/* PSPP - a program for statistical analysis.
- Copyright (C) 1997-2000, 2006-2007, 2009-2014 Free Software Foundation, Inc.
+ Copyright (C) 1997-2000, 2006-2007, 2009-2016 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 published by
#include <config.h>
-#include "data/sys-file-reader.h"
#include "data/sys-file-private.h"
#include <errno.h>
#include <sys/stat.h>
#include <zlib.h>
+#include "data/any-reader.h"
#include "data/attributes.h"
#include "data/case.h"
#include "data/casereader-provider.h"
#include "libpspp/assertion.h"
#include "libpspp/compiler.h"
#include "libpspp/i18n.h"
+#include "libpspp/ll.h"
#include "libpspp/message.h"
#include "libpspp/misc.h"
#include "libpspp/pool.h"
int weight_idx; /* 0 if unweighted, otherwise a var index. */
int nominal_case_size; /* Number of var positions. */
- /* These correspond to the members of struct sfm_file_info or a dictionary
+ /* These correspond to the members of struct any_file_info or a dictionary
but in the system file's encoding rather than ASCII. */
char creation_date[10]; /* "dd mmm yy". */
char creation_time[9]; /* "hh:mm:ss". */
{
off_t pos;
int width;
- char name[8];
+ char name[9];
int print_format;
int write_format;
int missing_value_code;
size_t n_lines;
};
+struct sfm_mrset
+ {
+ const char *name; /* Name. */
+ const char *label; /* Human-readable label for group. */
+ enum mrset_type type; /* Group type. */
+ const char **vars; /* Constituent variables' names. */
+ size_t n_vars; /* Number of constituent variables. */
+
+ /* MRSET_MD only. */
+ enum mrset_md_cat_source cat_source; /* Source of category labels. */
+ bool label_from_var_label; /* 'label' taken from variable label? */
+ const char *counted; /* Counted value, as string. */
+ };
+
struct sfm_extension_record
{
+ struct ll ll; /* In struct sfm_reader 'var_attrs' list. */
int subtype; /* Record subtype. */
off_t pos; /* Starting offset in file. */
- size_t size; /* Size of data elements. */
- size_t count; /* Number of data elements. */
+ unsigned int size; /* Size of data elements. */
+ unsigned int count; /* Number of data elements. */
void *data; /* Contents. */
};
/* System file reader. */
struct sfm_reader
{
+ struct any_reader any_reader;
+
/* Resource tracking. */
struct pool *pool; /* All system file state. */
+ /* File data. */
+ struct any_read_info info;
+ struct sfm_header_record header;
+ struct sfm_var_record *vars;
+ size_t n_vars;
+ struct sfm_value_label_record *labels;
+ size_t n_labels;
+ struct sfm_document_record *document;
+ struct sfm_mrset *mrsets;
+ size_t n_mrsets;
+ struct sfm_extension_record *extensions[32];
+ struct ll_list var_attrs; /* Contains "struct sfm_extension_record"s. */
+
/* File state. */
struct file_handle *fh; /* File handle. */
struct fh_lock *lock; /* Mutual exclusion for file handle. */
size_t sfm_var_cnt; /* Number of variables. */
int case_cnt; /* Number of cases */
const char *encoding; /* String encoding. */
+ bool written_by_readstat; /* From https://github.com/WizardMac/ReadStat? */
/* Decompression. */
- enum sfm_compression compression;
+ enum any_compression compression;
double bias; /* Compression bias, usually 100.0. */
uint8_t opcodes[8]; /* Current block of opcodes. */
size_t opcode_idx; /* Next opcode to interpret, 8 if none left. */
static const struct casereader_class sys_file_casereader_class;
-static bool close_reader (struct sfm_reader *);
+static struct sfm_reader *
+sfm_reader_cast (const struct any_reader *r_)
+{
+ assert (r_->klass == &sys_file_reader_class);
+ return UP_CAST (r_, struct sfm_reader, any_reader);
+}
+
+static bool sfm_close (struct any_reader *);
static struct variable *lookup_var_by_index (struct sfm_reader *, off_t,
const struct sfm_var_record *,
static char *fix_line_ends (const char *);
-static int parse_int (struct sfm_reader *, const void *data, size_t ofs);
-static double parse_float (struct sfm_reader *, const void *data, size_t ofs);
+static int parse_int (const struct sfm_reader *, const void *data, size_t ofs);
+static double parse_float (const struct sfm_reader *,
+ const void *data, size_t ofs);
static bool read_variable_record (struct sfm_reader *,
struct sfm_var_record *);
static bool read_value_label_record (struct sfm_reader *,
- struct sfm_value_label_record *,
- size_t n_vars);
-static struct sfm_document_record *read_document_record (struct sfm_reader *);
+ struct sfm_value_label_record *);
+static bool read_document_record (struct sfm_reader *);
static bool read_extension_record (struct sfm_reader *, int subtype,
struct sfm_extension_record **);
static bool skip_extension_record (struct sfm_reader *, int subtype);
-static const char *choose_encoding (
- struct sfm_reader *,
- const struct sfm_header_record *,
- const struct sfm_extension_record *ext_integer,
- const struct sfm_extension_record *ext_encoding);
-
static struct text_record *open_text_record (
struct sfm_reader *, const struct sfm_extension_record *,
bool recode_to_utf8);
struct text_record *,
struct variable **var, char **value);
static void text_warn (struct sfm_reader *r, struct text_record *text,
- const char *format, ...)
- PRINTF_FORMAT (3, 4);
+ const char *format, ...) PRINTF_FORMAT (3, 4);
static char *text_get_token (struct text_record *,
struct substring delimiters, char *delimiter);
static bool text_match (struct text_record *, char c);
struct text_record *);
static size_t text_pos (const struct text_record *);
static const char *text_get_all (const struct text_record *);
-
-static bool close_reader (struct sfm_reader *r);
\f
/* Dictionary reader. */
WRITE_FORMAT
};
-static bool read_header (struct sfm_reader *, struct sfm_read_info *,
+static bool read_dictionary (struct sfm_reader *);
+static bool read_record (struct sfm_reader *, int type,
+ size_t *allocated_vars, size_t *allocated_labels);
+static bool read_header (struct sfm_reader *, struct any_read_info *,
struct sfm_header_record *);
static void parse_header (struct sfm_reader *,
const struct sfm_header_record *,
- struct sfm_read_info *, struct dictionary *);
+ struct any_read_info *, struct dictionary *);
static bool parse_variable_records (struct sfm_reader *, struct dictionary *,
struct sfm_var_record *, size_t n);
static void parse_format_spec (struct sfm_reader *, off_t pos,
struct dictionary *);
static bool parse_machine_integer_info (struct sfm_reader *,
const struct sfm_extension_record *,
- struct sfm_read_info *);
+ struct any_read_info *);
static void parse_machine_float_info (struct sfm_reader *,
const struct sfm_extension_record *);
static void parse_extra_product_info (struct sfm_reader *,
const struct sfm_extension_record *,
- struct sfm_read_info *);
+ struct any_read_info *);
static void parse_mrsets (struct sfm_reader *,
const struct sfm_extension_record *,
- struct dictionary *);
+ size_t *allocated_mrsets);
+static void decode_mrsets (struct sfm_reader *, struct dictionary *);
static void parse_long_var_name_map (struct sfm_reader *,
const struct sfm_extension_record *,
struct dictionary *);
const struct sfm_extension_record *,
struct dictionary *);
static void assign_variable_roles (struct sfm_reader *, struct dictionary *);
-static bool parse_long_string_value_labels (struct sfm_reader *,
+static void parse_long_string_value_labels (struct sfm_reader *,
const struct sfm_extension_record *,
struct dictionary *);
-static bool parse_long_string_missing_values (
+static void parse_long_string_missing_values (
struct sfm_reader *, const struct sfm_extension_record *,
struct dictionary *);
/* Frees the strings inside INFO. */
void
-sfm_read_info_destroy (struct sfm_read_info *info)
+any_read_info_destroy (struct any_read_info *info)
{
if (info)
{
}
}
-/* Opens the system file designated by file handle FH for reading. Reads the
- system file's dictionary into *DICT.
-
- Ordinarily the reader attempts to automatically detect the character
- encoding based on the file's contents. This isn't always possible,
- especially for files written by old versions of SPSS or PSPP, so specifying
- a nonnull ENCODING overrides the choice of character encoding.
-
- If INFO is non-null, then it receives additional info about the system file,
- which the caller must eventually free with sfm_read_info_destroy() when it
- is no longer needed. */
-struct casereader *
-sfm_open_reader (struct file_handle *fh, const char *volatile encoding,
- struct dictionary **dictp, struct sfm_read_info *infop)
+/* Tries to open FH for reading as a system file. Returns an sfm_reader if
+ successful, otherwise NULL. */
+static struct any_reader *
+sfm_open (struct file_handle *fh)
{
- struct sfm_reader *r = NULL;
- struct sfm_read_info *volatile info;
-
- struct sfm_header_record header;
-
- struct sfm_var_record *vars;
- size_t n_vars, allocated_vars;
-
- struct sfm_value_label_record *labels;
- size_t n_labels, allocated_labels;
-
- struct sfm_document_record *document;
-
- struct sfm_extension_record *extensions[32];
-
- struct dictionary *dict = NULL;
- size_t i;
+ size_t allocated_mrsets = 0;
+ struct sfm_reader *r;
/* Create and initialize reader. */
- r = pool_create_container (struct sfm_reader, pool);
+ r = xzalloc (sizeof *r);
+ r->any_reader.klass = &sys_file_reader_class;
+ r->pool = pool_create ();
+ pool_register (r->pool, free, r);
r->fh = fh_ref (fh);
- r->lock = NULL;
- r->file = NULL;
- r->pos = 0;
- r->error = false;
r->opcode_idx = sizeof r->opcodes;
- r->corruption_warning = false;
- r->zin_buf = r->zout_buf = NULL;
-
- info = infop ? infop : xmalloc (sizeof *info);
- memset (info, 0, sizeof *info);
+ ll_init (&r->var_attrs);
/* TRANSLATORS: this fragment will be interpolated into
messages in fh_lock() that identify types of files. */
if (r->lock == NULL)
goto error;
- r->file = fn_open (fh_get_file_name (fh), "rb");
+ r->file = fn_open (fh, "rb");
if (r->file == NULL)
{
msg (ME, _("Error opening `%s' for reading as a system file: %s."),
goto error;
}
- /* Read header. */
- if (!read_header (r, info, &header))
+ if (!read_dictionary (r))
goto error;
- vars = NULL;
- n_vars = allocated_vars = 0;
+ if (r->extensions[EXT_MRSETS] != NULL)
+ parse_mrsets (r, r->extensions[EXT_MRSETS], &allocated_mrsets);
- labels = NULL;
- n_labels = allocated_labels = 0;
+ if (r->extensions[EXT_MRSETS2] != NULL)
+ parse_mrsets (r, r->extensions[EXT_MRSETS2], &allocated_mrsets);
- document = NULL;
+ return &r->any_reader;
- memset (extensions, 0, sizeof extensions);
+error:
+ if (r)
+ sfm_close (&r->any_reader);
+ return NULL;
+}
+
+static bool
+read_dictionary (struct sfm_reader *r)
+{
+ size_t allocated_vars;
+ size_t allocated_labels;
+ if (!read_header (r, &r->info, &r->header))
+ return false;
+
+ allocated_vars = 0;
+ allocated_labels = 0;
for (;;)
{
- int subtype;
int type;
- bool ok;
if (!read_int (r, &type))
- goto error;
+ return false;
if (type == 999)
+ break;
+ if (!read_record (r, type, &allocated_vars, &allocated_labels))
+ return false;
+ }
+
+ if (!skip_bytes (r, 4))
+ return false;
+
+ if (r->compression == ANY_COMP_ZLIB && !read_zheader (r))
+ return false;
+
+ return true;
+}
+
+static bool
+read_record (struct sfm_reader *r, int type,
+ size_t *allocated_vars, size_t *allocated_labels)
+{
+ int subtype;
+
+ switch (type)
+ {
+ case 2:
+ if (r->n_vars >= *allocated_vars)
+ r->vars = pool_2nrealloc (r->pool, r->vars, allocated_vars,
+ sizeof *r->vars);
+ return read_variable_record (r, &r->vars[r->n_vars++]);
+
+ case 3:
+ if (r->n_labels >= *allocated_labels)
+ r->labels = pool_2nrealloc (r->pool, r->labels, allocated_labels,
+ sizeof *r->labels);
+ return read_value_label_record (r, &r->labels[r->n_labels++]);
+
+ case 4:
+ /* A Type 4 record is always immediately after a type 3 record,
+ so the code for type 3 records reads the type 4 record too. */
+ sys_error (r, r->pos, _("Misplaced type 4 record."));
+ return false;
+
+ case 6:
+ if (r->document != NULL)
{
- int dummy;
- if (!read_int (r, &dummy))
- goto error;
- break;
+ sys_error (r, r->pos, _("Duplicate type 6 (document) record."));
+ return false;
}
+ return read_document_record (r);
- switch (type)
+ case 7:
+ if (!read_int (r, &subtype))
+ return false;
+ else if (subtype < 0
+ || subtype >= sizeof r->extensions / sizeof *r->extensions)
{
- case 2:
- if (n_vars >= allocated_vars)
- vars = pool_2nrealloc (r->pool, vars, &allocated_vars,
- sizeof *vars);
- ok = read_variable_record (r, &vars[n_vars++]);
- break;
+ sys_warn (r, r->pos,
+ _("Unrecognized record type 7, subtype %d. For help, "
+ "please send this file to %s and mention that you were "
+ "using %s."),
+ subtype, PACKAGE_BUGREPORT, PACKAGE_STRING);
+ return skip_extension_record (r, subtype);
+ }
+ else if (subtype == 18)
+ {
+ /* System files written by "Stata 14.1/-savespss- 1.77 by S.Radyakin"
+ put each variable attribute into a separate record with subtype
+ 18. I'm surprised that SPSS puts up with this. */
+ struct sfm_extension_record *ext;
+ bool ok = read_extension_record (r, subtype, &ext);
+ if (ok && ext)
+ ll_push_tail (&r->var_attrs, &ext->ll);
+ return ok;
+ }
+ else if (r->extensions[subtype] != NULL)
+ {
+ sys_warn (r, r->pos,
+ _("Record type 7, subtype %d found here has the same "
+ "type as the record found near offset 0x%llx. For "
+ "help, please send this file to %s and mention that "
+ "you were using %s."),
+ subtype, (long long int) r->extensions[subtype]->pos,
+ PACKAGE_BUGREPORT, PACKAGE_STRING);
+ return skip_extension_record (r, subtype);
+ }
+ else
+ return read_extension_record (r, subtype, &r->extensions[subtype]);
+
+ default:
+ sys_error (r, r->pos, _("Unrecognized record type %d."), type);
+ return false;
+ }
+
+ NOT_REACHED ();
+}
+
+/* Returns the character encoding obtained from R, or a null pointer if R
+ doesn't have an indication of its character encoding. */
+static const char *
+sfm_get_encoding (const struct sfm_reader *r)
+{
+ /* The EXT_ENCODING record is the best way to determine dictionary
+ encoding. */
+ if (r->extensions[EXT_ENCODING])
+ return r->extensions[EXT_ENCODING]->data;
+
+ /* But EXT_INTEGER is better than nothing as a fallback. */
+ if (r->extensions[EXT_INTEGER])
+ {
+ int codepage = parse_int (r, r->extensions[EXT_INTEGER]->data, 7 * 4);
+ const char *encoding;
+
+ switch (codepage)
+ {
+ case 1:
+ return "EBCDIC-US";
+ case 2:
case 3:
- if (n_labels >= allocated_labels)
- labels = pool_2nrealloc (r->pool, labels, &allocated_labels,
- sizeof *labels);
- ok = read_value_label_record (r, &labels[n_labels++], n_vars);
+ /* These ostensibly mean "7-bit ASCII" and "8-bit ASCII"[sic]
+ respectively. However, many files have character code 2 but data
+ which are clearly not ASCII. Therefore, ignore these values. */
break;
case 4:
- /* A Type 4 record is always immediately after a type 3 record,
- so the code for type 3 records reads the type 4 record too. */
- sys_error (r, r->pos, _("Misplaced type 4 record."));
- ok = false;
- break;
+ return "MS_KANJI";
- case 6:
- if (document != NULL)
- {
- sys_error (r, r->pos, _("Duplicate type 6 (document) record."));
- ok = false;
- break;
- }
- document = read_document_record (r);
- ok = document != NULL;
+ default:
+ encoding = sys_get_encoding_from_codepage (codepage);
+ if (encoding != NULL)
+ return encoding;
break;
+ }
+ }
- case 7:
- if (!read_int (r, &subtype))
- goto error;
- if (subtype < 0 || subtype >= sizeof extensions / sizeof *extensions)
- {
- sys_warn (r, r->pos,
- _("Unrecognized record type 7, subtype %d. Please "
- "send a copy of this file, and the syntax which "
- "created it to %s."),
- subtype, PACKAGE_BUGREPORT);
- ok = skip_extension_record (r, subtype);
- }
- else if (extensions[subtype] != NULL)
- {
- sys_warn (r, r->pos,
- _("Record type 7, subtype %d found here has the same "
- "type as the record found near offset 0x%llx. "
- "Please send a copy of this file, and the syntax "
- "which created it to %s."),
- subtype, (long long int) extensions[subtype]->pos,
- PACKAGE_BUGREPORT);
- ok = skip_extension_record (r, subtype);
- }
- else
- ok = read_extension_record (r, subtype, &extensions[subtype]);
- break;
+ /* If the file magic number is EBCDIC then its character data is too. */
+ if (!strcmp (r->header.magic, EBCDIC_MAGIC))
+ return "EBCDIC-US";
- default:
- sys_error (r, r->pos, _("Unrecognized record type %d."), type);
- ok = false;
- break;
+ return NULL;
+}
+
+struct get_strings_aux
+ {
+ struct pool *pool;
+ char **titles;
+ char **strings;
+ bool *ids;
+ size_t allocated;
+ size_t n;
+ };
+
+static void
+add_string__ (struct get_strings_aux *aux,
+ const char *string, bool id, char *title)
+{
+ if (aux->n >= aux->allocated)
+ {
+ aux->allocated = 2 * (aux->allocated + 1);
+ aux->titles = pool_realloc (aux->pool, aux->titles,
+ aux->allocated * sizeof *aux->titles);
+ aux->strings = pool_realloc (aux->pool, aux->strings,
+ aux->allocated * sizeof *aux->strings);
+ aux->ids = pool_realloc (aux->pool, aux->ids,
+ aux->allocated * sizeof *aux->ids);
+ }
+
+ aux->titles[aux->n] = title;
+ aux->strings[aux->n] = pool_strdup (aux->pool, string);
+ aux->ids[aux->n] = id;
+ aux->n++;
+}
+
+static void PRINTF_FORMAT (3, 4)
+add_string (struct get_strings_aux *aux,
+ const char *string, const char *title, ...)
+{
+ va_list args;
+
+ va_start (args, title);
+ add_string__ (aux, string, false, pool_vasprintf (aux->pool, title, args));
+ va_end (args);
+}
+
+static void PRINTF_FORMAT (3, 4)
+add_id (struct get_strings_aux *aux, const char *id, const char *title, ...)
+{
+ va_list args;
+
+ va_start (args, title);
+ add_string__ (aux, id, true, pool_vasprintf (aux->pool, title, args));
+ va_end (args);
+}
+
+/* Retrieves significant string data from R in its raw format, to allow the
+ caller to try to detect the encoding in use.
+
+ Returns the number of strings retrieved N. Sets each of *TITLESP, *IDSP,
+ and *STRINGSP to an array of N elements allocated from POOL. For each I in
+ 0...N-1, UTF-8 string *TITLESP[I] describes *STRINGSP[I], which is in
+ whatever encoding system file R uses. *IDS[I] is true if *STRINGSP[I] must
+ be a valid PSPP language identifier, false if *STRINGSP[I] is free-form
+ text. */
+static size_t
+sfm_get_strings (const struct any_reader *r_, struct pool *pool,
+ char ***titlesp, bool **idsp, char ***stringsp)
+{
+ struct sfm_reader *r = sfm_reader_cast (r_);
+ const struct sfm_mrset *mrset;
+ struct get_strings_aux aux;
+ size_t var_idx;
+ size_t i, j, k;
+
+ aux.pool = pool;
+ aux.titles = NULL;
+ aux.strings = NULL;
+ aux.ids = NULL;
+ aux.allocated = 0;
+ aux.n = 0;
+
+ var_idx = 0;
+ for (i = 0; i < r->n_vars; i++)
+ if (r->vars[i].width != -1)
+ add_id (&aux, r->vars[i].name, _("Variable %zu"), ++var_idx);
+
+ var_idx = 0;
+ for (i = 0; i < r->n_vars; i++)
+ if (r->vars[i].width != -1)
+ {
+ var_idx++;
+ if (r->vars[i].label)
+ add_string (&aux, r->vars[i].label, _("Variable %zu Label"),
+ var_idx);
+ }
+
+ k = 0;
+ for (i = 0; i < r->n_labels; i++)
+ for (j = 0; j < r->labels[i].n_labels; j++)
+ add_string (&aux, r->labels[i].labels[j].label,
+ _("Value Label %zu"), k++);
+
+ add_string (&aux, r->header.creation_date, _("Creation Date"));
+ add_string (&aux, r->header.creation_time, _("Creation Time"));
+ add_string (&aux, r->header.eye_catcher, _("Product"));
+ add_string (&aux, r->header.file_label, _("File Label"));
+
+ if (r->extensions[EXT_PRODUCT_INFO])
+ add_string (&aux, r->extensions[EXT_PRODUCT_INFO]->data,
+ _("Extra Product Info"));
+
+ if (r->document)
+ {
+ size_t i;
+
+ for (i = 0; i < r->document->n_lines; i++)
+ {
+ char line[81];
+
+ memcpy (line, r->document->documents + i * 80, 80);
+ line[80] = '\0';
+
+ add_string (&aux, line, _("Document Line %zu"), i + 1);
}
- if (!ok)
- goto error;
}
- if (r->compression == SFM_COMP_ZLIB && !read_zheader (r))
- goto error;
+ for (mrset = r->mrsets; mrset < &r->mrsets[r->n_mrsets]; mrset++)
+ {
+ size_t mrset_idx = mrset - r->mrsets + 1;
+
+ add_id (&aux, mrset->name, _("MRSET %zu"), mrset_idx);
+ if (mrset->label[0])
+ add_string (&aux, mrset->label, _("MRSET %zu Label"), mrset_idx);
+
+ /* Skip the variables because they ought to be duplicates. */
+
+ if (mrset->counted)
+ add_string (&aux, mrset->counted, _("MRSET %zu Counted Value"),
+ mrset_idx);
+ }
+
+ /* data file attributes */
+ /* variable attributes */
+ /* long var map */
+ /* long string value labels */
+ /* long string missing values */
+
+ *titlesp = aux.titles;
+ *idsp = aux.ids;
+ *stringsp = aux.strings;
+ return aux.n;
+}
+
+/* Decodes the dictionary read from R, saving it into into *DICT. Character
+ strings in R are decoded using ENCODING, or an encoding obtained from R if
+ ENCODING is null, or the locale encoding if R specifies no encoding.
+
+ If INFOP is non-null, then it receives additional info about the system
+ file, which the caller must eventually free with any_read_info_destroy()
+ when it is no longer needed.
+
+ This function consumes R. The caller must use it again later, even to
+ destroy it with sfm_close(). */
+static struct casereader *
+sfm_decode (struct any_reader *r_, const char *encoding,
+ struct dictionary **dictp, struct any_read_info *infop)
+{
+ struct sfm_reader *r = sfm_reader_cast (r_);
+ struct dictionary *dict;
+ size_t i;
- /* Now actually parse what we read.
+ if (encoding == NULL)
+ {
+ encoding = sfm_get_encoding (r);
+ if (encoding == NULL)
+ {
+ sys_warn (r, -1, _("This system file does not indicate its own "
+ "character encoding. Using default encoding "
+ "%s. For best results, specify an encoding "
+ "explicitly. Use SYSFILE INFO with "
+ "ENCODING=\"DETECT\" to analyze the possible "
+ "encodings."),
+ locale_charset ());
+ encoding = locale_charset ();
+ }
+ }
- First, figure out the correct character encoding, because this determines
- how the rest of the header data is to be interpreted. */
- dict = dict_create (encoding
- ? encoding
- : choose_encoding (r, &header, extensions[EXT_INTEGER],
- extensions[EXT_ENCODING]));
+ dict = dict_create (encoding);
r->encoding = dict_get_encoding (dict);
/* These records don't use variables at all. */
- if (document != NULL)
- parse_document (dict, document);
+ if (r->document != NULL)
+ parse_document (dict, r->document);
- if (extensions[EXT_INTEGER] != NULL
- && !parse_machine_integer_info (r, extensions[EXT_INTEGER], info))
+ if (r->extensions[EXT_INTEGER] != NULL
+ && !parse_machine_integer_info (r, r->extensions[EXT_INTEGER], &r->info))
goto error;
- if (extensions[EXT_FLOAT] != NULL)
- parse_machine_float_info (r, extensions[EXT_FLOAT]);
+ if (r->extensions[EXT_FLOAT] != NULL)
+ parse_machine_float_info (r, r->extensions[EXT_FLOAT]);
- if (extensions[EXT_PRODUCT_INFO] != NULL)
- parse_extra_product_info (r, extensions[EXT_PRODUCT_INFO], info);
+ if (r->extensions[EXT_PRODUCT_INFO] != NULL)
+ parse_extra_product_info (r, r->extensions[EXT_PRODUCT_INFO], &r->info);
- if (extensions[EXT_FILE_ATTRS] != NULL)
- parse_data_file_attributes (r, extensions[EXT_FILE_ATTRS], dict);
+ if (r->extensions[EXT_FILE_ATTRS] != NULL)
+ parse_data_file_attributes (r, r->extensions[EXT_FILE_ATTRS], dict);
- parse_header (r, &header, info, dict);
+ parse_header (r, &r->header, &r->info, dict);
/* Parse the variable records, the basis of almost everything else. */
- if (!parse_variable_records (r, dict, vars, n_vars))
+ if (!parse_variable_records (r, dict, r->vars, r->n_vars))
goto error;
/* Parse value labels and the weight variable immediately after the variable
records. These records use indexes into var_recs[], so we must parse them
before those indexes become invalidated by very long string variables. */
- for (i = 0; i < n_labels; i++)
- if (!parse_value_labels (r, dict, vars, n_vars, &labels[i]))
+ for (i = 0; i < r->n_labels; i++)
+ if (!parse_value_labels (r, dict, r->vars, r->n_vars, &r->labels[i]))
goto error;
- if (header.weight_idx != 0)
+ if (r->header.weight_idx != 0)
{
- struct variable *weight_var = lookup_var_by_index (r, 76, vars, n_vars,
- header.weight_idx);
+ struct variable *weight_var;
+
+ weight_var = lookup_var_by_index (r, 76, r->vars, r->n_vars,
+ r->header.weight_idx);
if (weight_var != NULL)
{
if (var_is_numeric (weight_var))
}
}
- if (extensions[EXT_DISPLAY] != NULL)
- parse_display_parameters (r, extensions[EXT_DISPLAY], dict);
+ if (r->extensions[EXT_DISPLAY] != NULL)
+ parse_display_parameters (r, r->extensions[EXT_DISPLAY], dict);
/* The following records use short names, so they need to be parsed before
parse_long_var_name_map() changes short names to long names. */
- if (extensions[EXT_MRSETS] != NULL)
- parse_mrsets (r, extensions[EXT_MRSETS], dict);
-
- if (extensions[EXT_MRSETS2] != NULL)
- parse_mrsets (r, extensions[EXT_MRSETS2], dict);
+ decode_mrsets (r, dict);
- if (extensions[EXT_LONG_STRINGS] != NULL
- && !parse_long_string_map (r, extensions[EXT_LONG_STRINGS], dict))
+ if (r->extensions[EXT_LONG_STRINGS] != NULL
+ && !parse_long_string_map (r, r->extensions[EXT_LONG_STRINGS], dict))
goto error;
/* Now rename variables to their long names. */
- parse_long_var_name_map (r, extensions[EXT_LONG_NAMES], dict);
+ parse_long_var_name_map (r, r->extensions[EXT_LONG_NAMES], dict);
/* The following records use long names, so they need to follow renaming. */
- if (extensions[EXT_VAR_ATTRS] != NULL)
+ if (!ll_is_empty (&r->var_attrs))
{
- parse_variable_attributes (r, extensions[EXT_VAR_ATTRS], dict);
+ struct sfm_extension_record *ext;
+ ll_for_each (ext, struct sfm_extension_record, ll, &r->var_attrs)
+ parse_variable_attributes (r, ext, dict);
/* Roles use the $@Role attribute. */
assign_variable_roles (r, dict);
}
-
- if (extensions[EXT_LONG_LABELS] != NULL
- && !parse_long_string_value_labels (r, extensions[EXT_LONG_LABELS],
- dict))
- goto error;
- if (extensions[EXT_LONG_MISSING] != NULL
- && !parse_long_string_missing_values (r, extensions[EXT_LONG_MISSING],
- dict))
- goto error;
+ if (r->extensions[EXT_LONG_LABELS] != NULL)
+ parse_long_string_value_labels (r, r->extensions[EXT_LONG_LABELS], dict);
+ if (r->extensions[EXT_LONG_MISSING] != NULL)
+ parse_long_string_missing_values (r, r->extensions[EXT_LONG_MISSING],
+ dict);
/* Warn if the actual amount of data per case differs from the
amount that the header claims. SPSS version 13 gets this
wrong when very long strings are involved, so don't warn in
that case. */
- if (header.nominal_case_size != -1 && header.nominal_case_size != n_vars
- && info->version_major != 13)
+ if (r->header.nominal_case_size > 0
+ && r->header.nominal_case_size != r->n_vars
+ && r->info.version_major != 13)
sys_warn (r, -1, _("File header claims %d variable positions but "
"%zu were read from file."),
- header.nominal_case_size, n_vars);
+ r->header.nominal_case_size, r->n_vars);
/* Create an index of dictionary variable widths for
sfm_read_case to use. We cannot use the `struct variable's
r->proto = caseproto_ref_pool (dict_get_proto (dict), r->pool);
*dictp = dict;
- if (infop != info)
+ if (infop)
{
- sfm_read_info_destroy (info);
- free (info);
+ *infop = r->info;
+ memset (&r->info, 0, sizeof r->info);
}
return casereader_create_sequential
&sys_file_casereader_class, r);
error:
- if (infop != info)
- {
- sfm_read_info_destroy (info);
- free (info);
- }
-
- close_reader (r);
+ sfm_close (r_);
dict_destroy (dict);
*dictp = NULL;
return NULL;
}
-/* Closes a system file after we're done with it.
+/* Closes R, which should have been returned by sfm_open() but not already
+ closed with sfm_decode() or this function.
Returns true if an I/O error has occurred on READER, false
otherwise. */
static bool
-close_reader (struct sfm_reader *r)
+sfm_close (struct any_reader *r_)
{
+ struct sfm_reader *r = sfm_reader_cast (r_);
bool error;
- if (r == NULL)
- return true;
-
if (r->file)
{
- if (fn_close (fh_get_file_name (r->fh), r->file) == EOF)
+ if (fn_close (r->fh, r->file) == EOF)
{
msg (ME, _("Error closing system file `%s': %s."),
fh_get_file_name (r->fh), strerror (errno));
r->file = NULL;
}
+ any_read_info_destroy (&r->info);
fh_unlock (r->lock);
fh_unref (r->fh);
sys_file_casereader_destroy (struct casereader *reader UNUSED, void *r_)
{
struct sfm_reader *r = r_;
- close_reader (r);
+ sfm_close (&r->any_reader);
}
-/* Returns true if FILE is an SPSS system file,
- false otherwise. */
-bool
+/* Detects whether FILE is an SPSS system file. Returns 1 if so, 0 if not, and
+ a negative errno value if there is an error reading FILE. */
+static int
sfm_detect (FILE *file)
{
char magic[5];
+ if (fseek (file, 0, SEEK_SET) != 0)
+ return -errno;
if (fread (magic, 4, 1, file) != 1)
- return false;
+ return ferror (file) ? -errno : 0;
magic[4] = '\0';
return (!strcmp (ASCII_MAGIC, magic)
except for the string fields in *INFO, which parse_header() will initialize
later once the file's encoding is known. */
static bool
-read_header (struct sfm_reader *r, struct sfm_read_info *info,
+read_header (struct sfm_reader *r, struct any_read_info *info,
struct sfm_header_record *header)
{
uint8_t raw_layout_code[4];
if (!read_string (r, header->magic, sizeof header->magic)
|| !read_string (r, header->eye_catcher, sizeof header->eye_catcher))
return false;
+ r->written_by_readstat = strstr (header->eye_catcher,
+ "https://github.com/WizardMac/ReadStat");
if (!strcmp (ASCII_MAGIC, header->magic)
|| !strcmp (EBCDIC_MAGIC, header->magic))
if (!zmagic)
{
if (compressed == 0)
- r->compression = SFM_COMP_NONE;
+ r->compression = ANY_COMP_NONE;
else if (compressed == 1)
- r->compression = SFM_COMP_SIMPLE;
+ r->compression = ANY_COMP_SIMPLE;
else if (compressed != 0)
{
sys_error (r, 0, "System file header has invalid compression "
else
{
if (compressed == 2)
- r->compression = SFM_COMP_ZLIB;
+ r->compression = ANY_COMP_ZLIB;
else
{
sys_error (r, 0, "ZLIB-compressed system file header has invalid "
|| !read_int (r, &record->missing_value_code)
|| !read_int (r, &record->print_format)
|| !read_int (r, &record->write_format)
- || !read_bytes (r, record->name, sizeof record->name))
+ || !read_string (r, record->name, sizeof record->name))
return false;
if (has_variable_label == 1)
{
- enum { MAX_LABEL_LEN = 255 };
+ enum { MAX_LABEL_LEN = 65536 };
unsigned int len, read_len;
if (!read_uint (r, &len))
/* Reads value labels from R into RECORD. */
static bool
read_value_label_record (struct sfm_reader *r,
- struct sfm_value_label_record *record,
- size_t n_vars)
+ struct sfm_value_label_record *record)
{
size_t i;
int type;
record->pos = r->pos;
if (!read_uint (r, &record->n_labels))
return false;
- if (record->n_labels > SIZE_MAX / sizeof *record->labels)
+ if (record->n_labels > UINT_MAX / sizeof *record->labels)
{
- sys_error (r, r->pos - 4, _("Invalid number of labels %zu."),
+ sys_error (r, r->pos - 4, _("Invalid number of labels %u."),
record->n_labels);
return false;
}
record. */
if (!read_uint (r, &record->n_vars))
return false;
- if (record->n_vars < 1 || record->n_vars > n_vars)
+ if (record->n_vars < 1 || record->n_vars > r->n_vars)
{
sys_error (r, r->pos - 4,
- _("Number of variables associated with a value label (%zu) "
+ _("Number of variables associated with a value label (%u) "
"is not between 1 and the number of variables (%zu)."),
- record->n_vars, n_vars);
+ record->n_vars, r->n_vars);
return false;
}
return true;
}
-/* Reads a document record from R and returns it. */
-static struct sfm_document_record *
+/* Reads a document record from R. Returns true if successful, false on
+ error. */
+static bool
read_document_record (struct sfm_reader *r)
{
- struct sfm_document_record *record;
int n_lines;
-
- record = pool_malloc (r->pool, sizeof *record);
- record->pos = r->pos;
-
if (!read_int (r, &n_lines))
- return NULL;
- if (n_lines <= 0 || n_lines >= INT_MAX / DOC_LINE_LENGTH)
+ return false;
+ else if (n_lines == 0)
+ return true;
+ else if (n_lines < 0 || n_lines >= INT_MAX / DOC_LINE_LENGTH)
{
- sys_error (r, record->pos,
+ sys_error (r, r->pos,
_("Number of document lines (%d) "
"must be greater than 0 and less than %d."),
n_lines, INT_MAX / DOC_LINE_LENGTH);
- return NULL;
+ return false;
}
+ struct sfm_document_record *record;
+ record = pool_malloc (r->pool, sizeof *record);
+ record->pos = r->pos;
record->n_lines = n_lines;
record->documents = pool_malloc (r->pool, DOC_LINE_LENGTH * n_lines);
if (!read_bytes (r, record->documents, DOC_LINE_LENGTH * n_lines))
- return NULL;
+ return false;
- return record;
+ r->document = record;
+ return true;
}
static bool
{
if (type->size > 0 && record->size != type->size)
sys_warn (r, record->pos,
- _("Record type 7, subtype %d has bad size %zu "
+ _("Record type 7, subtype %d has bad size %u "
"(expected %d)."), subtype, record->size, type->size);
else if (type->count > 0 && record->count != type->count)
sys_warn (r, record->pos,
- _("Record type 7, subtype %d has bad count %zu "
+ _("Record type 7, subtype %d has bad count %u "
"(expected %d)."), subtype, record->count, type->count);
else if (type->count == 0 && type->size == 0)
{
}
sys_warn (r, record->pos,
- _("Unrecognized record type 7, subtype %d. Please send a "
- "copy of this file, and the syntax which created it to %s."),
- subtype, PACKAGE_BUGREPORT);
+ _("Unrecognized record type 7, subtype %d. For help, please "
+ "send this file to %s and mention that you were using %s."),
+ subtype, PACKAGE_BUGREPORT, PACKAGE_STRING);
skip:
return skip_bytes (r, n_bytes);
static void
parse_header (struct sfm_reader *r, const struct sfm_header_record *header,
- struct sfm_read_info *info, struct dictionary *dict)
+ struct any_read_info *info, struct dictionary *dict)
{
const char *dict_encoding = dict_get_encoding (dict);
struct substring product;
size_t i;
name = recode_string_pool ("UTF-8", dict_encoding,
- rec->name, 8, r->pool);
+ rec->name, -1, r->pool);
name[strcspn (name, " ")] = '\0';
if (!dict_id_is_valid (dict, name, false)
"`%s' to `%s'."),
name, new_name);
var = rec->var = dict_create_var_assert (dict, new_name, rec->width);
+ var_set_short_name (var, 0, new_name);
free (new_name);
}
- /* Set the short name the same as the long name. */
- var_set_short_name (var, 0, name);
+ /* Set the short name the same as the long name (even if we renamed
+ it). */
+ var_set_short_name (var, 0, var_get_name (var));
/* Get variable label, if any. */
if (rec->label)
utf8_label = recode_string_pool ("UTF-8", dict_encoding,
rec->label, -1, r->pool);
- var_set_label (var, utf8_label, false);
+ var_set_label (var, utf8_label);
}
/* Set missing values. */
}
}
else
- {
- union value value;
-
- value_init_pool (r->pool, &value, width);
- value_set_missing (&value, width);
- for (i = 0; i < rec->missing_value_code; i++)
- mv_add_str (&mv, rec->missing + 8 * i, MIN (width, 8));
- }
+ for (i = 0; i < rec->missing_value_code; i++)
+ mv_add_str (&mv, rec->missing + 8 * i, MIN (width, 8));
var_set_missing_values (var, &mv);
}
static bool
parse_machine_integer_info (struct sfm_reader *r,
const struct sfm_extension_record *record,
- struct sfm_read_info *info)
+ struct any_read_info *info)
{
int float_representation, expected_float_format;
int integer_representation, expected_integer_format;
return true;
}
-static const char *
-choose_encoding (struct sfm_reader *r,
- const struct sfm_header_record *header,
- const struct sfm_extension_record *ext_integer,
- const struct sfm_extension_record *ext_encoding)
-{
- /* The EXT_ENCODING record is a more reliable way to determine dictionary
- encoding. */
- if (ext_encoding)
- return ext_encoding->data;
-
- /* But EXT_INTEGER is better than nothing as a fallback. */
- if (ext_integer)
- {
- int codepage = parse_int (r, ext_integer->data, 7 * 4);
- const char *encoding;
-
- switch (codepage)
- {
- case 1:
- return "EBCDIC-US";
-
- case 2:
- case 3:
- /* These ostensibly mean "7-bit ASCII" and "8-bit ASCII"[sic]
- respectively. However, there are known to be many files in the wild
- with character code 2, yet have data which are clearly not ASCII.
- Therefore we ignore these values. */
- break;
-
- case 4:
- return "MS_KANJI";
-
- default:
- encoding = sys_get_encoding_from_codepage (codepage);
- if (encoding != NULL)
- return encoding;
- break;
- }
- }
-
- /* If the file magic number is EBCDIC then its character data is too. */
- if (!strcmp (header->magic, EBCDIC_MAGIC))
- return "EBCDIC-US";
-
- return locale_charset ();
-}
-
/* Parses record type 7, subtype 4. */
static void
parse_machine_float_info (struct sfm_reader *r,
static void
parse_extra_product_info (struct sfm_reader *r,
const struct sfm_extension_record *record,
- struct sfm_read_info *info)
+ struct any_read_info *info)
{
struct text_record *text;
/* Parses record type 7, subtype 7 or 19. */
static void
parse_mrsets (struct sfm_reader *r, const struct sfm_extension_record *record,
- struct dictionary *dict)
+ size_t *allocated_mrsets)
{
struct text_record *text;
- struct mrset *mrset;
text = open_text_record (r, record, false);
for (;;)
{
- const char *counted = NULL;
- const char *name;
- const char *label;
- struct stringi_set var_names;
+ struct sfm_mrset *mrset;
size_t allocated_vars;
char delimiter;
- int width;
/* Skip extra line feeds if present. */
while (text_match (text, '\n'))
continue;
- mrset = xzalloc (sizeof *mrset);
+ if (r->n_mrsets >= *allocated_mrsets)
+ r->mrsets = pool_2nrealloc (r->pool, r->mrsets, allocated_mrsets,
+ sizeof *r->mrsets);
+ mrset = &r->mrsets[r->n_mrsets];
+ memset(mrset, 0, sizeof *mrset);
- name = text_get_token (text, ss_cstr ("="), NULL);
- if (name == NULL)
+ mrset->name = text_get_token (text, ss_cstr ("="), NULL);
+ if (mrset->name == NULL)
break;
- mrset->name = recode_string ("UTF-8", r->encoding, name, -1);
-
- if (mrset->name[0] != '$')
- {
- sys_warn (r, record->pos,
- _("`%s' does not begin with `$' at offset %zu "
- "in MRSETS record."), mrset->name, text_pos (text));
- break;
- }
if (text_match (text, 'C'))
{
mrset->label_from_var_label = true;
else if (strcmp (number, "1"))
sys_warn (r, record->pos,
- _("Unexpected label source value `%s' following `E' "
+ _("Unexpected label source value following `E' "
"at offset %zu in MRSETS record."),
- number, text_pos (text));
+ text_pos (text));
}
else
{
if (mrset->type == MRSET_MD)
{
- counted = text_parse_counted_string (r, text);
- if (counted == NULL)
+ mrset->counted = text_parse_counted_string (r, text);
+ if (mrset->counted == NULL)
break;
}
- label = text_parse_counted_string (r, text);
- if (label == NULL)
+ mrset->label = text_parse_counted_string (r, text);
+ if (mrset->label == NULL)
break;
- if (label[0] != '\0')
- mrset->label = recode_string ("UTF-8", r->encoding, label, -1);
- stringi_set_init (&var_names);
allocated_vars = 0;
- width = INT_MAX;
do
{
- const char *raw_var_name;
- struct variable *var;
- char *var_name;
+ const char *var;
- raw_var_name = text_get_token (text, ss_cstr (" \n"), &delimiter);
- if (raw_var_name == NULL)
+ var = text_get_token (text, ss_cstr (" \n"), &delimiter);
+ if (var == NULL)
{
if (delimiter != '\n')
sys_warn (r, record->pos,
text_pos (text));
break;
}
- var_name = recode_string ("UTF-8", r->encoding, raw_var_name, -1);
+
+ if (mrset->n_vars >= allocated_vars)
+ mrset->vars = pool_2nrealloc (r->pool, mrset->vars,
+ &allocated_vars,
+ sizeof *mrset->vars);
+ mrset->vars[mrset->n_vars++] = var;
+ }
+ while (delimiter != '\n');
+
+ r->n_mrsets++;
+ }
+ close_text_record (r, text);
+}
+
+static void
+decode_mrsets (struct sfm_reader *r, struct dictionary *dict)
+{
+ const struct sfm_mrset *s;
+
+ for (s = r->mrsets; s < &r->mrsets[r->n_mrsets]; s++)
+ {
+ struct stringi_set var_names;
+ struct mrset *mrset;
+ char *name;
+ int width;
+ size_t i;
+
+ name = recode_string ("UTF-8", r->encoding, s->name, -1);
+ if (!mrset_is_valid_name (name, dict_get_encoding (dict), false))
+ {
+ sys_warn (r, -1, _("Invalid multiple response set name `%s'."),
+ name);
+ free (name);
+ continue;
+ }
+
+ mrset = xzalloc (sizeof *mrset);
+ mrset->name = name;
+ mrset->type = s->type;
+ mrset->cat_source = s->cat_source;
+ mrset->label_from_var_label = s->label_from_var_label;
+ if (s->label[0] != '\0')
+ mrset->label = recode_string ("UTF-8", r->encoding, s->label, -1);
+
+ stringi_set_init (&var_names);
+ mrset->vars = xmalloc (s->n_vars * sizeof *mrset->vars);
+ width = INT_MAX;
+ for (i = 0; i < s->n_vars; i++)
+ {
+ struct variable *var;
+ char *var_name;
+
+ var_name = recode_string ("UTF-8", r->encoding, s->vars[i], -1);
var = dict_lookup_var (dict, var_name);
if (var == NULL)
}
if (!stringi_set_insert (&var_names, var_name))
{
- sys_warn (r, record->pos,
- _("Duplicate variable name %s "
- "at offset %zu in MRSETS record."),
- var_name, text_pos (text));
+ sys_warn (r, -1,
+ _("MRSET %s contains duplicate variable name %s."),
+ mrset->name, var_name);
free (var_name);
continue;
}
if (mrset->n_vars
&& var_get_type (var) != var_get_type (mrset->vars[0]))
{
- sys_warn (r, record->pos,
+ sys_warn (r, -1,
_("MRSET %s contains both string and "
- "numeric variables."), name);
+ "numeric variables."), mrset->name);
continue;
}
width = MIN (width, var_get_width (var));
- if (mrset->n_vars >= allocated_vars)
- mrset->vars = x2nrealloc (mrset->vars, &allocated_vars,
- sizeof *mrset->vars);
mrset->vars[mrset->n_vars++] = var;
}
- while (delimiter != '\n');
if (mrset->n_vars < 2)
{
- sys_warn (r, record->pos,
- _("MRSET %s has only %zu variables."), mrset->name,
- mrset->n_vars);
+ if (mrset->n_vars == 0)
+ sys_warn (r, -1, _("MRSET %s has no variables."), mrset->name);
+ else
+ sys_warn (r, -1, _("MRSET %s has only one variable."),
+ mrset->name);
mrset_destroy (mrset);
stringi_set_destroy (&var_names);
continue;
mrset->width = width;
value_init (&mrset->counted, width);
if (width == 0)
- mrset->counted.f = c_strtod (counted, NULL);
+ mrset->counted.f = c_strtod (s->counted, NULL);
else
value_copy_str_rpad (&mrset->counted, width,
- (const uint8_t *) counted, ' ');
+ (const uint8_t *) s->counted, ' ');
}
dict_add_mrset (dict, mrset);
- mrset = NULL;
stringi_set_destroy (&var_names);
}
- mrset_destroy (mrset);
- close_text_record (r, text);
}
/* Read record type 7, subtype 11, which specifies how variables
else
{
sys_warn (r, record->pos,
- _("Extension 11 has bad count %zu (for %zu variables)."),
+ _("Extension 11 has bad count %u (for %zu variables)."),
record->count, n_vars);
return;
}
}
static void
-rename_var_and_save_short_names (struct dictionary *dict, struct variable *var,
- const char *new_name)
+rename_var_and_save_short_names (struct sfm_reader *r, off_t pos,
+ struct dictionary *dict,
+ struct variable *var, const char *new_name)
{
size_t n_short_names;
char **short_names;
}
/* Set long name. */
- dict_rename_var (dict, var, new_name);
+ if (!dict_try_rename_var (dict, var, new_name))
+ sys_warn (r, pos, _("Duplicate long variable name `%s'."), new_name);
/* Restore short names. */
for (i = 0; i < n_short_names; i++)
char *new_name;
new_name = utf8_to_lower (var_get_name (var));
- rename_var_and_save_short_names (dict, var, new_name);
+ rename_var_and_save_short_names (r, -1, dict, var, new_name);
free (new_name);
}
while (read_variable_to_value_pair (r, dict, text, &var, &long_name))
{
/* Validate long name. */
- if (!dict_id_is_valid (dict, long_name, false))
+ if (!dict_id_is_valid (dict, long_name, false)
+ || long_name[0] == '$' || long_name[0] == '#')
{
sys_warn (r, record->pos,
_("Long variable mapping from %s to invalid "
continue;
}
- /* Identify any duplicates. */
- if (utf8_strcasecmp (var_get_short_name (var, 0), long_name)
- && dict_lookup_var (dict, long_name) != NULL)
- {
- sys_warn (r, record->pos,
- _("Duplicate long variable name `%s'."), long_name);
- continue;
- }
-
- rename_var_and_save_short_names (dict, var, long_name);
+ rename_var_and_save_short_names (r, record->pos, dict, var, long_name);
}
close_text_record (r, text);
}
if (!var_add_value_label (var, &value, utf8_labels[j]))
{
- if (var_is_numeric (var))
+ if (r->written_by_readstat)
+ {
+ /* Ignore the problem. ReadStat is buggy and emits value
+ labels whose values are longer than string variables'
+ widths, that are identical in the actual width of the
+ variable, e.g. both values "ABC123" and "ABC456" for a
+ string variable with width 3. */
+ }
+ else if (var_is_numeric (var))
sys_warn (r, record->pos,
_("Duplicate value label for %g on %s."),
value.f, var_get_name (var));
text_warn (r, text, _("Error parsing attribute value %s[%d]."),
key, index);
break;
- }
+ }
length = strlen (value);
- if (length >= 2 && value[0] == '\'' && value[length - 1] == '\'')
+ if (length >= 2 && value[0] == '\'' && value[length - 1] == '\'')
{
value[length - 1] = '\0';
- attribute_add_value (attr, value + 1);
+ attribute_add_value (attr, value + 1);
}
- else
+ else
{
text_warn (r, text,
_("Attribute value %s[%d] is not quoted: %s."),
key, index, value);
- attribute_add_value (attr, value);
+ attribute_add_value (attr, value);
}
/* Was this the last value for this attribute? */
break;
}
if (attrs != NULL)
- attrset_add (attrs, attr);
+ {
+ if (!attrset_try_add (attrs, attr))
+ {
+ text_warn (r, text, _("Duplicate attribute %s."),
+ attribute_get_name (attr));
+ attribute_destroy (attr);
+ }
+ }
else
attribute_destroy (attr);
}
size_t end = record->size * record->count;
if (length >= end || ofs + length > end)
{
- sys_error (r, record->pos + end,
- _("Extension record subtype %d ends unexpectedly."),
- record->subtype);
+ sys_warn (r, record->pos + end,
+ _("Extension record subtype %d ends unexpectedly."),
+ record->subtype);
return false;
}
return true;
}
-static bool
+static void
parse_long_string_value_labels (struct sfm_reader *r,
const struct sfm_extension_record *record,
struct dictionary *dict)
/* Parse variable name length. */
if (!check_overflow (r, record, ofs, 4))
- return false;
+ return;
var_name_len = parse_int (r, record->data, ofs);
ofs += 4;
/* Parse variable name, width, and number of labels. */
- if (!check_overflow (r, record, ofs, var_name_len + 8))
- return false;
+ if (!check_overflow (r, record, ofs, var_name_len)
+ || !check_overflow (r, record, ofs, var_name_len + 8))
+ return;
var_name = recode_string_pool ("UTF-8", dict_encoding,
(const char *) record->data + ofs,
var_name_len, r->pool);
/* Parse value length. */
if (!check_overflow (r, record, ofs, 4))
- return false;
+ return;
value_length = parse_int (r, record->data, ofs);
ofs += 4;
/* Parse value. */
if (!check_overflow (r, record, ofs, value_length))
- return false;
+ return;
if (!skip)
{
if (value_length == width)
/* Parse label length. */
if (!check_overflow (r, record, ofs, 4))
- return false;
+ return;
label_length = parse_int (r, record->data, ofs);
ofs += 4;
/* Parse label. */
if (!check_overflow (r, record, ofs, label_length))
- return false;
+ return;
if (!skip)
{
char *label;
ofs += label_length;
}
}
-
- return true;
}
-static bool
+static void
parse_long_string_missing_values (struct sfm_reader *r,
const struct sfm_extension_record *record,
struct dictionary *dict)
/* Parse variable name length. */
if (!check_overflow (r, record, ofs, 4))
- return false;
+ return;
var_name_len = parse_int (r, record->data, ofs);
ofs += 4;
/* Parse variable name. */
- if (!check_overflow (r, record, ofs, var_name_len + 1))
- return false;
+ if (!check_overflow (r, record, ofs, var_name_len)
+ || !check_overflow (r, record, ofs, var_name_len + 1))
+ return;
var_name = recode_string_pool ("UTF-8", dict_encoding,
(const char *) record->data + ofs,
var_name_len, r->pool);
/* Parse value length. */
if (!check_overflow (r, record, ofs, 4))
- return false;
+ return;
value_length = parse_int (r, record->data, ofs);
ofs += 4;
/* Parse value. */
if (!check_overflow (r, record, ofs, value_length))
- return false;
+ return;
if (var != NULL
&& i < 3
&& !mv_add_str (&mv, (const uint8_t *) record->data + ofs,
if (var != NULL)
var_set_missing_values (var, &mv);
}
-
- return true;
}
\f
/* Case reader. */
int retval;
int i;
- if (r->error)
+ if (r->error || !r->sfm_var_cnt)
return NULL;
c = case_create (r->proto);
static bool
read_case_number (struct sfm_reader *r, double *d)
{
- if (r->compression == SFM_COMP_NONE)
+ if (r->compression == ANY_COMP_NONE)
{
uint8_t number[8];
if (!try_read_bytes (r, number, sizeof number))
static int
read_opcode (struct sfm_reader *r)
{
- assert (r->compression != SFM_COMP_NONE);
+ assert (r->compression != ANY_COMP_NONE);
for (;;)
{
int opcode;
read_whole_strings (struct sfm_reader *r, uint8_t *s, size_t length)
{
assert (length % 8 == 0);
- if (r->compression == SFM_COMP_NONE)
+ if (r->compression == ANY_COMP_NONE)
return try_read_bytes (r, s, length);
else
{
}
/* Closes TEXT, frees its storage, and issues a final warning
- about suppressed warnings if necesary. */
+ about suppressed warnings if necessary. */
static void
close_text_record (struct sfm_reader *r, struct text_record *text)
{
{
if (!text_read_short_name (r, dict, text, ss_cstr ("="), var))
return false;
-
+
*value = text_get_token (text, ss_buffer ("\t\0", 2), NULL);
if (*value == NULL)
return false;
text_warn (struct sfm_reader *r, struct text_record *text,
const char *format, ...)
{
- if (text->n_warnings++ < MAX_TEXT_WARNINGS)
+ if (text->n_warnings++ < MAX_TEXT_WARNINGS)
{
va_list args;
static bool
text_match (struct text_record *text, char c)
{
- if (text->buffer.string[text->pos] == c)
+ if (text->pos >= text->buffer.length)
+ return false;
+
+ if (text->buffer.string[text->pos] == c)
{
text->pos++;
return true;
va_end (args);
}
-/* Displays an error for the current file position,
- marks it as in an error state,
- and aborts reading it using longjmp. */
+/* Displays an error for the current file position and marks it as in an error
+ state. */
static void
sys_error (struct sfm_reader *r, off_t offset, const char *format, ...)
{
}
static int
-parse_int (struct sfm_reader *r, const void *data, size_t ofs)
+parse_int (const struct sfm_reader *r, const void *data, size_t ofs)
{
return integer_get (r->integer_format, (const uint8_t *) data + ofs, 4);
}
static double
-parse_float (struct sfm_reader *r, const void *data, size_t ofs)
+parse_float (const struct sfm_reader *r, const void *data, size_t ofs)
{
return float_get_double (r->float_format, (const uint8_t *) data + ofs);
}
if (fstat (fileno (r->file), &s))
{
- sys_error (ME, 0, _("%s: stat failed (%s)."),
+ sys_error (r, 0, _("%s: stat failed (%s)."),
fh_get_file_name (r->fh), strerror (errno));
return false;
}
static int
read_compressed_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
{
- if (r->compression == SFM_COMP_SIMPLE)
+ if (r->compression == ANY_COMP_SIMPLE)
return read_bytes (r, buf, byte_cnt);
else
{
static int
try_read_compressed_bytes (struct sfm_reader *r, void *buf, size_t byte_cnt)
{
- if (r->compression == SFM_COMP_SIMPLE)
+ if (r->compression == ANY_COMP_SIMPLE)
return try_read_bytes (r, buf, byte_cnt);
else
return read_bytes_zlib (r, buf, byte_cnt);
NULL,
NULL,
};
+
+const struct any_reader_class sys_file_reader_class =
+ {
+ N_("SPSS System File"),
+ sfm_detect,
+ sfm_open,
+ sfm_close,
+ sfm_decode,
+ sfm_get_strings,
+ };
projects / pspp / blobdiff