Add a small rough XML interface based on libxml2

gmp/CMakeLists.txt

@@ -27,8 +27,9 @@ endif (NOT PKG_CONFIG_FOUND)
 ## Dependency checks
 
 pkg_check_modules (GLIB REQUIRED glib-2.0>=2.42)
+pkg_check_modules (LIBXML2 REQUIRED libxml-2.0>=2.0)
 
-include_directories (${GLIB_INCLUDE_DIRS})
+include_directories (${GLIB_INCLUDE_DIRS} ${LIBXML2_INCLUDE_DIRS})
 
 set (FILES gmp.c)
 

osp/CMakeLists.txt

@@ -27,8 +27,9 @@ endif (NOT PKG_CONFIG_FOUND)
 ## Dependency checks
 
 pkg_check_modules (GLIB REQUIRED glib-2.0>=2.42)
+pkg_check_modules (LIBXML2 REQUIRED libxml-2.0>=2.0)
 
-include_directories (${GLIB_INCLUDE_DIRS})
+include_directories (${GLIB_INCLUDE_DIRS} ${LIBXML2_INCLUDE_DIRS})
 
 set (FILES osp.c)
 

util/CMakeLists.txt

@@ -24,6 +24,8 @@ if (NOT PKG_CONFIG_FOUND)
   message(FATAL_ERROR "pkg-config executable not found. Aborting.")
 endif (NOT PKG_CONFIG_FOUND)
 
+set (CMAKE_VERBOSE_MAKEFILE     ON)
+
 ## Dependency checks
 
 # for all modules we need glib
@@ -47,6 +49,9 @@ pkg_check_modules (LIBSSH REQUIRED libssh>=0.6.0)
 # for kb we need libhiredis
 pkg_check_modules (REDIS REQUIRED hiredis>=0.10.1)
 
+# for fast XML we need libxml2
+pkg_check_modules (LIBXML2 REQUIRED libxml-2.0>=2.0)
+
 # Set NVTICACHE name with the version
 set (NVTICACHE_STR "nvticache${PROJECT_VERSION}")
 add_definitions (-DNVTICACHE_STR="${NVTICACHE_STR}")
@@ -131,7 +136,7 @@ if (BUILD_WITH_LDAP)
   endif (NOT LIBLDAP)
 endif (BUILD_WITH_LDAP)
 
-include_directories (${GLIB_INCLUDE_DIRS} ${GPGME_INCLUDE_DIRS} ${GCRYPT_INCLUDE_DIRS})
+include_directories (${GLIB_INCLUDE_DIRS} ${GPGME_INCLUDE_DIRS} ${GCRYPT_INCLUDE_DIRS} ${LIBXML2_INCLUDE_DIRS})
 
 set (FILES authutils.c compressutils.c fileutils.c gpgmeutils.c kb.c ldaputils.c
            nvticache.c radiusutils.c serverutils.c sshutils.c uuidutils.c
@@ -160,9 +165,30 @@ if (BUILD_SHARED)
                          ${GIO_LDFLAGS} ${GPGME_LDFLAGS} ${ZLIB_LDFLAGS}
                          ${RADIUS_LDFLAGS} ${LIBSSH_LDFLAGS} ${GNUTLS_LDFLAGS}
                          ${GCRYPT_LDFLAGS} ${LDAP_LDFLAGS} ${REDIS_LDFLAGS}
-                         ${UUID_LDFLAGS} ${LINKER_HARDENING_FLAGS})
+                         ${LIBXML2_LDFLAGS} ${UUID_LDFLAGS}
+                         ${LINKER_HARDENING_FLAGS})
 endif (BUILD_SHARED)
 
+
+## Tests
+
+add_executable (xmlutils-test
+                EXCLUDE_FROM_ALL
+                xmlutils_tests.c)
+
+add_test (xmlutils-test xmlutils-test)
+
+target_link_libraries (xmlutils-test cgreen
+                       ${GLIB_LDFLAGS} ${GIO_LDFLAGS} ${GPGME_LDFLAGS} ${ZLIB_LDFLAGS}
+					   ${RADIUS_LDFLAGS} ${LIBSSH_LDFLAGS} ${GNUTLS_LDFLAGS}
+					   ${GCRYPT_LDFLAGS} ${LDAP_LDFLAGS} ${REDIS_LDFLAGS}
+					   ${LIBXML2_LDFLAGS} ${UUID_LDFLAGS}
+					   ${LINKER_HARDENING_FLAGS})
+
+add_custom_target (tests-xmlutils
+                   DEPENDS xmlutils-test)
+
+
 ## Install
 configure_file (libgvm_util.pc.in ${CMAKE_BINARY_DIR}/libgvm_util.pc @ONLY)
 

util/xmlutils.c

@@ -1087,6 +1087,251 @@ read_entity_c (gvm_connection_t *connection, entity_t *entity)
   return try_read_entity_c (connection, 0, entity);
 }
 
+/**
+ * @brief Get the name an entity.
+ *
+ * @param[in]  entity  Entity.
+ *
+ * @return Entity name, which is freed by free_entity2.
+ */
+void
+xml_doc_free (xml_doc_t doc)
+{
+  xmlFreeDoc (doc);
+}
+
+/**
+ * @brief Read an XML entity tree from a string.
+ *
+ * @param[in]   string  Input string.
+ * @param[out]  entity  Pointer to an entity tree.
+ *
+ * @return 0 success, -1 read error, -2 parse error, -3 XML ended prematurely.
+ */
+int
+parse_entity2 (const char *string, xml_doc_t *xml_doc, entity2_t *entity)
+{
+  LIBXML_TEST_VERSION
+
+  /* XML_PARSE_NOCDATA makes this skip CDATA to match our glib parser, but
+   * it makes this extremely slow, so instead I stripped the CDATA out of the
+   * XML in my feed (for now, so that I can compare the resulting scap dbs). */
+  *xml_doc = xmlReadMemory (string, strlen (string), "noname.xml", NULL, /* XML_PARSE_NOCDATA */ 0);
+  if (*xml_doc == NULL)
+    return -2;
+  //xmlFreeDoc(doc);
+
+  *entity = xmlDocGetRootElement (*xml_doc);
+  return 0;
+}
+
+/**
+ * @brief Get the name an entity.
+ *
+ * @param[in]  entity  Entity.
+ *
+ * @return Entity name, which is freed by free_entity2.
+ */
+const char *
+entity2_name (entity2_t entity)
+{
+  if (entity
+      && (entity->type == XML_ELEMENT_NODE))
+    return (const char *) entity->name;
+
+  // FIX because we have text nodes now too
+  // FIX maybe should skip over in entity2_children and next_entities2
+  return "";
+}
+
+/**
+ * @brief Get children of an entity.
+ *
+ * @param[in]  entity  Entity.
+ *
+ * @return Children if found, else NULL.
+ */
+entities2_t
+entity2_children (entity2_t entity)
+{
+  if (!entity)
+    return NULL;
+  return entity->children;
+}
+
+static entity2_t
+find_child (entity2_t entity, const char *name)
+{
+  for (xmlNode *node = entity->children; node; node = node->next)
+    if (xmlStrcmp (node->name, (const xmlChar *) name) == 0)
+      return node;
+  return NULL;
+}
+
+/**
+ * @brief Get a child of an entity.
+ *
+ * @param[in]  entity  Entity.
+ * @param[in]  name    Name of the child.
+ *
+ * @return Entity if found, else NULL.
+ */
+entity2_t
+entity2_child (entity2_t entity, const char *name)
+{
+  const char *stripped_name;
+
+  if (!entity)
+    return NULL;
+
+  stripped_name = strchr (name, ':');
+  if (stripped_name)
+    {
+       entity2_t child;
+
+       /* There was a namespace in the name.
+        *
+        * First try without the namespace, because libxml2 doesn't consider the
+        * namespace in the name when the namespace is defined. */
+
+      stripped_name++;
+
+      if (*stripped_name == '\0')
+        /* Don't search for child with empty stripped name, because we'll
+         * find text nodes.  But search with just the namespace for glib
+         * compatibility. */
+        return find_child (entity, name);
+
+      child = find_child (entity, stripped_name);
+      if (child)
+       return child;
+
+      /* Didn't find anything. */
+    }
+
+  /* There was no namespace, or we didn't find anything without the namespace.
+   *
+   * Try with the full name. */
+
+  return find_child (entity, name);
+}
+
+/**
+ * @brief Get text of an entity.
+ *
+ * If entity is not NULL then the return is guaranteed to be a string.
+ * So if the caller has NULL checked entity then there is no need for
+ * the caller to NULL check the return.
+ *
+ * @param[in]  doc     XML doc.
+ * @param[in]  entity  Entity.
+ *
+ * @return NULL if entity is NULL, else the text.
+ */
+char *
+entity2_text (xml_doc_t doc, entity2_t entity)
+{
+  char *string;
+
+  if (!entity)
+    return NULL;
+
+  string = (char *) xmlNodeListGetString (doc, entity->xmlChildrenNode, 1);
+  if (string)
+    return string;
+  string = xmlMalloc (1);
+  string[0] = '\0';
+  return string;
+}
+
+void
+entity2_text_free (char *text)
+{
+  xmlFree (text);
+}
+/**
+ * @brief Get a child of an entity.
+ *
+ * @param[in]  entity  Entity.
+ * @param[in]  name    Name of the child.
+ *
+ * @return Entity if found, else NULL.
+ */
+char *
+entity2_attribute (entity2_t entity, const char *name)
+{
+  if (!entity)
+    return NULL;
+
+  const char *stripped_name;
+
+  stripped_name = strchr (name, ':');
+  if (stripped_name)
+    {
+       char *attribute;
+
+       /* There was a namespace in the name.
+        *
+        * First try without the namespace, because libxml2 doesn't consider the
+        * namespace in the name when the namespace is defined. */
+
+      stripped_name++;
+
+      if (*stripped_name == '\0')
+        /* Don't search for child with empty stripped name, because we'll
+         * find text nodes.  But search with just the namespace for glib
+         * compatibility. */
+        return (char *) xmlGetProp (entity, (const xmlChar *) name);
+
+      attribute = (char *) xmlGetProp (entity, (const xmlChar *) stripped_name);
+      if (attribute)
+       return attribute;
+
+      /* Didn't find anything. */
+    }
+
+  /* There was no namespace, or we didn't find anything without the namespace.
+   *
+   * Try with the full name. */
+
+  return (char *) xmlGetProp (entity, (const xmlChar *) name);
+}
+
+void
+xml_string_free (xml_string_t string)
+{
+  if (string)
+    xmlFree (string);
+}
+
+/**
+ * @brief Get the name an entity.
+ *
+ * @param[in]  entity  Entity.
+ *
+ * @return Entity name, which is freed by free_entity2.
+ */
+entities2_t
+first_entity2 (entities2_t entities)
+{
+  return entities;
+}
+
+/**
+ * @brief Get the name an entity.
+ *
+ * @param[in]  entity  Entity.
+ *
+ * @return Entity name, which is freed by free_entity2.
+ */
+entities2_t
+next_entities2 (entities2_t entities)
+{
+  if (entities)
+    return entities->next;
+  return NULL;
+}
+
 /**
  * @brief Read an XML entity tree from a string.
  *

util/xmlutils.h

@@ -29,6 +29,9 @@
 
 #include <glib.h>
 #include <gnutls/gnutls.h>
+// FIX should be in .c only
+#include <libxml/parser.h>
+#include <libxml/tree.h>
 #include <stdio.h>
 
 /**
@@ -54,6 +57,54 @@ xml_handle_end_element (context_data_t *, const gchar *);
 void
 xml_handle_text (context_data_t *, const gchar *, gsize);
 
+/**
+ * @brief XML doc.
+ */
+typedef xmlDoc *xml_doc_t;
+
+typedef xmlNode *entities2_t;
+/* equivalent now, kept for compat. */
+typedef xmlNode *entity2_t;
+
+typedef xmlChar *xml_string_t;
+typedef const xmlChar *const_xml_string_t;
+
+void
+xml_doc_free (xml_doc_t);
+
+int
+parse_entity2 (const char *, xml_doc_t *, entity2_t *);
+
+void
+xml_string_free (xml_string_t);
+
+entities2_t
+xml_doc_root (xml_doc_t);
+
+const char *
+entity2_name (entities2_t);
+
+char *
+entity2_attribute (entity2_t, const char *);
+
+char *
+entity2_text (xml_doc_t, entities2_t);
+
+void
+entity2_text_free (char *);
+
+entity2_t
+entity2_child (entity2_t, const char *);
+
+entities2_t
+entity2_children (entities2_t);
+
+entities2_t
+first_entity2 (entities2_t);
+
+entities2_t
+next_entities2 (entities2_t);
+
 /**
  * @brief Entities.
  */