MVMString * MVM_string_decodestream_get_chars(MVMThreadContext *tc, MVMDecodeStream *ds,
MVMint32 chars, MVMint64 eof) {
MVMint32 missing;
/* If we request nothing, give empty string. */
if (chars == 0)
return tc->instance->str_consts.empty;
/* If we don't already have enough chars, try and decode more. */
missing = missing_chars(tc, ds, chars);
ds->result_size_guess = missing;
if (missing)
run_decode(tc, ds, &missing, NULL, DECODE_NOT_EOF);
/* If we've got enough, assemble a string. Otherwise, flag EOF and retry,
* falling back to returning what's available. */
if (missing_chars(tc, ds, chars) == 0) {
return take_chars(tc, ds, chars, 0);
}
else if (eof) {
reached_eof(tc, ds);
return missing_chars(tc, ds, chars) == 0
? take_chars(tc, ds, chars, 0)
: MVM_string_decodestream_get_all(tc, ds);
}
else {
return NULL;
}
}
MVMString * MVM_string_decodestream_get_until_sep(MVMThreadContext *tc, MVMDecodeStream *ds,
MVMDecodeStreamSeparators *sep_spec, MVMint32 chomp) {
MVMint32 sep_loc, sep_length;
/* Look for separator, trying more decoding if it fails. We get the place
* just beyond the separator, so can use take_chars to get what's need.
* Note that decoders are only responsible for finding the final char of
* the separator, so we may need to loop a few times around this. */
sep_loc = find_separator(tc, ds, sep_spec, &sep_length, 0);
while (!sep_loc) {
MVMuint32 decode_outcome = run_decode(tc, ds, NULL, sep_spec, DECODE_NOT_EOF);
if (decode_outcome == RUN_DECODE_NOTHING_DECODED)
break;
if (decode_outcome == RUN_DECODE_STOPPER_REACHED)
sep_loc = find_separator(tc, ds, sep_spec, &sep_length, 0);
}
if (sep_loc) {
/* Use this line length as a guesstimate of the next, unless it's tiny
* in which case we treat it as an outlier (probably an empty line or
* some such). Also round up and to a nice power of 2. */
if (sep_loc > 32)
ds->result_size_guess = (sep_loc << 1) & ~0xF;
return take_chars(tc, ds, sep_loc, chomp ? sep_length : 0);
}
else {
return NULL;
}
}
MVMString * MVM_string_decodestream_get_until_sep(MVMThreadContext *tc, MVMDecodeStream *ds, MVMGrapheme32 sep) {
MVMint32 sep_loc;
/* Look for separator, trying more decoding if it fails. We get the place
* just beyond the separator, so can use take_chars to get what's need. */
sep_loc = find_separator(tc, ds, sep);
if (!sep_loc) {
run_decode(tc, ds, NULL, &sep);
sep_loc = find_separator(tc, ds, sep);
}
if (sep_loc)
return take_chars(tc, ds, sep_loc);
else
return NULL;
}
/* Variant of MVM_string_decodestream_get_until_sep that is called when we
* reach EOF. Trims the final separator if there is one, or returns the last
* line without the EOF marker. */
MVMString * MVM_string_decodestream_get_until_sep_eof(MVMThreadContext *tc, MVMDecodeStream *ds,
MVMDecodeStreamSeparators *sep_spec, MVMint32 chomp) {
MVMint32 sep_loc, sep_length;
/* Decode anything remaining and flush normalization buffer. */
reached_eof(tc, ds);
/* Look for separator, which should by now be at the end, and chomp it
* off if needed. */
sep_loc = find_separator(tc, ds, sep_spec, &sep_length);
if (sep_loc)
return take_chars(tc, ds, sep_loc, chomp ? sep_length : 0);
/* Otherwise, take all remaining chars. */
return MVM_string_decodestream_get_all(tc, ds);
}
MVMString * MVM_string_decodestream_get_chars(MVMThreadContext *tc, MVMDecodeStream *ds, MVMint32 chars) {
MVMint32 missing;
/* If we request nothing, give empty string. */
if (chars == 0)
return tc->instance->str_consts.empty;
/* If we don't already have enough chars, try and decode more. */
missing = missing_chars(tc, ds, chars);
if (missing)
run_decode(tc, ds, &missing, NULL);
/* If we've got enough, assemble a string. Otherwise, give up. */
if (missing_chars(tc, ds, chars) == 0)
return take_chars(tc, ds, chars, 0);
else
return NULL;
}
MVMString * MVM_string_decodestream_get_until_sep(MVMThreadContext *tc, MVMDecodeStream *ds,
MVMDecodeStreamSeparators *sep_spec, MVMint32 chomp) {
MVMint32 sep_loc, sep_length;
/* Look for separator, trying more decoding if it fails. We get the place
* just beyond the separator, so can use take_chars to get what's need.
* Note that decoders are only responsible for finding the final char of
* the separator, so we may need to loop a few times around this. */
sep_loc = find_separator(tc, ds, sep_spec, &sep_length);
while (!sep_loc) {
if (!run_decode(tc, ds, NULL, sep_spec))
break;
sep_loc = find_separator(tc, ds, sep_spec, &sep_length);
}
if (sep_loc)
return take_chars(tc, ds, sep_loc, chomp ? sep_length : 0);
else
return NULL;
}
