// ----------------------------------------------------------------
lrec_reader_t* lrec_reader_csvex_alloc(byte_reader_t* pbr, char irs, char ifs) {
lrec_reader_t* plrec_reader = mlr_malloc_or_die(sizeof(lrec_reader_t));
lrec_reader_csvex_state_t* pstate = mlr_malloc_or_die(sizeof(lrec_reader_csvex_state_t));
pstate->ilno = 0LL;
pstate->irs = "\r\n"; // xxx multi-byte the cli irs/ifs/etc, and integrate here
pstate->ifs = ","; // xxx multi-byte the cli irs/ifs/etc, and integrate here
pstate->dquote_irs = mlr_paste_2_strings("\"", pstate->irs);
pstate->dquote_ifs = mlr_paste_2_strings("\"", pstate->ifs);
pstate->dquote_eof = "\"\xff";
pstate->dquote = "\"";
pstate->dquote_dquote = "\"\"";
pstate->ifs_eof = mlr_paste_2_strings(pstate->ifs, "\xff");
pstate->irs_len = strlen(pstate->irs);
pstate->ifs_len = strlen(pstate->ifs);
pstate->dquote_irs_len = strlen(pstate->dquote_irs);
pstate->dquote_ifs_len = strlen(pstate->dquote_ifs);
pstate->dquote_eof_len = strlen(pstate->dquote_eof);
pstate->dquote_len = strlen(pstate->dquote);
pstate->dquote_dquote_len = strlen(pstate->dquote_dquote);
pstate->ifs_eof_len = strlen(pstate->ifs_eof);
pstate->peek_buf_len = pstate->irs_len;
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->ifs_len);
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->dquote_irs_len);
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->dquote_ifs_len);
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->dquote_eof_len);
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->dquote_len);
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->dquote_dquote_len);
pstate->peek_buf_len = mlr_imax2(pstate->peek_buf_len, pstate->ifs_eof_len);
pstate->peek_buf_len += 2;
sb_init(&pstate->sb, STRING_BUILDER_INIT_SIZE);
pstate->psb = &pstate->sb;
pstate->pbr = pbr;
pstate->pfr = NULL;
// xxx allocate the parse-tries here -- one for dquote only,
// the second for non-dquote-after-that, the third for dquoted-after-that.
pstate->expect_header_line_next = TRUE;
pstate->pheader_keeper = NULL;
pstate->pheader_keepers = lhmslv_alloc();
plrec_reader->pvstate = (void*)pstate;
plrec_reader->popen_func = &file_reader_stdio_vopen;
plrec_reader->pclose_func = &file_reader_stdio_vclose;
plrec_reader->pprocess_func = &lrec_reader_csvex_process;
plrec_reader->psof_func = &lrec_reader_csvex_sof;
plrec_reader->pfree_func = &lrec_reader_csvex_free;
return plrec_reader;
}
// ----------------------------------------------------------------
static char * test_paste() {
mu_assert("error: paste 2", streq(mlr_paste_2_strings("ab", "cd"), "abcd"));
mu_assert("error: paste 3", streq(mlr_paste_3_strings("ab", "cd", "ef"), "abcdef"));
mu_assert("error: paste 4", streq(mlr_paste_4_strings("ab", "cd", "ef", "gh"), "abcdefgh"));
mu_assert("error: paste 5", streq(mlr_paste_5_strings("ab", "cd", "ef", "gh", "ij"), "abcdefghij"));
return 0;
}
// ----------------------------------------------------------------
lrec_reader_t* lrec_reader_mmap_csv_alloc(char* irs, char* ifs, int use_implicit_header) {
lrec_reader_t* plrec_reader = mlr_malloc_or_die(sizeof(lrec_reader_t));
lrec_reader_mmap_csv_state_t* pstate = mlr_malloc_or_die(sizeof(lrec_reader_mmap_csv_state_t));
pstate->ilno = 0LL;
pstate->eof = "\xff";
pstate->irs = irs;
pstate->ifs = ifs;
pstate->ifs_eof = mlr_paste_2_strings(pstate->ifs, "\xff");
pstate->dquote = "\"";
pstate->dquote_irs = mlr_paste_2_strings("\"", pstate->irs);
pstate->dquote_ifs = mlr_paste_2_strings("\"", pstate->ifs);
pstate->dquote_eof = "\"\xff";
pstate->dquote_dquote = "\"\"";
pstate->dquotelen = strlen(pstate->dquote);
pstate->pno_dquote_parse_trie = parse_trie_alloc();
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->irs, IRS_STRIDX);
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->ifs, IFS_STRIDX);
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->dquote, DQUOTE_STRIDX);
pstate->pdquote_parse_trie = parse_trie_alloc();
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_irs, DQUOTE_IRS_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_ifs, DQUOTE_IFS_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_dquote, DQUOTE_DQUOTE_STRIDX);
pstate->pfields = rslls_alloc();
pstate->psb = sb_alloc(STRING_BUILDER_INIT_SIZE);
pstate->expect_header_line_next = use_implicit_header ? FALSE : TRUE;
pstate->use_implicit_header = use_implicit_header;
pstate->pheader_keeper = NULL;
pstate->pheader_keepers = lhmslv_alloc();
plrec_reader->pvstate = (void*)pstate;
plrec_reader->popen_func = file_reader_mmap_vopen;
plrec_reader->pclose_func = file_reader_mmap_vclose;
plrec_reader->pprocess_func = lrec_reader_mmap_csv_process;
plrec_reader->psof_func = lrec_reader_mmap_csv_sof;
plrec_reader->pfree_func = lrec_reader_mmap_csv_free;
return plrec_reader;
}
static step_t* step_counter_alloc(char* input_field_name) {
step_t* pstep = mlr_malloc_or_die(sizeof(step_t));
step_counter_state_t* pstate = mlr_malloc_or_die(sizeof(step_counter_state_t));
pstate->counter = 0LL;
pstate->output_field_name = mlr_paste_2_strings(input_field_name, "_counter");
pstep->pvstate = (void*)pstate;
pstep->psprocess_func = &step_counter_sprocess;
pstep->pdprocess_func = NULL;
return pstep;
}
static step_t* step_rsum_alloc(char* input_field_name) {
step_t* pstep = mlr_malloc_or_die(sizeof(step_t));
step_rsum_state_t* pstate = mlr_malloc_or_die(sizeof(step_rsum_state_t));
pstate->rsum = 0.0;
pstate->output_field_name = mlr_paste_2_strings(input_field_name, "_rsum");
pstep->pvstate = (void*)pstate;
pstep->psprocess_func = NULL;
pstep->pdprocess_func = &step_rsum_dprocess;
return pstep;
}
static step_t* step_delta_alloc(char* input_field_name) {
step_t* pstep = mlr_malloc_or_die(sizeof(step_t));
step_delta_state_t* pstate = mlr_malloc_or_die(sizeof(step_delta_state_t));
pstate->prev = -999.0;
pstate->have_prev = FALSE;
pstate->output_field_name = mlr_paste_2_strings(input_field_name, "_delta");
pstep->pvstate = (void*)pstate;
pstep->psprocess_func = NULL;
pstep->pdprocess_func = &step_delta_dprocess;
return pstep;
}
// ----------------------------------------------------------------
lrec_reader_t* lrec_reader_stdio_csv_alloc(char* irs, char* ifs, int use_implicit_header,
comment_handling_t comment_handling, char* comment_string)
{
lrec_reader_t* plrec_reader = mlr_malloc_or_die(sizeof(lrec_reader_t));
lrec_reader_stdio_csv_state_t* pstate = mlr_malloc_or_die(sizeof(lrec_reader_stdio_csv_state_t));
pstate->ilno = 0LL;
pstate->do_auto_line_term = FALSE;
if (streq(irs, "auto")) {
irs = "\n";
pstate->do_auto_line_term = TRUE;
}
pstate->comment_handling = comment_handling;
pstate->comment_string = comment_string;
pstate->comment_string_length = comment_string == NULL ? 0 : strlen(comment_string);
pstate->eof = "\xff";
pstate->irs = irs;
pstate->ifs = ifs;
pstate->ifs_eof = mlr_paste_2_strings(pstate->ifs, "\xff");
pstate->dquote = "\"";
pstate->dquote_ifs = mlr_paste_2_strings("\"", pstate->ifs);
pstate->dquote_eof = "\"\xff";
pstate->dquote_dquote = "\"\"";
pstate->dquotelen = strlen(pstate->dquote);
// Parse trie for UTF-8 BOM
pstate->putf8_bom_parse_trie = parse_trie_alloc();
parse_trie_add_string(pstate->putf8_bom_parse_trie, UTF8_BOM, UTF8_BOM_STRIDX);
// Parse trie for non-double-quoted fields
pstate->pno_dquote_parse_trie = parse_trie_alloc();
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->eof, EOF_STRIDX);
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->irs, IRS_STRIDX);
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->ifs_eof, IFS_EOF_STRIDX);
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->ifs, IFS_STRIDX);
parse_trie_add_string(pstate->pno_dquote_parse_trie, pstate->dquote, DQUOTE_STRIDX);
// Parse trie for double-quoted fields
pstate->pdquote_parse_trie = parse_trie_alloc();
if (pstate->do_auto_line_term) {
pstate->dquote_irs = mlr_paste_2_strings("\"", "\n");
pstate->dquote_irs2 = mlr_paste_2_strings("\"", "\r\n");
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_irs, DQUOTE_IRS_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_irs2, DQUOTE_IRS2_STRIDX);
} else {
pstate->dquote_irs = mlr_paste_2_strings("\"", pstate->irs);
pstate->dquote_irs2 = NULL;
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_irs, DQUOTE_IRS_STRIDX);
}
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->eof, EOF_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_irs, DQUOTE_IRS_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_ifs, DQUOTE_IFS_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_eof, DQUOTE_EOF_STRIDX);
parse_trie_add_string(pstate->pdquote_parse_trie, pstate->dquote_dquote, DQUOTE_DQUOTE_STRIDX);
pstate->pfields = rslls_alloc();
pstate->psb = sb_alloc(STRING_BUILDER_INIT_SIZE);
pstate->pbr = stdio_byte_reader_alloc();
pstate->pfr = pfr_alloc(pstate->pbr, mlr_imax3(
pstate->putf8_bom_parse_trie->maxlen,
pstate->pno_dquote_parse_trie->maxlen,
pstate->pdquote_parse_trie->maxlen));
pstate->expect_header_line_next = use_implicit_header ? FALSE : TRUE;
pstate->use_implicit_header = use_implicit_header;
pstate->pheader_keeper = NULL;
pstate->pheader_keepers = lhmslv_alloc();
plrec_reader->pvstate = (void*)pstate;
plrec_reader->popen_func = lrec_reader_stdio_csv_open;
plrec_reader->pclose_func = lrec_reader_stdio_csv_close;
plrec_reader->pprocess_func = lrec_reader_stdio_csv_process;
plrec_reader->psof_func = lrec_reader_stdio_csv_sof;
plrec_reader->pfree_func = lrec_reader_stdio_csv_free;
return plrec_reader;
}
