X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fsys-file-reader.c;h=bc58e4e45de0fe1b0513b7c323d3b3df094bfd0c;hb=2766c2d47448010527d52dd304213d0bb563dd00;hp=ebf0a7b1bb7194f1cf1ea9ecaac61bbe7c4699fa;hpb=1b47c4afe993f715c03e7392427b0b8919a26d4e;p=pspp-builds.git diff --git a/src/data/sys-file-reader.c b/src/data/sys-file-reader.c index ebf0a7b1..bc58e4e4 100644 --- a/src/data/sys-file-reader.c +++ b/src/data/sys-file-reader.c @@ -30,6 +30,7 @@ #include #include #include +#include #include "sys-file-reader.h" #include "sfm-private.h" @@ -40,49 +41,50 @@ #include "format.h" #include "value-labels.h" #include "variable.h" +#include "value.h" #include "gettext.h" #define _(msgid) gettext (msgid) /* System file reader. */ struct sfm_reader - { - struct file_handle *fh; /* File handle. */ - FILE *file; /* File stream. */ - - int reverse_endian; /* 1=file has endianness opposite us. */ - int fix_specials; /* 1=SYSMIS/HIGHEST/LOWEST differs from us. */ - int value_cnt; /* Number of `union values's per case. */ - long case_cnt; /* Number of cases, -1 if unknown. */ - int compressed; /* 1=compressed, 0=not compressed. */ - double bias; /* Compression bias, usually 100.0. */ - int weight_idx; /* 0-based index of weighting variable, or -1. */ - bool ok; /* False after an I/O error or corrupt data. */ - - /* Variables. */ - struct sfm_var *vars; /* Variables. */ - - /* File's special constants. */ - flt64 sysmis; - flt64 highest; - flt64 lowest; - - /* Decompression buffer. */ - flt64 *buf; /* Buffer data. */ - flt64 *ptr; /* Current location in buffer. */ - flt64 *end; /* End of buffer data. */ - - /* Compression instruction octet. */ - unsigned char x[8]; /* Current instruction octet. */ - unsigned char *y; /* Location in current instruction octet. */ - }; +{ + struct file_handle *fh; /* File handle. */ + FILE *file; /* File stream. */ + + int reverse_endian; /* 1=file has endianness opposite us. */ + int fix_specials; /* 1=SYSMIS/HIGHEST/LOWEST differs from us. */ + int value_cnt; /* Number of `union values's per case. */ + long case_cnt; /* Number of cases, -1 if unknown. */ + int compressed; /* 1=compressed, 0=not compressed. */ + double bias; /* Compression bias, usually 100.0. */ + int weight_idx; /* 0-based index of weighting variable, or -1. */ + bool ok; /* False after an I/O error or corrupt data. */ + + /* Variables. */ + struct sfm_var *vars; /* Variables. */ + + /* File's special constants. */ + flt64 sysmis; + flt64 highest; + flt64 lowest; + + /* Decompression buffer. */ + flt64 *buf; /* Buffer data. */ + flt64 *ptr; /* Current location in buffer. */ + flt64 *end; /* End of buffer data. */ + + /* Compression instruction octet. */ + unsigned char x[8]; /* Current instruction octet. */ + unsigned char *y; /* Location in current instruction octet. */ +}; /* A variable in a system file. */ struct sfm_var - { - int width; /* 0=numeric, otherwise string width. */ - int fv; /* Index into case. */ - }; +{ + int width; /* 0=numeric, otherwise string width. */ + int fv; /* Index into case. */ +}; /* Utilities. */ @@ -119,22 +121,26 @@ static void corrupt_msg (int class, const char *format,...) PRINTF_FORMAT (2, 3); -/* Displays a corrupt sysfile error. */ -static void -corrupt_msg (int class, const char *format,...) + /* Displays a corrupt sysfile error. */ + static void + corrupt_msg (int class, const char *format,...) { - struct error e; + struct msg m; va_list args; + struct string text; - e.category = msg_class_to_category (class); - e.severity = msg_class_to_severity (class); - e.where.file_name = NULL; - e.where.line_number = 0; - e.title = _("corrupt system file: "); - + ds_create (&text, _("corrupt system file: ")); va_start (args, format); - err_vmsg (&e, format, args); + ds_vprintf (&text, format, args); va_end (args); + + m.category = msg_class_to_category (class); + m.severity = msg_class_to_severity (class); + m.where.file_name = NULL; + m.where.line_number = 0; + m.text = ds_c_str (&text); + + msg_emit (&m); } /* Closes a system file after we're done with it. */ @@ -154,7 +160,7 @@ sfm_close_reader (struct sfm_reader *r) if (r->fh != NULL) fh_close (r->fh, "system file", "rs"); - + free (r->vars); free (r->buf); free (r); @@ -197,6 +203,69 @@ static int fread_ok (struct sfm_reader *, void *, size_t); goto error; \ } while (0) + +struct name_pair +{ + char *shortname; + char *longname; +}; + +static int +pair_sn_compare(const void *_p1, const void *_p2, void *aux UNUSED) +{ + int i; + + const struct name_pair *p1 = _p1; + const struct name_pair *p2 = _p2; + + char buf1[SHORT_NAME_LEN + 1]; + char buf2[SHORT_NAME_LEN + 1]; + + memset(buf1, 0, SHORT_NAME_LEN + 1); + memset(buf2, 0, SHORT_NAME_LEN + 1); + + for (i = 0 ; i <= SHORT_NAME_LEN ; ++i ) + { + buf1[i] = p1->shortname[i]; + if ( '\0' == buf1[i]) + break; + } + + for (i = 0 ; i <= SHORT_NAME_LEN ; ++i ) + { + buf2[i] = p2->shortname[i]; + if ( '\0' == buf2[i]) + break; + } + + return strncmp(buf1, buf2, SHORT_NAME_LEN); +} + +static unsigned int +pair_sn_hash(const void *_p, void *aux UNUSED) +{ + int i; + const struct name_pair *p = _p; + char buf[SHORT_NAME_LEN + 1]; + + memset(buf, 0, SHORT_NAME_LEN + 1); + for (i = 0 ; i <= SHORT_NAME_LEN ; ++i ) + { + buf[i] = p->shortname[i]; + if ( '\0' == buf[i]) + break; + } + + return hsh_hash_bytes(buf, strlen(buf)); +} + +static void +pair_sn_free(void *p, void *aux UNUSED) +{ + free(p); +} + + /* Opens the system file designated by file handle FH for reading. Reads the system file's dictionary into *DICT. If INFO is non-null, then it receives additional info about the @@ -208,6 +277,13 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, struct sfm_reader *r = NULL; struct variable **var_by_idx = NULL; + /* The data in record 7(14) */ + char *subrec14data = 0; + + /* A hash table of long variable names indexed by short name */ + struct hsh_table *short_to_long = NULL; + + *dict = dict_create (); if (!fh_open (fh, FH_REF_FILE, "system file", "rs")) goto error; @@ -265,7 +341,7 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, if (weight_var == NULL) lose ((ME, _("%s: Weighting variable may not be a continuation of " - "a long string variable."), fh_get_file_name (fh))); + "a long string variable."), fh_get_file_name (fh))); else if (weight_var->type == ALPHA) lose ((ME, _("%s: Weighting variable may not be a string variable."), fh_get_file_name (fh))); @@ -284,6 +360,7 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, if (r->reverse_endian) bswap_int32 (&rec_type); + switch (rec_type) { case 3: @@ -305,11 +382,11 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, case 7: { struct - { - int32_t subtype P; - int32_t size P; - int32_t count P; - } + { + int32_t subtype P; + int32_t size P; + int32_t count P; + } data; unsigned long bytes; @@ -323,6 +400,7 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, bswap_int32 (&data.count); } bytes = data.size * data.count; + if (bytes < data.size || bytes < data.count) lose ((ME, "%s: Record type %d subtype %d too large.", fh_get_file_name (r->fh), rec_type, data.subtype)); @@ -348,12 +426,13 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, { const int n_vars = data.count / 3 ; int i; - if ( data.count % 3 || n_vars > dict_get_var_cnt(*dict) ) + if ( data.count % 3 || n_vars != dict_get_var_cnt(*dict) ) { msg (MW, _("%s: Invalid subrecord length. " "Record: 7; Subrecord: 11"), fh_get_file_name (r->fh)); skip = 1; + break; } for ( i = 0 ; i < min(n_vars, dict_get_var_cnt(*dict)) ; ++i ) @@ -381,23 +460,29 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, case 13: /* SPSS 12.0 Long variable name map */ { - char *buf, *short_name, *save_ptr; + char *short_name, *save_ptr; int idx; /* Read data. */ - buf = xmalloc (bytes + 1); - if (!buf_read (r, buf, bytes, 0)) + subrec14data = xmalloc (bytes + 1); + if (!buf_read (r, subrec14data, bytes, 0)) { - free (buf); goto error; } - buf[bytes] = '\0'; + subrec14data[bytes] = '\0'; + + short_to_long = hsh_create(4, + pair_sn_compare, + pair_sn_hash, + pair_sn_free, + 0); /* Parse data. */ - for (short_name = strtok_r (buf, "=", &save_ptr), idx = 0; + for (short_name = strtok_r (subrec14data, "=", &save_ptr), idx = 0; short_name != NULL; short_name = strtok_r (NULL, "=", &save_ptr), idx++) { + struct name_pair *pair ; char *long_name = strtok_r (NULL, "\t", &save_ptr); struct variable *v; @@ -429,7 +514,7 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, /* Identify any duplicates. */ if ( compare_var_names(short_name, long_name, 0) && - NULL != dict_lookup_var (*dict, long_name)) + NULL != dict_lookup_var (*dict, long_name)) lose ((ME, _("%s: Duplicate long variable name `%s' " "within system file."), fh_get_file_name (r->fh), long_name)); @@ -442,6 +527,17 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, dict_rename_var (*dict, v, long_name); var_set_short_name (v, short_name); + pair = xmalloc(sizeof *pair); + pair->shortname = short_name; + pair->longname = long_name; + hsh_insert(short_to_long, pair); +#if 0 + /* This messes up the processing of subtype 14 (below). + I'm not sure if it is needed anyway, so I'm removing it for + now. If it's needed, then it will need to be done after all the + records have been processed. --- JMD 27 April 2006 + */ + /* For compatability, make sure dictionary is in long variable name map order. In the common case, this has no effect, @@ -449,10 +545,115 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, variable name map are already in the same order. */ dict_reorder_var (*dict, v, idx); +#endif } + + } + break; - /* Free data. */ - free (buf); + case 14: + { + int j = 0; + bool eq_seen = false; + int i; + + /* Read data. */ + char *buffer = xmalloc (bytes + 1); + if (!buf_read (r, buffer, bytes, 0)) + { + free (buffer); + goto error; + } + buffer[bytes] = '\0'; + + + /* Note: SPSS v13 terminates this record with 00, + whereas SPSS v14 terminates it with 00 09. We must + accept either */ + for(i = 0; i < bytes ; ++i) + { + long int length; + static char name[SHORT_NAME_LEN + 1] = {0}; + static char len_str[6] ={0}; + + switch( buffer[i] ) + { + case '=': + eq_seen = true; + j = 0; + break; + case '\0': + length = strtol(len_str, 0, 10); + if ( length != LONG_MAX && length != LONG_MIN) + { + char *lookup_name = name; + int l; + int idx; + struct variable *v; + + if ( short_to_long ) + { + struct name_pair pair; + struct name_pair *p; + + pair.shortname = name; + p = hsh_find(short_to_long, &pair); + if ( p ) + lookup_name = p->longname; + } + + v = dict_lookup_var(*dict, lookup_name); + if ( !v ) + { + corrupt_msg(MW, + _("%s: No variable called %s but it is listed in length table."), + fh_get_file_name (r->fh), lookup_name); + + goto error; + + } + + l = length; + if ( v->width > EFFECTIVE_LONG_STRING_LENGTH ) + l -= EFFECTIVE_LONG_STRING_LENGTH; + else + l -= v->width; + + idx = v->index; + while ( l > 0 ) + { + struct variable *v_next; + v_next = dict_get_var(*dict, idx + 1); + + if ( v_next->width > EFFECTIVE_LONG_STRING_LENGTH ) + l -= EFFECTIVE_LONG_STRING_LENGTH; + else + l -= v_next->width; + + dict_delete_var(*dict, v_next); + } + + v->width = length; + v->print.w = v->width; + v->write.w = v->width; + } + eq_seen = false; + memset(name, 0, SHORT_NAME_LEN+1); + memset(len_str, 0, 6); + j = 0; + break; + case '\t': + break; + default: + if ( eq_seen ) + len_str[j] = buffer[i]; + else + name[j] = buffer[i]; + j++; + break; + } + } + free(buffer); } break; @@ -478,24 +679,31 @@ sfm_open_reader (struct file_handle *fh, struct dictionary **dict, int32_t filler; assertive_buf_read (r, &filler, sizeof filler, 0); + goto success; } default: corrupt_msg(MW, _("%s: Unrecognized record type %d."), - fh_get_file_name (r->fh), rec_type); + fh_get_file_name (r->fh), rec_type); } } -success: + success: /* Come here on successful completion. */ + + free (var_by_idx); + hsh_destroy(short_to_long); + free (subrec14data); return r; -error: + error: /* Come here on unsuccessful completion. */ sfm_close_reader (r); free (var_by_idx); + hsh_destroy(short_to_long); + free (subrec14data); if (*dict != NULL) { dict_destroy (*dict); @@ -546,7 +754,7 @@ read_machine_int32_info (struct sfm_reader *r, int size, int count) fh_get_file_name (r->fh), file_bigendian ? _("big-endian") : _("little-endian"), data[6] == 1 ? _("big-endian") : (data[6] == 2 ? _("little-endian") - : _("unknown")))); + : _("unknown")))); /* PORTME: Character representation code. */ if (data[7] != 2 && data[7] != 3) @@ -558,7 +766,7 @@ read_machine_int32_info (struct sfm_reader *r, int size, int count) return 1; -error: + error: return 0; } @@ -596,7 +804,7 @@ read_machine_flt64_info (struct sfm_reader *r, int size, int count) return 1; -error: + error: return 0; } @@ -659,7 +867,7 @@ read_header (struct sfm_reader *r, fh_get_file_name (r->fh), hdr.layout_code)); r->reverse_endian = 1; - bswap_int32 (&hdr.case_size); + bswap_int32 (&hdr.nominal_case_size); bswap_int32 (&hdr.compress); bswap_int32 (&hdr.weight_idx); bswap_int32 (&hdr.case_cnt); @@ -668,9 +876,10 @@ read_header (struct sfm_reader *r, /* Copy basic info and verify correctness. */ - r->value_cnt = hdr.case_size; + r->value_cnt = hdr.nominal_case_size; - /* If value count is rediculous, then force it to -1 (a sentinel value) */ + /* If value count is ridiculous, then force it to -1 (a + sentinel value). */ if ( r->value_cnt < 0 || r->value_cnt > (INT_MAX / (int) sizeof (union value) / 2)) r->value_cnt = -1; @@ -739,7 +948,7 @@ read_header (struct sfm_reader *r, return 1; -error: + error: return 0; } @@ -777,9 +986,6 @@ read_variables (struct sfm_reader *r, int nv; int j; - if ( r->value_cnt != -1 && i >= r->value_cnt ) - break; - assertive_buf_read (r, &sv, sizeof sv, 0); if (r->reverse_endian) @@ -800,11 +1006,8 @@ read_variables (struct sfm_reader *r, break; } - if ( -1 == r->value_cnt ) - { - *var_by_idx = xnrealloc (*var_by_idx, i + 1, sizeof **var_by_idx); - r->vars = xnrealloc (r->vars, i + 1, sizeof *r->vars); - } + *var_by_idx = xnrealloc (*var_by_idx, i + 1, sizeof **var_by_idx); + r->vars = xnrealloc (r->vars, i + 1, sizeof *r->vars); /* If there was a long string previously, make sure that the continuations are present; otherwise make sure there aren't @@ -833,7 +1036,7 @@ read_variables (struct sfm_reader *r, fh_get_file_name (r->fh), i, sv.type)); if (sv.has_var_label != 0 && sv.has_var_label != 1) lose ((ME, _("%s: position %d: Variable label indicator field is not " - "0 or 1."), fh_get_file_name (r->fh), i)); + "0 or 1."), fh_get_file_name (r->fh), i)); if (sv.n_missing_values < -3 || sv.n_missing_values > 3 || sv.n_missing_values == -1) lose ((ME, _("%s: position %d: Missing value indicator field is not " @@ -869,6 +1072,7 @@ read_variables (struct sfm_reader *r, lose ((ME, _("%s: Duplicate variable name `%s' within system file."), fh_get_file_name (r->fh), name)); + /* Set the short name the same as the long name */ var_set_short_name (vv, vv->name); /* Case reading data. */ @@ -964,13 +1168,13 @@ read_variables (struct sfm_reader *r, if (next_value != r->value_cnt) corrupt_msg(MW, _("%s: System file header indicates %d variable positions but " - "%d were read from file."), - fh_get_file_name (r->fh), r->value_cnt, next_value); + "%d were read from file."), + fh_get_file_name (r->fh), r->value_cnt, next_value); return 1; -error: + error: return 0; } @@ -1005,7 +1209,7 @@ parse_format_spec (struct sfm_reader *r, int32_t s, } return 1; -error: + error: return 0; } @@ -1016,11 +1220,11 @@ read_value_labels (struct sfm_reader *r, struct dictionary *dict, struct variable **var_by_idx) { struct label - { - char raw_value[8]; /* Value as uninterpreted bytes. */ - union value value; /* Value. */ - char *label; /* Null-terminated label string. */ - }; + { + char raw_value[8]; /* Value as uninterpreted bytes. */ + union value value; /* Value. */ + char *label; /* Null-terminated label string. */ + }; struct label *labels = NULL; int32_t n_labels; /* Number of labels. */ @@ -1190,7 +1394,7 @@ read_value_labels (struct sfm_reader *r, free (var); return 1; -error: + error: if (labels) { for (i = 0; i < n_labels; i++) @@ -1228,6 +1432,7 @@ buf_read (struct sfm_reader *r, void *buf, size_t byte_cnt, size_t min_alloc) r->ok = false; return NULL; } + return buf; } @@ -1273,7 +1478,7 @@ read_documents (struct sfm_reader *r, struct dictionary *dict) free (documents); return 1; -error: + error: return 0; } @@ -1394,13 +1599,13 @@ read_compressed_data (struct sfm_reader *r, flt64 *buf) abort (); -success: + success: /* We have filled up an entire record. Update state and return successfully. */ r->y = ++p; return 1; -error: + error: /* I/O error. */ r->ok = false; return 0;