Navigation

← Previous Change
Next Change →

Changeset 2009 for libswish3/trunk/src

Timestamp:

02/03/08 23:29:35 (10 months ago)

Author:

karpet

Message:

rename some vars for clarity

Files:

libswish3/trunk/src/libswish3/libswish3.h (modified) (7 diffs)
libswish3/trunk/src/libswish3/parser.c (modified) (11 diffs)

Legend:

: Unmodified
: Added
: Removed
: Modified
: Copied
: Moved

libswish3/trunk/src/libswish3/libswish3.h

r1955	r2009
50	50
51	51	/* default config hash key names */
52		#define SWISH_INCLUDE_FILE "IncludeConfigFile"
53		#define SWISH_PROP "PropertyNames"
54		#define SWISH_PROP_ASIS "nostripchars"
55		#define SWISH_PROP_MAX "PropertyNamesMaxLength"
56		#define SWISH_PROP_SORT "PropertyNamesSortKeyLength"
57		#define SWISH_META "MetaNames"
58		#define SWISH_MIME "MIME"
59		#define SWISH_PARSERS "Parsers"
60		#define SWISH_INDEX "Index"
61		#define SWISH_ALIAS "TagAlias"
62		#define SWISH_WORDS "Words"
63		#define SWISH_DEFAULT_PARSER "default"
64		#define SWISH_PARSER_TXT "TXT"
65		#define SWISH_PARSER_XML "XML"
66		#define SWISH_PARSER_HTML "HTML"
67		#define SWISH_DEFAULT_PARSER_TYPE "HTML"
68		#define SWISH_INDEX_FORMAT "Format"
69		#define SWISH_INDEX_NAME "Name"
70		#define SWISH_INDEX_LOCALE "Locale"
71		#define SWISH_DEFAULT_VALUE "1"
72		#define SWISH_PARSE_WORDS "Tokenize"
	52	#define SWISH_INCLUDE_FILE "IncludeConfigFile"
	53	#define SWISH_PROP "PropertyNames"
	54	#define SWISH_PROP_ASIS "nostripchars"
	55	#define SWISH_PROP_MAX "PropertyNamesMaxLength"
	56	#define SWISH_PROP_SORT "PropertyNamesSortKeyLength"
	57	#define SWISH_META "MetaNames"
	58	#define SWISH_MIME "MIME"
	59	#define SWISH_PARSERS "Parsers"
	60	#define SWISH_INDEX "Index"
	61	#define SWISH_ALIAS "TagAlias"
	62	#define SWISH_WORDS "Words"
	63	#define SWISH_DEFAULT_PARSER "default"
	64	#define SWISH_PARSER_TXT "TXT"
	65	#define SWISH_PARSER_XML "XML"
	66	#define SWISH_PARSER_HTML "HTML"
	67	#define SWISH_DEFAULT_PARSER_TYPE "HTML"
	68	#define SWISH_INDEX_FORMAT "Format"
	69	#define SWISH_INDEX_NAME "Name"
	70	#define SWISH_INDEX_LOCALE "Locale"
	71	#define SWISH_DEFAULT_VALUE "1"
	72	#define SWISH_PARSE_WORDS "Tokenize"
73	73
74	74	/* tags */
…	…
154	154	typedef struct swish_MetaStackElement *swish_MetaStackElementPtr;
155	155	typedef struct swish_MetaStack swish_MetaStack;
	156	typedef struct swish_MetaName swish_MetaName;
	157	typedef struct swish_Property swish_Property;
156	158	typedef struct swish_Word swish_Word;
157	159	typedef struct swish_WordList swish_WordList;
…	…
176	178	struct swish_Config
177	179	{
178		int ref_cnt; /* for ~~scripting language~~s */
179		void stash; / for ~~scripting language~~s */
	180	int ref_cnt; /* for bindings */
	181	void stash; / for bindings */
180	182	xmlHashTablePtr conf; /* the meat */
181	183	struct swish_ConfigFlags flags; / shortcuts for parsing */
…	…
200	202	struct swish_NamedBuffer
201	203	{
202		int ref_cnt; /* for ~~scripting language~~s */
203		void stash; / for ~~scripting language~~s */
	204	int ref_cnt; /* for bindings */
	205	void stash; / for bindings */
204	206	xmlHashTablePtr hash; /* the meat */
205	207	};
…	…
218	220	};
219	221
	222	struct swish_MetaName
	223	{
	224	unsigned int id;
	225	xmlChar *name;
	226	int bias;
	227	};
	228
	229	struct swish_Property
	230	{
	231	unsigned int id;
	232	xmlChar *name;
	233	};
220	234
221	235	struct swish_Word
222	236	{
223		unsigned int position; // word position in doc
224		xmlChar *metaname; // immediate metaname
225		xmlChar *context; // metaname ancestry
226		xmlChar *word; // the word itself (NOTE stored as multibyte not wchar)
227		unsigned int start_offset; // start byte
228		unsigned int end_offset; // end byte
	237	unsigned int position; // word position in doc
	238	xmlChar *metaname; // immediate metaname
	239	xmlChar *context; // metaname ancestry
	240	xmlChar *word; // the word itself (NOTE stored as multibyte not wchar)
	241	unsigned int start_offset; // start byte
	242	unsigned int end_offset; // end byte
229	243	struct swish_Word *next; // pointer to next swish_Word
230	244	struct swish_Word *prev; // pointer to prev swish_Word
…	…
233	247	struct swish_WordList
234	248	{
235		swish_Word *head;
236		swish_Word *tail;
237		swish_Word *current; // for iterating
238		unsigned int nwords;
239		unsigned int ~~ref_cnt; // for scripting language~~s
	249	swish_Word *head;
	250	swish_Word *tail;
	251	swish_Word *current; // for iterating
	252	unsigned int nwords;
	253	unsigned int ref_cnt; // for bindings
240	254	};
241	255
…	…
281	295	struct swish_ParseData
282	296	{
283		xmlBufferPtr ~~buf_ptr; // tmp text (MetaName)~~ buffer
	297	xmlBufferPtr meta_buf; // tmp MetaName buffer
284	298	xmlBufferPtr prop_buf; // tmp Property buffer
285	299	xmlChar *tag; // current tag name

libswish3/trunk/src/libswish3/parser.c

r1952	r2009
316	316	if (SWISH_DEBUG == SWISH_DEBUG_PARSER)
317	317	SWISH_DEBUG_MSG("buffer is >>%s<< before flush, word_pos = %d",
318		xmlBufferContent(parse_data->~~buf_ptr~~), parse_data->word_pos);
	318	xmlBufferContent(parse_data->meta_buf), parse_data->word_pos);
319	319
320	320	/* since we only flush the buffer when metaname changes, and
…	…
325	325	parse_data->word_pos++;
326	326
327		/* add ~~buf_ptr~~ as-is to metanames buffer under current tag.
	327	/* add meta_buf as-is to metanames buffer under current tag.
328	328	this gives us both tokens and raw text de-tagged but organized by metaname.
329	329	*/
330	330	swish_add_buf_to_nb( parse_data->metanames,
331	331	metaname,
332		parse_data->buf_ptr, (xmlChar*)SWISH_META_CONNECTOR, 0, 1);
	332	parse_data->meta_buf,
	333	(xmlChar*)SWISH_META_CONNECTOR,
	334	0,
	335	1);
333	336
334	337	if (parse_data->context_as_meta)
…	…
341	344	swish_add_buf_to_nb(parse_data->metanames,
342	345	s->temp->name,
343		parse_data->buf_ptr, (xmlChar*)SWISH_META_CONNECTOR, 0, 1);
	346	parse_data->meta_buf,
	347	(xmlChar*)SWISH_META_CONNECTOR,
	348	0,
	349	1);
344	350	}
345	351	}
…	…
349	355
350	356	tokenize( parse_data,
351		(xmlChar *)xmlBufferContent(parse_data->~~buf_ptr~~),
352		xmlBufferLength(parse_data->~~buf_ptr~~),
	357	(xmlChar *)xmlBufferContent(parse_data->meta_buf),
	358	xmlBufferLength(parse_data->meta_buf),
353	359	metaname,
354	360	context
…	…
356	362	}
357	363
358		xmlBufferEmpty(parse_data->~~buf_ptr~~);
	364	xmlBufferEmpty(parse_data->meta_buf);
359	365
360	366	}
…	…
517	523	int i;
518	524	xmlChar output[len];
519		xmlBufferPtr buf = parse_data->~~buf_ptr~~;
	525	xmlBufferPtr buf = parse_data->meta_buf;
520	526	/*
521	527	* why not wchar_t ? len is number of bytes, not number of
…	…
740	746	ptr->stash = stash;
741	747
742		ptr->~~buf_ptr~~ = xmlBufferCreateSize(SWISH_BUFFER_CHUNK_SIZE);
	748	ptr->meta_buf = xmlBufferCreateSize(SWISH_BUFFER_CHUNK_SIZE);
743	749	ptr->prop_buf = xmlBufferCreateSize(SWISH_BUFFER_CHUNK_SIZE);
744	750
…	…
849	855	SWISH_DEBUG_MSG("freeing swish_ParseData xmlBuffer");
850	856
851		xmlBufferFree( ptr->~~buf_ptr~~ );
	857	xmlBufferFree( ptr->meta_buf );
852	858
853	859
…	…
1271	1277	swish_debug_docinfo(parse_data->docinfo);
1272	1278	SWISH_DEBUG_MSG(" word buffer length: %d bytes",
1273		xmlBufferLength(parse_data->~~buf_ptr~~));
	1279	xmlBufferLength(parse_data->meta_buf));
1274	1280	SWISH_DEBUG_MSG(" (%d words)", parse_data->docinfo->nwords);
1275	1281	}
…	…
1402	1408	swish_debug_docinfo(parse_data->docinfo);
1403	1409	SWISH_DEBUG_MSG(" word buffer length: %d bytes",
1404		xmlBufferLength(parse_data->~~buf_ptr~~));
	1410	xmlBufferLength(parse_data->meta_buf));
1405	1411	SWISH_DEBUG_MSG(" (%d words)", parse_data->docinfo->nwords);
1406	1412	}
…	…
1454	1460	swish_debug_docinfo(parse_data->docinfo);
1455	1461	SWISH_DEBUG_MSG(" word buffer length: %d bytes",
1456		xmlBufferLength(parse_data->~~buf_ptr~~));
	1462	xmlBufferLength(parse_data->meta_buf));
1457	1463	SWISH_DEBUG_MSG(" (%d words)", parse_data->docinfo->nwords);
1458	1464	}

Navigation

Changeset 2009 for libswish3/trunk/src

Legend:

libswish3/trunk/src/libswish3/libswish3.h

libswish3/trunk/src/libswish3/parser.c

Download in other formats: