From e769d0a00230266eb9f140128840aae911dee1cc Mon Sep 17 00:00:00 2001
From: yihaoDeng <luomoxyz@126.com>
Date: Mon, 22 Nov 2021 18:40:44 +0800
Subject: [PATCH] refactor builder struct

---
 source/libs/index/inc/index_fst.h      |  9 ++++++++-
 source/libs/index/inc/index_fst_util.h |  6 +++---
 source/libs/index/src/index_fst.c      | 22 ++++++++++++++++++----
 source/libs/index/src/index_fst_util.c | 10 ++++------
 4 files changed, 33 insertions(+), 14 deletions(-)

diff --git a/source/libs/index/inc/index_fst.h b/source/libs/index/inc/index_fst.h
index cb60ef7cb8..86346f64ff 100644
--- a/source/libs/index/inc/index_fst.h
+++ b/source/libs/index/inc/index_fst.h
@@ -36,6 +36,8 @@ typedef struct FstRange {
 typedef enum { OneTransNext, OneTrans, AnyTrans, EmptyFinal} State;
 typedef enum { Included, Excluded, Unbounded} FstBound; 
 
+typedef enum {Ordered, OutOfOrdered, DuplicateKey} OrderType;
+
 
 
 /*
@@ -66,11 +68,16 @@ typedef struct FstBuilder {
   FstCountingWriter  *wrt;         // The FST raw data is written directly to `wtr`.  
   FstUnFinishedNodes *unfinished; // The stack of unfinished nodes   
   FstRegistry*        registry;    // A map of finished nodes.        
-  SArray*            last;        // The last word added 
+  FstSlice            last;        // The last word added 
   CompiledAddr       lastAddr;    // The address of the last compiled node  
   uint64_t           len;         // num of keys added
 } FstBuilder;
 
+FstBuilder *fstBuilderCreate(void *w, FstType ty);
+OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup);
+CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn);
+
+
 
 typedef struct FstTransitions {
   FstNode    *node; 
diff --git a/source/libs/index/inc/index_fst_util.h b/source/libs/index/inc/index_fst_util.h
index 6410df6d61..36639af6f3 100644
--- a/source/libs/index/inc/index_fst_util.h
+++ b/source/libs/index/inc/index_fst_util.h
@@ -70,11 +70,11 @@ CompiledAddr unpackDelta(char *data, uint64_t len, uint64_t nodeAddr);
 typedef struct FstSlice {
   uint8_t *data; 
   uint64_t dLen;
-  uint32_t start;
-  uint32_t end;
+  int32_t start;
+  int32_t end;
 } FstSlice;
 
-FstSlice fstSliceCopy(FstSlice *slice, uint32_t start, uint32_t end);
+FstSlice fstSliceCopy(FstSlice *slice, int32_t start, int32_t end);
 FstSlice fstSliceCreate(uint8_t *data, uint64_t dLen);
 bool fstSliceEmpty(FstSlice *slice);
 
diff --git a/source/libs/index/src/index_fst.c b/source/libs/index/src/index_fst.c
index 22cfe76de4..6206180668 100644
--- a/source/libs/index/src/index_fst.c
+++ b/source/libs/index/src/index_fst.c
@@ -300,16 +300,30 @@ FstBuilder *fstBuilderCreate(void *w, FstType ty) {
   b->wrt = fstCountingWriterCreate(w);
   b->unfinished = fstUnFinishedNodesCreate();   
   b->registry   = fstRegistryCreate(10000, 2) ;
-  b->last       = NULL;
+  b->last       = fstSliceCreate(NULL, 0);
   b->lastAddr   = NONE_ADDRESS; 
   b->len        = 0;
   return b;
 }
 
 
-bool fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDupe) {
-      
-  return true;
+
+OrderType fstBuilderCheckLastKey(FstBuilder *b, FstSlice bs, bool ckDup) {
+  FstSlice *input = &bs;
+  if (fstSliceEmpty(&b->last)) {
+    // deep copy or not
+    b->last = fstSliceCopy(&bs, input->start, input->end);
+  } else {
+    int comp = fstSliceCompare(&b->last, &bs);
+    if (comp == 0 && ckDup) {
+      return DuplicateKey;  
+    } else if (comp == 1) {
+      return OutOfOrdered;
+    }
+    // deep copy or not
+    b->last = fstSliceCopy(&bs, input->start, input->end); 
+  }       
+  return Ordered;
 } 
 CompiledAddr fstBuilderCompile(FstBuilder *b, FstBuilderNode *bn) {
   if (FST_BUILDER_NODE_IS_FINAL(bn) 
diff --git a/source/libs/index/src/index_fst_util.c b/source/libs/index/src/index_fst_util.c
index 23e7056c64..8e0a104b5f 100644
--- a/source/libs/index/src/index_fst_util.c
+++ b/source/libs/index/src/index_fst_util.c
@@ -94,7 +94,7 @@ FstSlice fstSliceCreate(uint8_t *data, uint64_t dLen) {
   FstSlice slice = {.data = data, .dLen = dLen, .start = 0, .end = dLen - 1};
   return slice;
 } 
-FstSlice fstSliceCopy(FstSlice *slice, uint32_t start, uint32_t end) {
+FstSlice fstSliceCopy(FstSlice *slice, int32_t start, int32_t end) {
   FstSlice t;
   if (start >= slice->dLen || end >= slice->dLen || start > end) {
     t.data = NULL;
@@ -111,12 +111,10 @@ bool fstSliceEmpty(FstSlice *slice) {
   return slice->data == NULL || slice->dLen <= 0;
 }
 
-
-
 int fstSliceCompare(FstSlice *a, FstSlice *b) {
-  uint32_t aLen = (a->end - a->start + 1);  
-  uint32_t bLen = (b->end - b->start + 1);
-  uint32_t mLen = (aLen < bLen ? aLen : bLen);
+  int32_t aLen = (a->end - a->start + 1);  
+  int32_t bLen = (b->end - b->start + 1);
+  int32_t mLen = (aLen < bLen ? aLen : bLen);
   for (int i = 0; i < mLen; i++) {
     uint8_t x = a->data[i + a->start];
     uint8_t y = b->data[i + b->start];