Merge pull request #299 from mattmundell/fast-xml-parser-11.0

Add a fast memory-only XML parser (9.0)

CHANGELOG.md

@@ -8,6 +8,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ### Added
 - Add option to set finished hosts in OSP targets [#298](https://github.com/greenbone/gvm-libs/pull/298)
+- Add a fast memory-only XML parser [#299](https://github.com/greenbone/gvm-libs/pull/299)
 
 ### Fixed
 - Fix sigsegv when no plugin_feed_info.inc file present. [#278](https://github.com/greenbone/gvm-libs/pull/278)

CMakeLists.txt

@@ -223,6 +223,9 @@ configure_file (VERSION.in ${CMAKE_BINARY_DIR}/VERSION @ONLY)
 
 enable_testing ()
 
+add_custom_target (tests
+                   DEPENDS array-test xmlutils-test)
+
 ## Program
 
 if (NOT SKIP_SRC)

base/CMakeLists.txt

@@ -80,9 +80,6 @@ add_test (array-test array-test)
 
 target_link_libraries (array-test cgreen ${GLIB_LDFLAGS} ${LINKER_HARDENING_FLAGS})
 
-add_custom_target (tests
-                   DEPENDS array-test)
-
 
 ## Install
 

util/CMakeLists.txt

@@ -47,6 +47,9 @@ 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 +134,8 @@ 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 +164,30 @@ 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

@@ -33,6 +33,8 @@
 #include <fcntl.h>       /* for fcntl, F_SETFL, O_NONBLOCK */
 #include <glib.h>        /* for g_free, GSList, g_markup_parse_context_free */
 #include <glib/gtypes.h> /* for GPOINTER_TO_INT, GINT_TO_POINTER, gsize */
+#include <libxml/parser.h>
+#include <libxml/tree.h>
 #include <string.h>      /* for strcmp, strerror, strlen */
 #include <time.h>        /* for time, time_t */
 #include <unistd.h>      /* for ssize_t */
@@ -1566,3 +1568,247 @@ find_element_in_xml_file (gchar *file_path, gchar *find_element,
   return search_data.found;
 }
 #undef XML_FILE_BUFFER_SIZE
+
+
+/* The new faster parser that uses libxml2. */
+
+/**
+ * @brief Read an XML element tree from a string.
+ *
+ * Caller must not free string until caller is finished using element.
+ *
+ * @param[in]   string   Input string.
+ * @param[out]  element  Location for parsed element tree, or NULL if not
+ *                       required.   If given, set to NULL on failure.
+ *                       Free with element_free.
+ *
+ * @return 0 success, -1 read error, -2 parse error, -3 XML ended prematurely,
+ *         -4 setup error.
+ */
+int
+parse_element (const gchar *string, element_t *element)
+{
+  xmlDocPtr doc;
+
+  LIBXML_TEST_VERSION
+
+  if (element)
+    *element = NULL;
+
+  if (xmlMemSetup (g_free, g_malloc, g_realloc, g_strdup))
+    return -4;
+
+  doc = xmlReadMemory (string, strlen (string), "noname.xml", NULL, 0);
+  if (doc == NULL)
+    return -2;
+
+  if (element)
+    *element = xmlDocGetRootElement (doc);
+
+  return 0;
+}
+
+/**
+ * @brief Free an entire element tree.
+ *
+ * Beware that this frees the entire tree that element is part of, including
+ * any ancestors.
+ *
+ * @param[in]  element  Element.
+ */
+void
+element_free (element_t element)
+{
+  if (element)
+    {
+      assert (element->doc);
+      xmlFreeDoc (element->doc);
+    }
+}
+
+/**
+ * @brief Get the name of an element.
+ *
+ * @param[in]  element  Element.
+ *
+ * @return Element name.
+ */
+const gchar *
+element_name (element_t element)
+{
+  if (element
+      && (element->type == XML_ELEMENT_NODE))
+    return (const gchar *) element->name;
+
+  return "";
+}
+
+/**
+ * @brief Find child in an element.
+ *
+ * @param[in]  element  Element.
+ * @param[in]  name     Name of child.
+ *
+ * @return Child if found, else NULL.
+ */
+static element_t
+find_child (element_t element, const gchar *name)
+{
+  for (xmlNode *node = element->children; node; node = node->next)
+    if (xmlStrcmp (node->name, (const xmlChar *) name) == 0)
+      return node;
+  return NULL;
+}
+
+/**
+ * @brief Get a child of an element.
+ *
+ * @param[in]  element  Element.
+ * @param[in]  name    Name of the child.
+ *
+ * @return Element if found, else NULL.
+ */
+element_t
+element_child (element_t element, const gchar *name)
+{
+  const gchar *stripped_name;
+
+  if (!element)
+    return NULL;
+
+  stripped_name = strchr (name, ':');
+  if (stripped_name)
+    {
+       element_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 (element, name);
+
+      child = find_child (element, 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 (element, name);
+}
+
+/**
+ * @brief Get text of an element.
+ *
+ * If element is not NULL then the return is guaranteed to be a string.
+ * So if the caller has NULL checked element then there is no need for
+ * the caller to NULL check the return.
+ *
+ * @param[in]  element  Element.
+ *
+ * @return NULL if element is NULL, else the text.  Caller must g_free.
+ */
+gchar *
+element_text (element_t element)
+{
+  gchar *string;
+
+  if (!element)
+    return NULL;
+
+  string = (gchar *) xmlNodeListGetString (element->doc, element->xmlChildrenNode, 1);
+  if (string)
+    return string;
+  string = xmlMalloc (1);
+  string[0] = '\0';
+  return string;
+}
+
+/**
+ * @brief Get an attribute of an element.
+ *
+ * @param[in]  element  Element.
+ * @param[in]  name     Name of the attribute.
+ *
+ * @return Attribute value if found, else NULL.  Caller must g_free.
+ */
+gchar *
+element_attribute (element_t element, const gchar *name)
+{
+  const gchar *stripped_name;
+
+  if (!element)
+    return NULL;
+
+  stripped_name = strchr (name, ':');
+  if (stripped_name)
+    {
+       gchar *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 (gchar *) xmlGetProp (element, (const xmlChar *) name);
+
+      attribute = (gchar *) xmlGetProp (element, (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 (gchar *) xmlGetProp (element, (const xmlChar *) name);
+}
+
+/**
+ * @brief Get the first child of an element.
+ *
+ * @param[in]  element  Element.
+ *
+ * @return Child if there is one, else NULL.
+ */
+element_t
+element_first_child (element_t element)
+{
+  if (element)
+    return element->children;
+  return NULL;
+}
+
+/**
+ * @brief Get the next sibling of an element
+ *
+ * @param[in]  element  Element.
+ *
+ * @return Next sibling element if there is one, else NULL.
+ */
+element_t
+element_next (element_t element)
+{
+  if (element)
+    return element->next;
+  return NULL;
+}

util/xmlutils.h

@@ -158,9 +158,39 @@ int xml_count_entities (entities_t);
 void
 xml_string_append (GString *, const char *, ...);
 
+
 /* XML file utilities */
 
 int
 find_element_in_xml_file (gchar *, gchar *, GHashTable *);
 
+
+/* The new faster parser that uses libxml2. */
+
+typedef struct _xmlNode *element_t;
+
+int
+parse_element (const gchar *, element_t *);
+
+void
+element_free (element_t);
+
+const gchar *
+element_name (element_t);
+
+gchar *
+element_attribute (element_t, const gchar *);
+
+gchar *
+element_text (element_t);
+
+element_t
+element_child (element_t, const gchar *);
+
+element_t
+element_first_child (element_t);
+
+element_t
+element_next (element_t);
+
 #endif /* not _GVM_XMLUTILS_H */