With string_view, stripping whitespace can be done without allocs.

This commit is contained in:
Henrik Rydgård 2023-09-26 13:19:19 +02:00
parent 20e703db83
commit 2bd2292bac
4 changed files with 47 additions and 20 deletions

View File

@ -37,7 +37,8 @@ bool StringViewEqualCaseInsensitive(const std::string_view lhs, const std::strin
}
// This unescapes # signs.
static bool ParseLineKey(const std::string &line, size_t &pos, std::string *keyOut) {
// NOTE: These parse functions can make better use of the string_view - the pos argument should not be needed, for example.
static bool ParseLineKey(std::string_view line, size_t &pos, std::string *keyOut) {
std::string key = "";
while (pos < line.size()) {
@ -52,7 +53,8 @@ static bool ParseLineKey(const std::string &line, size_t &pos, std::string *keyO
}
// Escaped.
key += line.substr(pos, next - pos - 1) + "#";
key += line.substr(pos, next - pos - 1);
key.push_back('#');
pos = next + 1;
} else if (line[next] == '=') {
// Hurray, done.
@ -68,11 +70,11 @@ static bool ParseLineKey(const std::string &line, size_t &pos, std::string *keyO
return true;
}
static bool ParseLineValue(const std::string &line, size_t &pos, std::string *valueOut) {
static bool ParseLineValue(std::string_view line, size_t &pos, std::string *valueOut) {
std::string value = "";
std::string strippedLine = StripSpaces(line.substr(pos));
if (strippedLine[0] == '"' && strippedLine[strippedLine.size()-1] == '"') {
std::string_view strippedLine = StripSpaces(line.substr(pos));
if (strippedLine.size() >= 2 && strippedLine[0] == '"' && strippedLine[strippedLine.size() - 1] == '"') {
// Don't remove comment if is surrounded by " "
value += line.substr(pos);
pos = line.npos; // Won't enter the while below
@ -92,7 +94,8 @@ static bool ParseLineValue(const std::string &line, size_t &pos, std::string *va
break;
} else {
// Escaped.
value += line.substr(pos, next - pos - 1) + "#";
value += line.substr(pos, next - pos - 1);
value.push_back('#');
pos = next + 1;
}
}
@ -104,7 +107,7 @@ static bool ParseLineValue(const std::string &line, size_t &pos, std::string *va
return true;
}
static bool ParseLineComment(const std::string& line, size_t &pos, std::string *commentOut) {
static bool ParseLineComment(std::string_view line, size_t &pos, std::string *commentOut) {
// Don't bother with anything if we don't need the comment data.
if (commentOut) {
// Include any whitespace/formatting in the comment.
@ -125,8 +128,7 @@ static bool ParseLineComment(const std::string& line, size_t &pos, std::string *
return true;
}
// Ugh, this is ugly.
static bool ParseLine(const std::string& line, std::string* keyOut, std::string* valueOut, std::string* commentOut)
static bool ParseLine(std::string_view line, std::string* keyOut, std::string* valueOut, std::string* commentOut)
{
// Rules:
// 1. A line starting with ; is commented out.
@ -150,7 +152,7 @@ static bool ParseLine(const std::string& line, std::string* keyOut, std::string*
return true;
}
static std::string EscapeHash(const std::string &value) {
static std::string EscapeHash(std::string_view value) {
std::string result = "";
for (size_t pos = 0; pos < value.size(); ) {
@ -159,7 +161,8 @@ static std::string EscapeHash(const std::string &value) {
result += value.substr(pos);
pos = value.npos;
} else {
result += value.substr(pos, next - pos) + "\\#";
result += value.substr(pos, next - pos);
result += "\\#";
pos = next + 1;
}
}
@ -167,9 +170,13 @@ static std::string EscapeHash(const std::string &value) {
return result;
}
void ParsedIniLine::ParseFrom(const std::string &line) {
// TODO: Skip whitespace.
if (line[0] == '#') {
void ParsedIniLine::ParseFrom(std::string_view line) {
line = StripSpaces(line);
if (line.empty()) {
key.clear();
value.clear();
comment.clear();
} else if (line[0] == '#') {
key.clear();
value.clear();
comment = line;

View File

@ -33,7 +33,7 @@ public:
}
// Comments only come from "ParseFrom".
void ParseFrom(const std::string &line);
void ParseFrom(std::string_view line);
void Reconstruct(std::string *output) const;
// Having these as views allows a more efficient internal representation, like one joint string.

View File

@ -239,18 +239,15 @@ std::string StringFromFormat(const char* format, ...)
return temp;
}
std::string StringFromInt(int value)
{
std::string StringFromInt(int value) {
char temp[16];
snprintf(temp, sizeof(temp), "%d", value);
return temp;
}
// Turns " hej " into "hej". Also handles tabs.
std::string StripSpaces(const std::string &str)
{
std::string StripSpaces(const std::string &str) {
const size_t s = str.find_first_not_of(" \t\r\n");
if (str.npos != s)
return str.substr(s, str.find_last_not_of(" \t\r\n") - s + 1);
else
@ -268,6 +265,25 @@ std::string StripQuotes(const std::string& s)
return s;
}
// Turns " hej " into "hej". Also handles tabs.
std::string_view StripSpaces(std::string_view str) {
const size_t s = str.find_first_not_of(" \t\r\n");
if (str.npos != s)
return str.substr(s, str.find_last_not_of(" \t\r\n") - s + 1);
else
return "";
}
// "\"hello\"" is turned to "hello"
// This one assumes that the string has already been space stripped in both
// ends, as done by StripSpaces above, for example.
std::string_view StripQuotes(std::string_view s) {
if (s.size() && '\"' == s[0] && '\"' == *s.rbegin())
return s.substr(1, s.size() - 2);
else
return s;
}
void SplitString(const std::string& str, const char delim, std::vector<std::string>& output)
{
size_t next = 0;

View File

@ -21,6 +21,7 @@
#include <cstdint>
#include <string>
#include <cstring>
#include <string_view>
#include <vector>
#ifdef _MSC_VER
@ -83,6 +84,9 @@ std::string StringFromInt(int value);
std::string StripSpaces(const std::string &s);
std::string StripQuotes(const std::string &s);
std::string_view StripSpaces(std::string_view s);
std::string_view StripQuotes(std::string_view s);
void SplitString(const std::string& str, const char delim, std::vector<std::string>& output);
void GetQuotedStrings(const std::string& str, std::vector<std::string>& output);