#!/bin/bash
set -euo pipefail

echo "== files =="
git ls-files | rg '(^|/)csv\.rs$|csv' || true

echo "== outline =="
ast-grep outline crates/stdlib/src/csv.rs --view compact || true

echo "== relevant csv lines =="
sed -n '1,380p' crates/stdlib/src/csv.rs | cat -n | sed -n '220,320p'

echo "== search parse_single_char/usages =="
rg -n "parse_single_char|parse_first_char|not a string of length|dialection|Delimiter|QuoteChar|DoubleQuote|error\\(" crates/stdlib/src/csv.rs crates -g '*.rs' || true

#!/bin/bash
set -euo pipefail

echo "== csv.rs relevant section =="
sed -n '220,320p' crates/stdlib/src/csv.rs | cat -n

echo "== parse_single_char/parse_first_char and nearby dialect fields =="
rg -n "pub struct PyDialect|Delimiter|QuoteChar|DoubleQuoteChars|escapechar|lineterminator|skipinitialspace|parse_single_char|parse_first_char|error\\(" crates/stdlib/src/csv.rs -C 3

echo "== focused tests/usages =="
rg -n "sniffer|PyDialect|dialect|delimiter|quotechar|doublequote|escapechar|skipinitialspace|lineterminator|ParserError" crates -g '*.rs' --glob '!crates/vm/src/**' | head -n 200

echo "== CPython docs/source snippets online =="
python3 - <<'PY'
# Behavioral probe: model Rust string semantics around the parsing pattern.
s = "€"
print("single unicode char", len(s), "bytes", len(s.encode('utf-8')))
code_point = ord("€")
print("code point", code_point, "fits in u8", code_point <= 0xff)
s_multi = "€1"
print("multi char chars", len(s_multi))
PY

#!/bin/bash
set -euo pipefail

python3 - <<'PY'
import csv
for attr, names in [
    ("delimiter", ("excel",)),
    ("quotechar", ("excel",)),
    ("escapechar", ("excel",)),
    ("lineterminator", ("excel",)),
]:
    dialect = csv.get_dialect(names[0])
    print(f"csv dialect {attr}: {getattr(dialect, attr)!r!s!U} type={type(getattr(dialect, attr)).__name__!r}")
    for value, param in [
        ("€", attr),
        ("€1", attr[:-1] if attr == "lineterminator" else attr),
        ("", attr),
        ("💙", attr),
        ("💙1", attr),
    ]:
        d = csv.registered_dialects[1]
        try:
            d.update(**{param: value})
            print(f"   OK {attr}={value!r!s!U} len_chars={len(value)}")
        except Exception as e:
            print(f"   {type(e).__name__}: {e}")

print("Python unicode len examples:")
for s in ["€", "💙", "🩵", "€1", "💙1"]:
    print(repr(s), "chars=", len(s), "bytes=", len(s.encode("utf-8")), "ord=", ord(s[0]))
PY

#!/bin/bash
set -euo pipefail

python3 - <<'PY'
import csv
for attr, names in [
    ("delimiter", ("excel",)),
    ("quotechar", ("excel",)),
    ("escapechar", ("excel",)),
    ("lineterminator", ("excel",)),
]:
    dialect = csv.get_dialect(names[0])
    val = getattr(dialect, attr)
    print(f"csv dialect {attr}: {val!r} type={type(val).__name__}")
    for value, param in [
        ("€", attr),
        ("€1", attr if attr != "lineterminator" else attr[:-1]),
        ("", attr),
        ("💙", attr),
        ("💙1", attr if attr != "lineterminator" else attr[:-1]),
    ]:
        try:
            dialect.update(**{param: value})
            print(f"   OK {attr}={value!r} chars={len(value)}")
        except Exception as e:
            print(f"   {type(e).__name__}: {e}")

print("Python unicode len examples:")
for s in ["€", "💙", "🩵", "€1", "💙1"]:
    print(repr(s), "chars=", len(s), "bytes=", len(s.encode("utf-8")), "ord=", ord(s[0]))
PY

#!/bin/bash
set -euo pipefail

repo="RustPython/RustPython"
body=$(cat <<'EOF'
## Summary

Add support for single non-ASCII Unicode code points in CSV dialect character attributes, matching CPython behavior.

## Rationale

`csv` dialect options such as `delimiter`, `quotechar`, and `escapechar` accept exactly one Unicode character in CPython. RustPython currently stores these attributes as `u8` in `crates/stdlib/src/csv.rs`, so valid one-character values outside the byte range (for example, `€`) are rejected.

## Affected area

- `crates/stdlib/src/csv.rs`
- CSV dialect parsing, validation, reader, and writer configuration

## Required changes

Replace the byte-only representation and parsing path for dialect characters with a representation that preserves a full Unicode code point, then propagate that support through the CSV reader and writer implementation. Keep existing validation that rejects empty and multi-character values, while reporting conversion/representation failures accurately.

## Acceptance criteria

- `delimiter`, `quotechar`, and `escapechar` accept a single non-ASCII Unicode code point such as `€`.
- Empty and multi-character strings remain rejected according to CPython-compatible behavior.
- Reader and writer behavior correctly handles the supported Unicode dialect characters.
- Relevant RustPython CSV tests cover successful non-ASCII dialect-character usage and invalid lengths.

## Backlinks

- Follow-up from PR `#8402`: https://github.com/RustPython/RustPython/pull/8402
- Review discussion: https://github.com/RustPython/RustPython/pull/8402#discussion_r3658464157

Requested by `@hyoinandout`.
EOF
)

gh issue create \
  --repo "$repo" \
  --title "csv: support non-ASCII Unicode dialect characters" \
  --body "$body" \
  --assignee "hyoinandout"