Fixes after check by the Coverity Scan

10 files changed, 177 insertions, 81 deletions

diff --git a/source/myhtml/Makefile b/source/myhtml/Makefile
new file mode 100644
index 0000000..f24973c
--- /dev/null
+++ b/source/myhtml/Makefile
@@ -0,0 +1,53 @@
+TARGET := source/myhtml
+SRCDIR := source/myhtml
+
+CC ?= gcc
+
+LIBPOSTFIX := .so
+LIBNAME    := myhtml
+LIBSTATIC_POSTFIX  := _static
+
+MyHTML_OPTIMIZATION_LEVEL ?= -O2
+
+CFLAGS ?= -Wall -Werror
+CFLAGS += $(MyHTML_OPTIMIZATION_LEVEL) -fPIC --std=c99 -I..
+
+MyHTML_BUILD_WITHOUT_THREADS ?= NO
+ifeq ($(MyHTML_BUILD_WITHOUT_THREADS),YES)
+    $(info Build without POSIX Threads)
+	CFLAGS += -DMyHTML_BUILD_WITHOUT_THREADS
+else
+    $(info Build with POSIX Threads)
+	CFLAGS += -pthread
+endif
+
+ifeq ($(OS),Windows_NT)
+else
+	UNAM := $(shell uname -s)
+	ifeq ($(UNAM),Darwin)
+		LIBPOSTFIX := .dylib
+	else
+		CFLAGS += -D_POSIX_C_SOURCE=199309L
+	endif
+endif
+
+SRCS := $(wildcard *.c)
+SRCS += $(wildcard utils/*.c)
+HDRS := $(wildcard *.h)
+HDRS += $(wildcard utils/*.h)
+OBJS := $(patsubst %.c, %.o, $(SRCS))
+
+all: shared static
+
+shared: $(OBJS) $(HDRS)
+	$(CC) -shared $(LDFLAGS) $(OBJS) -o lib$(LIBNAME)$(LIBPOSTFIX)
+
+static: shared
+	$(AR) crus lib$(LIBNAME)$(LIBSTATIC_POSTFIX).a $(OBJS)
+
+clean:
+	rm -rf *.o
+	rm -rf utils/*.o
+	rm -rf *lib$(LIBNAME)*
+
+.PHONY: all clean
diff --git a/source/myhtml/api.h b/source/myhtml/api.h
index ad94e84..794b9c7 100644..100755
--- a/source/myhtml/api.h
+++ b/source/myhtml/api.h
@@ -36,7 +36,7 @@
 
 #define MyHTML_VERSION_MAJOR 1
 #define MyHTML_VERSION_MINOR 0
-#define MyHTML_VERSION_PATCH 4
+#define MyHTML_VERSION_PATCH 5
 
 #include <stdbool.h>
 #include <stddef.h>
@@ -577,6 +577,7 @@ typedef myhtml_version_t;
 // callback functions
 typedef void* (*myhtml_callback_token_f)(myhtml_tree_t* tree, myhtml_token_node_t* token, void* ctx);
 typedef void (*myhtml_callback_tree_node_f)(myhtml_tree_t* tree, myhtml_tree_node_t* node, void* ctx);
+typedef void (*myhtml_callback_serialize_f)(const char* buffer, size_t size, void* ctx);
 
 /***********************************************************************************
  *
@@ -2714,28 +2715,74 @@ myhtml_strncasecmp(const char* str1, const char* str2, size_t size);
  ***********************************************************************************/
 
 /**
- * Tree fragment serialization
+ * Tree fragment serialization 
+ * The same as myhtml_serialization_tree_buffer function
+ */
+bool
+myhtml_serialization(myhtml_tree_t* tree, myhtml_tree_node_t* scope_node,
+                     myhtml_string_raw_t* str);
+
+/**
+ * Only one tree node serialization
+ * The same as myhtml_serialization_node_buffer function
+ */
+bool
+myhtml_serialization_node(myhtml_tree_t* tree, myhtml_tree_node_t* node,
+                          myhtml_string_raw_t* str);
+
+/**
+ * Serialize tree to an output string
  *
  * @param[in] myhtml_tree_t*
- * @param[in] scope node, myhtml_tree_node_t*
- * @param[in] myhtml_string_raw_t* (date to be created if str_raw.data == NULL)
+ * @param[in] scope node
+ * @param[in] myhtml_string_raw_t*
  *
  * @return true if successful, otherwise false
  */
 bool
-myhtml_serialization(myhtml_tree_t* tree, myhtml_tree_node_t* scope_node, myhtml_string_raw_t* str);
+myhtml_serialization_tree_buffer(myhtml_tree_t* tree, myhtml_tree_node_t* scope_node,
+                                 myhtml_string_raw_t* str);
 
 /**
- * Only one tree node serialization
+ * Serialize node to an output string
  *
  * @param[in] myhtml_tree_t*
- * @param[in] myhtml_tree_node_t*
- * @param[in] myhtml_string_raw_t* (date to be created if str_raw.data == NULL)
+ * @param[in] node
+ * @param[in] myhtml_string_raw_t*
+ *
+ * @return true if successful, otherwise false
+ */
+bool
+myhtml_serialization_node_buffer(myhtml_tree_t* tree, myhtml_tree_node_t* node,
+                                 myhtml_string_raw_t* str);
+
+/**
+ * The serialize function for an entire tree
+ *
+ * @param[in] tree        the tree to be serialized
+ * @param[in] scope_node  the scope_node
+ * @param[in] callback    function that will be called for all strings that have to be printed
+ * @param[in] ptr         user-supplied pointer
+ *
+ * @return true if successful, otherwise false
+ */
+bool
+myhtml_serialization_tree_callback(myhtml_tree_t* tree, myhtml_tree_node_t* scope_node,
+                                   myhtml_callback_serialize_f callback, void* ptr);
+
+/**
+ * The serialize function for a single node
+ *
+ * @param[in] tree        the tree to be serialized
+ * @param[in] node        the node that is going to be serialized
+ * @param[in] callback    function that will be called for all strings that have to be printed
+ * @param[in] ptr         user-supplied pointer
  *
  * @return true if successful, otherwise false
  */
 bool
-myhtml_serialization_node(myhtml_tree_t* tree, myhtml_tree_node_t* node, myhtml_string_raw_t* str);
+myhtml_serialization_node_callback(myhtml_tree_t* tree, myhtml_tree_node_t* node,
+                                   myhtml_callback_serialize_f callback, void* ptr);
 
 /***********************************************************************************
  *
diff --git a/source/myhtml/encoding.c b/source/myhtml/encoding.c
index 1c7560a..078ac5c 100644
--- a/source/myhtml/encoding.c
+++ b/source/myhtml/encoding.c
@@ -398,7 +398,7 @@ unsigned long myhtml_encoding_index_gb18030_ranges_code_point(unsigned long poin
         return 0xe7c7;
     
     // step 3
-    unsigned long offset = sizeof(myhtml_encoding_map_gb18030_ranges) / sizeof(unsigned long);
+    unsigned long offset = sizeof(myhtml_encoding_map_gb18030_ranges) / (sizeof(unsigned long) * 2);
     
     unsigned long code_point_offset = 0;
     while (offset) {
diff --git a/source/myhtml/myhtml.c b/source/myhtml/myhtml.c
index 7edeaf1..6b3bbd2 100644
--- a/source/myhtml/myhtml.c
+++ b/source/myhtml/myhtml.c
@@ -1434,35 +1434,50 @@ void myhtml_queue_add(myhtml_tree_t *tree, size_t begin, myhtml_token_node_t* to
         }
     }
     
-    qnode->token = token;
-    
 #ifndef MyHTML_BUILD_WITHOUT_THREADS
     
     if(tree->flags & MyHTML_TREE_FLAGS_SINGLE_MODE) {
-        myhtml_parser_worker(0, qnode);
-        myhtml_parser_stream(0, qnode);
+        if(qnode) {
+            qnode->token = token;
+            
+            myhtml_parser_worker(0, qnode);
+            myhtml_parser_stream(0, qnode);
+        }
         
         tree->current_qnode = mythread_queue_node_malloc_limit(tree->myhtml->thread, tree->queue, 4, NULL);
     }
     else {
+        if(qnode)
+            qnode->token = token;
+            
         tree->current_qnode = mythread_queue_node_malloc_round(tree->myhtml->thread, tree->queue_entry, NULL);
     }
     
 #else
     
-    myhtml_parser_worker(0, qnode);
-    myhtml_parser_stream(0, qnode);
+    f(qnode) {
+        qnode->token = token;
+        
+        myhtml_parser_worker(0, qnode);
+        myhtml_parser_stream(0, qnode);
+    }
     
     tree->current_qnode = mythread_queue_node_malloc_limit(tree->myhtml->thread, tree->queue, 4, NULL);
     
 #endif /* MyHTML_BUILD_WITHOUT_THREADS */
     
+    if(tree->current_qnode == NULL) {
+        // TODO: add return status
+        return;
+    }
+    
     tree->current_qnode->tree = tree;
     tree->current_qnode->prev = qnode;
     
-    if(qnode)
+    if(qnode && token)
         myhtml_tokenizer_calc_current_namespace(tree, token);
     
+    // TODO: add check created node
     myhtml_token_node_malloc(tree->token, tree->current_token_node, tree->token->mcasync_token_id);
     
     tree->current_token_node->raw_begin = tree->current_token_node->element_begin = (tree->global_offset + begin);
diff --git a/source/myhtml/myosi.h b/source/myhtml/myosi.h
index 530dd72..4c516a8 100644
--- a/source/myhtml/myosi.h
+++ b/source/myhtml/myosi.h
@@ -31,7 +31,7 @@
 
 #define MyHTML_VERSION_MAJOR 1
 #define MyHTML_VERSION_MINOR 0
-#define MyHTML_VERSION_PATCH 4
+#define MyHTML_VERSION_PATCH 5
 
 #if (defined(_WIN32) || defined(_WIN64)) && !defined(__WINPTHREADS_VERSION)
 #define IS_OS_WINDOWS
diff --git a/source/myhtml/serialization.c b/source/myhtml/serialization.c
index ef76b94..9685cd0 100644..100755
--- a/source/myhtml/serialization.c
+++ b/source/myhtml/serialization.c
@@ -269,9 +269,15 @@ void myhtml_serialization_append(const char *data, size_t size, myhtml_callback_
             notwritten = 0;
             break;
         case 0xA0:
-            if (notwritten) callback(data + i - notwritten, notwritten, ptr);
-            callback("&nbsp;", 6, ptr);
-            notwritten = 0;
+            if(i > 0 && (unsigned char)(data[(i - 1)]) == 0xC2) {
+                if (notwritten) callback(data + i - notwritten, (notwritten - 1), ptr);
+                callback("&nbsp;", 6, ptr);
+                notwritten = 0;
+            }
+            else {
+                ++notwritten;
+            }
+            
             break;
         default:
             ++notwritten;
@@ -308,9 +314,15 @@ void myhtml_serialization_append_attr(const char* data, size_t size, myhtml_call
             notwritten = 0;
             break;
         case 0xA0:
-            if (notwritten) callback(data + i - notwritten, notwritten, ptr);
-            callback("&nbsp;", 6, ptr);
-            notwritten = 0;
+            if(i > 0 && (unsigned char)(data[(i - 1)]) == 0xC2) {
+                if (notwritten) callback(data + i - notwritten, (notwritten - 1), ptr);
+                callback("&nbsp;", 6, ptr);
+                notwritten = 0;
+            }
+            else {
+                ++notwritten;
+            }
+            
             break;
         default:
             ++notwritten;
diff --git a/source/myhtml/tree.c b/source/myhtml/tree.c
index cd5e929..6bf7772 100644
--- a/source/myhtml/tree.c
+++ b/source/myhtml/tree.c
@@ -71,14 +71,6 @@ myhtml_status_t myhtml_tree_init(myhtml_tree_t* tree, myhtml_t* myhtml)
     if(mcstatus)
         return MyHTML_STATUS_TREE_ERROR_MCOBJECT_INIT;
     
-    if(status) {
-        myhtml->parse_state_func = NULL;
-        myhtml->insertion_func = NULL;
-        myhtml->thread = NULL;
-        
-        return status;
-    }
-    
     tree->mchar              = mchar_async_create(128, (4096 * 5));
     tree->active_formatting  = myhtml_tree_active_formatting_init(tree);
     tree->open_elements      = myhtml_tree_open_elements_init(tree);
@@ -880,7 +872,7 @@ void myhtml_tree_list_insert_by_index(myhtml_tree_list_t* list, myhtml_tree_node
     
     myhtml_tree_node_t** node_list = list->list;
     
-    memmove(&node_list[(index + 1)], &node_list[index], sizeof(myhtml_tree_node_t**) * (list->length - index));
+    memmove(&node_list[(index + 1)], &node_list[index], sizeof(myhtml_tree_node_t*) * (list->length - index));
     
     list->list[index] = node;
     list->length++;
@@ -961,7 +953,7 @@ void myhtml_tree_open_elements_remove(myhtml_tree_t* tree, myhtml_tree_node_t* n
         
         if(list[el_idx] == node)
         {
-            memmove(&list[el_idx], &list[el_idx + 1], sizeof(myhtml_tree_node_t**) * (tree->open_elements->length - el_idx));
+            memmove(&list[el_idx], &list[el_idx + 1], sizeof(myhtml_tree_node_t*) * (tree->open_elements->length - el_idx));
             tree->open_elements->length--;
             
             break;
@@ -1451,7 +1443,7 @@ void myhtml_tree_active_formatting_remove(myhtml_tree_t* tree, myhtml_tree_node_
         
         if(list[el_idx] == node)
         {
-            memmove(&list[el_idx], &list[el_idx + 1], sizeof(myhtml_tree_node_t**) * (tree->active_formatting->length - el_idx));
+            memmove(&list[el_idx], &list[el_idx + 1], sizeof(myhtml_tree_node_t*) * (tree->active_formatting->length - el_idx));
             tree->active_formatting->length--;
             
             break;
@@ -1469,7 +1461,7 @@ void myhtml_tree_active_formatting_remove_by_index(myhtml_tree_t* tree, size_t i
 {
     myhtml_tree_node_t** list = tree->active_formatting->list;
     
-    memmove(&list[idx], &list[idx + 1], sizeof(myhtml_tree_node_t**) * (tree->active_formatting->length - idx));
+    memmove(&list[idx], &list[idx + 1], sizeof(myhtml_tree_node_t*) * (tree->active_formatting->length - idx));
     tree->active_formatting->length--;
     
 #ifdef DEBUG_MODE
diff --git a/source/myhtml/utils.c b/source/myhtml/utils.c
index 2134270..eb5c741 100644
--- a/source/myhtml/utils.c
+++ b/source/myhtml/utils.c
@@ -54,14 +54,8 @@ static const unsigned char myhtml_utils_chars_map[] = {
 
 size_t myhtml_strncasecmp(const char* str1, const char* str2, size_t size)
 {
-    if(str1 == NULL) {
-        if(str2 == NULL)
-            return 0;
-        
-        return 1;
-    }
-    else if(str2 == NULL) {
-        if(str1 == NULL)
+    if(str1 == NULL || str2 == NULL) {
+        if(str1 == str2)
             return 0;
         
         return 1;
@@ -86,14 +80,8 @@ size_t myhtml_strncasecmp(const char* str1, const char* str2, size_t size)
 
 size_t myhtml_strcasecmp(const char* str1, const char* str2)
 {
-    if(str1 == NULL) {
-        if(str2 == NULL)
-            return 0;
-        
-        return 1;
-    }
-    else if(str2 == NULL) {
-        if(str1 == NULL)
+    if(str1 == NULL || str2 == NULL) {
+        if(str1 == str2)
             return 0;
         
         return 1;
@@ -119,14 +107,8 @@ size_t myhtml_strcasecmp(const char* str1, const char* str2)
 
 size_t myhtml_strncmp(const char* str1, const char* str2, size_t size)
 {
-    if(str1 == NULL) {
-        if(str2 == NULL)
-            return 0;
-        
-        return 1;
-    }
-    else if(str2 == NULL) {
-        if(str1 == NULL)
+    if(str1 == NULL || str2 == NULL) {
+        if(str1 == str2)
             return 0;
         
         return 1;
@@ -151,14 +133,8 @@ size_t myhtml_strncmp(const char* str1, const char* str2, size_t size)
 
 size_t myhtml_strcmp(const char* str1, const char* str2)
 {
-    if(str1 == NULL) {
-        if(str2 == NULL)
-            return 0;
-        
-        return 1;
-    }
-    else if(str2 == NULL) {
-        if(str1 == NULL)
+    if(str1 == NULL || str2 == NULL) {
+        if(str1 == str2)
             return 0;
         
         return 1;
@@ -184,14 +160,8 @@ size_t myhtml_strcmp(const char* str1, const char* str2)
 
 size_t myhtml_strcmp_ws(const char* str1, const char* str2)
 {
-    if(str1 == NULL) {
-        if(str2 == NULL)
-            return 0;
-        
-        return 1;
-    }
-    else if(str2 == NULL) {
-        if(str1 == NULL)
+    if(str1 == NULL || str2 == NULL) {
+        if(str1 == str2)
             return 0;
         
         return 1;
diff --git a/source/myhtml/utils/mcobject_async.c b/source/myhtml/utils/mcobject_async.c
index c62b4d8..21e36e8 100644
--- a/source/myhtml/utils/mcobject_async.c
+++ b/source/myhtml/utils/mcobject_async.c
@@ -354,13 +354,20 @@ void mcobject_async_node_delete(mcobject_async_t *mcobj_async, size_t node_idx)
     while (chunk)
     {
         if(mcobj_async->chunk_cache_length >= mcobj_async->chunk_cache_size) {
-            mcobj_async->chunk_cache_size <<= 1;
+            size_t new_size = mcobj_async->chunk_cache_size << 1;
             
             mcobject_async_chunk_t **tmp = (mcobject_async_chunk_t**)myhtml_realloc(mcobj_async->chunk_cache,
-                                                                               sizeof(mcobject_async_chunk_t*) * mcobj_async->chunk_cache_size);
+                                                                               sizeof(mcobject_async_chunk_t*) * new_size);
             
-            if(tmp)
+            if(tmp) {
+                mcobj_async->chunk_cache_size = new_size;
                 mcobj_async->chunk_cache = tmp;
+            }
+            else {
+                // TODO: add return status
+                mcsync_unlock(mcobj_async->mcsync);
+                return;
+            }
         }
         
         mcobj_async->chunk_cache[ mcobj_async->chunk_cache_length ] = chunk;
diff --git a/source/myhtml/utils/mctree.h b/source/myhtml/utils/mctree.h
index e3c2b42..bd3054f 100644
--- a/source/myhtml/utils/mctree.h
+++ b/source/myhtml/utils/mctree.h
@@ -45,8 +45,8 @@ extern "C" {
 
 #define mctree_node_add(mctree)                                     \
     mctree->nodes_length++;                                         \
-    if(mctree->nodes_length == mctree->nodes_size) {                \
-        mctree->nodes_size += 4096;                                 \
+    if(mctree->nodes_length >= mctree->nodes_size) {                \
+        mctree->nodes_size = mctree->nodes_length + 4096;           \
         mctree->nodes = (mctree_node_t*)myhtml_realloc(mctree->nodes,    \
             sizeof(mctree_node_t) * mctree->nodes_size);            \
     }                                                               \