C++ification of bml parser.

bml.cpp

@@ -1,5 +1,7 @@
 #include <vector>
-#include <string.h>
+#include <iostream>
+#include <fstream>
+#include <stack>
 #include <stdio.h>
 
 #include "port.h"
@@ -21,6 +23,11 @@ static inline int isblank(char c)
     return (c == ' ' || c == '\t');
 }
 
+static inline int isblankorlf(char c)
+{
+    return (islf(c) || isblank(c));
+}
+
 static inline int isalnum(char c)
 {
     return ((c >= 'a' && c <= 'z') ||
@@ -28,173 +35,151 @@ static inline int isalnum(char c)
             (c >= '0' && c <= '9'));
 }
 
-static inline int bml_valid (char c)
+static inline int bml_valid(char c)
 {
-    return (isalnum (c) || c == '-');
+    return (isalnum(c) || c == '-');
 }
 
 static std::string trim(std::string str)
 {
     int start;
     int end;
-    for (start = 0; str[start] && start != (int)str.length() && isblank (str[start]); start++) {}
+    for (start = 0; str[start] && start != (int)str.length() && isblank(str[start]); start++) {}
     if (start >= (int)str.length())
         return std::string("");
-    for (end = str.length() - 1; isblank (str[end]) || str[end] == '\n' || str[end] == '\r'; end--) {}
+    for (end = str.length() - 1; isblank(str[end]) || str[end] == '\n' || str[end] == '\r'; end--) {}
     return str.substr(start, end - start + 1);
 }
 
-static inline unsigned int bml_read_depth(char *data)
+static std::string trimcomments(std::string str)
 {
-    unsigned int depth;
+    int end = str.length();
-    for (depth = 0; isblank(data[depth]); depth++) {}
+    size_t comment = str.find("//");
-    return depth;
+    if (comment != std::string::npos)
+        end = comment;
+
+    for (int i = end - 1; i >= 0; i--)
+    {
+        if (!isblankorlf(str[i]))
+        {
+            end = i + 1;
+            break;
+        }
+    }
+
+    return str.substr(0, end);
 }
 
-static void bml_parse_depth(bml_node &node, char **data)
+static inline int bml_read_depth(std::string &data)
 {
-    unsigned int depth = bml_read_depth(*data);
+    size_t depth;
-    *data += depth;
+    for (depth = 0; isblank(data[depth]) && depth < data.length(); depth++) {}
+    return depth == data.length() ? -1 : depth;
+}
+
+static void bml_parse_depth(bml_node &node, std::string &line)
+{
+    unsigned int depth = bml_read_depth(line);
+    line.erase(0, depth);
     node.depth = depth;
 }
 
-static void bml_parse_name(bml_node &node, char **data)
+static void bml_parse_name(bml_node &node, std::string &line)
 {
     int len;
 
-    for (len = 0; bml_valid(*(*data + len)); len++) {};
+    for (len = 0; bml_valid(line[len]); len++) {};
 
-    node.name = trim(std::string(*data, len));
+    node.name = trim(line.substr(0, len));
-    *data += len;
+    line.erase(0, len);
 }
 
-static void bml_parse_data(bml_node &node, char **data)
+static void bml_parse_data(bml_node &node, std::string &line)
 {
-    char *p = *data;
     int len;
 
-    if (p[0] == '=' && p[1] == '\"')
+    if (line[0] == '=' && line[1] == '\"')
     {
         len = 2;
-        while (p[len] && p[len] != '\"' && !islf(p[len]))
+        while (line[len] && line[len] != '\"' && !islf(line[len]))
             len++;
-        if (p[len] != '\"')
+        if (line[len] != '\"')
             return;
 
-        node.data = std::string(p + 2, len - 2);
+        node.data = line.substr(2, len - 2);
-        *data += len + 1;
+        line.erase(0, len + 1);
     }
-    else if (*p == '=')
+    else if (line[0] == '=')
     {
         len = 1;
-        while (p[len] && !islf(p[len]) && p[len] != '"' && p[len] != ' ')
+        while (line[len] && !islf(line[len]) && line[len] != '"' && line[len] != ' ')
             len++;
-        if (p[len] == '\"')
+        if (line[len] == '\"')
             return;
-        node.data = std::string(p + 1, len - 1);
+        node.data = line.substr(1, len - 1);
-        *data += len;
+        line.erase(0, len);
     }
-    else if (*p == ':')
+    else if (line[0] == ':')
     {
         len = 1;
-        while (p[len] && !islf(p[len]))
+        while (line[len] && !islf(line[len]))
             len++;
-        node.data = std::string(p + 1, len - 1);
+        node.data = line.substr(1, len - 1);
-        *data += len;
+        line.erase(0, len);
     }
 
     return;
 }
 
-static void bml_skip_empty(char **data)
+static std::string bml_read_line(std::ifstream &fd)
 {
-    char *p = *data;
+    std::string line;
 
-    while (*p)
+    while (fd)
     {
-        for (; *p && isblank (*p) ; p++) {}
+        std::getline(fd, line);
-
+        line = trimcomments(line);
-        if (!islf(p[0]) && (p[0] != '/' && p[1] != '/'))
+        if (!line.empty())
-            return;
-
-        /* Skip comment data */
-        while (*p && *p != '\r' && *p != '\n')
-            p++;
-
-        /* If we found new-line, try to skip more */
-        if (*p)
         {
-            p++;
-            *data = p;
-        }
-    }
-}
-
-static char *bml_read_line (char **data)
-{
-    char *line;
-    char *p;
-
-    bml_skip_empty (data);
-
-    line = *data;
-
-    if (line == NULL || *line == '\0')
-        return NULL;
-
-    p = strpbrk (line, "\r\n\0");
-
-    if (p == NULL)
-        return NULL;
-
-    if (islf (*p))
-    {
-        *p = '\0';
-        p++;
-    }
-
-    *data = p;
-
             return line;
+        }
+    }
+
+    return std::string("");
 }
 
-static void bml_parse_attr(bml_node &node, char **data)
+static void bml_parse_attr(bml_node &node, std::string &line)
 {
-    char *p = *data;
     int len;
 
-    while (*p && !islf(*p))
+    while (line.length() > 0)
     {
-        if (*p != ' ')
+        if (!isblank(line[0]))
             return;
 
-        while (isblank(*p))
+        while (isblank(line[0]))
-            p++;
+            line.erase(0, 1);
-        if (p[0] == '/' && p[1] == '/')
-            break;
 
         bml_node n;
         len = 0;
-        while (bml_valid(p[len]))
+        while (bml_valid(line[len]))
             len++;
         if (len == 0)
             return;
-        n.name = trim(std::string(p, len));
+        n.name = trim(line.substr(0, len));
-        p += len;
+        line.erase(0, len);
-        bml_parse_data(n, &p);
+        bml_parse_data(n, line);
         n.depth = node.depth + 1;
-        n.type = bml_node::CHILD;
+        n.type = bml_node::ATTRIBUTE;
         node.child.push_back(n);
     }
-
-    *data = p;
 }
 
-static int contains_space (const char *str)
+static int contains_space(const char *str)
 {
     for (int i = 0; str[i]; i++)
     {
-        if (isblank (str[i]))
+        if (isblank(str[i]))
             return 1;
     }
 
@@ -207,44 +192,44 @@ static void bml_print_node(bml_node &node, int depth)
 
     for (i = 0; i < depth * 2; i++)
     {
-        printf (" ");
+        printf(" ");
     }
 
     if (!node.name.empty())
-        printf ("%s", node.name.c_str());
+        printf("%s", node.name.c_str());
 
     if (!node.data.empty())
     {
         if (contains_space(node.data.c_str()))
-            printf ("=\"%s\"", node.data.c_str());
+            printf("=\"%s\"", node.data.c_str());
         else
-            printf (": %s", node.data.c_str());
+            printf(": %s", node.data.c_str());
     }
-    for (i = 0; i < (int) node.child.size () && node.child[i].type == bml_node::CHILD; i++)
+    for (i = 0; i < (int)node.child.size() && node.child[i].type == bml_node::ATTRIBUTE; i++)
     {
         if (!node.child[i].name.empty())
         {
-            printf (" %s", node.child[i].name.c_str());
+            printf(" %s", node.child[i].name.c_str());
             if (!node.child[i].data.empty())
             {
                 if (contains_space(node.child[i].data.c_str()))
-                    printf ("=\"%s\"", node.child[i].data.c_str());
+                    printf("=\"%s\"", node.child[i].data.c_str());
                 else
-                    printf ("=%s", node.child[i].data.c_str());
+                    printf("=%s", node.child[i].data.c_str());
             }
         }
     }
 
     if (depth >= 0)
-        printf ("\n");
+        printf("\n");
 
-    for (; i < (int) node.child.size(); i++)
+    for (; i < (int)node.child.size(); i++)
     {
-        bml_print_node (node.child[i], depth + 1);
+        bml_print_node(node.child[i], depth + 1);
     }
 
     if (depth == 0)
-        printf ("\n");
+        printf("\n");
 }
 
 void bml_node::print()
@@ -252,53 +237,39 @@ void bml_node::print()
     bml_print_node(*this, -1);
 }
 
-static bml_node bml_parse_node(char **doc)
+void bml_node::parse(std::ifstream &fd)
 {
-    char *line;
+    std::stack<bml_node *> nodestack;
-    bml_node node;
+    nodestack.push(this);
 
-    if ((line = bml_read_line(doc)))
+    while (fd)
     {
-        bml_parse_depth(node, &line);
+        bml_node newnode;
-        bml_parse_name(node, &line);
+        std::string line = bml_read_line(fd);
-        bml_parse_data(node, &line);
+        if (line.empty())
-        bml_parse_attr(node, &line);
+            return;
-    }
-    else
-        return node;
 
-    bml_skip_empty(doc);
+        int line_depth = bml_read_depth(line);
-    while (*doc && (int)bml_read_depth(*doc) > node.depth)
+        while (line_depth <= nodestack.top()->depth && nodestack.size() > 1)
-    {
+            nodestack.pop();
-        bml_node child = bml_parse_node(doc);
 
-        if (child.depth != -1)
+        bml_parse_depth(newnode, line);
-            node.child.push_back(child);
+        bml_parse_name(newnode, line);
+        bml_parse_data(newnode, line);
+        bml_parse_attr(newnode, line);
 
-        bml_skip_empty(doc);
+        nodestack.top()->child.push_back(newnode);
-    }
+        nodestack.push(&nodestack.top()->child.back());
-
-    return node;
-}
-
-void bml_node::parse(char *doc)
-{
-    bml_node node;
-    char *ptr = doc;
-
-    while ((node = bml_parse_node (&ptr)).depth != -1)
-    {
-        child.push_back(node);
     }
 
     return;
 }
 
-bml_node *bml_node::find_subnode (std::string name)
+bml_node *bml_node::find_subnode(std::string name)
 {
     unsigned int i;
 
-    for (i = 0; i < child.size (); i++)
+    for (i = 0; i < child.size(); i++)
     {
         if (name.compare(child[i].name) == 0)
             return &child[i];
@@ -307,28 +278,14 @@ bml_node *bml_node::find_subnode (std::string name)
     return NULL;
 }
 
-bool bml_node::parse_file(const char *filename)
+bool bml_node::parse_file(std::string filename)
 {
-    FILE *file = NULL;
+    std::ifstream file(filename, std::ios_base::binary);
-    char *buffer = NULL;
-    int file_size = 0;
-
-    file = fopen(filename, "rb");
 
     if (!file)
         return false;
 
-    fseek(file, 0, SEEK_END);
+    parse(file);
-    file_size = ftell(file);
-    fseek(file, 0, SEEK_SET);
-
-    buffer = new char[file_size + 1];
-    fread(buffer, file_size, 1, file);
-    buffer[file_size] = '\0';
-    fclose(file);
-
-    parse(buffer);
-    delete[] buffer;
 
     return true;
 }

bml.h

@@ -2,6 +2,7 @@
 #define __BML_H
 #include <vector>
 #include <string>
+#include <fstream>
 
 struct bml_node
 {
@@ -11,8 +12,8 @@ struct bml_node
     };
 
     bml_node();
-    bool parse_file(const char *filename);
+    bool parse_file(std::string filename);
-    void parse(char *buffer);
+    void parse(std::ifstream &fd);
     bml_node *find_subnode(std::string name);
     void print();