From e86518b7167ab67455364bd0ab79df1d8278b070 Mon Sep 17 00:00:00 2001
From: Danila Fedorin <danila.fedorin@gmail.com>
Date: Thu, 19 Jan 2017 19:34:23 -0800
Subject: [PATCH] Write a general header file and a header file for regular
 expressions.

---
 include/liblex.h  |  24 +++++++
 include/pattern.h | 169 ++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 193 insertions(+)
 create mode 100644 include/liblex.h
 create mode 100644 include/pattern.h

diff --git a/include/liblex.h b/include/liblex.h
new file mode 100644
index 0000000..63ee7ba
--- /dev/null
+++ b/include/liblex.h
@@ -0,0 +1,24 @@
+#ifndef LIBLEX_H
+#define LIBLEX_H
+
+/**
+ * Represents a result of a liblex operation.
+ */
+enum liblex_result_e {
+  /**
+   * Represents a succesful operation.
+   */
+  LIBLEX_SUCCESS,
+  /**
+   * Represents a memory allocation failure.
+   */
+  LIBLEX_MALLOC,
+  /**
+   * Represents an invalid pattern.
+   */
+  LIBLEX_INVALID
+};
+
+typedef enum liblex_result_e liblex_result;
+
+#endif
diff --git a/include/pattern.h b/include/pattern.h
new file mode 100644
index 0000000..b512d9d
--- /dev/null
+++ b/include/pattern.h
@@ -0,0 +1,169 @@
+#ifndef LIBLEX_PATTERN_H
+#define LIBLEX_PATTERN_H
+
+#include "liblex.h"
+
+/**
+ * Enum that represents the type
+ * of a NFA node / state.
+ */
+enum pattern_node_type_e {
+  /**
+   * Represents a clear, un-configured node.
+   */
+  PNODE_CLEAR,
+  /**
+   * Represents a single value to be matched.
+   */
+  PNODE_VALUE,
+  /**
+   * Represents a range of values, inclusive.
+   */
+  PNODE_RANGE,
+  /**
+   * Represents any character other than the NULL terminator.
+   */
+  PNODE_ANY,
+  /**
+   * Represents a "connection" between nodes, does not 
+   * match any value.
+   * */
+  PNODE_CONNECT,
+  /**
+   * Represents a fork / divide in possible "next" states.
+   */
+  PNODE_FORK,
+  /**
+   * Represents a succesfully matched pattern.
+   */
+  PNODE_END
+};
+
+/**
+ * Struct representing a single NFA node / state.
+ */
+struct pattern_node_s {
+  /**
+   * The type of this node.
+   */
+  enum pattern_node_type_e type;
+  /**
+   * The id of this node.
+   * The id is unique within the pattern.
+   */
+  int id;
+
+  /**
+   * The node's data that varies based on type.
+   */
+  union {
+    /**
+     * Data for a "value" node.
+     */
+    struct {
+      /**
+       * The value this node matches.
+       */
+      char value;
+      /**
+       * The next state in the pattern.
+       */
+      struct pattern_node_s* next;
+    } value_s;
+
+    /**
+     * Data for a "range" node.
+     */
+    struct {
+      /**
+       * The beginning of the range, inclusive, of matched values.
+       */
+      char from;
+      /**
+       * The end of the range, inclusive, of matched values.
+       */
+      char to;
+      /**
+       * The next state in the pattern.
+       */
+      struct pattern_node_s* next;
+    } range_s;
+
+    /**
+     * Data for an "any" node.
+     */
+    struct {
+      /**
+       * The next state in the pattern.
+       */
+      struct pattern_node_s* next;
+    } any_s;
+
+    /**
+     * Data for a "connect" node.
+     */
+    struct {
+      /**
+       * The next node in the pattern.
+       */
+      struct pattern_node_s* next;
+    } connect_s;
+
+    /**
+     * Data for a "fork" node.
+     */
+    struct {
+      /**
+       * The left node in the fork.
+       */
+      struct pattern_node_s* left;
+      /**
+       * The right node in the fork.
+       */
+      struct pattern_node_s* right;
+    } fork_s;
+
+    /**
+     * Data for an "end" node.
+     */
+    struct {
+      /**
+       * The ID of the pattern that just finished matching.
+       */
+      int pattern_id;
+    } end_s;
+  } data_u;
+};
+
+/**
+ * A chain of pattern nodes.
+ */
+struct pattern_chain_s {
+  /**
+   * The beginning of the chain, its first state.
+   */
+  struct pattern_node_s* head;
+  /**
+   * The end of the chain, its last state.
+   */
+  struct pattern_node_s* tail;
+};
+
+typedef enum pattern_node_type_e pattern_node_type;
+typedef struct pattern_node_s pattern_node;
+typedef struct pattern_chain_s pattern_chain;
+
+/**
+ * Compiles a string representation of a pattern into a pattern NFA,
+ * and stores the first node of the new NFA in root.
+ * @param root the node to store the resulting pattern into.
+ * @param expression the pattern represented as a string.
+ * @return LIBLEX_SUCCESS if all goes well, otherwise some other liblex_result.
+ */
+liblex_result pattern_compile(pattern_node** root, char* expression);
+/**
+ * Frees a pattern NFA allocated by pattern_compile.
+ */
+void pattern_free(pattern_node* root);
+
+#endif