From a8a80f0dc449863c924161223579e1d66a8a8e0e Mon Sep 17 00:00:00 2001
From: Yi Zhang <zhangyiss@icloud.com>
Date: Thu, 2 Jan 2025 19:29:24 +0800
Subject: [PATCH] add filter to dsv_io

---
 example/text_io_ex.cpp |  4 +++
 lib/io/dsv_io.cpp      | 81 ++++++++++++++++++++++++++++++++++++++++++
 lib/io/dsv_io.h        | 20 +++++++++++
 3 files changed, 105 insertions(+)

diff --git a/example/text_io_ex.cpp b/example/text_io_ex.cpp
index d5bd8fd..ed22339 100644
--- a/example/text_io_ex.cpp
+++ b/example/text_io_ex.cpp
@@ -82,6 +82,10 @@ int main(int argc, char const *argv[]) try
     tc.add_row(3);
     tc.fill_row(array<double>{5.5, 4.4, 3.3, 2.2, 1.1}, 3);
 
+    geodsv_io out_table;
+    tc.filter("dist > 1000", {"dist"}, {"x (m)", "y (m)", "elev (m)"}, out_table);
+    out_table.save_text("out2");
+
     _1s_vector s = tc.get_tags();
     s.push_back("Elev = 1000");
     tc.set_tags(s);
diff --git a/lib/io/dsv_io.cpp b/lib/io/dsv_io.cpp
index 977a228..470f9bc 100644
--- a/lib/io/dsv_io.cpp
+++ b/lib/io/dsv_io.cpp
@@ -362,6 +362,28 @@ void gctl::dsv_io::init_table(int row, int col, table_headtype_e t)
     return;
 }
 
+void gctl::dsv_io::init_table(const std::vector<std::vector<std::string> > &str_table, table_headtype_e t)
+{
+    row_num_ = str_table.size();
+    col_num_ = str_table[0].size();
+    
+    // 初始的列头和行头均为空白
+    table_.resize(row_num_ + 1);
+    for (size_t i = 0; i < row_num_ + 1; i++)
+    {
+        table_[i].resize(col_num_ + 1);
+    }
+
+    for (size_t i = 1; i <= row_num_; i++)
+    {
+        for (size_t j = 1; j <= col_num_; j++)
+        {
+            table_[i][j].str_ = str_table[i - 1][j - 1];
+        }
+    }
+    return;
+}
+
 void gctl::dsv_io::info(table_headtype_e t)
 {
     std::clog << "File: " << file_ << "\n------------\n";
@@ -637,6 +659,65 @@ void gctl::dsv_io::cal_column(std::string expr_str, const std::vector<std::strin
     return;
 }
 
+void gctl::dsv_io::filter(std::string cnd_str, const std::vector<std::string> &cnd_col, 
+	const std::vector<std::string> &out_col, dsv_io& out_table)
+{
+    array<int> idx(cnd_col.size());
+    for (size_t i = 0; i < cnd_col.size(); i++)
+    {
+        idx[i] = name_index(cnd_col[i]);
+    }
+
+    array<int> odx(out_col.size());
+    for (size_t i = 0; i < out_col.size(); i++)
+    {
+        odx[i] = name_index(out_col[i]);
+    }
+    
+    exprtk::symbol_table<double> symbol_table;
+	array<double> var(cnd_col.size());
+
+    for (size_t i = 0; i < var.size(); i++)
+	{
+		symbol_table.add_variable(cnd_col[i], var[i]);
+	}
+
+    exprtk::expression<double> expression;
+	expression.register_symbol_table(symbol_table);
+
+	exprtk::parser<double> parser;
+	if (!parser.compile(cnd_str, expression))
+    {
+        throw std::runtime_error("[gctl::dsv_io] Fail to compile the math expression.");
+    }   
+
+    std::vector<std::string> str_line;
+    std::vector<std::vector<std::string> > str_table; 
+
+    for (size_t i = 1; i <= row_num_; i++)
+    {
+        for (size_t j = 0; j < var.size(); j++)
+        {
+            var[j] = table_[i][idx[j]].value<double>();
+        }
+
+        if (expression.value() > 0.5) // return 1 if matched or 0 if dismatched
+        {
+            str_line.clear();
+            for (size_t j = 0; j < odx.size(); j++)
+            {
+                str_line.push_back(table_[i][odx[j]].str_);
+            }
+            
+            str_table.push_back(str_line);
+        }
+    }
+
+    out_table.init_table(str_table);
+    out_table.set_column_names(out_col);
+    return;
+}
+
 #endif // GCTL_EXPRTK
 
 gctl::geodsv_io::geodsv_io(){}
diff --git a/lib/io/dsv_io.h b/lib/io/dsv_io.h
index aeab508..f789d07 100644
--- a/lib/io/dsv_io.h
+++ b/lib/io/dsv_io.h
@@ -352,9 +352,18 @@ namespace gctl
 		 * 
 		 * @param row 数据行数
 		 * @param col 数据列数
+		 * @param t 表格头类型
 		 */
 		void init_table(int row, int col, table_headtype_e t = ColumnHead);
 
+		/**
+		 * @brief 初始化表格
+		 * 
+		 * @param str_table 字符串表格
+		 * @param t 表格头类型
+		 */
+		void init_table(const std::vector<std::vector<std::string> > &str_table, table_headtype_e t = ColumnHead);
+
 		/**
 		 * @brief 返回表格信息
 		 * 
@@ -463,6 +472,17 @@ namespace gctl
 		 */
 		void cal_column(std::string expr_str, const std::vector<std::string> &col_list, int p = 6);
 
+		/**
+		 * @brief 按行过滤并返回符合条件的列数据
+		 * 
+		 * @param cnd_str 条件表达式
+		 * @param cnd_col 用于条件表达式的列索引列表
+		 * @param out_col 输出的列索引列表，即条件判断为真是即筛选这些行与列上对应的数据
+		 * @param out_table 输出的表格
+		 */
+		void filter(std::string cnd_str, const std::vector<std::string> &cnd_col, 
+			const std::vector<std::string> &out_col, dsv_io &out_table);
+
 #endif // GCTL_EXPRTK
 
 		/**