Implement URI Handling Functions

[Razican]

ECMASCript feature We currently lack the implementation of some built-in global functions used for URI encoding/decoding. You can find these functions in the spec here.

The 4 functions to implement would be these:

• encodeURI()
• encodeURIComponent()
• decodeURI()
• decodeURIComponent()

In those links you can find example code that should work after this implementation.

Implementation tips We should probably create a new module here for these functions. An example of a built-in global function implementation can be seen here.

[sidntrivedi012]

Hey @Razican , I would like to work on it. Thanks :)

[Razican]

Hi @sidntrivedi012 how is this going?

[captain-yossarian]

I will be happy to take a look if @sidntrivedi012 don't mind

[tofpie]

Go ahead @captain-yossarian

[captain-yossarian]

@Razican @tofpie Should/Can I use this percent_encoding crrate ?

[Razican]

@Razican @tofpie Should/Can I use this percent_encoding crrate ?

Hi, yes, if it makes sense to use it, feel free :)

[jedel1043]

Unassigned the issue by @captain-yossarian's request. For anyone interested, feel free to ping us if you want to work on this😁

[jtara1]

I'll take a look at this, please assign to me.

[jedel1043]

@jtara1 I've asigned it to you. Let us know if you need some pointers to get started :)

[jtara1]

I got something working with that crate ^ but I'm trying to just follow their algo directly in the ecma spec now

What's the return values for this function? https://github.com/jtara1/boa/blob/8ba500a26afdad8e200c9990b375664b5c04a97a/boa/src/builtins/string/mod.rs#L36

e: looks like @joshwd36 implemented that, what do the return values of (u32, u8, bool) represent for code_point_at func?

[jedel1043]

code_point_at just represents the operation CodePointAt from the spec. But maybe we could rewrite it to be more idiomatic with descriptive enum returns instead of bools and u8s

[jtara1]

the spec answered my question with

Return the Record { [[CodePoint]], [[CodeUnitCount]], [[IsUnpairedSurrogate]] }.

thanks

[jtara1]

So far the hardest part of this is figuring out what the transformation algo (for singles & pairs) for code point -> utf-8 encoded byte(s) is. https://en.wikipedia.org/wiki/UTF-8 which actually seems like the most thorough explanation for this

I'm also checking the node.js implementation of this https://github.com/nodejs/node/blob/e46c680bf2b211bbd52cf959ca17ee98c7f657f5/deps/v8/src/strings/uri.cc#L279

and

void AddEncodedOctetToBuffer(uint8_t octet, std::vector<uint8_t>* buffer) {
  buffer->push_back('%');
  buffer->push_back(HexCharOfValue(octet >> 4));
  buffer->push_back(HexCharOfValue(octet & 0x0F));
}

void EncodeSingle(uc16 c, std::vector<uint8_t>* buffer) {
  char s[4] = {};
  int number_of_bytes;
  number_of_bytes =
      unibrow::Utf8::Encode(s, c, unibrow::Utf16::kNoPreviousCharacter, false);
  for (int k = 0; k < number_of_bytes; k++) {
    AddEncodedOctetToBuffer(s[k], buffer);
  }
}

void EncodePair(uc16 cc1, uc16 cc2, std::vector<uint8_t>* buffer) {
  char s[4] = {};
  int number_of_bytes =
      unibrow::Utf8::Encode(s, unibrow::Utf16::CombineSurrogatePair(cc1, cc2),
                            unibrow::Utf16::kNoPreviousCharacter, false);
  for (int k = 0; k < number_of_bytes; k++) {
    AddEncodedOctetToBuffer(s[k], buffer);
  }
}

e: my answer may lie here https://github.com/nodejs/node/blob/e46c680bf2b211bbd52cf959ca17ee98c7f657f5/deps/v8/src/strings/unicode-inl.h#L144 still sorting through things

[joshwd36]

Would it be worth looking into #736 again? It seems that the implementation will be pretty closely tied to the encoding

[jedel1043]

Would it be worth looking into #736 again? It seems that the implementation will be pretty closely tied to the encoding

Yeah, maybe. We could reevaluate if there's a way to adapt our regex to consume UTF-16, then change the definition of JsString.