1
0
mirror of https://github.com/janet-lang/janet synced 2024-11-28 11:09:54 +00:00

peg replacement functions have access to captures

When peg/replace or peg/replace-all are given a function to serve as the text
replacement, any captures produced by the PEG are passed as additional
arguments to that function.
This commit is contained in:
Ian Henry 2023-04-23 09:09:14 -07:00
parent 485099fd6e
commit 9dc7e8ed3a
No known key found for this signature in database
5 changed files with 56 additions and 21 deletions

View File

@ -1738,7 +1738,7 @@ static Janet cfun_peg_replace_generic(int32_t argc, Janet *argv, int only_one) {
trail = i;
}
int32_t nexti = (int32_t)(result - c.bytes.bytes);
JanetByteView subst = janet_text_substitution(&c.subst, c.bytes.bytes + i, nexti - i);
JanetByteView subst = janet_text_substitution(&c.subst, c.bytes.bytes + i, nexti - i, c.s.captures);
janet_buffer_push_bytes(ret, subst.bytes, subst.len);
trail = nexti;
if (nexti == i) nexti++;
@ -1758,8 +1758,8 @@ JANET_CORE_FN(cfun_peg_replace_all,
"(peg/replace-all peg subst text &opt start & args)",
"Replace all matches of `peg` in `text` with `subst`, returning a new buffer. "
"The peg does not need to make captures to do replacement. "
"If `subst` is a function, it will be called once for each match "
"and should return the actual replacement text to use.") {
"If `subst` is a function, it will be called with the "
"matching text followed by any captures.") {
return cfun_peg_replace_generic(argc, argv, 0);
}
@ -1767,8 +1767,8 @@ JANET_CORE_FN(cfun_peg_replace,
"(peg/replace peg repl text &opt start & args)",
"Replace first match of `peg` in `text` with `subst`, returning a new buffer. "
"The peg does not need to make captures to do replacement. "
"If `subst` is a function, it will be called with the matching text, "
"and should return the actual replacement text to use. "
"If `subst` is a function, it will be called with the "
"matching text followed by any captures. "
"If no matches are found, returns the input string in a new buffer.") {
return cfun_peg_replace_generic(argc, argv, 1);
}

View File

@ -397,7 +397,7 @@ JANET_CORE_FN(cfun_string_replace,
kmp_deinit(&s.kmp);
return janet_stringv(s.kmp.text, s.kmp.textlen);
}
JanetByteView subst = janet_text_substitution(&s.subst, s.kmp.text + result, s.kmp.patlen);
JanetByteView subst = janet_text_substitution(&s.subst, s.kmp.text + result, s.kmp.patlen, NULL);
buf = janet_string_begin(s.kmp.textlen - s.kmp.patlen + subst.len);
safe_memcpy(buf, s.kmp.text, result);
safe_memcpy(buf + result, subst.bytes, subst.len);
@ -422,7 +422,7 @@ JANET_CORE_FN(cfun_string_replaceall,
replacesetup(argc, argv, &s);
janet_buffer_init(&b, s.kmp.textlen);
while ((result = kmp_next(&s.kmp)) >= 0) {
JanetByteView subst = janet_text_substitution(&s.subst, s.kmp.text + result, s.kmp.patlen);
JanetByteView subst = janet_text_substitution(&s.subst, s.kmp.text + result, s.kmp.patlen, NULL);
janet_buffer_push_bytes(&b, s.kmp.text + lastindex, result - lastindex);
janet_buffer_push_bytes(&b, subst.bytes, subst.len);
lastindex = result + s.kmp.patlen;

View File

@ -688,19 +688,32 @@ static JanetByteView to_byte_view(Janet value) {
return result;
}
JanetByteView janet_text_substitution(Janet *subst, const uint8_t *bytes, uint32_t len) {
switch (janet_type(*subst)) {
case JANET_CFUNCTION: {
Janet matched = janet_stringv(bytes, len);
return to_byte_view(janet_unwrap_cfunction(*subst)(1, &matched));
}
case JANET_FUNCTION: {
Janet matched = janet_stringv(bytes, len);
return to_byte_view(janet_call(janet_unwrap_function(*subst), 1, &matched));
}
default:
return memoize_byte_view(subst);
}
JanetByteView janet_text_substitution(
Janet *subst,
const uint8_t *bytes,
uint32_t len,
JanetArray *extra_argv) {
int32_t extra_argc = extra_argv == NULL ? 0 : extra_argv->count;
JanetType type = janet_type(*subst);
switch (type) {
case JANET_FUNCTION:
case JANET_CFUNCTION: {
int32_t argc = 1 + extra_argc;
Janet *argv = janet_tuple_begin(argc);
argv[0] = janet_stringv(bytes, len);
for (int32_t i = 0; i < extra_argc; i++) {
argv[i + 1] = extra_argv->data[i];
}
janet_tuple_end(argv);
if (type == JANET_FUNCTION) {
return to_byte_view(janet_call(janet_unwrap_function(*subst), argc, argv));
} else {
return to_byte_view(janet_unwrap_cfunction(*subst)(argc, argv));
}
}
default:
return memoize_byte_view(subst);
}
}
JanetBinding janet_resolve_ext(JanetTable *env, const uint8_t *sym) {

View File

@ -93,7 +93,11 @@ void janet_buffer_format(
Janet *argv);
Janet janet_next_impl(Janet ds, Janet key, int is_interpreter);
JanetBinding janet_binding_from_entry(Janet entry);
JanetByteView janet_text_substitution(Janet *subst, const uint8_t *bytes, uint32_t len);
JanetByteView janet_text_substitution(
Janet *subst,
const uint8_t *bytes,
uint32_t len,
JanetArray *extra_args);
/* Registry functions */
void janet_registry_put(

View File

@ -340,6 +340,7 @@ neldb\0\0\0\xD8\x05printG\x01\0\xDE\xDE\xDE'\x03\0marshal_tes/\x02
(check-replacer "aba" "" "ababababababa")
(check-replacer "aba" string/ascii-upper "ababababababa")
(check-replacer "aba" 123 "ababababababa")
(assert (= (string (peg/replace-all ~(set "ab") string/ascii-upper "abcaa"))
"ABcAA")
"peg/replace-all cfunction")
@ -347,6 +348,23 @@ neldb\0\0\0\xD8\x05printG\x01\0\xDE\xDE\xDE'\x03\0marshal_tes/\x02
"abcaa")
"peg/replace-all function")
(defn peg-test [name f peg subst text expected]
(assert (= (string (f peg subst text)) expected) name))
(peg-test "peg/replace has access to captures"
peg/replace
~(sequence "." (capture (set "ab")))
(fn [str char] (string/format "%s -> %s, " str (string/ascii-upper char)))
".a.b.c"
".a -> A, .b.c")
(peg-test "peg/replace-all has access to captures"
peg/replace-all
~(sequence "." (capture (set "ab")))
(fn [str char] (string/format "%s -> %s, " str (string/ascii-upper char)))
".a.b.c"
".a -> A, .b -> B, .c")
# Peg bug
(assert (deep= @[] (peg/match '(any 1) @"")) "peg empty pattern 1")
(assert (deep= @[] (peg/match '(any 1) (buffer))) "peg empty pattern 2")