get()) continue;
94
95	if (stop_mode == STOPWORDS_INDEX_UNSTEMMED_ONLY && (*stopper)(term))
96	continue;
97	@@ -218,7 +218,7 @@
98	// Add stemmed form without positional information.
99	string stem("Z");
100	stem += prefix;
101	- stem += stemmer(term);
102	+ stem += (*stemmer)(term);
103	doc.add_term(stem, weight);
104	}
105	}
106	Index: tests/nocompile_stem1.cc
107	===================================================================
108	--- tests/nocompile_stem1.cc (revision 0)
109	+++ tests/nocompile_stem1.cc (revision 0)
110	@@ -0,0 +1,29 @@
111	+/** @file nocompile_stem1.cc
112	+ * @brief Check that it's not possible to instantiate an instance of one of
113	+ * the built-in BaseStem subclasses on the stack.
114	+ */
115	+/* Copyright 2008 Lemur Consulting Ltd
116	+ *
117	+ * This program is free software; you can redistribute it and/or
118	+ * modify it under the terms of the GNU General Public License as
119	+ * published by the Free Software Foundation; either version 2 of the
120	+ * License, or (at your option) any later version.
121	+ *
122	+ * This program is distributed in the hope that it will be useful,
123	+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
124	+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
125	+ * GNU General Public License for more details.
126	+ *
127	+ * You should have received a copy of the GNU General Public License
128	+ * along with this program; if not, write to the Free Software
129	+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301
130	+ * USA
131	+ */
132	+
133	+#include <config.h>
134	+#include <xapian.h>
135	+
136	+int main()
137	+{
138	+ Xapian::SnowballStem stem("en");
139	+}
140
141	Property changes on: tests/nocompile_stem1.cc
142	___________________________________________________________________
143	Name: svn:eol-style
144	+ native
145
146	Index: tests/nocompile_stem2.cc
147	===================================================================
148	--- tests/nocompile_stem2.cc (revision 0)
149	+++ tests/nocompile_stem2.cc (revision 0)
150	@@ -0,0 +1,42 @@
151	+/** @file nocompile_stem2.cc
152	+ * @brief Check that it's not possible to instantiate an instance of a
153	+ * SnowballStem subclass on the stack.
154	+ */
155	+/* Copyright 2008 Lemur Consulting Ltd
156	+ *
157	+ * This program is free software; you can redistribute it and/or
158	+ * modify it under the terms of the GNU General Public License as
159	+ * published by the Free Software Foundation; either version 2 of the
160	+ * License, or (at your option) any later version.
161	+ *
162	+ * This program is distributed in the hope that it will be useful,
163	+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
164	+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
165	+ * GNU General Public License for more details.
166	+ *
167	+ * You should have received a copy of the GNU General Public License
168	+ * along with this program; if not, write to the Free Software
169	+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301
170	+ * USA
171	+ */
172	+
173	+#include <config.h>
174	+#include <xapian.h>
175	+
176	+class SnowballStemSubclass : public Xapian::SnowballStem {
177	+ protected:
178	+ // All user derived subclasses have to make their destructor
179	+ // restricted, so that the class can't be instantiated on the stack.
180	+ ~SnowballStemSubclass() {}
181	+ public:
182	+ SnowballStemSubclass(const std::string & language)
183	+ : Xapian::SnowballStem(language) {}
184	+
185	+ std::string operator()(const std::string &word) const { return "!2" + Xapian::SnowballStem::operator()(word); }
186	+ std::string get_description() const { return "StemSubclass"; }
187	+};
188	+
189	+int main()
190	+{
191	+ SnowballStemSubclass stem("en");
192	+}
193
194	Property changes on: tests/nocompile_stem2.cc
195	___________________________________________________________________
196	Name: svn:eol-style
197	+ native
198
199	Index: tests/nocompiletest.in
200	===================================================================
201	--- tests/nocompiletest.in (revision 0)
202	+++ tests/nocompiletest.in (revision 0)
203	@@ -0,0 +1,61 @@
204	+#!/bin/sh
205	+# @configure_input@
206	+#
207	+# Copyright (C) 2002,2003,2004,2005,2006,2007 Olly Betts
208	+#
209	+# This program is free software; you can redistribute it and/or
210	+# modify it under the terms of the GNU General Public License as
211	+# published by the Free Software Foundation; either version 2 of the
212	+# License, or (at your option) any later version.
213	+#
214	+# This program is distributed in the hope that it will be useful,
215	+# but WITHOUT ANY WARRANTY; without even the implied warranty of
216	+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
217	+# GNU General Public License for more details.
218	+#
219	+# You should have received a copy of the GNU General Public License
220	+# along with this program; if not, write to the Free Software
221	+# Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301
222	+# USA
223	+
224	+# Test that various programs don't compile.
225	+
226	+EXEEXT="@EXEEXT@"
227	+SRCDIR="@srcdir@"
228	+out="nocompiletest_stdout.tmp"
229	+err="nocompiletest_stderr.tmp"
230	+
231	+TESTS="stem1 stem2"
232	+exitstatus=0
233	+failures=0
234	+
235	+echo "Attempting to build valid test program"
236	+# First, check that building "nocompile_pass" works, to ensure that the
237	+# platform actually allows anything to compile.
238	+rm -f "nocompile_pass${EXEEXT}"
239	+make "nocompile_pass${EXEEXT}" >"$out" 2>"$err"
240	+if [ $? -ne 0 ] ; then
241	+ cat < "$out"
242	+ echo
243	+ echo "Failed to compile valid test program, so unable to test for compilation"
244	+ echo "problems with invalid test programs."
245	+ echo
246	+ cat < "$err"
247	+ rm -f $out $err
248	+ exit 1
249	+fi
250	+
251	+# Now, try compiling each of the test programs, checking that compilation fails.
252	+for test in $TESTS
253	+do
254	+ echo "Testing build of nocompile_${test}"
255	+ rm -f "nocompile_${test}${EXEEXT}"
256	+ make "nocompile_${test}${EXEEXT}" >"$out" 2>"$err"
257	+ if [ $? -eq 0 ] ; then
258	+ echo "FAIL: Compilation of invalid test program succeeded."
259	+ exitstatus=1
260	+ fi
261	+done
262	+
263	+rm -f $out $err
264	+exit $exitstatus
265	Index: tests/nocompile_pass.cc
266	===================================================================
267	--- tests/nocompile_pass.cc (revision 0)
268	+++ tests/nocompile_pass.cc (revision 0)
269	@@ -0,0 +1,29 @@
270	+/** @file nocompile_pass.cc
271	+ * @brief Simple program using Xapian which should compile and link correctly.
272	+ */
273	+/* Copyright 2008 Lemur Consulting Ltd
274	+ *
275	+ * This program is free software; you can redistribute it and/or
276	+ * modify it under the terms of the GNU General Public License as
277	+ * published by the Free Software Foundation; either version 2 of the
278	+ * License, or (at your option) any later version.
279	+ *
280	+ * This program is distributed in the hope that it will be useful,
281	+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
282	+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
283	+ * GNU General Public License for more details.
284	+ *
285	+ * You should have received a copy of the GNU General Public License
286	+ * along with this program; if not, write to the Free Software
287	+ * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301
288	+ * USA
289	+ */
290	+
291	+#include <config.h>
292	+#include <xapian.h>
293	+
294	+int main()
295	+{
296	+ Xapian::Stem stem("en");
297	+ stem("hello");
298	+}
299
300	Property changes on: tests/nocompile_pass.cc
301	___________________________________________________________________
302	Name: svn:eol-style
303	+ native
304
305	Index: tests/api_nodb.cc
306	===================================================================
307	--- tests/api_nodb.cc (revision 10133)
308	+++ tests/api_nodb.cc (working copy)
309	@@ -284,6 +284,37 @@
310	return true;
311	}
312
313	+class StemSubclass : public Xapian::BaseStem {
314	+ private:
315	+ virtual ~StemSubclass() {}
316	+ public:
317	+ std::string operator()(const std::string &word) const { return "!" + word; }
318	+ std::string get_description() const { return "StemSubclass"; }
319	+};
320	+
321	+class SnowballStemSubclass : public Xapian::SnowballStem {
322	+ public:
323	+ SnowballStemSubclass(const std::string & language)
324	+ : Xapian::SnowballStem(language) {}
325	+
326	+ std::string operator()(const std::string &word) const { return "!2" + Xapian::SnowballStem::operator()(word); }
327	+ std::string get_description() const { return "StemSubclass"; }
328	+};
329	+
330	+// Test subclassing of stem.
331	+DEFINE_TESTCASE(stemsubclass1, !backend) {
332	+ Xapian::Stem stemmer(new StemSubclass());
333	+ TEST_EQUAL(stemmer("foods"), "!foods");
334	+ TEST_EQUAL(stemmer("food"), "!food");
335	+ TEST_EQUAL(stemmer(""), "!");
336	+
337	+ Xapian::Stem stemmer2(new SnowballStemSubclass("en"));
338	+ TEST_EQUAL(stemmer2("foods"), "!2food");
339	+ TEST_EQUAL(stemmer2("food"), "!2food");
340	+ TEST_EQUAL(stemmer2(""), "!2");
341	+ return true;
342	+}
343	+
344	// Some simple tests of the built in weighting schemes.
345	DEFINE_TESTCASE(weight1, !backend) {
346	Xapian::Weight * wt;
347	Index: tests/Makefile.am
348	===================================================================
349	--- tests/Makefile.am (revision 10133)
350	+++ tests/Makefile.am (working copy)
351	@@ -17,11 +17,13 @@
352
353	check-none: apitest$(EXEEXT) \
354	internaltest$(EXEEXT) \
355	+ nocompiletest \
356	stemtest$(EXEEXT) \
357	queryparsertest$(EXEEXT) \
358	termgentest$(EXEEXT)
359	$(TESTS_ENVIRONMENT) ./apitest$(EXEEXT) -b none
360	$(TESTS_ENVIRONMENT) ./internaltest$(EXEEXT)
361	+ $(TESTS_ENVIRONMENT) ./nocompiletest
362	$(TESTS_ENVIRONMENT) ./stemtest$(EXEEXT)
363	$(TESTS_ENVIRONMENT) ./queryparsertest$(EXEEXT)
364	$(TESTS_ENVIRONMENT) ./termgentest$(EXEEXT)
365	@@ -62,9 +64,18 @@
366	INCLUDES += -I$(top_srcdir)/backends/quartz
367	endif
368
369	+nocompile_programs = \
370	+ nocompile_pass \
371	+ nocompile_stem1 \
372	+ nocompile_stem2
373	+
374	## Test programs to be run
375	-TESTS = $(quartz_tests) \
376	- apitest$(EXEEXT) internaltest$(EXEEXT) stemtest$(EXEEXT) \
377	+TESTS = \
378	+ nocompiletest \
379	+ $(quartz_tests) \
380	+ apitest$(EXEEXT) \
381	+ internaltest$(EXEEXT) \
382	+ stemtest$(EXEEXT) \
383	queryparsertest$(EXEEXT) \
384	termgentest$(EXEEXT)
385
386	@@ -72,9 +83,13 @@
387	check_PROGRAMS = $(quartz_check_programs) \
388	apitest internaltest stemtest queryparsertest termgentest
389
390	-# Make sure runtest is up to date before running tests
391	-check_SCRIPTS = runtest
392	+## We want to have rules for the nocompile programs, but not try building
393	+## ourselves (nocompiletest calls back to the makefile to build them).
394	+EXTRA_PROGRAMS = $(nocompile_programs)
395
396	+# We include runtest here to make sure it is up to date before running tests
397	+check_SCRIPTS = nocompiletest runtest
398	+
399	## Sources:
400
401	noinst_HEADERS = apitest.h
402	@@ -113,6 +128,13 @@
403	termgentest_LDFLAGS = -no-install $(ldflags)
404	termgentest_LDADD = ../libgetopt.la ../libxapian.la
405
406	+nocompile_pass_SOURCES = nocompile_pass.cc
407	+nocompile_pass_LDADD = ../libxapian.la
408	+nocompile_stem1_SOURCES = nocompile_stem1.cc
409	+nocompile_stem1_LDADD = ../libxapian.la
410	+nocompile_stem2_SOURCES = nocompile_stem2.cc
411	+nocompile_stem2_LDADD = ../libxapian.la
412	+
413	BUILT_SOURCES =
414
415	if MAINTAINER_MODE
416	@@ -150,6 +172,7 @@
417
418	## Distribute test data:
419	EXTRA_DIST +=\
420	+ nocompiletest.in \
421	runtest.in \
422	valgrind.supp \
423	testdata/apitest_onedoc.txt \
424	Index: configure.ac
425	===================================================================
426	--- configure.ac (revision 10133)
427	+++ configure.ac (working copy)
428	@@ -1114,6 +1114,7 @@
429	docs/doxygen_full_conf
430	xapian-core.spec
431	])
432	+AC_CONFIG_FILES([tests/nocompiletest], [chmod +x tests/nocompiletest])
433	AC_CONFIG_FILES([tests/runtest], [chmod +x tests/runtest])
434	AC_CONFIG_FILES([tests/runsrv], [chmod +x tests/runsrv])
435	AC_CONFIG_FILES([xapian-config], [chmod +x xapian-config])
436	Index: include/xapian/stem.h
437	===================================================================
438	--- include/xapian/stem.h (revision 10133)
439	+++ include/xapian/stem.h (working copy)
440	@@ -2,6 +2,7 @@
441	* \brief stemming algorithms
442	*/
443	/* Copyright (C) 2005,2007 Olly Betts
444	+ * Copyright (C) 2008 Lemur Consulting Ltd
445	*
446	* This program is free software; you can redistribute it and/or
447	* modify it under the terms of the GNU General Public License as
448	@@ -28,27 +29,62 @@
449
450	namespace Xapian {
451
452	-/// Class representing a stemming algorithm.
453	-class XAPIAN_VISIBILITY_DEFAULT Stem {
454	- public:
455	- /// @private @internal Class representing the stemmer internals.
456	- class Internal;
457	- /// @private @internal Reference counted internals.
458	- Xapian::Internal::RefCntPtr<Internal> internal;
459	+/// Base representing a stemming algorithm.
460	+class XAPIAN_VISIBILITY_DEFAULT BaseStem : public Xapian::Internal::RefCntBase {
461	+ /// No copying allowed.
462	+ BaseStem(const BaseStem & o);
463
464	- /// Copy constructor.
465	- Stem(const Stem & o);
466	+ /// No assignment allowed.
467	+ void operator=(const BaseStem & o);
468
469	- /// Assignment.
470	- void operator=(const Stem & o);
471	+ protected:
472	+ /** Destructor is protected since it should only be called by subclasses
473	+ * and RefCntPtr. Subclasses should make their destructors protected,
474	+ * to force users to use a RefCntPtr to reference them.
475	+ */
476	+ virtual ~BaseStem() {}
477
478	- /** Construct a Xapian::Stem object which doesn't change terms.
479	+ friend class Xapian::Internal::RefCntPtr<BaseStem>;
480	+
481	+ public:
482	+ BaseStem() {}
483	+
484	+ /** Stem a word.
485	*
486	- * Equivalent to Stem("none").
487	+ * @param word a word to stem.
488	+ * @return the stemmed form of the word.
489	*/
490	- Stem();
491	+ virtual std::string operator()(const std::string &word) const = 0;
492
493	- /** Construct a Xapian::Stem object for a particular language.
494	+ /// Return a string describing this object.
495	+ virtual std::string get_description() const = 0;
496	+};
497	+
498	+/// Class representing one of the snowball stemming algorithms.
499	+class XAPIAN_VISIBILITY_DEFAULT SnowballStem : public BaseStem {
500	+ /// No copying allowed.
501	+ SnowballStem(const SnowballStem & o);
502	+
503	+ /// No assignment allowed.
504	+ void operator=(const SnowballStem & o);
505	+
506	+ protected:
507	+ /** Destructor is protected since it should only be called by subclasses
508	+ * and RefCntPtr. Subclasses should make their destructors protected,
509	+ * to force users to use a RefCntPtr to reference them.
510	+ */
511	+ virtual ~SnowballStem();
512	+
513	+ public:
514	+ /// @private @internal Class representing the snowball stemmer internals.
515	+ class Internal;
516	+
517	+ private:
518	+ /// @private @internal Snowball stemmer internals.
519	+ Internal * internal;
520	+
521	+ public:
522	+ /** Construct a Xapian::SnowballStem object for a particular language.
523	*
524	* @param language Either the English name for the language
525	* or the two letter ISO639 code.
526	@@ -56,7 +92,6 @@
527	* The following language names are understood (aliases follow the
528	* name):
529	*
530	- * - none - don't stem terms
531	* - danish (da)
532	* - dutch (nl)
533	* - english (en) - Martin Porter's 2002 revision of his stemmer
534	@@ -76,15 +111,12 @@
535	* @exception Xapian::InvalidArgumentError is thrown if
536	* language isn't recognised.
537	*/
538	- explicit Stem(const std::string &language);
539	+ explicit SnowballStem(const std::string &language);
540
541	- /// Destructor.
542	- ~Stem();
543	-
544	/** Stem a word.
545	*
546	- * @param word a word to stem.
547	- * @return the stem
548	+ * @param word a word to stem.
549	+ * @return the stemmed form of the word.
550	*/
551	std::string operator()(const std::string &word) const;
552
553	@@ -104,6 +136,74 @@
554	static std::string get_available_languages();
555	};
556
557	+/// Class wrapping a reference counted stemming algorithm.
558	+class XAPIAN_VISIBILITY_DEFAULT Stem {
559	+ public:
560	+ /// @private @internal Reference counted internals.
561	+ Xapian::Internal::RefCntPtr<Xapian::BaseStem> internal;
562	+
563	+ /// Copy constructor.
564	+ Stem(const Stem & o) : internal(o.internal) { }
565	+
566	+ /// Assignment.
567	+ void operator=(const Stem & o) { internal = o.internal; }
568	+
569	+ /** Construct a Xapian::Stem object from a pointer to a BaseStem.
570	+ */
571	+ Stem(Xapian::Internal::RefCntPtr<Xapian::BaseStem> internal_)
572	+ : internal(internal_) {}
573	+
574	+ /** Construct a Xapian::Stem object which doesn't change terms.
575	+ */
576	+ Stem() : internal(0) {}
577	+
578	+ /** Construct a Xapian::Stem object for a particular language.
579	+ *
580	+ * This constructor is included for convenience, and is equivalent to
581	+ * Stem(new SnowballStem(language)) - except that a language parameter of
582	+ * "none" will produce a stemmer which doesn't remove any stems.
583	+ *
584	+ * See Xapian::SnowballStem for details.
585	+ */
586	+ explicit Stem(const std::string &language)
587	+ {
588	+ if (language == "none")
589	+ internal = 0;
590	+ else
591	+ internal = new Xapian::SnowballStem(language);
592	+ }
593	+
594	+ /** Stem a word.
595	+ *
596	+ * @param word a word to stem.
597	+ * @return the stemmed form of the word.
598	+ */
599	+ std::string operator()(const std::string &word) const
600	+ {
601	+ if (!internal.get()) return word;
602	+ return internal->operator()(word);
603	+ }
604	+
605	+ /// Return a string describing this object.
606	+ std::string get_description() const
607	+ {
608	+ if (!internal.get()) return "Xapian::Stem()";
609	+ return "Xapian::Stem(" + internal->get_description() + ")";
610	+ }
611	+
612	+ /** Return a list of available languages.
613	+ *
614	+ * This is included for convenience, and is equivalent to
615	+ * SnowballStem.get_available_languages().
616	+ *
617	+ * See Xapian::SnowballStem for details.
618	+ */
619	+ static std::string get_available_languages()
620	+ {
621	+ return SnowballStem::get_available_languages();
622	+ }
623	+};
624	+
625	}
626
627	#endif // XAPIAN_INCLUDED_STEM_H
628	Index: languages/Makefile.mk
629	===================================================================
630	--- languages/Makefile.mk (revision 10133)
631	+++ languages/Makefile.mk (working copy)
632	@@ -56,10 +56,10 @@
633	$(CC_FOR_BUILD) -o languages/snowball -DDISABLE_JAVA `for f in $(snowball_sources) ; do test -f $$f && echo $$f \|\| echo $(srcdir)/$$f ; done`
634
635	.sbl.cc:
636	- languages/snowball $< -o `echo $@\|sed 's!\.cc$$!!'` -c++ -u -n InternalStem`echo $<\|sed 's!./$.$.!\1!'\|tr a-z A-Z``echo $<\|sed 's!.*/.!!;s!\.sbl!!'` -p Stem::Internal
637	+ languages/snowball $< -o `echo $@\|sed 's!\.cc$$!!'` -c++ -u -n InternalStem`echo $<\|sed 's!./$.$.!\1!'\|tr a-z A-Z``echo $<\|sed 's!.*/.!!;s!\.sbl!!'` -p SnowballStem::Internal
638
639	.sbl.h:
640	- languages/snowball $< -o `echo $@\|sed 's!\.h$$!!'` -c++ -u -n InternalStem`echo $<\|sed 's!./$.$.!\1!'\|tr a-z A-Z``echo $<\|sed 's!.*/.!!;s!\.sbl!!'` -p Stem::Internal
641	+ languages/snowball $< -o `echo $@\|sed 's!\.h$$!!'` -c++ -u -n InternalStem`echo $<\|sed 's!./$.$.!\1!'\|tr a-z A-Z``echo $<\|sed 's!.*/.!!;s!\.sbl!!'` -p SnowballStem::Internal
642
643	languages/allsnowballheaders.h: languages/generate-allsnowballheaders languages/Makefile.mk
644	languages/generate-allsnowballheaders $(snowball_built_sources)
645	Index: languages/steminternal.cc
646	===================================================================
647	--- languages/steminternal.cc (revision 10133)
648	+++ languages/steminternal.cc (working copy)
649	@@ -131,18 +131,18 @@
650
651	namespace Xapian {
652
653	-Stem::Internal::Internal()
654	+SnowballStem::Internal::Internal()
655	: p(create_s()), c(0), l(0), lb(0), bra(0), ket(0)
656	{
657	}
658
659	-Stem::Internal::~Internal()
660	+SnowballStem::Internal::~Internal()
661	{
662	lose_s(p);
663	}
664
665	string
666	-Stem::Internal::operator()(const string & word)
667	+SnowballStem::Internal::operator()(const string & word)
668	{
669	const symbol * s = reinterpret_cast<const symbol *>(word.data());
670	replace_s(0, l, word.size(), s);
671	@@ -156,7 +156,7 @@
672
673	/* Code for character groupings: utf8 cases */
674
675	-int Stem::Internal::get_utf8(int * slot) {
676	+int SnowballStem::Internal::get_utf8(int * slot) {
677	int b0, b1;
678	int tmp = c;
679	if (tmp >= l) return 0;
680	@@ -171,7 +171,7 @@
681	* slot = (b0 & 0xF) << 12 \| (b1 & 0x3F) << 6 \| (p[tmp] & 0x3F); return 3;
682	}
683
684	-int Stem::Internal::get_b_utf8(int * slot) {
685	+int SnowballStem::Internal::get_b_utf8(int * slot) {
686	int b0, b1;
687	int tmp = c;
688	if (tmp <= lb) return 0;
689	@@ -186,7 +186,7 @@
690	* slot = (p[tmp] & 0xF) << 12 \| (b1 & 0x3F) << 6 \| (b0 & 0x3F); return 3;
691	}
692
693	-int Stem::Internal::in_grouping_U(const unsigned char * s, int min, int max, int repeat) {
694	+int SnowballStem::Internal::in_grouping_U(const unsigned char * s, int min, int max, int repeat) {
695	do {
696	int ch;
697	int w = get_utf8(&ch);
698	@@ -198,7 +198,7 @@
699	return 0;
700	}
701
702	-int Stem::Internal::in_grouping_b_U(const unsigned char * s, int min, int max, int repeat) {
703	+int SnowballStem::Internal::in_grouping_b_U(const unsigned char * s, int min, int max, int repeat) {
704	do {
705	int ch;
706	int w = get_b_utf8(&ch);
707	@@ -210,7 +210,7 @@
708	return 0;
709	}
710
711	-int Stem::Internal::out_grouping_U(const unsigned char * s, int min, int max, int repeat) {
712	+int SnowballStem::Internal::out_grouping_U(const unsigned char * s, int min, int max, int repeat) {
713	do {
714	int ch;
715	int w = get_utf8(&ch);
716	@@ -222,7 +222,7 @@
717	return 0;
718	}
719
720	-int Stem::Internal::out_grouping_b_U(const unsigned char * s, int min, int max, int repeat) {
721	+int SnowballStem::Internal::out_grouping_b_U(const unsigned char * s, int min, int max, int repeat) {
722	do {
723	int ch;
724	int w = get_b_utf8(&ch);
725	@@ -234,21 +234,21 @@
726	return 0;
727	}
728
729	-int Stem::Internal::eq_s(int s_size, const symbol * s) {
730	+int SnowballStem::Internal::eq_s(int s_size, const symbol * s) {
731	if (l - c < s_size \|\| memcmp(p + c, s, s_size * sizeof(symbol)) != 0)
732	return 0;
733	c += s_size;
734	return 1;
735	}
736
737	-int Stem::Internal::eq_s_b(int s_size, const symbol * s) {
738	+int SnowballStem::Internal::eq_s_b(int s_size, const symbol * s) {
739	if (c - lb < s_size \|\| memcmp(p + c - s_size, s, s_size * sizeof(symbol)) != 0)
740	return 0;
741	c -= s_size;
742	return 1;
743	}
744
745	-int Stem::Internal::find_among(const struct among * v, int v_size, const unsigned char * fnum, const among_function * f) {
746	+int SnowballStem::Internal::find_among(const struct among * v, int v_size, const unsigned char * fnum, const among_function * f) {
747	int i = 0;
748	int j = v_size;
749
750	@@ -302,7 +302,7 @@
751	}
752
753	/* find_among_b is for backwards processing. Same comments apply */
754	-int Stem::Internal::find_among_b(const struct among * v, int v_size, const unsigned char * fnum, const among_function * f) {
755	+int SnowballStem::Internal::find_among_b(const struct among * v, int v_size, const unsigned char * fnum, const among_function * f) {
756	int i = 0;
757	int j = v_size;
758
759	@@ -351,7 +351,7 @@
760	}
761
762	int
763	-Stem::Internal::replace_s(int c_bra, int c_ket, int s_size, const symbol * s)
764	+SnowballStem::Internal::replace_s(int c_bra, int c_ket, int s_size, const symbol * s)
765	{
766	int adjustment;
767	int len;
768	@@ -377,7 +377,7 @@
769	return adjustment;
770	}
771
772	-int Stem::Internal::slice_check() {
773	+int SnowballStem::Internal::slice_check() {
774	Assert(p);
775	if (bra < 0 \|\| bra > ket \|\| ket > l) {
776	#if 0
777	@@ -389,19 +389,19 @@
778	return 0;
779	}
780
781	-int Stem::Internal::slice_from_s(int s_size, const symbol * s) {
782	+int SnowballStem::Internal::slice_from_s(int s_size, const symbol * s) {
783	if (slice_check()) return -1;
784	replace_s(bra, ket, s_size, s);
785	return 0;
786	}
787
788	-void Stem::Internal::insert_s(int c_bra, int c_ket, int s_size, const symbol * s) {
789	+void SnowballStem::Internal::insert_s(int c_bra, int c_ket, int s_size, const symbol * s) {
790	int adjustment = replace_s(c_bra, c_ket, s_size, s);
791	if (c_bra <= bra) bra += adjustment;
792	if (c_bra <= ket) ket += adjustment;
793	}
794
795	-symbol * Stem::Internal::slice_to(symbol * v) {
796	+symbol * SnowballStem::Internal::slice_to(symbol * v) {
797	if (slice_check()) return NULL;
798	{
799	int len = ket - bra;
800	@@ -414,7 +414,7 @@
801	return v;
802	}
803
804	-symbol * Stem::Internal::assign_to(symbol * v) {
805	+symbol * SnowballStem::Internal::assign_to(symbol * v) {
806	int len = l;
807	if (CAPACITY(v) < len) {
808	v = increase_size(v, len);
809	@@ -425,7 +425,7 @@
810	}
811
812	#if 0
813	-void Stem::Internal::debug(int number, int line_count) {
814	+void SnowballStem::Internal::debug(int number, int line_count) {
815	int i;
816	int limit = SIZE(p);
817	/if (number >= 0) printf("%3d (line %4d): '", number, line_count);/
818	Index: languages/steminternal.h
819	===================================================================
820	--- languages/steminternal.h (revision 10133)
821	+++ languages/steminternal.h (working copy)
822	@@ -41,7 +41,7 @@
823	#define CAPACITY(P) ((const int )(const void )(P))[-2]
824	#define SET_CAPACITY(P, N) ((int )(void )(P))[-2] = N
825
826	-typedef int (among_function)(Xapian::Stem::Internal );
827	+typedef int (among_function)(Xapian::SnowballStem::Internal );
828
829	struct among {
830	int s_size; /* length of search string (in symbols) */
831	@@ -60,7 +60,7 @@
832
833	namespace Xapian {
834
835	-class Stem::Internal : public Xapian::Internal::RefCntBase {
836	+class SnowballStem::Internal {
837	int slice_check();
838
839	protected:
840	Index: languages/stem.cc
841	===================================================================
842	--- languages/stem.cc (revision 10133)
843	+++ languages/stem.cc (working copy)
844	@@ -33,17 +33,9 @@
845
846	namespace Xapian {
847
848	-Stem::Stem(const Stem & o) : internal(o.internal) { }
849	-
850	-void
851	-Stem::operator=(const Stem & o)
852	+SnowballStem::SnowballStem(const std::string &language)
853	+ : internal(0)
854	{
855	- internal = o.internal;
856	-}
857	-
858	-Stem::Stem() : internal(0) { }
859	-
860	-Stem::Stem(const std::string &language) : internal(0) {
861	if (language.empty()) return;
862	switch (language[0]) {
863	case 'd':
864	@@ -119,9 +111,6 @@
865	internal = new InternalStemNorwegian;
866	return;
867	}
868	- if (language == "none") {
869	- return;
870	- }
871	break;
872	case 'p':
873	if (language == "pt" \|\| language == "portuguese") {
874	@@ -163,20 +152,23 @@
875	throw Xapian::InvalidArgumentError("Language code " + language + " unknown");
876	}
877
878	-Stem::~Stem() { }
879	+SnowballStem::~SnowballStem()
880	+{
881	+ delete internal;
882	+}
883
884	string
885	-Stem::operator()(const std::string &word) const
886	+SnowballStem::operator()(const std::string &word) const
887	{
888	- if (!internal.get() \|\| word.empty()) return word;
889	+ if (word.empty()) return word;
890	return internal->operator()(word);
891	}
892
893	string
894	-Stem::get_description() const
895	+SnowballStem::get_description() const
896	{
897	- string desc = "Xapian::Stem(";
898	- if (internal.get()) {
899	+ string desc = "Xapian::SnowballStem(";
900	+ if (internal) {
901	desc += internal->get_description();
902	desc += ')';
903	} else {
904	@@ -186,7 +178,7 @@
905	}
906
907	string
908	-Stem::get_available_languages()
909	+SnowballStem::get_available_languages()
910	{
911	return LANGSTRING;
912	}
913	Index: languages/compiler/generator.c
914	===================================================================
915	--- languages/compiler/generator.c (revision 10133)
916	+++ languages/compiler/generator.c (working copy)
917	@@ -1481,7 +1481,7 @@
918	if (q->type == t_routine && q->routine_called_from_among) {
919	q->among_func_count = ++among_func_count;
920	g->V[0] = q;
921	- w(g, "static int t~V0(Xapian::Stem::Internal * this_ptr) {~N"
922	+ w(g, "static int t~V0(Xapian::SnowballStem::Internal * this_ptr) {~N"
923	" return (static_cast<Xapian::~S0 *>(this_ptr))->~V0();~N"
924	"}~N"
925	"~N");

Download in other formats:

Original Format