QSEAWK Language {#awk-lang} ================================================================================ Overview -------- QSEAWK implements the language described in the [The AWK Programming Language][awkbook] with extensions. QSEAWK reads an AWK program, recognizes various tokens contained while skipping comments and whitespaces that don't constinute a token, analyses syntax, and tranforms them to an internal form for execution. An QSEAWK program can be composed of the following elements at the top level. - *BEGIN* blocks - *END* blocks - pattern-action blocks - user-defined functions - comments - \@global variables - \@include statements The following code snippet is a valid QSEAWK program that print the string *hello, world* to the console. it is composed of a single *BEGIN* block. ~~~~~{.awk} BEGIN { print "hello, world"; } ~~~~~ In general, QSEAWK starts executing the *BEGIN* blocks. For each input record from an input stream, it executes the pattern-action blocks if the pattern evaluates to true. Finally, it executes the *END* blocks. By default, each line in the input stream is an input record. None of these blocks are mandatory. However, a useful program needs at least 1 block to be present. For the following input records, ~~~~~{.txt} abcdefgahijklmn 1234567890 opqrstuvwxyz ~~~~~ this AWK program produces ~~~~~{.awk} BEGIN { mr=0; } /abc|vwx/ { print $0; mr++; } END { print "total records: " NR; print "matching records: " mr; } ~~~~~ this output text. ~~~~~{.txt} abcdefgahijklmn opqrstuvwxyz total records: 3 matching records: 2 ~~~~~ The QSEAWK library provides a capability to use a use a user-defined function as an entry point instead of executing these blocks. See \ref awk-embed for how to change the entry point. Comments -------- A single-line comment is introduced by a hash character #, and is terminated at the end of the same line. Additionally, it supports a C-style multi-line comment enclosed in /* and */. The multi-line comment can't nest and can't appear within string literals and regular expressions. ~~~~~{.awk} x = y; # assign y to x. /* this line is ignored. this line is ignored too. */ ~~~~~ Tokens ------ When QSEAWK parses a program, it classifies a series of input characters into meaningful tokens. It can extract the smallest meaningful unit through this tokenization process. ### Reserved Words ### The following words are reserved and cannot be used as a variable name, a parameter name, or a function name. - BEGIN - END - function - @local - @global - @include - if - else - while - for - do - break - continue - return - exit - @abort - delete - @reset - next - nextfile - nextofile - print - printf - getline However, these words can be used as normal names in the context of a module call. In practice, the predefined names used for built-in commands, functions, and variables are treated as if they are reserved since you can't create another denifition with the same name. ### Numbers ### An integer begins with a numeric digit between 0 and 9 inclusive and can be followed by more numeric digits. If an integer is immediately followed by a floating point, and optionally a series of numeric digits without whitespaces, it becomes a floting-point number. An integer or a simple floating-point number can be followed by e or E, and optionally a series of numeric digits with a optional single sign letter. A floating-point number may begin with a floting point with a preceeding number. 369 # integer 3.69 # floating-pointe number 13. # 13.0 .369 # 0.369 34e-2 # 34 * (10 ** -2) 34e+2 # 34 * (10 ** 2) 34.56e # 34.56 34.56E3 An integer can be prefixed with 0x, 0, 0b for a hexa-decimal number, an octal number, and a binary number respectively. For a hexa-decimal number, letters from A to F can form a number case-insenstively in addition to numeric digits. 0xA1 # 161 0xB0b0 # 45232 020 # 16 0b101 # 5 If the prefix is not followed by any numeric digits, it is still a valid token and represents the value of 0. 0x # 0x0 but not desirable. 0b # 0b0 but not desirable. ### Strings ### A string is enclosed in a pair of double quotes or single quotes. A character in a string enclosed in the double-quotes, when preceded with a back-slash, changes the meaning. - \\a - alert - \\b - backspace - \\f - formfeed - \\n - newline - \\r - carriage return - \\t - horizontal tab - \\v - vertical tab - \\\\ - backslash - \\" - double quote You can specify a character with an octal number or a hexadecimal number. The actual value can range between 0 and 255 inclusive. - \\OOO - O is an octal digit. - \\xXX - X is a hexadecimal digit. In the octal sequence, you can specify up to 3 octal digits after \\; In the hexadecimal sequence, you can specify as many hexadecimal digits as possible after \\x. ~~~~~{.awk} BEGIN { print "\xC720\xB2C8\xCF54\xB4DC \x7D71\x4E00\x78BC"; } ~~~~~ This program should print \em 유니코드 \em 統一碼 if the character type can represent the numbers in the sequence. If the number doesn't fit in the range that the current character type can represent, the character generated from the sequence is undefined. The \\u and \\U sequences, unlike ths \\x sequence, limits the maximum number of hexadecimal digits. It is available if the [Character Type](@ref installation) chosen for building is the wide character type. - \\uXXXX - X is a hexadecimal digit. up to 4 digits - \\UXXXXXXXX - X is a hexadecimal digit. up to 8 digits The program above can be rewritten like this. ~~~~~{.awk} BEGIN { print "\uC720\uB2C8\uCF54\uB4DC \U00007D71\U00004E00\U000078BC"; } ~~~~~ If \\x, \\u, \\U are not followed by a hexadecimal digit, *x*, *u*, *U* are produced respectively. There are no special sequences supported for a string enclosed in the single quotes. For that reason, you can't specify the single quote itself within a single-quoted string. The following program prints *awk* in double quotes. ~~~~~{.awk} BEGIN { print '"awk"'; } ~~~~~ ### Regular Expressions ### A regular expression is enclosed in a pair of forward slashes. The special sequences for a double-quoted string are all supported in a regular expression. TBD. Octal character notation is not supported in a regular expression literal since it conflicts with the backreference notation. ### Note ### QSEAWK forms a token with the lognest valid sequences. Tokenization cab confusing, especially for the implicit concatention. Let's take this as an example. 0xT Since 0x not followed by a digit is a valid token, and T is an identifier, it is the same expression as 0x concatenated with T (0x @@ T). Commands -------- AWK has the following statement constructs. - if - while - for - do .. while - break - continue - return - exit - abort - next - nextfile - nextofile - delete - reset - print - printf - expression @subsection awk_litvar LITERAL AND VARIABLE Value type - Scalar -- String -- Integer -- Floating-Pointer number - Hashed Map - Regular expression Scalar values are immutable while a hashed map value is mutable. A regular expression value is specially treated. A variable is tied to a value when it is assigned with a value. If the variable is tied to a map value, it can't be assigned again. You can use 'reset' to untie the variable from the value, and thus restore the variable to the 'nil' state. .... @subsection awk_ext_teq TEQ OPERATOR The === operator compares two values and evaluates to a non-zero value if both have the same internal type and the actual values are the same. so 1 is not equal to 1.0 for the === operator. A map comparison for the === operator is a bit special. The contents of the map is never inspected. Comparing two maps always result in inequality. However, if two variables points to the same map value, it can evaluate to a non-zero value. This is possible if you allow assigning a map to another non-map variable with #QSE_AWK_MAPTOVAR. In this case, a map is not deep-copied but the reference to it is copied. ~~~~~{.awk} BEGIN { a[10]=20; b=a; b[20]=40; for (i in a) print i, a[i]; print a===b; } ~~~~~ The === operator may be also useful when you want to indicate an error with an uninitialized variable. The following code check if the function returned a map. Since the variable 'nil' has never been assigned, its internal type is 'NIL' and ~~~~~{.awk} function a () { x[10] = 2; return x; } BEGIN { t = a(); if (t === nil) print "nil"; else print "ok"; } ~~~~~ The !== operator is a negated form of the === operator. ### Variable Declaration ### Variables declared are accessed directly bypassing the global named map that stores undeclared variables. The keyword \@global introduces a global variable and the keyword \@local introduces local variable. Local variable declaraion in a block must be located before an expression or a statement appears. @global g1, g2; #declares two global variables g1 and g2 BEGIN { @local a1, a2, a3; # declares three local variables g1 = 300; a1 = 200; { @local a1; # a1 here hides the a1 at the outer scope @local g1; # g1 here hides the global g1 a1 = 10; g1 = 5; print a1, g1; # it prints 10 and 5 } print a1, g1; # it prints 200 and 300 } To disable named variables, you must turn off #QSE_AWK_IMPLICIT. ### \@include ### The \@include directive inserts the contents of the object specified in the following string, typically a file name, as if they appeared in the source stream being processed. The directive can only be used at the outmost scope where global variable declarations, *BEGIN*, *END*, and/or pattern-action blocks appear. ~~~~~{.awk} @include "abc.awk" BEGIN { func_in_abc (); } ~~~~~ A semicolon is optional after the included file name. The following is the same as the sample above. ~~~~~{.awk} @include "abc.awk"; BEGIN { func_in_abc(); } ~~~~~ If #QSE_AWK_NEWLINE is off, the semicolon is required. ### Function Call ### name(1); if there is no space between 'name' and the left parenthesis, the name is treated as a function name. name (1); If there is a space, the name is treated as a function name if the name has been declared as the function or if #QSE_AWK_IMPLICIT is on, it may be 'name' concatenated with the expression in the parentheses. The following is a valid program. BEGIN { name (1); } function name(a) { print a; }' However, in this program, the first 'name' becomes a named global variable. so the function declaration with 'name' triggers the variable redefinition error. BEGIN { name (1); } function name(a) { print a; }' ### GROUPED EXPRESSION ### When #QSE_AWK_TOLERANT is on, you can use a grouped expression without the 'in' operator. A grouped expression is a parentheses-enclosed list of expressions separated with a comma. Each expression in the group is evaluated in the appearing order. The evaluation result of the last expression in the group is returned as that of the group. ~~~~~{.awk} BEGIN { c = (1, 2, 9); a=((1*c, 3*c), (3 - c), ((k = 6+(c+1, c+2)), (-7 * c))); print c; # 9; print a; # -63 print k; # 17 } ~~~~~ ### RETURN ### The return statement is valid in pattern-action blocks as well as in functions. The execution of a calling block is aborted once the return statement is executed. ~~~~~ $ qseawk 'BEGIN { return 20; }' ; echo $? 20 ~~~~~ If #QSE_AWK_MAPTOVAR is on, you can return an arrayed value from a function. ~~~~~{.awk} function getarray() { @local a; a["one"] = 1; a["two"] = 2; a["three"] = 3; return a; } BEGIN { @local x; x = getarray(); for (i in x) print i, x[i]; } ~~~~~ ### RESET ### The reset statement resets an array variable back to the initial state. After that, the array variable can also be used as a scalar variable again. You must have #QSE_AWK_RESET on to be able to be able to use this statement. ~~~~~{.awk} BEGIN { a[1] = 20; reset a; a = 20; # this is legal print a; } ~~~~~ ### ABORT ### The abort statment is similar to the exit statement except that it skips executing the END block. You must have #QSE_AWK_ABORT on to be able to use this statement. @code BEGIN { print "--- BEGIN ---"; abort 10; } END { print "--- END ---"; # this must not be printed } @endcode ### EXTENDED FUNCTIONS ### index() and match() can accept the third parameter indicating the position where the search begins. A negative value indicates a position from the back. @code BEGIN { xstr = "abcdefabcdefabcdef"; xsub = "abc"; xlen = length(xsub); i = 1; while ((i = index(xstr, xsub, i)) > 0) { print i, substr(xstr, i, xlen); i += xlen; } } @endcode ### EXTENDED FS ### If the value for FS begins with a question mark followed by 4 additional letters, QSEAWK can split a record with quoted fields delimited by a single-letter separator. The 4 additional letters are composed of a field separator, an escaper, a opening quote, and a closing quote. @code $ cat x.awk BEGIN { FS="?:\\[]"; } { for (i = 1; i <= NF; i++) print "$" i ": " $i; print "---------------"; } @endcode The value of FS above means the following. - : is a field separator. - a backslash is an escaper. - a left bracket is an opening quote. - a right bracket is a closing quote. See the following output. @code $ cat x.dat [fx1]:[fx2]:[f\[x\]3] abc:def:[a b c] $ qseawk -f x.awk x.dat $1: fx1 $2: fx2 $3: f[x]3 --------------- $1: abc $2: def $3: a b c --------------- @endcode ## Built-in I/O ## QSEAWK comes with built-in I/O commands and functions in addition to the implicit input streams for pattern-action blocks. The built-in I/O facility is available only if QSEAWK is set with #QSE_AWK_RIO. ### getline ### The *getline* command has multiple forms of usage. It can be used with or without a variable name and can also be associated with a pipe or a file redirection. The default association is the console when no pipe and file redirection is specified. In principle, it reads a record from the associated input stream and updates $0 or a variable with the record. If it managed to perform this successfully, it return 1; it if detected EOF, it returns 0; it return -1 on failure. *getline* without a following variable reads a record from an associated input stream, updates $0 with the value and increments *FNR*, *NR*. Updating $0 also causes changes in *NF* and fields from $1 to $NF. The sample below reads records from the console and prints them. BEGIN { while (getline > 0) print $0; } It is equivalent to { print $0 } but performs the task in the *BEGIN* block. *getline* with a variable reads a record from an associated input stream and updates the variable with the value. It updates *FNR* and *NR*, too. BEGIN { while (getline line > 0) print line; } You can change the stream association to a pipe or a file. If *getline* or *getline variable* is followed by a input redirection operator(<) and an expression, the evaluation result of the expression becomes the name of the file to read records from. The file is opened at the first occurrence and can be closed with the *close* function. BEGIN { filename = "/etc/passwd"; while ((getline line < filename) > 0) print line; close (filename); } When *getline* or *getline variable* is preceded with an expression and a pipe operator(|), the evaluation result of the expression becomes the name of the external command to execute. The command is executed at the first occurrence and can be terminated with the *close* function. The example below reads the output of the *ls -laF* command and prints it to the console. BEGIN { procname = "ls -laF"; while ((procname | getline line) > 0) print line; close (procname); } The two-way pipe operator(||) can also be used to read records from an external command. There is no visible chanages to the end-user in case of the example above if you switch the operator. BEGIN { procname = "ls -laF"; while ((procname || getline line) > 0) print line; close (procname); } The *getline* command acts like a function in that it returns a value. But you can't place an empty parentheses when no variable name is specified nor can you parenthesize the optional variable name. For example, *getline(a)* is different from *getline a* and means the concatenation of the return value of *getline* and the variable *a*. Besides, it is not clear if getline a < b is (getline a) < b or (getline) (a < b) For this reason, you are advised to parenthesize *getline* and its related components to avoid confusion whenever necessary. The example reading into the variable *line* can be made clearer with parenthesization. ~~~~~{.awk} BEGIN { while ((getline line) > 0) print line; } ~~~~~ ### print ### **TODO** ### printf ### When #QSE_AWK_TOLERANT is on, print and printf are treated as if they are function calls. In this mode, they return a negative number on failure and a zero on success and any I/O failure doesn't abort a running program. ~~~~~{.awk} BEGIN { a = print "hello, world" > "/dev/null"; print a; a = print ("hello, world") > "/dev/null"; print a; } ~~~~~ Since print and printf are like function calls, you can use them in any context where a normal expression is allowed. For example, printf is used as a conditional expression in an 'if' statement in the sample code below. ~~~~~{.awk} BEGIN { if ((printf "hello, world\n" || "tcp://127.0.0.1:9999") <= -1) print "FAILURE"; else print "SUCCESS"; } ~~~~~ ### close (io-name, what) ### The *close* function closes a stream indicated by the name *io-name*. It takes an optional parameter *what* indicating whether input or output should be closed. If *io-name* is a file, it closes the file handle associated; If *io-name* is a command, it may kill the running process from the command, reclaims other sytstem resources, and closes the pipe handles; If *io-name* is a network stream, it tears down connections to the network peer and closes the socket handles. The optional paramenter *what* must be one of *r* or *w* when used is useful when *io-name* is a command invoked for the two-way pipe operator. The value of *r* causes the function to close the read-end of the pipe and the value of *w* causes the function to close the write-end of the pipe. The function returns 0 on success and -1 on failure. Though not so useful, it is possible to create more than 1 streams of different kinds under the same name. The following program generates a shell script /tmp/x containing a command *ls -laF* and executes it without closing the script file being generated. It reads the execution output via a pipe and prints it to the console. It is undefined which stream the last *close* should close assuming the first *close* is commented out and the program works. ~~~~~{.awk} BEGIN { print "ls -laF" > "/tmp/x"; # file stream system ("chmod ugo+x /tmp/x"); #close ("/tmp/x"); while(("/tmp/x" | getline y) > 0) print y; # pipe stream close ("/tmp/x"); # which stream to close? } ~~~~~ Note that the execution of generated script fails if the script file is open on some platforms. That's what the first *close* commented out is actually for. ### fflush (io-name) ### The *fflush* function flushes the output stream indicated by *io-name*. If *io-name* is not specified, it flushes the open console output stream. If *io-name* is an empty stream, it flushes all open output streams. It returns 0 on success and -1 on failure. QSEAWK doesn't open the console output stream before it executes any output commands like *print* or *printf*. so fflush() returns -1 in the following program. ~~~~~{.awk} BEGIN { fflush(); } ~~~~~ The *print* command is executed before fflush() in the following program. When fflush() is executed, the output stream is open. so fflush() returns 0. ~~~~~{.awk} BEGIN { print 1; fflush(); } ~~~~~ Though not so useful, it is possible to create more than 1 output streams of different kinds under the same name. *fflush* in the following program flushes both the file stream and the pipe stream. ~~~~~{.awk} BEGIN { print 1 | "/tmp/x"; # file stream print 1 > "/tmp/x"; # pipe stream fflush ("/tmp/x"); } ~~~~~ ### setioattr (io-name, attr-name, attr-value) ### The *setioattr* function changes the I/O attribute of the name *attr-name* to the value *attr-value* for a stream identified by *io-name*. It returns 0 on success and -1 on failure. - *io-name* is a source or target name used in *getline*, *print*, *printf* combined with |, ||, >, <, >>. - *attr-name* is one of *codepage*, *ctimeout*, *atimeout*, *rtimeout*, *wtimeout*. - *attr-value* varies depending on *attr-name*. + codepage: *cp949*, *cp950*, *utf8*, *slmb*, *mb8* + ctimeout, atimeout, rtimeout, wtimeout: the number of seconds. effective on socket based streams only. you may use a floating-point number for lower resoluation than a second. a negative value turns off timeout. See this sample that prints the contents of a document encoded in cp949. ~~~~~{.awk} BEGIN { setioattr ("README.TXT", "codepage", "cp949"); while ((getline x < "README.TXT") > 0) print x; } ~~~~~ ### getioattr (io-name, attr-name, attr-value) ### The getioattr() function retrieves the current attribute value of the attribute named *attr-name* for the stream identified by *io-name*. The value retrieved is set to the variable referenced by *attr-value*. See *setioattr* for description on *io-name* and *attr-name*. It returns 0 on success and -1 on failure. ~~~~~{.awk} BEGIN { setioattr ("README.TXT", "codepage", "cp949"); if (getioattr ("README.TXT", "codepage", codepage) <= -1) print "codepage unknown"; else print "codepage: " codepage; } ~~~~~ ### Two-way Pipe ### The two-way pipe is indicated by the two-way pipe operator(||) and QSEAWK must be set with #QSE_AWK_RWPIPE to be able to use the two-way pipe. The example redirects the output of *print* to the external *sort* command and reads back the output. ~~~~~{.awk} BEGIN { print "15" || "sort"; print "14" || "sort"; print "13" || "sort"; print "12" || "sort"; print "11" || "sort"; # close the input side of the pipe as 'sort' starts emitting result # once the input is closed. close ("sort", "r"); while (("sort" || getline x) > 0) print x; } ~~~~~ This two-way pipe can create a TCP or UDP connection if the pipe command string is prefixed with one of the followings: - tcp:// - establishes a TCP connection to a specified IP address/port. - udp:// - establishes a TCP connection to a specified IP address/port. - tcpd:// - binds a TCP socket to a specified IP address/port and waits for the first connection. - udpd:// - binds a TCP socket to a specified IP address/port and waits for the first sender. See this example. ~~~~~{.awk} BEGIN { # it binds a TCP socket to the IPv6 address :: and the port number # 9999 and waits for the first coming connection. It repeats writing # "hello world" to the first connected peer and reading a line from # it until the session is torn down. do { print "hello world" || "tcpd://[::]:9999"; if (("tcpd://[::]:9999" || getline x) <= 0) break; print x; } while(1); } ~~~~~ You can manipulate TCP or UDP timeouts for connection, accepting, reading, and writing with the *setioattr* function and the *getioattr* function. See the example below. ~~~~~{.awk} BEGIN { setioattr ("tcp://127.0.0.1:9999", "ctimeout", 3); setioattr ("tcp://127.0.0.1:9999", "rtimeout", 5.5); print "hello world" || "tcp://127.0.0.1:9999"; "tcp://127.0.0.1:9999" || getline x; print x; } ~~~~~ Here is an interesting example adopting Michael Sanders' AWK web server, modified for QSEAWK. ~~~~~{.awk} # # Michael Sanders' AWK web server for QSEAWK. # Orginal code in http://awk.info/?tools/server # # qseawk --tolerant=on --rwpipe=on webserver.awk # BEGIN { x = 1 # script exits if x < 1 port = 8080 # port number host = "tcpd://0.0.0.0:" port # host string url = "http://localhost:" port # server url status = 200 # 200 == OK reason = "OK" # server response RS = ORS = "\r\n" # header line terminators doc = Setup() # html document len = length(doc) + length(ORS) # length of document while (x) { if ($1 == "GET") RunApp(substr($2, 2)) if (! x) break print "HTTP/1.0", status, reason || host print "Connection: Close" || host print "Pragma: no-cache" || host print "Content-length:", len || host print ORS doc || host close(host) # close client connection host || getline # wait for new client request } # server terminated... doc = Bye() len = length(doc) + length(ORS) print "HTTP/1.0", status, reason || host print "Connection: Close" || host print "Pragma: no-cache" || host print "Content-length:", len || host print ORS doc || host close(host) } function Setup() { tmp = "\ Simple gawk server\ \

xterm\

xcalc\

xload\

terminate script\ \ " return tmp } function Bye() { tmp = "\ Simple gawk server\

Script Terminated...\ " return tmp } function RunApp(app) { if (app == "xterm") {system("xterm&"); return} if (app == "xcalc" ) {system("xcalc&"); return} if (app == "xload" ) {system("xload&"); return} if (app == "exit") {x = 0} } ~~~~~ ### I/O Character Encoding ### You can change the character encoding encoding of a stream. See qse_findcmgr() for a list of supported encoding names. Let's say you run this simple echoing script on a WIN32 platform that has the active code page of 949 and is reachable at the IP address 192.168.2.8. C:\> chcp Active code page: 949 C:\> type s.awk BEGIN { sock = "tcpd://0.0.0.0:9999"; setioattr (sock, "codepage", "cp949"); do { if ((sock || getline x) <= 0) break; print "PEER: " x; print x || sock; } while(1); } C:\> qseawk -f r.awk PEER: 안녕 PEER: ?好! Now you run the following script on a UTF-8 console of a Linux box. $ echo $LANG en_US.UTF-8 $ cat c.awk BEGIN { peer = "tcp://192.168.2.8:9999"; setioattr (peer, "codepage", "cp949"); do { printf "> "; if ((getline x) <= 0) break; print x || peer; if ((peer || getline line) <= -1) break; print "PEER: " line; } while (1); } $ qseawk --rwpipe=on -f c.awk > 안녕 PEER: 안녕 > 你好! PEER: ?好! Note that 你 has been converted to a question mark since the letter is not supported by cp949. Modules ------- QSEAWK supports various external modules. ## String ## The *str* module provides an extensive set of string manipulation functions. - str::index - str::isalnum - str::isalpha - str::isblank - str::iscntrl - str::isdigit - str::isgraph - str::islower - str::isprint - str::ispunct - str::isspace - str::isupper - str::isxdigit - str::ltrim - str::normspace - str::rindex - str::rtrim - str::trim ## Directory ## The *dir* module provides an interface to read file names in a specified directory. - dir::open - dir::close - dir::read - dir::reset - dir::errno - dir::errstr ~~~~~{.awk} BEGIN { x = dir::open ("."); while ((dir::read(x, file)) > 0) print file; dir::close(x); }' ~~~~~ ## SED ## The *sed* module provides built-in sed capabilities. - sed::file_to_file - sed::str_to_str ~~~~~{.awk} BEGIN { sed::file_to_file ("s/[a-z]/#/g", "in.txt", "out.txt"); }' ~~~~~ [awkbook]: http://cm.bell-labs.com/cm/cs/awkbook/