From 3a660c1783ea276b2d4540397b63ff15ebf65a78 Mon Sep 17 00:00:00 2001
From: vaclavt <vt@stocksriddle.com>
Date: Tue, 11 Jan 2022 23:34:49 +0100
Subject: [PATCH] csv parsing a bit simplified

---
 csvreader.cpp | 76 +++++++++++++++++++--------------------------------
 csvreader.h   |  4 +--
 2 files changed, 30 insertions(+), 50 deletions(-)

diff --git a/csvreader.cpp b/csvreader.cpp
index 2d6b72a..f5c6754 100644
--- a/csvreader.cpp
+++ b/csvreader.cpp
@@ -18,19 +18,15 @@ CsvReader::CsvReader(bool skip_hdr, char field_sep, char quote_ch, char line_sep
 }
 
 
-int CsvReader::parseCSV(const std::string &filename, std::vector<ColDefNode> &cols_def, Table &table) {
+size_t CsvReader::parseCSVFile(const std::string &filename, std::vector<ColDefNode> &cols_def, Table &table) {
+	size_t row_cnt = 0;
 	std::vector<ColDefNode> cdefs;
 	cdefs.reserve(cols_def.size());
 	for (auto &cd : cols_def) {
 		cdefs.emplace_back(table.get_column_def(cd.name));
 	}
 
-	int row_cnt = 0;
 	bool inQuote(false);
-	std::string field;
-
-	std::vector<std::string> line;
-	line.reserve(32);
 
 	errno = 0;
 	FILE* fp = fopen(filename.c_str(), "r");
@@ -40,67 +36,50 @@ int CsvReader::parseCSV(const std::string &filename, std::vector<ColDefNode> &co
 	char* line_str = NULL;
 	size_t len = 0;
 
-
-	long read_chars;
-	while ((read_chars = getline(&line_str, &len, fp)) != -1) {
-		if (skip_header && !header_skiped) {
-			header_skiped = true;
-			continue;
-		}
-		if (read_chars > 0 && line_str[read_chars - 1] == '\n') {
- 		     line_str[read_chars - 1] = '\0';
- 		     --read_chars;
-  		}
-		std::string csvSource{line_str};
-		
-		std::string::const_iterator aChar = csvSource.begin();
-		std::string::const_iterator strEnd = csvSource.end();
-		while (aChar != strEnd) {
-			if (*aChar == quote_character) {
-				inQuote = !inQuote;
-			} else if (*aChar == field_separator) {
-				if (inQuote) {
-					field += *aChar;
-				} else {
-					line.push_back(field);
-					field.clear();
-				}
-			} else {
-				field.push_back(*aChar);
+	try {
+		long read_chars;
+		while ((read_chars = getline(&line_str, &len, fp)) != -1) {
+			if (skip_header && !header_skiped) {
+				header_skiped = true;
+				continue;
+			}
+			if (read_chars > 0 && line_str[read_chars - 1] == '\n') {
+				line_str[read_chars - 1] = '\0';
+				--read_chars;
 			}
 
-			++aChar;
+			row_cnt += parseCSVString(line_str, cols_def, table);
 		}
 
-		if (!field.empty())
-			line.push_back(field);
+		fclose(fp);
 
-		table.create_row_from_vector(cols_def, line);
-		row_cnt++;
+	} catch (const std::exception &e) {
+		if (line_str)
+			free(line_str);
 
-		field.clear();
-		line.clear();
+		throw e;
 	}
 
-	fclose(fp);
 	if (line_str)
-	    free(line_str);
+		free(line_str);
 
 
 	return row_cnt;
 }
 
-int CsvReader::parseCSV2(const std::string &csvSource, std::vector<ColDefNode> &cols_def, Table& table) {
-	int row_cnt = 0;
+size_t CsvReader::parseCSVString(const std::string &csvSource, std::vector<ColDefNode> &cols_def, Table& table) {
+	size_t row_cnt = 0;
 	bool inQuote(false);
 	bool newLine(false);
-	std::string field;
 
 	std::vector<std::string> line;
-	line.reserve(32);
+	std::string field;
+	line.reserve(256);
+	field.reserve(64);
 
 	std::string::const_iterator aChar = csvSource.begin();
-	while (aChar != csvSource.end()) {
+	std::string::const_iterator aEnd = csvSource.end();
+	while (aChar != aEnd) {
 		if (*aChar == quote_character) {
 			newLine = false;
 			inQuote = !inQuote;
@@ -136,7 +115,8 @@ int CsvReader::parseCSV2(const std::string &csvSource, std::vector<ColDefNode> &
 		aChar++;
 	}
 
-	if (!field.empty()) line.push_back(field);
+	if (!field.empty())
+		line.push_back(field);
 
 	if (header_skiped) {
 		table.create_row_from_vector(cols_def, line);
diff --git a/csvreader.h b/csvreader.h
index 1778437..5f5c8e6 100644
--- a/csvreader.h
+++ b/csvreader.h
@@ -26,9 +26,9 @@ namespace usql {
     public:
 	explicit CsvReader(bool skip_hdr = true, char field_sep = ',', char quote_ch = '"', char line_sep = '\r', char line_sep2 = '\n');
 
-	int parseCSV2(const std::string &csvSource, std::vector<ColDefNode> &cols_def, Table& table);
+	size_t parseCSVString(const std::string &csvSource, std::vector<ColDefNode> &cols_def, Table& table);
 
-	int parseCSV(const std::string &filename, std::vector<ColDefNode> &cols_def, Table& table);
+	size_t parseCSVFile(const std::string &filename, std::vector<ColDefNode> &cols_def, Table& table);
 
     };