X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=perl-module%2FPSPP.xs;h=f6655b8ded0dbf8d24266a47c3cf6858e4244822;hb=6615bd0654fe1d821b158db2d343c958ef86e3d7;hp=237ae95f541a9564829ff93bff2ce33d8fcb2314;hpb=8953baa61127d6d3b91f763663ea647bf3e4e793;p=pspp diff --git a/perl-module/PSPP.xs b/perl-module/PSPP.xs index 237ae95f54..f6655b8ded 100644 --- a/perl-module/PSPP.xs +++ b/perl-module/PSPP.xs @@ -1,5 +1,5 @@ /* PSPP - computes sample statistics. - Copyright (C) 2007, 2008, 2009 Free Software Foundation, Inc. + Copyright (C) 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 Free Software Foundation, Inc. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as @@ -17,30 +17,44 @@ 02110-1301, USA. */ +#undef VERSION +#include + +/* The Gnulib "strftime" module defines my_strftime in for use by + gl/strftime.c. Perl also defines my_strftime in embed.h for some other + purpose. The former definition doesn't matter in this file, so suppress it + to avoid a compiler warning. */ +#undef my_strftime + #include "EXTERN.h" #include "perl.h" #include "XSUB.h" -#include - #include "ppport.h" #include "minmax.h" +#include +#include #include #include +#include #include #include #include #include +#include #include #include #include +#include +#include #include -#include #include +#include #include #include #include +#include #include typedef struct fmt_spec input_format ; @@ -48,7 +62,7 @@ typedef struct fmt_spec output_format ; /* A thin wrapper around sfm_writer */ -struct sysfile_info +struct syswriter_info { bool opened; @@ -56,7 +70,7 @@ struct sysfile_info struct casewriter *writer; /* A pointer to the dictionary. Owned externally */ - const struct dictionary *dict; + const struct pspp_dict *dict; /* The scalar containing the dictionary */ SV *dict_sv; @@ -66,35 +80,47 @@ struct sysfile_info /* A thin wrapper around sfm_reader */ struct sysreader_info { - struct sfm_read_info opts; + struct any_read_info opts; /* A pointer to the reader. The reader is owned by the struct */ struct casereader *reader; /* A pointer to the dictionary. */ - struct dictionary *dict; + struct pspp_dict *dict; +}; + + +struct input_format { + struct hmap_node hmap_node; /* In struct pspp_dict's input_formats map. */ + const struct variable *var; + struct fmt_spec input_format; }; +/* A thin wrapper around struct dictionary.*/ +struct pspp_dict { + struct dictionary *dict; + struct hmap input_formats; /* Contains struct input_format. */ +}; /* A message handler which writes messages to PSPP::errstr */ static void -message_handler (const struct msg *m) +message_handler (const struct msg *m, void *aux) { SV *errstr = get_sv("PSPP::errstr", TRUE); sv_setpv (errstr, m->text); } static int -sysfile_close (struct sysfile_info *sfi) +sysfile_close (struct syswriter_info *swi) { int retval ; - if ( ! sfi->opened ) + if ( ! swi->opened ) return 0; - retval = casewriter_destroy (sfi->writer); + retval = casewriter_destroy (swi->writer); if (retval > 0 ) - sfi->opened = false; + swi->opened = false; return retval; } @@ -113,8 +139,9 @@ scalar_to_value (union value *val, SV *scalar, const struct variable *var) { STRLEN len; const char *p = SvPV (scalar, len); - memset (val->s, ' ', var_get_width (var)); - memcpy (val->s, p, len); + int width = var_get_width (var); + value_set_missing (val, width); + memcpy (value_str_rw (val, width), p, len); } } @@ -130,24 +157,40 @@ value_to_scalar (const union value *val, const struct variable *var) return newSVnv (val->f); } else - return newSVpvn (val->s, var_get_width (var)); + { + int width = var_get_width (var); + return newSVpvn (value_str (val, width), width); + } } static void -var_set_input_format (struct variable *v, input_format ip_fmt) +make_value_from_scalar (union value *uv, SV *val, const struct variable *var) { - struct fmt_spec *if_copy = malloc (sizeof (*if_copy)); - memcpy (if_copy, &ip_fmt, sizeof (ip_fmt)); - var_attach_aux (v, if_copy, var_dtor_free); + value_init (uv, var_get_width (var)); + scalar_to_value (uv, val, var); } -static union value * -make_value_from_scalar (SV *val, const struct variable *var) +static struct pspp_dict * +create_pspp_dict (struct dictionary *dict) { - union value *uv = value_create (var_get_width (var)); - scalar_to_value (uv, val, var); - return uv; + struct pspp_dict *pspp_dict = xmalloc (sizeof *pspp_dict); + pspp_dict->dict = dict; + hmap_init (&pspp_dict->input_formats); + return pspp_dict; +} + +static const struct fmt_spec * +find_input_format (const struct pspp_dict *dict, const struct variable *var) +{ + struct input_format *input_format; + + HMAP_FOR_EACH_IN_BUCKET (input_format, struct input_format, hmap_node, + hash_pointer (var, 0), &dict->input_formats) + if (input_format->var == var) + return &input_format->input_format; + + return NULL; } @@ -155,13 +198,20 @@ MODULE = PSPP MODULE = PSPP PACKAGE = PSPP +PROTOTYPES: ENABLE + void onBoot (ver) const char *ver CODE: - assert (0 == strcmp (ver, bare_version)); - msg_init (NULL, message_handler); - settings_init (0, 0); + /* Check that the version is correct up to the length of 'ver'. + This allows PSPP autobuilders to add a "-build#" suffix to the + PSPP version without causing failures here. */ + assert (0 == strncmp (ver, bare_version, strlen (ver))); + + i18n_init (); + msg_set_handler (message_handler, NULL); + settings_init (); fh_init (); SV * @@ -171,12 +221,11 @@ format_value (val, var) CODE: SV *ret; const struct fmt_spec *fmt = var_get_print_format (var); - union value *uv = make_value_from_scalar (val, var); + union value uv; char *s; - s = malloc (fmt->w); - memset (s, '\0', fmt->w); - data_out (uv, fmt, s); - free (uv); + make_value_from_scalar (&uv, val, var); + s = data_out (&uv, var_get_encoding (var), fmt); + value_destroy (&uv, var_get_width (var)); ret = newSVpv (s, fmt->w); free (s); RETVAL = ret; @@ -189,9 +238,11 @@ value_is_missing (val, var) SV *val struct variable *var CODE: - union value *uv = make_value_from_scalar (val, var); - int ret = var_is_value_missing (var, uv, MV_ANY); - free (uv); + union value uv; + int ret; + make_value_from_scalar (&uv, val, var); + ret = var_is_value_missing (var, &uv, MV_ANY); + value_destroy (&uv, var_get_width (var)); RETVAL = ret; OUTPUT: RETVAL @@ -200,94 +251,106 @@ RETVAL MODULE = PSPP PACKAGE = PSPP::Dict -struct dictionary * +struct pspp_dict * pxs_dict_new() CODE: - RETVAL = dict_create (); + RETVAL = create_pspp_dict (dict_create ("UTF-8")); OUTPUT: RETVAL void DESTROY (dict) - struct dictionary *dict + struct pspp_dict *dict CODE: - dict_destroy (dict); + if (dict != NULL) + { + struct input_format *input_format, *next_input_format; + HMAP_FOR_EACH_SAFE (input_format, next_input_format, + struct input_format, hmap_node, &dict->input_formats) + { + hmap_delete (&dict->input_formats, &input_format->hmap_node); + free (input_format); + } + hmap_destroy (&dict->input_formats); + dict_unref (dict->dict); + free (dict); + } int get_var_cnt (dict) - struct dictionary *dict + struct pspp_dict *dict CODE: - RETVAL = dict_get_var_cnt (dict); + RETVAL = dict_get_var_cnt (dict->dict); OUTPUT: RETVAL void set_label (dict, label) - struct dictionary *dict + struct pspp_dict *dict char *label CODE: - dict_set_label (dict, label); + dict_set_label (dict->dict, label); void set_documents (dict, docs) - struct dictionary *dict + struct pspp_dict *dict char *docs CODE: - dict_set_documents (dict, docs); + dict_set_documents_string (dict->dict, docs); void add_document (dict, doc) - struct dictionary *dict + struct pspp_dict *dict char *doc CODE: - dict_add_document_line (dict, doc); + dict_add_document_line (dict->dict, doc, false); void clear_documents (dict) - struct dictionary *dict + struct pspp_dict *dict CODE: - dict_clear_documents (dict); + dict_clear_documents (dict->dict); void set_weight (dict, var) - struct dictionary *dict + struct pspp_dict *dict struct variable *var CODE: - dict_set_weight (dict, var); + dict_set_weight (dict->dict, var); struct variable * pxs_get_variable (dict, idx) - struct dictionary *dict + struct pspp_dict *dict SV *idx INIT: SV *errstr = get_sv("PSPP::errstr", TRUE); sv_setpv (errstr, ""); - if ( SvIV (idx) >= dict_get_var_cnt (dict)) + if ( SvIV (idx) >= dict_get_var_cnt (dict->dict)) { sv_setpv (errstr, "The dictionary doesn't have that many variables."); XSRETURN_UNDEF; } CODE: - RETVAL = dict_get_var (dict, SvIV (idx)); + RETVAL = dict_get_var (dict->dict, SvIV (idx)); OUTPUT: RETVAL struct variable * pxs_get_var_by_name (dict, name) - struct dictionary *dict + struct pspp_dict *dict const char *name INIT: SV *errstr = get_sv("PSPP::errstr", TRUE); sv_setpv (errstr, ""); CODE: - struct variable *var = dict_lookup_var (dict, name); + struct variable *var = dict_lookup_var (dict->dict, name); if ( ! var ) sv_setpv (errstr, "No such variable."); RETVAL = var; @@ -300,23 +363,24 @@ MODULE = PSPP PACKAGE = PSPP::Var struct variable * pxs_dict_create_var (dict, name, ip_fmt) - struct dictionary * dict + struct pspp_dict * dict char *name input_format ip_fmt INIT: SV *errstr = get_sv("PSPP::errstr", TRUE); sv_setpv (errstr, ""); - if ( ! var_is_plausible_name (name, false)) + if ( ! id_is_plausible (name, false)) { sv_setpv (errstr, "The variable name is not valid."); XSRETURN_UNDEF; } CODE: struct fmt_spec op_fmt; + struct input_format *input_format; struct variable *v; op_fmt = fmt_for_output_from_input (&ip_fmt); - v = dict_create_var (dict, name, + v = dict_create_var (dict->dict, name, fmt_is_string (op_fmt.type) ? op_fmt.w : 0); if ( NULL == v ) { @@ -324,7 +388,13 @@ CODE: XSRETURN_UNDEF; } var_set_both_formats (v, &op_fmt); - var_set_input_format (v, ip_fmt); + + input_format = xmalloc (sizeof *input_format); + input_format->var = v; + input_format->input_format = ip_fmt; + hmap_insert (&dict->input_formats, &input_format->hmap_node, + hash_pointer (v, 0)); + RETVAL = v; OUTPUT: RETVAL @@ -365,6 +435,37 @@ clear_value_labels (var) CODE: var_clear_value_labels (var); +SV * +get_write_format (var) + struct variable *var +CODE: + HV *fmthash = (HV *) sv_2mortal ((SV *) newHV()); + const struct fmt_spec *fmt = var_get_write_format (var); + + hv_store (fmthash, "fmt", 3, newSVnv (fmt->type), 0); + hv_store (fmthash, "decimals", 8, newSVnv (fmt->d), 0); + hv_store (fmthash, "width", 5, newSVnv (fmt->w), 0); + + RETVAL = newRV ((SV *) fmthash); + OUTPUT: +RETVAL + +SV * +get_print_format (var) + struct variable *var +CODE: + HV *fmthash = (HV *) sv_2mortal ((SV *) newHV()); + const struct fmt_spec *fmt = var_get_print_format (var); + + hv_store (fmthash, "fmt", 3, newSVnv (fmt->type), 0); + hv_store (fmthash, "decimals", 8, newSVnv (fmt->d), 0); + hv_store (fmthash, "width", 5, newSVnv (fmt->w), 0); + + RETVAL = newRV ((SV *) fmthash); + OUTPUT: +RETVAL + + void pxs_set_write_format (var, fmt) struct variable *var @@ -398,26 +499,27 @@ INIT: sv_setpv (errstr, ""); CODE: union value the_value; + int width = var_get_width (var); + int ok; + value_init (&the_value, width); if ( var_is_numeric (var)) { if ( ! looks_like_number (key)) { sv_setpv (errstr, "Cannot add label with string key to a numeric variable"); + value_destroy (&the_value, width); XSRETURN_IV (0); } the_value.f = SvNV (key); } else { - if ( var_is_long_string (var) ) - { - sv_setpv (errstr, "Cannot add label to a long string variable"); - XSRETURN_IV (0); - } - strncpy (the_value.s, SvPV_nolen(key), MAX_SHORT_STRING); + value_copy_str_rpad (&the_value, width, SvPV_nolen(key), ' '); } - if (! var_add_value_label (var, &the_value, label) ) + ok = var_add_value_label (var, &the_value, label); + value_destroy (&the_value, width); + if (!ok) { sv_setpv (errstr, "Something went wrong"); XSRETURN_IV (0); @@ -486,20 +588,19 @@ get_value_labels (var) struct variable *var CODE: HV *labelhash = (HV *) sv_2mortal ((SV *) newHV()); - struct val_lab *vl; - struct val_labs_iterator *viter = NULL; + const struct val_lab *vl; const struct val_labs *labels = var_get_value_labels (var); if ( labels ) { - for (vl = val_labs_first (labels, &viter); + for (vl = val_labs_first (labels); vl; - vl = val_labs_next (labels, &viter)) + vl = val_labs_next (labels, vl)) { SV *sv = value_to_scalar (&vl->value, var); STRLEN len; const char *s = SvPV (sv, len); - hv_store (labelhash, s, len, newSVpv (vl->label, 0), 0); + hv_store (labelhash, s, len, newSVpv (val_lab_get_label (vl), 0), 0); } } @@ -512,14 +613,13 @@ RETVAL MODULE = PSPP PACKAGE = PSPP::Sysfile -struct sysfile_info * -pxs_create_sysfile (name, dict_ref, opts_hr) +struct syswriter_info * +pxs_create_sysfile (name, dict, opts_hr) char *name - SV *dict_ref + struct pspp_dict *dict; SV *opts_hr INIT: - SV *dict_sv = SvRV (dict_ref); - struct dictionary *dict = (void *) SvIV (dict_sv); + SV *dict_sv = ST(1); struct sfm_write_options opts; if (!SvROK (opts_hr)) { @@ -533,42 +633,44 @@ INIT: SV** version = hv_fetch(opt_h, "version", 7, 0); opts.create_writeable = readonly ? ! SvIV (*readonly) : true; - opts.compress = compress ? SvIV (*compress) : false; + opts.compression = (compress && SvIV (*compress) + ? ANY_COMP_SIMPLE + : ANY_COMP_NONE); opts.version = version ? SvIV (*version) : 3 ; } CODE: struct file_handle *fh = - fh_create_file (NULL, name, fh_default_properties () ); - struct sysfile_info *sfi = xmalloc (sizeof (*sfi)); - sfi->writer = sfm_open_writer (fh, dict, opts); - sfi->dict = dict; - sfi->opened = true; - sfi->dict_sv = dict_sv; - SvREFCNT_inc (sfi->dict_sv); + fh_create_file (NULL, name, "UTF-8", fh_default_properties () ); + struct syswriter_info *swi = xmalloc (sizeof (*swi)); + swi->writer = sfm_open_writer (fh, dict->dict, opts); + swi->dict = dict; + swi->opened = true; + swi->dict_sv = dict_sv; + SvREFCNT_inc (swi->dict_sv); - RETVAL = sfi; + RETVAL = swi; OUTPUT: RETVAL int -close (sfi) - struct sysfile_info *sfi +close (swi) + struct syswriter_info *swi CODE: - RETVAL = sysfile_close (sfi); + RETVAL = sysfile_close (swi); OUTPUT: RETVAL void -DESTROY (sfi) - struct sysfile_info *sfi +DESTROY (swi) + struct syswriter_info *swi CODE: - sysfile_close (sfi); - SvREFCNT_dec (sfi->dict_sv); - free (sfi); + sysfile_close (swi); + SvREFCNT_dec (swi->dict_sv); + free (swi); int -append_case (sfi, ccase) - struct sysfile_info *sfi +append_case (swi, ccase) + struct syswriter_info *swi SV *ccase INIT: SV *errstr = get_sv("PSPP::errstr", TRUE); @@ -586,17 +688,18 @@ CODE: struct ccase *c; SV *sv; - if ( av_len (av_case) >= dict_get_var_cnt (sfi->dict)) + if ( av_len (av_case) >= dict_get_var_cnt (swi->dict->dict)) XSRETURN_UNDEF; - c = case_create (dict_get_next_value_idx (sfi->dict)); + c = case_create (dict_get_proto (swi->dict->dict)); - dict_get_vars (sfi->dict, &vv, &nv, 1u << DC_ORDINARY | 1u << DC_SYSTEM); + dict_get_vars (swi->dict->dict, &vv, &nv, + 1u << DC_ORDINARY | 1u << DC_SYSTEM); for (sv = av_shift (av_case); SvOK (sv); sv = av_shift (av_case)) { const struct variable *v = vv[i++]; - const struct fmt_spec *ifmt = var_get_aux (v); + const struct fmt_spec *ifmt = find_input_format (swi->dict, v); /* If an input format has been set, then use it. Otherwise just convert the raw value. @@ -604,9 +707,16 @@ CODE: if ( ifmt ) { struct substring ss = ss_cstr (SvPV_nolen (sv)); - if ( ! data_in (ss, LEGACY_NATIVE, ifmt->type, 0, 0, 0, - case_data_rw (c, v), - var_get_width (v)) ) + char *error; + bool ok; + + error = data_in (ss, SvUTF8(sv) ? UTF8: "iso-8859-1", ifmt->type, + case_data_rw (c, v), var_get_width (v), + dict_get_encoding (swi->dict->dict)); + ok = error == NULL; + free (error); + + if ( !ok ) { RETVAL = 0; goto finish; @@ -619,16 +729,14 @@ CODE: } /* The remaining variables must be sysmis or blank string */ - while (i < dict_get_var_cnt (sfi->dict)) + while (i < dict_get_var_cnt (swi->dict->dict)) { const struct variable *v = vv[i++]; union value *val = case_data_rw (c, v); - if ( var_is_numeric (v)) - val->f = SYSMIS; - else - memset (val->s, ' ', var_get_width (v)); + value_set_missing (val, var_get_width (v)); } - RETVAL = casewriter_write (sfi->writer, c); + casewriter_write (swi->writer, c); + RETVAL = 1; finish: free (vv); OUTPUT: @@ -643,26 +751,27 @@ struct sysreader_info * pxs_open_sysfile (name) char * name CODE: - struct casereader *reader; struct sysreader_info *sri = NULL; struct file_handle *fh = - fh_create_file (NULL, name, fh_default_properties () ); + fh_create_file (NULL, name, "UTF-8", fh_default_properties () ); + struct dictionary *dict; sri = xmalloc (sizeof (*sri)); - sri->reader = sfm_open_reader (fh, &sri->dict, &sri->opts); - - if ( NULL == sri->reader) - { - free (sri); - sri = NULL; - } + sri->reader = any_reader_open_and_decode (fh, NULL, &dict, &sri->opts); + if (sri->reader) + sri->dict = create_pspp_dict (dict); + else + { + free (sri); + sri = NULL; + } RETVAL = sri; OUTPUT: RETVAL -struct dictionary * +struct pspp_dict * pxs_get_dict (reader) struct sysreader_info *reader; CODE: @@ -670,6 +779,21 @@ CODE: OUTPUT: RETVAL +SV * +get_case_cnt (sfr) + struct sysreader_info *sfr; +CODE: + SV *ret; + casenumber n = casereader_get_case_cnt (sfr->reader); + if (n == CASENUMBER_MAX) + ret = &PL_sv_undef; + else + ret = newSViv (n); + RETVAL = ret; + OUTPUT: +RETVAL + + void get_next_case (sfr) @@ -677,14 +801,14 @@ get_next_case (sfr) PPCODE: struct ccase *c; - if (c = casereader_read (sfr->reader)) + if ((c = casereader_read (sfr->reader)) != NULL) { int v; - EXTEND (SP, dict_get_var_cnt (sfr->dict)); - for (v = 0; v < dict_get_var_cnt (sfr->dict); ++v ) + EXTEND (SP, dict_get_var_cnt (sfr->dict->dict)); + for (v = 0; v < dict_get_var_cnt (sfr->dict->dict); ++v ) { - const struct variable *var = dict_get_var (sfr->dict, v); + const struct variable *var = dict_get_var (sfr->dict->dict, v); const union value *val = case_data (c, var); PUSHs (sv_2mortal (value_to_scalar (val, var)));