Elaborate the construction of state machine; add states diagram.

3 files changed, 44 insertions, 1 deletions

diff --git a/doc/diagram/iterative-parser-states-diagram.dot b/doc/diagram/iterative-parser-states-diagram.dot
new file mode 100644
index 00000000..57daa0db
--- /dev/null
+++ b/doc/diagram/iterative-parser-states-diagram.dot
@@ -0,0 +1,37 @@
+digraph {
+    splines=true;
+    node [shape = doublecircle]; Start; Finish;
+    node [shape = circle];
+
+    Start -> ArrayInitial [label="["];
+    Start -> ObjectInitial [label="{"];
+
+    ObjectInitial -> ObjectFinish [label="}"];
+    ObjectInitial -> MemberKey [label="string"];
+
+    MemberKey -> KeyValueDelimiter [label=":"];
+
+    KeyValueDelimiter -> ArrayInitial [label="[ (push MemberValue)"];
+    KeyValueDelimiter -> ObjectInitial [label="{ (push MemberValue)"];
+    KeyValueDelimiter -> MemberValue [label="string|false|true|null|number"];
+
+    MemberValue -> ObjectFinish [label="}"];
+    MemberValue -> MemberDelimiter [label=","];
+
+    MemberDelimiter -> MemberKey [label="string"];
+
+    ArrayInitial -> ArrayInitial [label="[ (push Element)"];
+    ArrayInitial -> ArrayFinish [label="]"];
+    ArrayInitial -> ObjectInitial [label="{ (push Element)"];
+    ArrayInitial -> Element [label="string|flase|true|null|number"];
+
+    Element -> ArrayFinish [label="]"];
+    Element -> ElementDelimiter [label=","];
+
+    ElementDelimiter -> ArrayInitial [label="[ (push Element)"];
+    ElementDelimiter -> ObjectInitial [label="{ (push Element)"];
+    ElementDelimiter -> Element [label="string|false|true|null|number"];
+
+    ArrayFinish -> Finish;
+    ObjectFinish -> Finish;
+}
diff --git a/doc/diagram/iterative-parser-states-diagram.png b/doc/diagram/iterative-parser-states-diagram.png
new file mode 100644
index 00000000..d209514a
--- /dev/null
+++ b/doc/diagram/iterative-parser-states-diagram.png
diff --git a/doc/internals.md b/doc/internals.md
index 85fcf338..fd68b1f2 100644
--- a/doc/internals.md
+++ b/doc/internals.md
@@ -112,10 +112,16 @@ while generating a production could work.
 In RapidJSON, several modifications(or adaptations to current design) are made to a direct implementation.
 
 First, the parsing table is encoded in a state machine in RapidJSON.
-Extra states are added for productions involved with `array` and `object`.
+States are constructed by the head and body of production.
+State transitions are constructed by production rules.
+Besides, extra states are added for productions involved with `array` and `object`.
 In this way the generation of array values or object members would be a single state transition,
 rather than several pop/push operations in the direct implementation.
 This also makes the estimation of stack size more easier.
 
+The final states diagram is shown below:
+
+![States Diagram](diagram/iterative-parser-states-diagram.png)
+
 Second, the iterative parser also keeps track of array's value count and object's member count
 in its internal stack, which may be different from a conventional implementation.