Changeset 2041

Timestamp:

02/29/08 23:18:18 (3 months ago)

Author:

karpet

Message:

major reconstruction of config object.
basically, let go of the naive idea that the config code could be a general
purpose XML config structure and instead customized it for libswish3.
still TODO is to update the perl bindings and move code from swish_header.c
into config.c, and to write/serialize header.

Files:

• libswish3/trunk/src/Makefile.am (modified) (1 diff)
• libswish3/trunk/src/libswish3/Makefile.am (modified) (1 diff)
• libswish3/trunk/src/libswish3/analyzer.c (modified) (2 diffs)
• libswish3/trunk/src/libswish3/config.c (modified) (18 diffs)
• libswish3/trunk/src/libswish3/hash.c (modified) (1 diff)
• libswish3/trunk/src/libswish3/libswish3.h (modified) (11 diffs)
• libswish3/trunk/src/libswish3/metaname.c (added)
• libswish3/trunk/src/libswish3/mime_types.c (modified) (2 diffs)
• libswish3/trunk/src/libswish3/namedbuffer.c (modified) (4 diffs)
• libswish3/trunk/src/libswish3/parser.c (modified) (5 diffs)
• libswish3/trunk/src/libswish3/property.c (added)
• libswish3/trunk/src/libswish3/string.c (modified) (10 diffs)
• libswish3/trunk/src/swish_header.c (added)

libswish3/trunk/src/Makefile.am

@@ -9 +9 @@
-# -pg is for profiling -- don't use in production
-
-
+ swish_header
-check_PROGRAMS = swish_lint
-swish_lint_SOURCES = swish_lint.c $(myheaders)
-swish_words_SOURCES = swish_words.c $(myheaders)
-
+
+
-utf8test_SOURCES = utf8test.c $(myheaders)
-

libswish3/trunk/src/libswish3/Makefile.am

@@ -26 +26 @@
-                        swish.c \
-                        analyzer.c \
+                        property.c \
+                        metaname.c \
-                        $(myheaders) 
-

libswish3/trunk/src/libswish3/analyzer.c

@@ -33 +33 @@
-    a = swish_xmalloc(sizeof(swish_Analyzer));
-    
-is
+ese all
-    a->maxwordlen = SWISH_MAX_WORD_LEN;
-    a->minwordlen = SWISH_MIN_WORD_LEN;
-    a->lc         = 1;
-    a->ref_cnt    = 0;
+    a->tokenize   = config->flags->tokenize;
-        
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
-    /* tokenizer set in the parse* function */
-    a->tokenizer = NULL;
@@ -66 +52 @@
-    a->regex     = NULL;
-    
- 
+
-    
-    return a;

libswish3/trunk/src/libswish3/config.c

@@ -24 +24 @@
- */
-
-
-/* #include <libxml/hash.h> */
-#include <libxml/xmlstring.h>
-#include <sys/param.h>
@@ -40 +38 @@
-
-
-
-
-
-
+
+
-static xmlDocPtr parse_xml_config(xmlChar * conf);
-
-
-
-
-
-
+
-static xmlNode *get_dom_root(xmlDocPtr doc);
-
-
-
-
-
-
+
+
+
+
-
-static void
-config2(void
+string(xmlChar
-{
-    if (SWISH_DEBUG >= SWISH_DEBUG_CONFIG)
-(xmlChar *) 
-
-(xmlChar *) 
+
+
+
-}
-
-static void
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-}
-
@@ -87 +82 @@
-swish_free_config(swish_Config * config)
-{
-            
+
-
-    if (SWISH_DEBUG >= SWISH_DEBUG_CONFIG)
@@ -96 +91 @@
-    }
-
-
+
+
+
+
+
+
+
-    swish_xfree(config->flags);
-
@@ -121 +122 @@
-    swish_Config  *config;
-    swish_ConfigFlags *flags;
-
+
+
+
-
-    if (SWISH_DEBUG >= SWISH_DEBUG_CONFIG)
@@ -139 +142 @@
-    /* default metanames */
-    metas = xmlHashCreate(8);
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
-    /* default MIME types */
-swish_hash_add(c, (xmlChar *) SWISH_MIME, swish_mime_hash()
+config->mimes = swish_mime_hash(
-
-
@@ -167 +181 @@
-            swish_xstrdup((xmlChar *) SWISH_DEFAULT_PARSER_TYPE));
-
-swish_hash_add(c, (xmlChar *) SWISH_PARSERS, parsers)
+config->parsers = parsers
-
-
@@ -183 +197 @@
-            swish_xstrdup((xmlChar *) setlocale(LC_ALL, NULL)));
-
-swish_hash_add(c, (xmlChar *) SWISH_INDEX, index)
+config->index = index
-
-    /* properties hash: each property is "propertyname" => type these match tag
-     * (meta) names and are aggregated for each doc, in propHash */
-    prop = xmlHashCreate(8);
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
-
-
@@ -209 +233 @@
-            swish_xstrdup((xmlChar *) SWISH_PROP_DESCRIPTION));
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
-    config->conf = c;
-    config->ref_cnt = 0;
@@ -244 +263 @@
-    return config;
-
-}
-
-swish_ConfigValue *
-swish_init_ConfigValue()
-{
-    swish_ConfigValue *cv;
-    cv = swish_xmalloc(sizeof(swish_ConfigValue));
-    cv->ref_cnt = 1;
-    cv->multi = 0;
-    cv->equal = 0;
-    cv->key = NULL;
-    cv->value = NULL;
-    return cv;
-}
-
-void
-swish_free_ConfigValue(swish_ConfigValue * cv)
-{
-    if (cv->key != NULL)
-        swish_xfree(cv->key);
-
-    if (cv->value != NULL)
-        swish_xfree(cv->value);
-
-    swish_xfree(cv);
-}
-
-swish_ConfigValue *
-swish_keys(swish_Config * config,...)
-{
-    va_list         args;
-    swish_ConfigValue *cv = swish_init_ConfigValue();
-
-    va_start(args, config);
-
-
-    va_end(args);
-    return cv;
-}
-
-swish_ConfigValue *
-swish_value(swish_Config * config, xmlChar * key,...)
-{
-    va_list         args;
-    swish_ConfigValue *cv = swish_init_ConfigValue();
-
-    va_start(args, key);
-
-
-    va_end(args);
-    return cv;
-}
-
-
-/* a xml node is multi if: - it is a predefined top-level key (MetaNames, PropertyNames,
- * etc.) - it has the multi attribute present (value is ignored) - has one or more spaces
- * in the value */
-static int 
-is_multi(xmlNode * node)
-{
-    xmlChar        *v;
-
-    if (xmlHasProp(node, (const xmlChar *) "multi") != NULL)
-        return 1;
-
-    if (is_top_level(node))
-        return 1;
-
-    v = get_node_value(node);
-    if (v != NULL && (xmlStrchr(v, ' ') || xmlStrchr(v, '\n')))
-    {
-        xmlFree(v);
-        return 1;
-    }
-
-    return 0;
-}
-
-static int 
-is_top_level(xmlNode * node)
-{
-    if (xmlStrEqual(node->name, (xmlChar *) SWISH_META))
-        return 1;
-
-    if (xmlStrEqual(node->name, (xmlChar *) SWISH_PROP))
-        return 1;
-
-    if (xmlStrEqual(node->name, (xmlChar *) SWISH_PROP_MAX))
-        return 1;
-
-    if (xmlStrEqual(node->name, (xmlChar *) SWISH_PROP_SORT))
-        return 1;
-
-    return 0;
-}
-
-
-/* a xml node is equal if: - no attributes are present, just content - 'key' attr is
- * present and xmlStrEqual() with content - 'key' and 'value' attrs are both present and
- * xmlStrEqual() */
-static int 
-is_equal(xmlNode * node)
-{
-    xmlChar        *v, *k;
-    int             e;
-
-    if (!node_has_key(node) && !node_has_value(node))
-        return 1;
-
-    v = get_node_value(node);
-    k = xmlGetProp(node, (const xmlChar *) "key");
-    e = xmlStrEqual(k, v);
-    xmlFree(v);
-    xmlFree(k);
-
-    if (e)
-        return 1;
-
-
-    if (node_has_value(node) && !node_has_key(node))
-    {
-        SWISH_CROAK("config node with value but no key: %s", node->name);
-        return 0;
-    }
-
-    return 0;
-}
-
-static int 
-node_has_key(xmlNode * node)
-{
-    return node_has_attr(node, (const xmlChar *) "key");
-}
-
-static int 
-node_has_value(xmlNode * node)
-{
-    return node_has_attr(node, (const xmlChar *) "value");
-}
-
-static int 
-node_has_type(xmlNode * node)
-{
-    return node_has_attr(node, (const xmlChar *) "type");
-}
-
-static int 
-node_has_attr(xmlNode * node, const xmlChar * att)
-{
-    if (xmlHasProp(node, att))
-        return 1;
-    else
-        return 0;
-}
-
-
-
-/* key is attr value if present, otherwise get_node_value() */
-static xmlChar *
-get_node_key(xmlNode * node)
-{
-    xmlChar        *k;
-    k = xmlGetProp(node, (const xmlChar *) "key");
-    return k;        /* MUST be freed */
-}
-
-static xmlChar *
-get_node_type(xmlNode * node)
-{
-    xmlChar        *k;
-    k = xmlGetProp(node, (const xmlChar *) "type");
-    return k;        /* MUST be freed */
-}
-
-/* get node content and remove and leading or trailing spaces */
-static xmlChar *
-get_node_value(xmlNode * node)
-{
-    xmlChar        *str;
-    str = xmlNodeGetContent(node);
-    if (str == NULL)
-    {
-        str = xmlGetProp(node, (const xmlChar *) "value");
-        if (str == NULL)
-        {
-            SWISH_WARN("no value for config opt '%s'", node->name);
-            return NULL;
-        }
-    }
-
-
-    /* TODO strip spaces */
-
-    return str;        /* MUST be freed */
-}
-
@@ -473 +298 @@
-}
-
-static void 
-add_multi_node_to_cv(xmlNode * node, swish_ConfigValue * cv)
-{
-    xmlChar        *key, *tmp, *type;
-    swish_StringList *value;
-    int             i;
-
-    /* if node has an explicit key, use that with entire value else split up value
-     * into tokens on whitespace. if node has a 'type' attr, use that as hash value,
-     * else use str as both key/value. */
-
-    if (node_has_key(node))
-    {
-        key = get_node_key(node);
-        tmp = get_node_value(node);
-        if (xmlHashLookup(cv->value, key))
-            swish_hash_replace(cv->value, key, tmp);
-        else
-            swish_hash_add(cv->value, key, tmp);
-
-    }
-    else
-    {
-        tmp = get_node_value(node);
-        value = swish_make_StringList(tmp);
-        xmlFree(tmp);
-
-        if (node_has_type(node))
-            type = get_node_type(node);
-
-
-        for (i = 0; i < value->n; i++)
-        {
-            key = swish_xstrdup(value->word[i]);
-            if (type == NULL)
-                tmp = swish_xstrdup(key);
-            else
-                tmp = type;
-
-            if (xmlHashLookup(cv->value, key))
-                swish_hash_replace(cv->value, key, tmp);
-            else
-                swish_hash_add(cv->value, key, tmp);
-        }
-    }
-
-}
-
-static xmlNode *
@@ -544 +322 @@
-}
-
-
-/* this is too complicated right now. might be easier to just plunge in and define
- * different config opt types (MetaNames, etc.) rather than try and remain agnostic and
- * extensible about attribute names, etc. */
-swish_Config  *
-swish_parse_config_new(xmlChar * conf, swish_Config * config)
-{
-
-    xmlNode        *node, *root;
-    xmlDocPtr       doc;
-    swish_ConfigValue *cv;
-
-    doc = parse_xml_config(conf);
-    root = get_dom_root(doc);
-
-    /* -------------------------------------------------------------------------- get
-     * options/values format is:
-     * 
-     * <directive type="someval">some arg</directive>
-     * 
-     * where type="someval" attr is optional -- defaults to whatever 'some arg' is
-     * -------------------------------------------------------------------------- */
-
-    for (node = root->children; node != NULL; node = node->next)
-    {
-        if (node->type == XML_ELEMENT_NODE)
-        {
-
-            /* is this opt already in our config? if yes, and multi, append
-             * it if yes, and !multi, replace it if no, create it */
-
-            if (xmlHashLookup(config->conf, node->name))
-            {
-                cv = xmlHashLookup(config->conf, node->name);
-
-                if (cv->multi)    /* already flagged as multi */
-                {
-                    SWISH_DEBUG_MSG("%s is an existing multi-config", node->name);
-
-                    add_multi_node_to_cv(node, cv);
-
-                }
-                else
-                {
-                    SWISH_DEBUG_MSG("%s exists but is not a multi-config", node->name);
-
-                    /* free the existing one and replace it with new
-                     * one */
-                    swish_free_ConfigValue(cv);
-
-                    cv = swish_init_ConfigValue();
-                    cv->key = get_node_name(node);
-                    cv->value = get_node_value(node);
-
-                }
-            }
-            else
-            {
-                /* TODO - split all these if/else into separate functions */
-                cv = swish_init_ConfigValue();
-
-                if (is_multi(node))
-                {
-                    SWISH_DEBUG_MSG("%s is a new multi-config", node->name);
-                    cv->value = swish_new_hash(16);
-                    add_multi_node_to_cv(node, cv);
-
-                }
-
-                if (is_equal(node))
-                {
-                    SWISH_DEBUG_MSG("%s is an equal node", node->name);
-                    cv->equal = 1;
-                }
-
-
-            }
-
-        }
-
-    }
-
-    xmlFreeDoc(doc);
-    xmlCleanupParser();
-    return config;
-}
-
-swish_Config  *
@@ -722 +414 @@
-
-
-StringL
+stringl
-
-            for (i = 0; i < arg_list->n; i++)
@@ -750 +442 @@
-                        SWISH_DEBUG_MSG("tolower str: >%s<", tmp_arg);
-
-
+ // TODO mem leak !! ??
-
-                }
@@ -768 +460 @@
-            }
-
-StringL
+stringl
-
-            /* don't use our swish_xfree since it throws off memcount */
@@ -802 +494 @@
-
-static void
-val_printer(xmlChar * val, xmlChar * str, xmlChar * 
-
-  %s => %s"
+printer(xmlChar *val, xmlChar *str, xmlChar *
+
+%s:  %s => %s", str
-}
-
-static void
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+
-}
-
@@ -821 +517 @@
-swish_debug_config(swish_Config * config)
-{
-            
+
-
-    SWISH_DEBUG_MSG("config->ref_cnt = %d", config->ref_cnt);
@@ -828 +524 @@
-    SWISH_DEBUG_MSG("ptr addr: 0x%x  %d", (int) config->conf, (int) config->conf);
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
+
+
+
+

libswish3/trunk/src/libswish3/hash.c

@@ -69 +69 @@
-}
-
+boolean
+swish_hash_exists( xmlHashTablePtr hash, xmlChar *key )
+{
+    return xmlHashLookup(hash, key) ? 1 : 0;
+}
+
-xmlHashTablePtr swish_new_hash(int size)
-{

libswish3/trunk/src/libswish3/libswish3.h

@@ -70 +70 @@
-#define SWISH_INDEX_LOCALE          "Locale"
-#define SWISH_DEFAULT_VALUE         "1"
-PARSE_WORDS
+TOKENIZE   
-
-/* tags */
@@ -146 +146 @@
-#endif
-
+typedef char   boolean;
-typedef struct swish_3                  swish_3;
-typedef struct swish_Token              swish_Token;
@@ -203 +204 @@
-struct swish_Config
-{
-    /* for bindings */
+
-    void                        *stash;      /* for bindings */
-
+
+
+
+
+
+
+
-    struct swish_ConfigFlags    *flags;      /* shortcuts for parsing */
-};
@@ -211 +218 @@
-struct swish_ConfigFlags
-{
+    boolean         tokenize;
-    
-
-
-
-
-
-
-
-
-
-
-
-
+
-
-struct swish_NamedBuffer
@@ -248 +245 @@
-struct swish_MetaName
-{
+    int                 ref_cnt;
-    unsigned int        id;
-    xmlChar            *name;
-    int                 bias;
+    xmlChar            *alias_for;
-};
-
-struct swish_Property
-{
+    int                 ref_cnt;
-    unsigned int        id;
-    xmlChar            *name;
+    boolean             ignore_case;
+    boolean             type;
+    boolean             verbatim;
+    xmlChar            *alias_for;
+    unsigned int        max;
+    boolean             sort;
-};
-
@@ -316 +322 @@
-};
-
-// TODO maybe store swish_Parser * here instead of separate config and analyzer
-struct swish_ParserData
-{
@@ -368 +373 @@
-int         swish_hash_replace( xmlHashTablePtr hash, xmlChar *key, void *value );
-int         swish_hash_delete( xmlHashTablePtr hash, xmlChar *key );
+boolean     swish_hash_exists( xmlHashTablePtr hash, xmlChar *key );
-xmlHashTablePtr swish_new_hash(int size);
-/*
@@ -421 +427 @@
-xmlChar *           swish_str_skip_ws(xmlChar *s);
-void                swish_str_trim_ws(xmlChar *string);
+int                 swish_str_all_ws(xmlChar * s);
-void                swish_debug_wchars( const wchar_t * widechars );
-int                 swish_wchar_t_comp(const void *s1, const void *s2);
-int                 swish_sort_wchar(wchar_t *s);
-StringL
-StringL
-StringL
+stringl
+stringl
+stringl
-/*
-=cut
@@ -438 +445 @@
-swish_Config *  swish_add_config( xmlChar * conf, swish_Config * config );
-swish_Config *  swish_parse_config( xmlChar * conf, swish_Config * config );
-swish_Config *  swish_parse_config_new( xmlChar * conf, swish_Config * config );
-int             swish_debug_config( swish_Config * config );
-xmlHashTablePtr swish_subconfig_hash( swish_Config * config, xmlChar *key);
-int             swish_config_value_exists( swish_Config * config, xmlChar *key, xmlChar *value );
-xmlChar *       swish_get_config_value(swish_Config * config, xmlChar * key, xmlChar * value);
-void            swish_free_config(swish_Config * config);
-swish_ConfigValue * swish_keys( swish_Config * config, ... );
-swish_ConfigValue * swish_value( swish_Config * config, xmlChar * key, ... );
-swish_ConfigValue * swish_init_ConfigValue();
-void                swish_free_ConfigValue( swish_ConfigValue * cv );
-xmlHashTablePtr swish_mime_hash();
-xmlChar *       swish_get_mime_type( swish_Config * config, xmlChar * fileext );
@@ -553 +552 @@
-=head2 Buffer Functions
-*/
-swish_Config * config, xmlChar * configKey
+xmlHashTablePtr confhash
-void                swish_free_nb( swish_NamedBuffer * nb );
-void                swish_debug_nb( swish_NamedBuffer * nb, xmlChar * label );
@@ -574 +573 @@
-*/
-
+/*
+=head2 Property Functions
+*/
+swish_Property *    swish_init_property( xmlChar *name );
+void                swish_free_property( swish_Property *prop );
+void                swish_debug_property( swish_Property *prop );
+/*
+=cut
+*/
+
+/*
+=head2 MetaName Functions
+*/
+swish_MetaName *    swish_init_metaname( xmlChar *name);
+void                swish_free_metaname( swish_MetaName *m );
+void                swish_debug_metaname( swish_MetaName *m );
+/*
+=cut
+*/
+
-
-#ifdef __cplusplus

libswish3/trunk/src/libswish3/mime_types.c

@@ -220 +220 @@
-{
-    xmlChar * mime;
-
-
+
-    if ( mime == NULL )
-    {
@@ -236 +235 @@
-    xmlChar *parser;
-    xmlChar *deftype;
-
-
-
-
+
+
-    
-    if (SWISH_DEBUG > 9)
-        SWISH_DEBUG_MSG("using parser '%s' based on MIME '%s'", parser, mime );
-    
-
+config->
-        
-    if ( parser == NULL )

libswish3/trunk/src/libswish3/namedbuffer.c

@@ -35 +35 @@
-
-static void     free_name_from_hash(void *buffer, xmlChar * name);
-xmlChar * val, xmlHashTablePtr hash, xmlChar * 
- label, xmlChar * 
+void *ignored, xmlHashTablePtr nbhash, xmlChar *
+label, xmlChar *
-
-static void 
-xmlChar * val, xmlHashTablePtr 
+void *ignored, xmlHashTablePtr nb
-{    
-    /* make sure we don't already have it */
-xmlHashLookup(
+swish_hash_exists(nb
-    {
-        SWISH_WARN("%s is already in NamedBuffer hash -- ignoring", name);
@@ -52 +52 @@
-        SWISH_DEBUG_MSG("  adding %s to NamedBuffer\n", name);
-    
-
+nb
-}
-
@@ -65 +65 @@
-
-swish_NamedBuffer *
-swish_Config * config, xmlChar * confKey
+ xmlHashTablePtr confhash
-{    
-    swish_NamedBuffer * nb = swish_xmalloc(sizeof(swish_NamedBuffer));
@@ -72 +72 @@
-    nb->hash    = xmlHashCreate(8);    /* will grow as needed */
-    
-
-
-
-
+
+
-
-    return nb;

libswish3/trunk/src/libswish3/parser.c

@@ -293 +293 @@
-
-    /* change our internal name for this tag if it is aliased in config */
-swish_get_config_value(parser_data->s3->config, (xmlChar*)SWISH_ALIAS
+xmlHashLookup(parser_data->s3->config->tag_aliases
-    if (alias)
-    {
@@ -447 +447 @@
-
-    /* set property if this tag is configured for it */
-config_value_exists(parser_data->s3->config, (xmlChar*)SWISH_PROP
+hash_exists(parser_data->s3->config->properties
-    {
-        if (SWISH_DEBUG >= SWISH_DEBUG_PARSER)
-            SWISH_DEBUG_MSG(" %s = new property", parser_data->tag);
-
-
+ //TODO why NULL here ??
-        xmlBufferEmpty(parser_data->prop_buf);
-
-        parser_data->propstack = push_tag_stack(parser_data->propstack, parser_data->tag);
-
-
+
+
-    }
-
-    /* likewise for metastack */
-config_value_exists(parser_data->s3->config, (xmlChar*)SWISH_META
+hash_exists(parser_data->s3->config->metanames
-    {
-        if (SWISH_DEBUG >= SWISH_DEBUG_PARSER)
@@ -750 +751 @@
-    ptr->wordlist   = swish_init_wordlist();
-    ptr->wordlist->ref_cnt++;
-, (xmlChar*)SWISH_PROP
+->properties
-    ptr->properties->ref_cnt++;
-, (xmlChar*)SWISH_META
+->metanames
-    ptr->metanames->ref_cnt++;
-