CGATS file I/O library, V2.01 README file ----------------------------------------- Package contents: ----------------- cgatslib.zip ZIP archive of the following files Readme.txt This file. License4.txt Important! - Permissions for use of this package. cgats.c CGATS Library source code. cgatsstd.c I/O and malloc source code. cgats.h CGATS Library include file. Note machine dependent defines. pars.c Parser source code. parsstd.c I/O and malloc source code. pars.h Parser include file. Note machine dependent defines. Jampfile JAM style "makefile" see Makefile Makefile. Modify this to include one of the following rule sets. Makefile.WNT Makefile defines for Microsoft C++ on Windows NT Makefile.IBMNT Makefile defines for IBM C++ on Windows NT Makefile.UNIX Makefile defines for generic UNIX system Makefile.OSX Makefile defines for Apple MAC OSX Changes: -------- Changes since V2.00 Removed all exit()s from code - now return error values from all functions. Added abstract objects for File I/O and memory allocation to improve system compatibility. Separated the implimentations of the abstract I/O and memory objects that ise stdio and malloc into separate files so that a library can be compiled without reference to these system calls. --------------------------------------------------------------- DOCUMENTATION This library has been implemented from the CGATS.5 Data Exchange Format specification, in Annex J, of the ANSI CGATS.5-1993 standard. See This module attempts to make reading CGATS.5 and IT8.7 files easy and convenient. Since the standard is a less than clear on some points it is hard to know how much compatibility is to be expected. The module supports non-standard keywords and fields automatically. It does not support reading of comments. It supports reading and writing multiple tables within one file. Abreviated tables may be written. Non-standard fields written by this module will be recognized correcty when read by this module (data types are written unambiguously - reals all have a decimal point, no non-quoted strings are written that could be interpreted as a real or integer), but there is no certainty that non-standard fields written by other software will be recognized corectly (e.g. - how to you tell whether 1234 is an integer, real, or non-quoted string ?) To creat a file each element needs to be built up in turn: Create an empty cgats structure: new_cgats(). A non-standard memory allocator can be specified for use by the cgats object by passing an object that inherits from the cgatsAlloc class defined in parse.h, to the new_cgats_al(cgatsAlloc *al) constructor. Use the module methods to: Add a user defined file identifier to augment the standard identifiers: add_other(cgats *p, char *osym) This can be used to read or write table that are compatible with CGATS.5 syntax but have different file identifiers. Use a zero length string (ie. just "") for wildcard. Normaly returns 0, returns -ve if there was a system error, & sets p->errc & p->err appropriately. create an empty table entry: add_table(cgats *p, table_type tt, int oi) tt is the table type: it8_7_1, it8_7_2, it8_7_3, it8_7_4, cgats_5, cgats_X, tt_other or tt_none. cgat_X represents any "CGATS.XXX" table type. If tt_other is used for creating a file, then oi must be set to the index of the user defined table identifiers set by calls to p->add_other(); The table type read can be identified by looking at the index p->t[table_number].tt, which will be set to the table type. If the type is cgats_X type, the type actually found will be in p->cgats_type. If tt == tt_other, then the "other index" value is p->t[table_numberx].oi. For a wildcard tt_other, the actual type found will be at p->others[oi]. Tables are added in turn, and are indexed from 0 in other functions. Returns 0 normally. It returns -2 if there was a system error, & sets p->errc & p->err appropriately. Suppress the writing of the file identifier string, standard keyword definitions and/or field definitions for a second or subsequent table: set_table_flags(cgats *p, int table, int sup_id, int sup_kwords, int sup_fields); This only makes sense if the subsequent table has the same identifier type, keywords and field definitions as the preceding table. Returns 0 normally. It returns -1 if there was an error, & sets p->errc & p->err appropriately. Add keywords and their values: int add_kword(cgats *p, int table, char *ksym, char *kdata, char *kcom); The return value is the index of the new keyword, or -1, errc & err on error. Any non-standard keywords will automatically be declared in the file. The comment is optional and NULL should be passed if no comment is to be used after the keyword/value pair. Standard keywords are: ORIGINATOR System, organization or individual that created data DESCRIPTOR Purpose or contents of the data file CREATED Date of creation of the file MANUFACTURER Manufacturer of physical target PROD_DATE Year and month of physical target production yyyy:mm SERIAL Unique physical target serial number MATERIAL Material physical target was produced on INSTRUMENTATION Manufacturer and model number of measuring instrument MEASUREMENT_SOURCE Illumination used for spectral measurements PRINT_CONDITIONS Characteristics of printed sheet being reported Standard keywords will be created automatically if necessary for a legal file format. The following kewords are supplied automatically by the module, and cannot be used for other things: NUMBER_OF_FIELDS BEGIN_DATA_FORMAT END_DATA_FORMAT NUMBER_OF_SETS BEGIN_DATA END_DATA KEYWORD Add fields: int add_field(cgats *p, int table, char *fsym, data_type ftype); The return value is the index of the new field, or return -1, errc & err on error, -2, errc & err on system error. ftype defines the data type from: r_t, i_t, cs_t, nqcs_t. r_t is the real (double) type, i_t is the integer (int) type, cs_t is the character string (char*) type, nqcs_t is the same as cs_t except that it will be non-quoted if possible. Note that the type must agree with the standard type if the field is from the set default data format identifiers: SAMPLE_ID nqcs_t Identifies sample which data represents STRING cs_t Identifies label, or other non-machine readable value. CMYK_C r_t Cyan percentage of CMYK CMYK_M r_t Magenta percentage of CMYK CMYK_Y r_t Yellow percentage of CMYK CMYK_K r_t Black percentage of CMYK D_RED r_t Red filter reflection density D_GREEN r_t Green filter reflection density D_BLUE r_t Blue filter reflection density D_VIS r_t Visual filter reflection density RGB_R r_t Red component of RGB data RGB_G r_t Green component of RGB data RGB_B r_t Blue component of RGB data SPECTRAL_NM r_t Wavelength of measurement in nanometers SPECTRAL_PCT r_t Precentage reflectance/transmittance XYZ_X r_t X component of tristimulus data XYZ_Y r_t Y component of tristimulus data XYZ_Z r_t Z component of tristimulus data XYY_X r_t x component of chromaticity data XYY_Y r_t y component of chromaticity data XYY_CAPY r_t Y component of chromaticity data LAB_L r_t L* component of Lab data LAB_A r_t a* component of Lab data LAB_B r_t b* component of Lab data LAB_C r_t C*ab component of Lab data LAB_H r_t hab component of Lab data LAB_DE r_t CIA delta E STDEV_X r_t Standard deviation of X (tristimulous data) STDEV_Y r_t Standard deviation of Y (tristimulous data) STDEV_Z r_t Standard deviation of Z (tristimulous data) STDEV_L r_t Standard deviation of L* STDEV_A r_t Standard deviation of a* STDEV_B r_t Standard deviation of b* STDEV_DE r_t Standard deviation of CIE delta E Add a set of data: add_set(cgats *p, int table, ...) The data should be supplied as a varargs list in the appropropriate data format [char*, double or int]. Returns 0 normally, -1 errc & err if parameter error, -2 errc & err if system error. Add a set of data from union array: add_setarr(cgats *p, int table, cgats_set_elem *args); The data should be supplied as an array of cgats_set_elem unions. Returns 0 normally, -1 errc & err if parameter error, -2 errc & err if system error. Write the data out to a file. write_name(cgats *p, char *fname); The method will return non-zero on an error, with an error description in the err location of the structure. A non-standard destination of data can be read by passing an object that inherits from the cgatsFile class defined in parse.h to the write(cgats *p, cgatsFile *fp) method. To read in a data file, the cgats structure should be created as usual. The read method can then be called to read in the file: read_name(cgats *p, char *fname) Returns 0 normally, and -ve on an error, with an error description in the err location of the structure, and errc set with the return code. A non-standard source of data can be read by passing an object that inherits from the cgatsFile class defined in parse.h to the read(cgats *p, cgatsFile *fp) method. The reader will deal automaticaly with carry over of keywords and/or field definitions from one table to another, making each table appear independent once read. The data is accessed by refering to the following read-only structure entries: The number of tables will be in p->ntables The number of keywords will be in p->t[table_number].nkwords The number of fields will be in p->t[table_number].nfields The number of sets will be in p->t[table_number].nsets Tables, keywords, fields and sets index from 0. The keywords will be in p->t[table_number].ksym[keyword_index] The keywords character string value be in p->t[table_number].kdata[keyword_index] The field format identifiers of each field will be in p->t[table_number].fsym[field_index] The data type of each field will be in p->t[table_number].ftype[field_index] A void pointer to the data of each field of each set will be in p->t[table_number].fdata[set_index][field_index] Cast the void pointet according to its type to retrieve the data. Alternatively the p->get_setarr(struct _cgats *p, int table, int set_index, cgats_set_elem *ary) method can be used to fill in a suitable sized cgats_set_elem array with the value of all the fields at a particular index. Any character string type will be a pointer to the data in p->t[table_number].fdata[set_index][field_index]. To find the index to a particular keyword, use: find_kword(cgats *p, int table, char *ksym) -1 will be returned if no match is found. -2 will be returned, p->errc & p->err will be set if table is out of range. To find the index to a particular field, use: find_field(cgats *p, int table, char *fsym); -1 will be returned if no match is found. -2 will be returned, p->errc & p->err will be set if table is out of range. Rather than checking the error return codes from every method, the first error is "sticky", and recorded in the object. This can be checked after a series of operations by calling the error(cgats *p, char **mes) method, which will return the error code, and (optionaly) the error message. Once operations are finished, the object can be deleted by calling the delete method: del(cgats *p) Graeme Gill. ---------------------------------------------------------------