Changeset 2110

Timestamp:

04/03/08 22:44:09 (1 month ago)

Author:

karpet

Message:

Refactor duplicate id checks to use hash instead of array. Fixes bug with merging configs. Default ->id is now -1 (invalid value)

Files:

• libswish3/trunk/src/libswish3/config.c (modified) (13 diffs)
• libswish3/trunk/src/libswish3/header.c (modified) (14 diffs)
• libswish3/trunk/src/libswish3/libswish3.h (modified) (6 diffs)
• libswish3/trunk/src/libswish3/metaname.c (modified) (1 diff)
• libswish3/trunk/src/libswish3/property.c (modified) (1 diff)
• libswish3/trunk/src/libswish3/string.c (modified) (6 diffs)

libswish3/trunk/src/libswish3/config.c

@@ -138 +138 @@
-    }
-
- 
- 
- 
- 
- 
- 
- 
-xfree
+
+
+
+
+
+
+
+free_config_flags
-
-    if (config->ref_cnt != 0) {
@@ -162 +162 @@
-)
-{
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
-    return flags;
-}
-
-
+
+
+
+
+
+
+
+
+
+
+
+
-
-/* init config object */
@@ -204 +207 @@
-    config->ref_cnt = 0;
-    config->stash = NULL;
-
+
-    if (SWISH_DEBUG & SWISH_DEBUG_MEMORY) {
-        SWISH_DEBUG_MSG("config ptr 0x%x", (int)config);
@@ -220 +223 @@
-    swish_Property *tmpprop;
-    swish_MetaName *tmpmeta;
+    xmlChar *tmpbuf;
-
-    if (SWISH_DEBUG & SWISH_DEBUG_CONFIG)
@@ -230 +234 @@
-
-/* metanames */
-
-
-
-
-
-
-
-
-
-
-
-
-
-
+
+
-    tmpmeta->ref_cnt++;
-    tmpmeta->id = SWISH_META_DEFAULT_ID;
-
-
+
+
+
+
+
+
+
-    tmpmeta->ref_cnt++;
-    tmpmeta->id = SWISH_META_TITLE_ID;
+    tmpbuf = swish_int_to_string(SWISH_META_TITLE_ID);
+    swish_hash_add(config->flags->meta_ids, tmpbuf, tmpmeta);
+    swish_hash_add(config->metanames, (xmlChar *)SWISH_TITLE_METANAME, tmpmeta);
+    swish_xfree(tmpbuf);
+
+/* properties */
+    // description
+    tmpprop = swish_init_property(swish_xstrdup((xmlChar *)SWISH_PROP_DESCRIPTION));
+    tmpprop->ref_cnt++;
+    tmpprop->id = SWISH_PROP_DESCRIPTION_ID;
+    swish_hash_add(config->properties, (xmlChar *)SWISH_PROP_DESCRIPTION, tmpprop);
+    tmpbuf = swish_int_to_string(SWISH_PROP_DESCRIPTION_ID);
+    swish_hash_add(config->flags->prop_ids, tmpbuf, tmpprop);
+    swish_xfree(tmpbuf);
+
+    // title
+    tmpprop = swish_init_property(swish_xstrdup((xmlChar *)SWISH_PROP_TITLE));
+    tmpprop->ref_cnt++;
+    tmpprop->id = SWISH_PROP_TITLE_ID;
+    swish_hash_add(config->properties, (xmlChar *)SWISH_PROP_TITLE, tmpprop);
+    tmpbuf = swish_int_to_string(SWISH_PROP_TITLE_ID);
+    swish_hash_add(config->flags->prop_ids, tmpbuf, tmpprop);
+    swish_xfree(tmpbuf);
-
-/* parsers */
@@ -269 +289 @@
-                   swish_xstrdup((xmlChar *)setlocale(LC_ALL, "")));
-
-/* properties */
-    swish_hash_add(config->properties, (xmlChar *)SWISH_PROP_DESCRIPTION,
-                   swish_init_property(swish_xstrdup
-                                       ((xmlChar *)SWISH_PROP_DESCRIPTION))
-        );
-    swish_hash_add(config->properties, (xmlChar *)SWISH_PROP_TITLE,
-                   swish_init_property(swish_xstrdup
-                                       ((xmlChar *)SWISH_PROP_TITLE))
-        );
-
-/* same deal as metanames above */
-    tmpprop =
-        swish_hash_fetch(config->properties, (xmlChar *)SWISH_PROP_DESCRIPTION);
-    tmpprop->ref_cnt++;
-    tmpprop->id = SWISH_PROP_DESCRIPTION_ID;
-    tmpprop = swish_hash_fetch(config->properties, (xmlChar *)SWISH_PROP_TITLE);
-    tmpprop->ref_cnt++;
-    tmpprop->id = SWISH_PROP_TITLE_ID;
-
-/* aliases: other names a tag might be known as, for matching properties and
-     * metanames */
@@ -370 +371 @@
-    SWISH_DEBUG_MSG("ptr addr: 0x%x  %d", (int)config, (int)config);
-
-
-
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
-}
-
@@ -390 +388 @@
-{
-    swish_Property *prop1;
-    boolean in_hash;
-
-    if (swish_hash_exists(props1, prop2name)) {
-        prop1 = swish_hash_fetch(props1, prop2name);
-
+
+
+
+
-    }
-    else {
-        prop1 = swish_init_property(swish_xstrdup(prop2name));
-        prop1->ref_cnt++;
-
-
-
-
-
-
-
-
+
+
+
+
+
+
+
-    prop1->ignore_case = prop2->ignore_case;
-    prop1->type = prop2->type;
@@ -419 +418 @@
-    prop1->sort = prop2->sort;
-
-    if (!in_hash) {
-        swish_hash_add(props1, prop1->name, prop1);
-    }
-
-}
-
@@ -431 +426 @@
-)
-{
- 
+
-}
-
@@ -442 +437 @@
-{
-    swish_MetaName *meta1;
-
-
+
-    if (swish_hash_exists(metas1, meta2name)) {
-        meta1 = swish_hash_fetch(metas1, meta2name);
-
+
+
+
+
-    }
-    else {
-        meta1 = swish_init_metaname(swish_xstrdup(meta2name));
-        meta1->ref_cnt++;
-
-
-
+
+
+
+
+
+
+
-    meta1->id = meta2->id;
-    if (in_hash && meta1->name != NULL) {
-        swish_xfree(meta1->name);
-        meta1->name = swish_xstrdup(meta2->name);
-    }
-    meta1->bias = meta2->bias;
-    if (meta1->alias_for != NULL) {
@@ -467 +464 @@
-    }
-
-    if (!in_hash) {
-        swish_hash_add(metas1, meta1->name, meta1);
-    }
-}
-
@@ -478 +472 @@
-)
-{
- 
+
-}
-

libswish3/trunk/src/libswish3/header.c

@@ -108 +108 @@
-    xmlChar *name
-);
-static void test_meta_unique_ids(
-    swish_MetaName *meta,
-    swish_Config *c,
-    xmlChar *name
-);
-static void test_prop_unique_ids(
-    swish_Property *prop,
-    swish_Config *c,
-    xmlChar *name
-);
-static headmaker *init_headmaker(
-);
@@ -212 +202 @@
-                newmeta = swish_hash_fetch(h->config->metanames, newname);
-            }
+/* else new metaname */
-            else {
-                newmeta = swish_init_metaname(newname);
@@ -243 +234 @@
-)
-{
+    swish_MetaName *dupe;
-    if (xmlStrEqual(attr, (xmlChar *)"bias")) {
-(boolean) strtol((char *)attr_val, (char **)NULL, 10
+swish_string_to_int((char*)attr_val
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"id")) {
-
+
+
+
+
+
+
+
+
+
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"alias_for")) {
@@ -292 +292 @@
-
-/*  must have an id */
-!meta->id
+meta->id == -1
-        meta->id = h->meta_id++;
-    }
@@ -364 +364 @@
-)
-{
-
+
+
-    if (xmlStrEqual(attr, (xmlChar *)"ignore_case")) {
- 
+
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"max")) {
@@ -372 +373 @@
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"verbatim")) {
- 
+
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"sort")) {
- 
+
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"id")) {
-
+
+
+
+
+
+
+
+
+
-    }
-    else if (xmlStrEqual(attr, (xmlChar *)"type")) {
@@ -434 +443 @@
-    }
-
-!prop->id
+prop->id == -1
-        prop->id = h->prop_id++;
-    }
@@ -754 +763 @@
-}
-
-static void
-test_meta_unique_ids(
-    swish_MetaName *meta,
-    swish_Config *c,
-    xmlChar *name
-)
-{
-    c->flags->meta_ids[meta->id]++;
-}
-
-static void
-test_prop_unique_ids(
-    swish_Property *prop,
-    swish_Config *c,
-    xmlChar *name
-)
-{
-    c->flags->prop_ids[prop->id]++;
-}
-
-void
-swish_config_test_unique_ids(
-    swish_Config *c
-)
-{
-    int i;
-    xmlHashScan(c->metanames, (xmlHashScanner)test_meta_unique_ids, c);
-    xmlHashScan(c->properties, (xmlHashScanner)test_prop_unique_ids, c);
-    for (i = 0; i < SWISH_MAX_IDS; i++) {
-        if (c->flags->meta_ids[i] > 1) {
-            SWISH_WARN("meta id %d == %d", i, c->flags->meta_ids[i]);
-        }
-        if (c->flags->prop_ids[i] > 1) {
-            SWISH_WARN("prop id %d == %d", i, c->flags->prop_ids[i]);
-        }
-
-        /*
-           set back to 0 in case we are called again 
-         */
-        c->flags->prop_ids[i] = 0;
-        c->flags->meta_ids[i] = 0;
-    }
-}
-
-static headmaker *
@@ -831 +797 @@
-    swish_config_test_alias_fors(h->config);
-
-/*  make sure ids are all unique */
-    swish_config_test_unique_ids(h->config);
-
-    swish_debug_config(h->config);
-    swish_free_config(h->config);
@@ -861 +824 @@
-/*  test that all the alias_for links resolve ok */
-    swish_config_test_alias_fors(c);
-
-/*  make sure ids are all unique */
-    swish_config_test_unique_ids(c);
-
-    return 1;
@@ -930 +890 @@
-{
-    int rc;
-    boolean is_alias;
-    write_open_tag(writer, name);
-
-
+
+
+
+
+
-    if (meta->alias_for != NULL) {
-        rc = xmlTextWriterWriteAttribute(writer, BAD_CAST "alias_for", meta->alias_for);
-
-
-
-
-
-
-
+
+
+
+
+
+
-        rc = xmlTextWriterWriteFormatAttribute(writer, BAD_CAST "bias", "%d", meta->bias);
-        if (rc < 0) {
-            SWISH_CROAK("Error writing metaname bias attribute for %s", name);
-        }
-
-
-
-
-
-
-
+
+
-    write_close_tag(writer);
-}
@@ -984 +940 @@
-{
-    int rc;
-    boolean is_alias;
-    write_open_tag(writer, name);
-    rc = 0;
-    is_alias = 0;
-    if (prop->alias_for != NULL) {
-        rc = xmlTextWriterWriteAttribute(writer, BAD_CAST "alias_for", prop->alias_for);
-        is_alias = 1;
-    }
-    if (rc < 0) {
-        SWISH_CROAK("Error writing property alias_for attribute for %s", name);
-    }
-    rc = xmlTextWriterWriteFormatAttribute(writer, BAD_CAST "id", "%d", prop->id);
-    if (rc < 0) {
@@ -1000 +946 @@
-    }
-
+    if (prop->alias_for != NULL) {
+        rc = xmlTextWriterWriteAttribute(writer, BAD_CAST "alias_for", prop->alias_for);
+        if (rc < 0) {
+            SWISH_CROAK("Error writing property alias_for attribute for %s", name);
+        }
+    }
+    else {
+
-/* all other attrs are irrelevant if this is an alias */
-    if (!is_alias) {
-        rc = xmlTextWriterWriteFormatAttribute(writer, BAD_CAST "ignore_case", "%d",
-                                               prop->ignore_case);
@@ -1087 +1040 @@
-    things->thing3 = writer;
-    xmlHashScan(mimes, (xmlHashScanner)write_mime, things);
- things->thing1 
+things->thing1
-    swish_xfree(things);
-}

libswish3/trunk/src/libswish3/libswish3.h

@@ -41 +41 @@
-
-#define SWISH_CONTRACTIONS          1
-
-#define SWISH_MAX_IDS               1024
-#define SWISH_SPECIAL_ARG           1
-#define SWISH_MAX_SORT_STRING_LEN   100
@@ -246 +244 @@
-{
-    boolean         tokenize;
-
-
-
+
+
-};
-
@@ -275 +272 @@
-{
-    int                 ref_cnt;
-unsigned int
+int         
-    xmlChar            *name;
-    int                 bias;
@@ -284 +281 @@
-{
-    int                 ref_cnt;
-unsigned int
+int         
-    xmlChar            *name;
-    boolean             ignore_case;
@@ -475 +472 @@
-swish_StringList *  swish_init_stringlist();
-void                swish_free_stringlist(swish_StringList * sl);
+int                 swish_string_to_int( char *buf );
+xmlChar *           swish_int_to_string( int val );
+xmlChar *           swish_long_to_string( long val );
+xmlChar *           swish_double_to_string( double val );
+xmlChar *           swish_date_to_string( int y, int m, int d );
-/*
-=cut
@@ -493 +495 @@
-xmlChar *       swish_get_mime_type( swish_Config * config, xmlChar * fileext );
-xmlChar *       swish_get_parser( swish_Config * config, xmlChar *mime );
-void            swish_config_test_unique_ids( swish_Config *c );
-void            swish_config_test_alias_fors( swish_Config *c );
-swish_ConfigFlags * swish_init_config_flags();
-
+
-/*
-=cut

libswish3/trunk/src/libswish3/metaname.c

@@ -30 +30 @@
-    m = swish_xmalloc(sizeof(swish_MetaName));
-    m->ref_cnt = 0;
-0
+-1
-    m->name = name;
-    m->bias = 0;

libswish3/trunk/src/libswish3/property.c

@@ -30 +30 @@
-    p = swish_xmalloc(sizeof(swish_Property));
-    p->ref_cnt = 0;
-0
+-1
-    p->name = name;
-    p->ignore_case = 1;

libswish3/trunk/src/libswish3/string.c

@@ -54 +54 @@
-);
-
+/* these string conversion functions based on code from xapian-omega */
+#define BUFSIZE 100
+
+#ifdef SNPRINTF
+#define CONVERT_TO_STRING(FMT) \
+    char buf[BUFSIZE+1];\
+    int len = SNPRINTF(buf, BUFSIZE, (FMT), val);\
+    if (len == -1 || len > BUFSIZE) buf[BUFSIZE+1] = '\0';\
+    else buf[len+1] = '\0';\
+    return swish_xstrdup((xmlChar*)buf);
+#else
+#define CONVERT_TO_STRING(FMT) \
+    char buf[BUFSIZE+1];\
+    buf[BUFSIZE+1] = '\0';\
+    sprintf(buf, (FMT), val);\
+    if (buf[BUFSIZE]) abort();\
+    return swish_xstrdup((xmlChar*)buf);
+#endif
+
+int swish_string_to_int(
+    char *buf
+) 
+{
+    return (int)strtol(buf, (char **)NULL, 10);
+}
+
+xmlChar *
+swish_int_to_string(
+    int val
+)
+{
+    CONVERT_TO_STRING("%d")
+}
+
+xmlChar *
+swish_long_to_string(
+    long val
+)
+{
+    CONVERT_TO_STRING("%ld")
+}
+
+xmlChar *
+swish_double_to_string(
+    double val
+)
+{
+    CONVERT_TO_STRING("%f")
+}
+
+xmlChar *
+swish_date_to_string(
+    int y,
+    int m,
+    int d
+)
+{
+    char buf[11];
+    if (y < 0)
+        y = 0;
+    else if (y > 9999)
+        y = 9999;
+    if (m < 1)
+        m = 1;
+    else if (m > 12)
+        m = 12;
+    if (d < 1)
+        d = 1;
+    else if (d > 31)
+        d = 31;
+#ifdef SNPRINTF
+    int len = SNPRINTF(buf, sizeof(buf), "%04d%02d%02d", y, m, d);
+    if (len == -1 || len > BUFSIZE)
+        buf[BUFSIZE + 1] = '\0';
+#else
+    buf[BUFSIZE + 1] = '\0';
+    sprintf(buf, "%04d%02d%02d", y, m, d);
+    if (buf[BUFSIZE + 1])
+        abort();                /* Uh-oh, buffer overrun */
+#endif
+    return swish_xstrdup((xmlChar*)buf);
+}
+
+/* TODO need these ??
+inline uint32_t
+binary_string_to_int(
+    const std::string & s
+)
+{
+    if (s.size() != 4)
+        return (uint32_t) - 1;
+    uint32_t
+        v;
+    memcpy(&v, s.data(), 4);
+    return ntohl(v);
+}
+
+inline
+    std::string
+int_to_binary_string(
+    uint32_t v
+)
+{
+    v = htonl(v);
+    return std::string(reinterpret_cast < const char *>(&v), 4);
+}
+
+*/
+
-/* returns length of a UTF8 character, based on first byte (see below) */
-int
@@ -121 +230 @@
-    else {
-        if (SWISH_DEBUG)
-
-
+
-
-        enc = (xmlChar *)SWISH_DEFAULT_ENCODING;
@@ -141 +249 @@
-        if (SWISH_DEBUG)
-            SWISH_DEBUG_MSG
-
-loc, 
+ loc,
+
-
-        setlocale(LC_CTYPE, SWISH_LOCALE);
@@ -321 +429 @@
-    int i;
-    for (i = 0; widechars[i] != 0; i++) {
-
-int)widechars[i], (long 
+ (long int)widechars[i],
+
-    }
-}
@@ -451 +559 @@
-        if (cursize == maxsize) {
-            sl->word =
-
-
+
-        }
-
@@ -462 +569 @@
-    if (cursize == maxsize) {
-        sl->word =
-
-
+
-    }
-