UDUNITS 2.2.28 C API Guide

This manual describes how to use the C API of the UDUNITS-2 library. Among other things, the library allows C code to obtain a binary representation of a unit of a physical quantity, to operate on such units, and to convert numeric values between compatible units.

The library comes with an extensive database of units all referenced to the SI system of units.

Table of Contents

• 1 Synopsis
• 2 What’s a Unit Package Good For?
• 3 Unit-Systems
  • 3.1 Obtaining a Unit-System
  • 3.2 Extracting Units from a Unit-System
  • 3.3 Adding Units to a Unit-System
  • 3.4 Adding Unit-Prefixes to a Unit-System
  • 3.5 Miscellaneous Operations on Unit-Systems
• 4 Converting Values Between Units
• 5 Parsing a String into a Unit
• 6 Unit Syntax
  • 6.1 Unit Specification Examples
  • 6.2 Unit Grammar
• 7 Formatting a Unit into a String
• 8 Unit Operations
  • 8.1 Unary Unit Operations
  • 8.2 Binary Unit Operations
• 9 Mapping Between Identifiers and Units
  • 9.1 Names
  • 9.2 Symbols
• 10 The Handling of Time
• 11 Error Handling
  • 11.1 Status of Last Operation
  • 11.2 Error-Messages
• 12 The Units Database
• 13 Data Types

1 Synopsis

Coding:

      #include <udunits2.h>
    

Compiling:

      c89 -I includedir ...
    

Where includedir is the installation-directory for C header files (e.g., /usr/local/include).

Linking:

      c89 ... -Llibdir -ludunits2 -lexpat ... -lm
    

Where libdir is the installation-directory for object code libraries (e.g., /usr/local/lib).

2 What’s a Unit Package Good For?

The existence of a software package is justified by what you can do with it. The three main things you can do with the UDUNIT-2 package are:

1. Convert numeric values between compatible units.
2. Convert a string representation of a unit into a binary one — enable the programmatic manipulation of units. There are three ways to do this:
   • Get the unit from a unit-system. This requires that you know the unit’s name or symbol and that the unit is in a unit-system.
   • Parse a string representation of the unit into its binary representation. This requires that the string be parsable by ut_parse().
   • Explicity construct the unit from subcomponent units using unit operations.
3. Convert a binary representation of a unit into a string — enabling the printing and storing of units in a human-readable form.

While the above might seem to be trivial activities, their general availability at the time might have helped prevent the Mars Climate Orbiter fiasco.

3 Unit-Systems

A unit-system is a set of units that are all defined in terms of the same set of base units. In the SI system of units, for example, the base units are the meter, kilogram, second, ampere, kelvin, mole, and candela. (For definitions of these base units, see the NIST International System of Units (SI).)

In the UDUNITS-2 package, every accessible unit belongs to one and only one unit-system. It is not possible to convert numeric values between units of different unit-systems. Similarly, units belonging to different unit-systems always compare unequal.

There are several categories of operations on unit-systems:

• 3.1 Obtaining a Unit-System
• 3.2 Extracting Units from a Unit-System
• 3.3 Adding Units to a Unit-System
• 3.4 Adding Unit-Prefixes to a Unit-System
• 3.5 Miscellaneous Operations on Unit-Systems

3.1 Obtaining a Unit-System

Typically, you would obtain a unit-system of predefined units by reading the default unit database using ut_read_xml() with a NULL pathname argument. If this doesn’t quite match your needs, then there are alternatives. Together with the typical solution, the means for obtaining a useful unit-system are (in order of increasing complexity):

• Obtain the default unit-system using ut_read_xml(NULL).
• Copy and customize the unit database and then call ut_read_xml() with the pathname of the customized database to obtain a customized unit-system.
• Same as either of the above but then adding new units to the unit-system using ut_new_base_unit() and ut_new_dimensionless_unit().
• Same as the above but also deriving new units usingunit operations and then adding them to the unit-system using unit mapping.
• Same as the above but starting with an empty unit-system obtained from ut_new_system(), in which case you will definitely have to start with ut_new_base_unit() and ut_new_dimensionless_unit().

You should pass every unit-system pointer to ut_free_system() when you no longer need the corresponding unit-system.

Function: const char* ut_get_path_xml (const char* path, ut_status* status)

Returns the pathname of the XML-formatted unit-database corresponding to path. If path is non-NULL, then it is returned; otherwise, if the environment variable UDUNITS2_XML_PATH is set, then its value is returned; otherwise, the pathname of the default unit-database is returned. The value of *status indicates which of these possibilities occurred:

UT_OPEN_ARG

path is non-NULL and was returned.

UT_OPEN_ENV

path is NULL, the environment variable UDUNITS2_XML_PATH is set, and its value was returned.

UT_OPEN_DEFAULT

path is NULL, the environment variable UDUNITS2_XML_PATH is unset, and the pathname of the default unit-database was returned.

Function: ut_system* ut_read_xml (const char* path)

Reads the XML-formatted unit-database specified by path and returns the corresponding unit-system. If path is NULL, then the pathname specified by the environment variable UDUNITS2_XML_PATH is used if set; otherwise, the compile-time pathname of the installed, default, unit database is used. You should pass the returned pointer to ut_free_system() when you no longer need the unit-system. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_OPEN_ARG

path is non-NULL but the file couldn’t be opened. See errno for the reason.

UT_OPEN_ENV

path is NULL and environment variable UDUNITS2_XML_PATH is set but the file couldn’t be opened. See errno for the reason.

UT_OPEN_DEFAULT

path is NULL, environment variable UDUNITS2_XML_PATH is unset, and the installed, default, unit database couldn’t be opened. See errno for the reason.

UT_OS

Operating-system error. See errno.

UT_PARSE

The database file couldn’t be parsed.

Function: ut_system* ut_new_system (void)

Creates and returns a new unit-system. On success, the unit-system will be empty except for the dimensionless unit one. You should pass the returned pointer to ut_free_system() when you no longer need the unit-system.If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return the following:

UT_OS

Operating-system error. See errno.

3.2 Extracting Units from a Unit-System

NOTE: This section covers low-level access to the individual units of a unit-system. General parsing of arbitrary unit specifications is covered in the section Parsing.

A unit-system contains mappings from identifiers to units (and vice versa). Consequently, once you have a unit-system, you can easily obtain a unit for which you know the name or symbol using the function ut_get_unit_by_name() or ut_get_unit_by_symbol().

Function: ut_unit* ut_get_unit_by_name (const ut_system* system, const char* name)

Returns the unit to which name maps from the unit-system referenced by system or NULL if no such unit exists. Name comparisons are case-insensitive. If this function returns NULL, then ut_get_status() will return one of the following:

UT_SUCCESS

name doesn’t map to a unit of system.

UT_BAD_ARG

system or name is NULL.

You should pass the returned unit to ut_free() when it is no longer needed.

Function: ut_unit* ut_get_unit_by_symbol (const ut_system* system, const char* symbol)

Returns the unit to which symbol maps from the unit-system referenced by system or NULL if no such unit exists. Symbol comparisons are case-sensitive. If this function returns NULL, then ut_get_status() will return one of the following:

UT_SUCCESS

symbol doesn’t map to a unit of system.

UT_BAD_ARG

system or symbol is NULL.

You should pass the returned unit to ut_free() when it is no longer needed.

Function: ut_unit* ut_get_dimensionless_unit_one (const ut_system* system)

Returns the dimensionless unit one of the unit-system referenced by system. While not necessary, the returned pointer may be passed to ut_free() when you no longer need the unit. If system is NULL, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return UT_BAD_ARG.

3.3 Adding Units to a Unit-System

If you use ut_read_xml(), then you should not normally need to add any new units to a unit-system.

Because you get units via their names or symbols, adding a unit to a unit-system actually means mapping one or more identifiers (i.e., names or symbols) to the unit. Thereafter, you can use ut_get_unit_by_name() and ut_get_unit_by_symbol() to retrieve the unit. The mapping of identifiers to units is covered here.

Having said that, it is possible to create a new base or dimensionless unit within a unit-system using ut_new_base_unit() or ut_new_dimensionless_unit()—you’ll just also have to map identifiers to the newly-created unit in order to be able to retrieve it later by identifier.

Function: ut_unit* ut_new_base_unit (ut_system* system)

Creates and adds a new base-unit to the unit-system referenced by system. This function returns the new base-unit. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

system is NULL.

UT_OS

Operating-system failure. See errno.

If you use ut_read_xml(), then you should not normally need to call this function.

Function: ut_unit* ut_new_dimensionless_unit (ut_system* system)

Creates and adds a new dimensionless-unit to the unit-system referenced by system. This function returns the new dimensionless-unit. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

system is NULL.

UT_OS

Operating-system failure. See errno.

If you use ut_read_xml(), then you should not normally need to call this function.

3.4 Adding Unit-Prefixes to a Unit-System

A prefix is a word or symbol that is appended to the beginning of a word or symbol that represents a unit in order to modify the value of that unit. For example, the prefix “kilo” in the word “kiloamperes” changes the value from one ampere to one-thousand amperes.

If you use ut_read_xml(), then you should not normally need to add any new prefixes to a unit-system.

Function: ut_status ut_add_name_prefix (ut_system* system, const char* name, double value)

Adds the name-prefix name with the value value to the unit-system system. A name-prefix is something like “mega” or “milli”. Comparisons between name-prefixes are case-insensitive. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

system or name is NULL, or value is 0.

UT_EXISTS

name already maps to a different value.

UT_OS

Operating-system failure. See errno.

Function: ut_status ut_add_symbol_prefix (ut_system* system, const char* symbol, double value)

Adds the symbol-prefix symbol with the value value to the unit-system system. A symbol-prefix is something like “M” or “m”. Comparisons between symbol-prefixes are case-sensitive. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

system or symbol is NULL, or value is 0.

UT_EXISTS

symbol already maps to a different value.

UT_OS

Operating-system failure. See errno.

3.5 Miscellaneous Operations on Unit-Systems

Function: void ut_free_system (ut_system* system)

Frees the unit-system referenced by system. All unit-to-identifier and identifier-to-unit mappings are removed. Use of system after this function returns results in undefined behavior.

Function: ut_status ut_set_second (const ut_unit* second)

Sets the “second” unit of a unit-system. This function must be called before the first call to ut_offset_by_time() for a unit in the same unit-system. ut_read_xml() calls this function if the unit-system it’s reading contains a unit named “second”. This function returns one of the following:

UT_SUCCESS

The “second” unit of system was successfully set.

UT_EXISTS

The “second” unit of system is set to a different unit.

UT_BAD_ARG

second is NULL.

4 Converting Values Between Units

You can convert numeric values in one unit to equivalent values in another, compatible unit by means of a converter. For example:

      #include <udunits2.h>
      ...
        ut_unit*      from = ...;
        ut_unit*      to = ...;
        cv_converter* converter = ut_get_converter(from, to);
        double       fromValue = ...;
        double       toValue = cv_convert_double(converter, fromValue);

        cv_free(converter);
    

The converter API is declared in the header-file <converter.h>, which is automatically included by the UDUNITS-2 header-file (<udunits2.h>) so you don’t need to explicitly include it.

Function: int ut_are_convertible (const ut_unit* unit1, uconst t_unit* unit2)

Indicates if numeric values in unit unit1 are convertible to numeric values in unit unit2 via ut_get_converter(). In making this determination, dimensionless units are ignored. This function returns a non-zero value if conversion is possible; otherwise, 0 is returned and ut_get_status() will return one of the following:

UT_BAD_ARG

unit1 or unit2 is NULL.

UT_NOT_SAME_SYSTEM

unit1 and unit2 belong to different unit-systems.

UT_SUCCESS

Conversion between the units is not possible (e.g., unit1 refers to a meter and unit2 refers to a kilogram.

Function: cv_converter* ut_get_converter (ut_unit* const from, ut_unit* const to)

Creates and returns a converter of numeric values in the from unit to equivalent values in the to unit. You should pass the returned pointer to cv_free() when you no longer need the converter. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

from or to is NULL.

UT_NOT_SAME_SYSTEM

The units from and to don’t belong to the same unit-system.

UT_MEANINGLESS

The units belong to the same unit-system but conversion between them is meaningless (e.g., conversion between seconds and kilograms is meaningless).

UT_OS

Operating-system failure. See errno.

Function: float cv_convert_float (const cv_converter* converter, const float value)

Converts the single floating-point value value and returns the new value.

Function: double cv_convert_double (const cv_converter* converter, const double value)

Converts the single double-precision value value and returns the new value.

Function: float* cv_convert_floats (const cv_converter* converter, const float* in, size_t count, float* out)

Converts the count floating-point values starting at in, writing the new values starting at out and, as a convenience, returns out. The input and output arrays may overlap or be identical.

Function: double* cv_convert_doubles (const cv_converter* converter, const double* in, size_t count, double* out)

Converts the count double-precision values starting at in, writing the new values starting at out and, as a convenience, returns out. The input and output arrays may overlap or be identical.

Function: void cv_free (cv_converter* conv);

Frees resources associated with the converter referenced by conv. You should call this function when you no longer need the converter. Use of conv upon return results in undefined behavior.

5 Parsing a String into a Unit

Here’s an example of parsing a string representation of a unit into its binary representation:

      #include <stdlib.h>
      #include <udunits2.h>
      ...
        ut_system*   unitSystem = ut_read_xml(NULL);
        const char* string = "kg.m2/s3";
        ut_unit*     watt = ut_parse(unitSystem, string, UT_ASCII);

        if (watt == NULL) {
          /* Unable to parse string. */
        }
        else {
          /* Life is good. */
        }
    

Function: ut_unit* ut_parse (const ut_system* system, const char* string, ut_encoding encoding)

Returns the binary unit representation corresponding to the string unit representation string in the character-set encoding using the unit-system system. string must have no leading or trailing whitespace (see ut_trim()). If an error occurs, then this function returns NULL and ut_get_status() will return one of the following:

UT_BAD_ARG

system or string is NULL.

UT_SYNTAX

string contained a syntax error.

UT_UNKNOWN

string contained an unknown identifier.

UT_OS

Operating-system failure. See errno for the reason.

You should pass the returned unit to ut_free() when it is no longer needed.

Function: size_t ut_trim (char* string, ut_encoding encoding)

Removes all leading and trailing whitespace from the NUL-terminated string string. Returns string, which is modified if it contained leading or trailing whitespace.

6 Unit Syntax

For the most part, the UDUNITS-2 package follows the syntax for unit-strings promulgated by the US National Institute for Standards and Technology (NIST). Details, of which, can be found at the NIST International System of Units (SI). The one general exception to this is the invention of a syntax for “offset”-units (e.g., the definition of the degree Celsius is “K @ 273.15”).

• 6.1 Unit Specification Examples
• 6.2 Unit Grammar

6.1 Unit Specification Examples

The above may be combined, e.g., "0.1 lg(re m/(5 s)^2) @ 50".

You may also look at the <def> elements in the units database to see examples of string unit specifications.

You may use the (udunits2prog)udunits2 utility to experiment with string unit specifications.

6.2 Unit Grammar

Here is the unit-syntax understood by the UDUNITS-2 package. Words printed Thusly indicate non-terminals; words printed THUSLY indicate terminals; and words printed <thusly> indicate lexical elements.

      Unit-Spec: one of
              nothing
              Shift-Spec

      Shift-Spec: one of
              Product-Spec
              Product-Spec SHIFT REAL
              Product-Spec SHIFT INT
              Product-Spec SHIFT Timestamp

      Product-Spec: one of
              Power-Spec
              Product-Spec Power-Spec
              Product-Spec MULTIPLY Power-Spec
              Product-Spec DIVIDE Power-Spec

      Power-Spec: one of
              Basic-Spec
              Basic-Spec INT
              Basic-Spec EXPONENT
              Basic-Spec RAISE INT

      Basic-Spec: one of
              ID
              "(" Shift-Spec ")"
              LOGREF Product_Spec ")"
              Number

      Number: one of
              INT
              REAL

      Timestamp: one of
              DATE
              DATE CLOCK
              DATE CLOCK CLOCK
              DATE CLOCK INT
              DATE CLOCK ID
              TIMESTAMP
              TIMESTAMP INT
              TIMESTAMP ID

      SHIFT:
              <space>* <shift_op> <space>*

      <shift_op>: one of
              "@"
              "after"
              "from"
              "since"
              "ref"

      REAL:
              the usual floating-point format

      INT:
              the usual integer format

      MULTIPLY: one of
              "-"
              "."
              "*"
              <space>+
              <centered middot>

      DIVIDE:
              <space>* <divide_op> <space>*

      <divide_op>: one of
              per
              PER
              "/"

      EXPONENT:
              ISO-8859-9 or UTF-8 encoded exponent characters

      RAISE: one of
              "^"
              "**"

      ID: one of
              <id>
              "%"
              "'"
              "\""
              degree sign
              greek mu character

      <id>:
              <alpha> <alphanum>*

      <alpha>:
              [A-Za-z_]
              ISO-8859-1 alphabetic characters
              non-breaking space

      <alphanum>: one of
              <alpha>
              <digit>

      <digit>:
              [0-9]

      LOGREF:
              <log> <space>* <logref>

      <log>: one of
              "log"
              "lg"
              "ln"
              "lb"

      <logref>:
              "(" <space>* <re> ":"? <space>*

      DATE:
              <year> "-" <month> ("-" <day>)?

      <year>:
              [+-]?[0-9]{1,4}

      <month>:
              "0"?[1-9]|1[0-2]

      <day>:
              "0"?[1-9]|[1-2][0-9]|"30"|"31"

      CLOCK:
              <hour> ":" <minute> (":" <second>)?

      TIMESTAMP:
              <year> (<month> <day>?)? "T" <hour> (<minute> <second>?)?

      <hour>:
              [+-]?[0-1]?[0-9]|2[0-3]

      <minute>:
              [0-5]?[0-9]

      <second>:
              (<minute>|60) (\.[0-9]*)?

    

7 Formatting a Unit into a String

Use the ut_format() function to obtain the string representation of a binary unit. For example, the following gets the definition of the unit "watt" in ASCII characters using unit-symbols rather than unit-names:

      ut_unit*     watt = ...;
      char        buf[128];
      unsigned    opts = UT_ASCII | UT_DEFINITION;
      int         len = ut_format(watt, buf, sizeof(buf), opts);

      if (len == -1) {
        /* Couldn't get string */
      }
      else if (len == sizeof(buf)) {
        /* Entire buffer used: no terminating NUL */
      }
      else {
        /* Have string with terminating NUL */
      }
    

Function: int ut_format (const ut_unit* unit, char* buf, size_t size, unsigned opts)

Formats the unit unit (i.e., returns its string representation) into the buffer pointed-to by buf of size size. The argument opts specifies how the formatting is to be done and is a bitwise OR of a ut_encoding value and zero or more of the following:

UT_NAMES

Use unit names instead of symbols.

UT_DEFINITION

The formatted string should be the definition of unit in terms of basic-units instead of stopping any expansion at the highest level possible.

On success, this function returns either the number of bytes – excluding the terminating NUL – that were written into buf or the number of bytes that would have been written. The difference is due to the runtime snprinf() function that was used.

On failure, this function returns -1 and ut_get_status() will return one of the following:

UT_BAD_ARG

unit or buf is NULL, or opts contains the bit patterns of both UT_LATIN1 and UT_UTF8.

UT_CANT_FORMAT

unit can’t be formatted in the desired manner (e.g., opts contains UT_ASCII but unit doesn’t have an identifier in that character-set or opts doesn’t contain UT_NAMES and a necessary symbol doesn’t exist).

8 Unit Operations

You can use unit operations to construct new units, get information about units, or compare units.

• 8.1 Unary Unit Operations: Operations on a single unit
• 8.2 Binary Unit Operations: Operations on pairs of units

8.1 Unary Unit Operations

Function: void ut_free (ut_unit* unit)

Frees resources associated with unit. You should invoke this function on every unit that you no longer need. Use of unit upon return from this function results in undefined behavior.

Function: ut_unit* ut_scale (double factor, const ut_unit* unit)

Returns a unit equivalent to another unit scaled by a numeric factor. For example:

          const ut_unit*   meter = ...
          const ut_unit*   kilometer = ut_scale(1000, meter);
        

The returned unit is equivalent to unit multiplied by factor. You should pass the returned pointer to ut_free() when you no longer need the unit.

Function: ut_unit* ut_offset (const ut_unit* unit, double offset)

Returns a unit equivalent to another unit relative to a particular origin. For example:

          const ut_unit*   kelvin = ...
          const ut_unit*   celsius = ut_offset(kelvin, 273.15);
        

The returned unit is equivalent to unit with an origin of offset. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function returns NULL and ut_get_status() will return one of the following:

UT_BAD_ARG

unit is NULL.

UT_OS

Operating-system error. See errno for the reason.

Function: ut_unit* ut_offset_by_time (const ut_unit* const unit, const double origin)

Returns a timestamp-unit equivalent to the time unit unit referenced to the time-origin origin (as returned by ut_encode_time()). For example:

          const ut_unit*   second = ...
          const ut_unit*   secondsSinceTheEpoch =
              ut_offset_by_time(second, ut_encode_time(1970, 1, 1, 0, 0, 0.0));
        

Leap seconds are not taken into account. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function returns NULL and ut_get_status() will return one of the following:

UT_BAD_ARG

unit is NULL.

UT_OS

Operating-system error. See errno for the reason.

UT_MEANINGLESS

Creation of a timestamp unit based on unit is not meaningful. It might not be a time-unit, for example.

UT_NO_SECOND

The associated unit-system doesn’t contain a “second” unit. See ut_set_second().

CAUTION: The timestamp-unit was created to be analogous to, for example, the degree celsius—but for the time dimension. I’ve come to believe, however, that creating such a unit was a mistake, primarily because users try to use the unit in ways for which it was not designed (such as converting dates in a calendar whose year is exactly 365 days long). Such activities are much better handled by a dedicated calendar package. Please be careful about using timestamp-units. See also the section on The Handling of Time.

Function: ut_unit* ut_invert (const ut_unit* unit)

Returns the inverse (i.e., reciprocal) of the unit unit. This convenience function is equal to ut_raise(unit,-1). You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

unit is NULL.

UT_OS

Operating-system error. See errno for the reason.

Function: ut_unit* ut_raise (const ut_unit* unit, int power)

Returns the unit equal to unit unit raised to the power power. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

unit is NULL.

UT_OS

Operating-system error. See errno for the reason.

Function: ut_unit* ut_root (const ut_unit* unit, int root)

Returns the unit equal to the root root of unit unit. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

unit is NULL.

UT_MEANINGLESS

It’s meaningless to take the given root of the given unit. This could be because the resulting unit would have fractional (i.e., non-integral) dimensionality, or because the unit is, for example, a logarithmic unit.

UT_OS

Operating-system error. See errno for the reason.

Function: ut_unit* ut_log (double base, const ut_unit* reference)

Returns the logarithmic unit corresponding to the logarithmic base base and a reference level specified as the unit reference. For example, the following creates a decibel unit with a one milliwatt reference level:

          const ut_unit* milliWatt = ...;
          const ut_unit* bel_1_mW = ut_log(10.0, milliWatt);

          if (bel_1_mW != NULL) {
            const ut_unit* decibel_1_mW = ut_scale(0.1, bel_1_mW);

            ut_free(bel_1_mW);   /* no longer needed */

            if (decibel_1_mW != NULL) {
              /* Have decibel unit with 1 mW reference */
              ...
              ut_free(decibel_1_mW);
            }                 /* "decibel_1_mW" allocated */
          }
        

You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

reference is NULL.

UT_OS

Operating-system error. See errno for the reason.

UT_BAD_ARG

base is invalid (e.g., it must be greater than one).

Function: const char* ut_get_name (const ut_unit* unit, ut_encoding encoding)

Returns the name to which the unit referenced by unit maps in the character-encoding specified by encoding. If this function returns NULL, then ut_get_status() will return one of the following:

UT_BAD_ARG

name is NULL.

UT_SUCCESS

unit doesn’t map to a name in the given character-set.

Function: const char* ut_get_symbol (const ut_unit* unit, ut_encoding encoding)

Returns the symbol to which the unit referenced by unit maps in the character-encoding specified by encoding. If this function returns NULL, then ut_get_status() will return one of the following:

UT_BAD_ARG

symbol is NULL.

UT_SUCCESS

unit doesn’t map to a symbol in the given character-set.

Function: ut_system* ut_get_system (const ut_unit* unit)

Returns the unit-system to which the unit referenced by unit belongs. If unit is NULL, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return UT_BAD_ARG.

Function: int ut_is_dimensionless (const ut_unit* unit)

Indicates if unit unit is dimensionless (like “radian”). This function returns a non-zero value if the unit is dimensionless; otherwise, 0 is returned and ut_get_status() will return one of the following:

UT_BAD_ARG

unit1 is NULL.

UT_SUCCESS

The unit is dimensionless.

Function: ut_unit* ut_clone (const ut_unit* unit)

Returns a copy of the unit referenced by unit. You should pass the returned pointer to ut_free() when you no longer need the unit. If an error occurs, then this function writes an error-message using ut_handle_error_message() and returns NULL. Also, ut_get_status() will return one of the following:

UT_BAD_ARG

unit is NULL.

UT_OS

Operating-system failure. See errno.

If you use ut_read_xml(), then you should not normally need to call this function.

Function: ut_status ut_accept_visitor (const ut_unit* unit, const ut_visitor* visitor, void* arg)

Accepts the visitor visitor to the unit unit. The argument arg is passed to the visitor’s functions. This function returns one of the following:

UT_BAD_ARG

visitor or unit is NULL.

UT_VISIT_ERROR

An error occurred in visitor while visiting unit.

UT_SUCCESS

Success.

Data type: ut_visitor int foo(int) int bar(int, int)

You pass a pointer to a data object of this type if and when you call ut_accept_visitor(). It contains the following pointers to functions that implement your unit-visitor:

ut_status (*visit_basic)(const ut_unit* unit, void* arg);

Visits the basic-unit unit. A basic-unit is a base unit like “meter” or a non-dimensional but named unit like “radian”. This function returns UT_SUCCESS on and only on success.

ut_status (*visit_product)(const ut_unit* unit, int count, const ut_unit* const* basicUnits, const int* powers, void* arg);

Visits the product-unit unit. The product-unit is a product of the count basic-units referenced by basicUnits, each raised to their respective, non-zero power in powers. This function returns UT_SUCCESS on and only on success.

ut_status (*visit_galilean)(const ut_unit* unit, double scale, const ut_unit* underlyingUnit, double origin, void* arg);

Visits the Galilean-unit unit. The Galilean-unit has the underlying unit underlyingUnit and either the non-unity scale factor scale or the non-zero origin origin, or both. This function returns UT_SUCCESS on and only on success.

ut_status (*visit_timestamp)(const ut_unit* unit, const ut_unit* timeUnit, double origin, void* arg);

Visits the timestamp-unit unit. The timestamp-unit has the underlying unit of time timeUnit and the ut_encode_time()-encoded time-origin origin. This function returns UT_SUCCESS on and only on success.

ut_status (*visit_logarithmic)(const ut_unit* unit, double base, const ut_unit* reference, void* arg);

Visits the logarithmic-unit unit.The logarithmic-unit has the logarithmic base base and the reference-level is specified by the unit reference. This function returns UT_SUCCESS on and only on success.

8.2 Binary Unit Operations

Binary unit operations act on two units.

NOTE: The functions ut_are_convertible() and ut_get_converter() are also binary unit operations but are documented elsewhere.

Function: ut_unit* ut_multiply (const ut_unit* unit1, const ut_unit* unit2)

Returns the result of multiplying unit unit1 by unit unit2. You should pass the pointer to ut_free() when you no longer need the unit On failure, this function returns NULL and ut_get_status() will return one of the following:

UT_BAD_ARG

unit1 or unit2 is NULL.

UT_NOT_SAME_SYSTEM

unit1 and unit2 belong to different unit-systems.

UT_OS

Operating-system error. See errno for the reason.

Function: ut_unit* ut_divide (const ut_unit* numer, const ut_unit* denom)

Returns the result of dividing unit numer by unit denom. You should pass the pointer to ut_free() when you no longer need the unit On failure, this function returns NULL and ut_get_status() will return one of the following:

UT_BAD_ARG

numer or denom is NULL.

UT_NOT_SAME_SYSTEM

unit1 and unit2 belong to different unit-systems.

UT_OS

Operating-system error. See errno for the reason.

Function: int ut_compare (const ut_unit* unit1, const ut_unit* unit2)

Compares two units. Returns a value less than, equal to, or greater than zero as unit1 is considered less than, equal to, or greater than unit2, respectively. Units from different unit-systems never compare equal. The value zero is also returned if both unit pointers are NULL.

Function: int ut_same_system (const ut_unit* unit1, const ut_unit* unit2)

Indicates if two units belong to the same unit-system. This function returns a non-zero value if the two units belong to the same unit-system; otherwise, 0 is returned and ut_get_status() will return one of the following:

UT_BAD_ARG

unit1 or unit2 is NULL.

UT_SUCCESS

The units belong to different unit-systems.

9 Mapping Between Identifiers and Units

Within a unit-system, you can map an identifier to a unit and vice versa. If an identifier maps to a unit, then the unit can be retrieved from the unit-system via the identifier. Similarly, if a unit maps to an identifier, then the unit can be printed using the identifier.

There two kinds of identifiers: names and symbols.

• 9.1 Names: Mapping between units and names.
• 9.2 Symbols: Mapping between units and symbols.

9.1 Names

You can map a name to a unit and vice versa. If you use ut_read_xml(), then you shouldn’t normally need to do this.

Function: ut_status ut_map_name_to_unit (const char* name, const ut_encoding encoding, const ut_unit* unit)

Maps the name referenced by name, in character-set encoding, to the unit referenced by unit in the unit-system that contains unit. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

name or unit is NULL.

UT_OS

Operating-system failure. See errno.

UT_EXISTS

name already maps to a different unit.

Function: ut_status ut_unmap_name_to_unit (ut_system* system, const char* name, const ut_encoding encoding)

Removes any mapping from name name, in character-set encoding, to a unit in unit-system system. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

system or name is NULL.

Function: ut_status ut_map_unit_to_name (const ut_unit* unit, const char* name, ut_encoding encoding)

Maps the unit unit to the name name, which is in character-set encoding, in the unit-system that contains the unit. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

unit or name is NULL, or name is not in the character-set encoding.

UT_OS

Operating-system failure. See errno.

UT_EXISTS

unit already maps to a different name.

Function: ut_status ut_unmap_unit_to_name (const ut_unit* unit, ut_encoding encoding)

Removes any mapping from unit unit to a name in character-set encoding from the unit-system that contains the unit. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

unit is NULL.

9.2 Symbols

You can map a symbol to a unit and vice versa. If you use ut_read_xml(), then you shouldn’t normally need to do this.

Function: ut_status ut_map_symbol_to_unit (const char* symbol, const ut_encoding encoding, const ut_unit* unit)

Maps the symbol referenced by symbol, in character-set encoding, to the unit referenced by unit in the unit-system that contains unit. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

symbol or unit is NULL.

UT_OS

Operating-system failure. See errno.

UT_EXISTS

symbol already maps to a different unit.

Function: ut_status ut_unmap_symbol_to_unit (ut_system* system, const char* symbol, const ut_encoding encoding)

Removes any mapping from symbol symbol, in character-set encoding, to a unit in unit-system system. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

system or symbol is NULL.

Function: ut_status ut_map_unit_to_symbol (const ut_unit* unit, const char* symbol, ut_encoding encoding)

Maps the unit unit to the symbol symbol, which is in character-set encoding, in the unit-system that contains the unit. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

unit or symbol is NULL.

UT_BAD_ARG

Symbol symbol is not in the character-set encoding.

UT_OS

Operating-system failure. See errno.

UT_EXISTS

unit already maps to a different symbol.

Function: ut_status ut_unmap_unit_to_symbol (const ut_unit* unit, ut_encoding encoding)

Removes any mapping from unit unit to a symbol in character-set encoding from the unit-system that contains the unit. This function returns one of the following:

UT_SUCCESS

Success.

UT_BAD_ARG

unit is NULL.

10 The Handling of Time

You should use a true calendar package rather than the UDUNITS-2 package to handle time. Having said that, many people use the time-handling capabilities of the UDUNITS-2 package because it supports "units" like "seconds since 1970-01-01". You should be aware, however, that the hybrid Gregorian/Julian calendar used by the UDUNITS-2 package cannot be changed. Dates on or after 1582-10-15 are assumed to be Gregorian dates; dates before that are assumed to be Julian dates. In particular, the year 1 BCE is immediately followed by the year 1 CE.

In general, the UDUNITS-2 package handles time by encoding it as double-precision value, which can then be acted upon arithmetically.

Function: double ut_encode_time (int year, int month, int day, int hour, int minute, double second)

Encodes a time as a double-precision value. This convenience function is equivalent to: ut_encode_date(year,month,day) + ut_encode_clock(hour,minute,second)

Function: double ut_encode_date (int year, int month, int day)

Encodes a date as a double-precision value. You probably won’t use this function. Dates on or after 1582-10-15 are assumed to be Gregorian dates; dates before that are assumed to be Julian dates. In particular, the year 1 BCE is immediately followed by the year 1 CE.

Function: double ut_encode_clock (int hour, int minute, double second)

Encodes a clock-time as a double-precision value. abs(hour) must be less than 24; abs(minute) must be less than 60; and fabs(second) must be less than or equal to 62. You probably won’t use this function.

Function: void ut_decode_time (double time, int* year, int* month, int* day, int* hour, int* minute, double* second, double* resolution)

Decodes a time from a double-precision value into its individual components.The variable referenced by resolution will be set to the resolution (i.e., uncertainty) of the time in seconds.

11 Error Handling

Error-handling in the units module has two aspects: the status of the last operation performed by the module and the handling of error-messages:

• 11.1 Status Of Last Operation
• 11.2 Handling of Error Messages

11.1 Status of Last Operation

UDUNITS-2 functions set their status by calling ut_set_status(). You can use the function ut_get_status() to retrieve that status.

Function: ut_status ut_get_status (void)

Returns the value specified in the last call to ut_set_status()

Function: void ut_set_status (ut_status status)

Set the status of the units module to status.

Data type: ut_status

This enumeration has the following values:

UT_SUCCESS

Success

UT_BAD_ARG

An argument violates the the function’s contract (e.g., it’s NULL).

UT_EXISTS

Unit, prefix, or identifier already exists

UT_NO_UNIT

No such unit exists

UT_OS

Operating-system error. See errno for the reason.

UT_NOT_SAME_SYSTEM

The units belong to different unit-systems

UT_MEANINGLESS

The operation on the unit or units is meaningless

UT_NO_SECOND

The unit-system doesn’t have a unit named “second”

UT_VISIT_ERROR

An error occurred while visiting a unit

UT_CANT_FORMAT

A unit can’t be formatted in the desired manner

UT_SYNTAX

String unit representation contains syntax error

UT_UNKNOWN

String unit representation contains unknown word

UT_OPEN_ARG

Can’t open argument-specified unit database

UT_OPEN_ENV

Can’t open environment-specified unit database

UT_OPEN_DEFAULT

Can’t open installed, default, unit database

UT_PARSE

Error parsing unit database

11.2 Error-Messages

Function: int ut_handle_error_message (const char* fmt, ...)

Handles the error-message corresponding to the format-string fmt and any subsequent arguments referenced by it. The interpretation of the formatting-string is identical to that of the UNIX function printf(). On success, this function returns the number of bytes in the error-message; otherwise, this function returns -1.

Use the function ut_set_error_message_handler() to change how error-messages are handled.

Function: ut_error_message_handler ut_set_error_message_handler (ut_error_message_handler handler)

Sets the function that handles error-messages and returns the previous error-message handler. The initial error-message handler is ut_write_to_stderr().

Function: int ut_write_to_stderr (const char* fmt, va_list args)

Writes the variadic error-message corresponding to formatting-string fmt and arguments args to the standard-error stream and appends a newline. The interpretation of the formatting-string is identical to that of the UNIX function printf(). On success, this function returns the number of bytes in the error-message; otherwise, this function returns -1.

Function: int ut_ignore (const char* fmt, va_list args)

Does nothing. In particular, it ignores the variadic error-message corresponding to formatting-string fmt and arguments args. Pass this function to ut_set_error_message_handler() when you don’t want the unit module to print any error-messages.

Data type: ut_error_message_handler

This is the type of an error-message handler. It’s definition is:

          typedef int (*ut_error_message_handler)(const char* fmt, va_list args);
        

12 The Units Database

The database of units that comes with the UDUNITS-2 package is an XML-formatted file that is based on the SI system of units. It contains the names and symbols of most of the units that you will ever encounter. The pathname of the installed file is datadir/udunits2.xml, where datadir is the installation-directory for read-only, architecture-independent data (e.g., /usr/local/share). This pathname is the default that ut_read_xml() uses.

Naturally, because the database is a regular file, it can be edited to add new units or remove existing ones. Be very careful about doing this, however, because you might lose the benefit of exchanging unit-based information with others who haven’t modified their database.

13 Data Types

The data types ut_visitor, ut_status, and ut_error_message_handler are documented elsewhere.

Data type: ut_encoding

This enumeration has the following values:

UT_ASCII

US ASCII character-set.

UT_ISO_8859_1

The ISO-8859-1 character-set.

UT_LATIN1

Synonym for UT_ISO_8859_1.

UT_UTF8

The UTF-8 encoding of the Unicode character-set.