qse

Not a standalone awk, but an interesting library embedding an awk interpreter, from the announcement on comp.lang.awk :

• http://qse.googlecode.com/

QSE is a code library that implements various Unix utilities in an embeddable form and provides a set of APIs to embed them into an application. The APIs have been designed to be flexible enough to access various aspects of an embedding application and an embedded object from each other.

By embedding a Unix utility into an application, a developer is relieved of problems caused by interacting with external programs and can have tighter control over it.

Currently the library contains the following utilities:

• AWK Interpreter
• CUT Text Cutter
• SED Stream Editor

QSEAWK is an embeddable AWK interpreter and is a part of the QSE library. The interpreter implements the language described in the book The AWK Proramming Language (http://cm.bell-labs.com/cm/cs/awkbook/) with extensions. Its design focuses on building a flexible and robust embedding API with minimal platform dependency. An embedding application is capable of:

• adding new global variables and functions.
• getting and set the value of a global variable.
• calling a function with or without parameters and getting its return value.
• customizing I/O handlers for file, pipe, console I/O.
• creating multiple interpreters independent of each other.
• running a single script with different I/O streams independently.
• changing language features by setting options.
• and more

Embedding an interpreter typically involves the following steps:

• open a new interpreter
• parse in a source script
• open a new runtime context
• execute pattern-action blocks or call a function
• close the runtime context
• close the interpter

The code example below demonstrates the steps in C. It executes the one liner

BEGIN { print "hello, world" }

#include <qse/awk/std.h> 
#include <qse/cmn/stdio.h> 
#define FAIL(msg) do { qse_printf(QSE_T("ERR: %s\n"),msg); goto oops; } while(0) 
int main () 
{ 
        qse_awk_t* awk = QSE_NULL; 
        qse_awk_rtx_t* rtx = QSE_NULL; 
        qse_awk_val_t* retv; 
        qse_awk_parsestd_in_t psin; 
        int ret = -1; 
        awk = qse_awk_openstd (0); /* open a new interpreter */ 
        if (!awk) FAIL ("cannot open awk"); 
        /* parse the hello world script from a string */ 
        psin.type = QSE_AWK_PARSESTD_CP; 
        psin.u.cp = QSE_T("BEGIN { print \"hello, world\" }"); 
        if (qse_awk_parsestd (awk, &psin, QSE_NULL) <= -1) 
                FAIL (qse_awk_geterrmsg(awk)); 
        rtx = qse_awk_rtx_openstd ( /* open a runtime context */ 
                awk, 0,             /* no extension */ 
                QSE_T("hello"),     /* ARGV[0] */ 
                QSE_NULL,           /* stdin */ 
                QSE_NULL            /* stdout */ 
        ); 
        if (!rtx) FAIL (qse_awk_geterrmsg(awk)); 
        retv = qse_awk_rtx_loop (rtx); /* exeucte pattern-action blocks */ 
        if (!retv) FAIL (qse_awk_rtx_geterrmsg(rtx)); 
        qse_awk_rtx_refdownval (rtx, retv); /* destroy the return value */ 
        ret = 0; 
oops: 
        if (rtx) qse_awk_rtx_close (rtx); /* close the runtime context */ 
        if (awk) qse_awk_close (awk); /* close the interpreter */ 
        return ret; 
} 

Things can get simpler when you use C++ API. Note that the C++ API supports just a single runtime context for each interpreter.

#include <qse/awk/StdAwk.hpp> 
#include <iostream> 
#ifdef QSE_CHAR_IS_MCHAR 
#       define xcout std::cout 
#else 
#       define xcout std::wcout 
#endif 
struct MyAwk: public QSE::StdAwk { ~MyAwk () { QSE::StdAwk::close 
(); } }; 
#define FAIL(awk) do { \ 
        xcout << QSE_T("ERR: ") << awk.getErrorMessage() << std::endl; \ 
        return -1; \ 
} while (0) 

int main (int argc, char* argv[]) 
{ 
        MyAwk awk; 
        // open a new interpreter 
        if (awk.open () <= -1) FAIL (awk); 
        // set ARGV[0] 
        if (awk.addArgument (QSE_T("hello")) <= -1) FAIL (awk); 
        // parse the source script string 
        MyAwk::SourceString in(QSE_T("BEGIN { print \"hello, world \" }")); 
        if (awk.parse (in, MyAwk::Source::NONE) == QSE_NULL) FAIL (awk); 
        // execute the BEGIN, pattern-action, END blocks. 
        MyAwk::Value r; 
        if (awk.loop (&r) <= -1) FAIL (awk); 
        return 0; 
}