fix: lex_string flush bug ignores characters at start of input

[loremmoqi]

Bug: lex_string flush ignores characters at start of input

Description

In lex_string, the flush function was originally written as:

if start > 0 && end > start {
    buf.write_substring(ctx.input, start, end - start)
}

When parsing a string literal that starts at the beginning of the input (ctx.offset = 0), the first call to flush does not add any characters to the buffer, causing the returned string to be empty. This issue only occurs when the string starts at the beginning of the input or when start = 0 during the first flush.

Fix

Remove the start > 0 condition and use:

if end > start {
    buf.write_substring(ctx.input, start, end - start)
}

This ensures that characters at the start of the string are correctly added to the buffer.

[peter-jerry-ye-code-review]

Fix for string lexing bug that ignores characters at start of input

Category
Correctness
Code Snippet
Line 20: if start > 0 && end > start {
buf.write_substring(ctx.input, start, end - start)
}
Recommendation
Change condition to: if end > start {
buf.write_substring(ctx.input, start, end - start)
}
Reasoning
The original condition start > 0 prevents flushing characters when parsing starts at the beginning of input (offset 0), causing the lexer to skip the initial characters of string literals that appear at the start of the input buffer.

[peter-jerry-ye]

Can you please construct a test case?

[loremmoqi]

Can you please construct a test case?
Sure,I have added the test.
But I don't know why the nightly-builds of three platforms including window failed.

[peter-jerry-ye]

You added a test using internal API. What I would like to see is a test using external API because the "bug" you mentioned could have been an undocumented invariant.