Skip to content

Commit

Permalink
Refactor {escape,remove}ControlCharacters
Browse files Browse the repository at this point in the history
Use the new approach where we avoid adding each character one by one.
  • Loading branch information
zorael committed Jan 27, 2024
1 parent ca253bb commit e1af3dc
Showing 1 changed file with 58 additions and 59 deletions.
117 changes: 58 additions & 59 deletions source/lu/string.d
Original file line number Diff line number Diff line change
Expand Up @@ -1701,58 +1701,53 @@ unittest
Returns:
A new string with control characters escaped, or the original one unchanged.
+/
auto escapeControlCharacters(/*const*/ return scope string line) pure nothrow
string escapeControlCharacters(/*const*/ return scope string line) pure nothrow
{
import std.algorithm.comparison : among;
import std.array : Appender;
import std.string : representation;

if (!line.length) return line;

Appender!(char[]) sink;
bool dirty;
size_t lastEnd;
bool reserved;

immutable asBytes = line.representation;

foreach (immutable i, immutable c; line.representation)
void commitUpTo(const size_t i)
{
if (!dirty)
if (!reserved)
{
if (c.among!('\n', '\t', '\r', '\0'))
{
sink.reserve(line.length);
sink.put(line[0..i]);
dirty = true;

// Drop down into lower dirty switch
}
else
{
continue;
}
sink.reserve(asBytes.length + 16); // guesstimate
reserved = true;
}
sink.put(asBytes[lastEnd..i]);
}

switch (c)
{
case '\n':
sink.put("\\n");
break;

case '\t':
sink.put("\\t");
break;

case '\r':
sink.put("\\r");
break;
for (size_t i; i<asBytes.length; ++i)
{
import std.algorithm.comparison : among;

case '\0':
sink.put("\\0");
break;
if (asBytes[i].among!('\n', '\t', '\r', '\0'))
{
commitUpTo(i);
lastEnd = i+1;

default:
sink.put(c);
break;
switch (asBytes[i])
{
case '\n': sink.put(`\n`); break;
case '\t': sink.put(`\t`); break;
case '\r': sink.put(`\r`); break;
case '\0': sink.put(`\0`); break;
default: break;
}
}
}

return dirty ? sink.data : line;
if (!sink.data.length) return line;

sink.put(asBytes[lastEnd..$]);
return sink.data;
}

///
Expand Down Expand Up @@ -1798,40 +1793,44 @@ unittest
Returns:
A new string with control characters removed, or the original one unchanged.
+/
auto removeControlCharacters(/*const*/ return scope string line) pure nothrow
string removeControlCharacters(/*const*/ return scope string line) pure nothrow
{
import std.array : Appender;
import std.string : representation;

if (!line.length) return line;

Appender!(char[]) sink;
bool dirty;
size_t lastEnd;
bool reserved;

foreach (immutable i, immutable c; line.representation)
immutable asBytes = line.representation;

void commitUpTo(const size_t i)
{
switch (c)
if (!reserved)
{
case '\n':
case '\t':
case '\r':
case '\0':
if (!dirty)
{
sink.reserve(line.length);
sink.put(line[0..i]);
dirty = true;
}
break;
sink.reserve(asBytes.length);
reserved = true;
}
sink.put(asBytes[lastEnd..i]);
}

default:
if (dirty)
{
sink.put(c);
}
break;
for (size_t i; i<asBytes.length; ++i)
{
import std.algorithm.comparison : among;

if (asBytes[i].among!('\n', '\t', '\r', '\0'))
{
commitUpTo(i);
lastEnd = i+1;
}
}

return dirty ? sink.data : line;
if (lastEnd == 0) return line;

sink.put(asBytes[lastEnd..$]);
return sink.data;
}

///
Expand Down

0 comments on commit e1af3dc

Please sign in to comment.