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;
}
}
/* 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);
}
/* Decodes all the buffers, producing a string containing all the decoded
* characters. */
MVMString * MVM_string_decodestream_get_all(MVMThreadContext *tc, MVMDecodeStream *ds) {
MVMString *result = (MVMString *)MVM_repr_alloc_init(tc, tc->instance->VMString);
result->body.storage_type = MVM_STRING_GRAPHEME_32;
/* Decode anything remaining and flush normalization buffer. */
reached_eof(tc, ds);
/* If there's no codepoint buffer, then return the empty string. */
if (!ds->chars_head) {
result->body.storage.blob_32 = NULL;
result->body.num_graphs = 0;
}
/* If there's exactly one resulting codepoint buffer and we swallowed none
* of it, just use it. */
else if (ds->chars_head == ds->chars_tail && ds->chars_head_pos == 0) {
/* Set up result string. */
result->body.storage.blob_32 = ds->chars_head->chars;
result->body.num_graphs = ds->chars_head->length;
/* Don't free the buffer's memory itself, just the holder, as we
* stole that for the buffer into the string above. */
MVM_free(ds->chars_head);
ds->chars_head = ds->chars_tail = NULL;
}
/* Otherwise, need to assemble all the things. */
else {
/* Calculate length. */
MVMint32 length = 0, pos = 0;
MVMDecodeStreamChars *cur_chars = ds->chars_head;
while (cur_chars) {
if (cur_chars == ds->chars_head)
length += cur_chars->length - ds->chars_head_pos;
else
length += cur_chars->length;
cur_chars = cur_chars->next;
}
/* Allocate a result buffer of the right size. */
result->body.storage.blob_32 = MVM_malloc(length * sizeof(MVMGrapheme32));
result->body.num_graphs = length;
/* Copy all the things into the target, freeing as we go. */
cur_chars = ds->chars_head;
while (cur_chars) {
if (cur_chars == ds->chars_head) {
MVMint32 to_copy = ds->chars_head->length - ds->chars_head_pos;
memcpy(result->body.storage.blob_32 + pos, cur_chars->chars + ds->chars_head_pos,
cur_chars->length * sizeof(MVMGrapheme32));
pos += to_copy;
}
else {
memcpy(result->body.storage.blob_32 + pos, cur_chars->chars,
cur_chars->length * sizeof(MVMGrapheme32));
pos += cur_chars->length;
}
cur_chars = cur_chars->next;
}
ds->chars_head = ds->chars_tail = NULL;
}
return result;
}
/* Decodes all the buffers, signals EOF to flush any normalization buffers, and
* returns a string of all decoded chars. */
MVMString * MVM_string_decodestream_get_all(MVMThreadContext *tc, MVMDecodeStream *ds) {
reached_eof(tc, ds);
return get_all_in_buffer(tc, ds);
}
