test/chmindex.hhk @@ -4,8 +4,13 @@ #include <libxml/parser.h> #include <libxml/HTMLparser.h> -struct _ChmIndexPriv { +#include "utils/utils.h" +#include "models/link.h" + +#define selfp (self->priv) +struct _ChmIndexPriv { + GList* data; /* GList<Link> */ }; G_DEFINE_TYPE(ChmIndex, chmindex, G_TYPE_OBJECT); @@ -25,24 +30,107 @@ chmindex_init(ChmIndex* self) self->priv = G_TYPE_INSTANCE_GET_PRIVATE(self, TYPE_CHMINDEX, ChmIndexPriv); } +enum { + STATE_OUT, + STATE_IN_SITEMAP +}; + + typedef struct { + int state; + GList* names; /* GList<gchar*> */ + char* local; + GList* res; /* GList<Link> */ } SAXContext; -/** - * TODO - */ +SAXContext* saxcontext_new() { + return g_new0(SAXContext, 1); +} + +void saxcontext_free(SAXContext* self) { + g_list_foreach(self->names, (GFunc)g_free, NULL); + g_list_free(self->names); + + g_free(self->local); + + g_list_foreach(self->res, (GFunc)g_free, NULL); + g_list_free(self->res); + + g_free(self); +} + + static void chm_index_start_element (void *ctx, - const xmlChar *name, - const xmlChar **atts) { + const xmlChar *name_, + const xmlChar **attrs_) { + const char* name = (const char*) name_; + const char** attrs = (const char**) attrs_; + SAXContext* context = (SAXContext*) ctx; + + switch(context->state) { + case STATE_OUT: + if(g_ascii_strcasecmp(name, "object") == 0) { + const char* type = get_attr(attrs, "type"); + if(g_ascii_strcasecmp(type, "text/sitemap") == 0) { + context->state = STATE_IN_SITEMAP; + } + } + break; + case STATE_IN_SITEMAP: + if(g_ascii_strcasecmp(name, "param") == 0) { + const char* name = get_attr(attrs, "name"); + const char* value = get_attr(attrs, "value"); + + if(name != NULL && value != NULL) { + if(g_ascii_strcasecmp(name, "name") == 0) { + context->names = g_list_append(context->names, g_strdup(value)); + } else if(g_ascii_strcasecmp(name, "local") == 0) { + if(context->local != NULL) { + g_warning("more than one local in one sitemap"); + g_free(context->local); + } + context->local = g_strdup(value); + } + } else { + g_warning("name or value is null"); + } + } + break; + default: + g_return_if_reached(); + } + } -/** - * TODO - */ static void chm_index_end_element (void *ctx, - const xmlChar *name) { - + const xmlChar *name_) { + const char* name = (const char*) name_; + SAXContext* context = (SAXContext*) ctx; + + switch(context->state) { + case STATE_OUT: + break; + case STATE_IN_SITEMAP: + if(g_ascii_strcasecmp(name, "object") == 0) { + if(context->local != NULL && context->names != NULL) { + GList* iter; + for(iter = context->names; iter; iter = iter->next) { + Link* link = link_new(LINK_TYPE_PAGE, iter->data, context->local); + context->res = g_list_append(context->res, link); + } + } + + g_free(context->local); + context->local = NULL; + g_list_foreach(context->names, (GFunc)g_free, NULL); + g_list_free(context->names); + context->names = NULL; + } + break; + default: + g_return_if_reached(); + } } @@ -82,22 +170,39 @@ static htmlSAXHandler hhSAXHandler = { NULL /* xmlStructuredErrorFunc */ }; +static gint data_sort_func(gconstpointer lhs_, gconstpointer rhs_) { + Link* lhs = (Link*)lhs_; + Link* rhs = (Link*)rhs_; + + gint res = ncase_compare_utf8_string(lhs->name, rhs->name); + if(res == 0) { + res = ncase_compare_utf8_string(lhs->uri, rhs->uri); + } + return res; +} -/** - * TODO: - */ ChmIndex* chmindex_new(const char* filename, const char* encoding) { - SAXContext context; + SAXContext* context = saxcontext_new(); if(htmlSAXParseFile(filename, encoding, &hhSAXHandler, - &context) != NULL) { + context) != NULL) { g_error("parse %s with encoding %s failed.", filename, encoding); return NULL; } ChmIndex* self = g_object_new(TYPE_CHMINDEX, NULL); + selfp->data = g_list_sort(context->res, data_sort_func); + context->res = NULL; + + saxcontext_free(context); return self; } + +GList* chmindex_get_data(ChmIndex* self) { + g_return_val_if_fail(IS_CHMINDEX(self), NULL); + return selfp->data; +} + src/models/chmindex.c @@ -37,5 +37,7 @@ GType chmindex_get_type(void); * @return NULL if open or parse indexFname failed. */ ChmIndex *chmindex_new(const gchar* indexFname, const gchar* encoding); +/** @return GList<Link> */ +GList* chmindex_get_data(ChmIndex* self); #endif /* CHMINDEX_H_ */ src/models/chmindex.h @@ -1,15 +1,36 @@ #include <glib.h> #include "utils/utils.h" +#include "models/link.h" +#include "models/chmindex.h" void test_issue_19() { g_test_bug("19"); g_assert_cmpstr("WINDOWS-1256", ==, get_encoding_by_lcid(0xc01)); } +void test_chmindex() { + ChmIndex* chmIndex = chmindex_new("test/chmindex.hhk", "UTF-8"); + GList* data = chmindex_get_data(chmIndex); + g_assert_cmpint(3, ==, g_list_length(data)); + g_assert_cmpstr("a", ==, ((Link*)data->data)->name); + g_assert_cmpstr("a1", ==, ((Link*)data->data)->uri); + + data = data->next; + g_assert_cmpstr("a", ==, ((Link*)data->data)->name); + g_assert_cmpstr("a2", ==, ((Link*)data->data)->uri); + + data = data->next; + g_assert_cmpstr("b", ==, ((Link*)data->data)->name); + g_assert_cmpstr("a1", ==, ((Link*)data->data)->uri); +} + + int main(int argc, char* argv[]) { + gtk_init(&argc, &argv); g_test_init(&argc, &argv, NULL); g_test_bug_base("http://code.google.com/p/chmsee/issues/detail?id="); g_test_add_func("/chmsee/19", test_issue_19); + g_test_add_func("/chmsee/chmindex", test_chmindex); g_test_run(); return 0; } test/test.c e31807bf3ef02862ffed4cf79a00ca43184571dd LI Daobing lidaobing@gmail.com http://github.com/lidaobing/chmsee/commit/21daf09d38358e61014e6832b290e0ea154c2cb2 21daf09d38358e61014e6832b290e0ea154c2cb2 2009-06-14T03:18:26-07:00 2009-06-14T03:18:26-07:00 add chmindex and the testcase ab83f628f3a1dc86d2f0c589568b75ca32530a5f LI Daobing lidaobing@gmail.com