Refactor for interleaved shell commands and text

1 files changed, 84 insertions, 146 deletions

@@ -1,194 +1,132 @@
-# subtext -- text substitutor -*- awk -*-
+# subtext -*- awk -*-
 # (C) C Duckworth <acdw@acdw.net>
+## Subtext is an awk program that converts roff-like input to a shell script
+## that can output markup.  Lines prefixed with . are wrapped in $(...)
+## constructs within the `body' function, but lines prefixed with # are
+## transported to the top of the file, where they're executed within the shell
+## environment.
 
 BEGIN {
-        true = 1; false = 0
         ## Tuneables
-        runbody = bool(runbody ? runbody : true)
-        postproc = postproc ? postproc : "shexpand"
-        bodyfunc = bodyfunc ? bodyfunc : "body"
-        sopath = get_value(sopath, ".:"ENVIRON["HOME"]"/.subtext")
-        ## Globals
-        ## Wrap the text in a function
-        pretext = "### begin text\n"bodyfunc"(){\n"
-        posttext = "}\n### end text"
-        ## Prelude function strings
+        dryrun = dryrun ? dryrun : 0
+        sopath = sopath ? sopath : ".:" ENVIRON["HOME"] "/.subtext"
+        bodyend = bodyend ? bodyend : "_@end@_"
+        ## Kernel
         # Ask sed to do these b/c awk has no capture groups ;_;
         shellfix = "sed -E" \
                 " -e 's/`/\\\\`/g'" \
                 " -e 's/(^|[^\\$])\\$([^\\$]|$)/\\1\\\\$\\2/g'" \
                 " -e 's/(^|[^\\$])\\$(\\$+)([^\\$]|$)/\\1\\2\\3/g'"
         shxwrap = " -e 's/^/:/'"
-        htmlfix = "sed -E" \
-                " -e 's#([^\\\\]|^)&#\\1\\&amp;#g'" \
-                " -e 's#([^\\\\]|^)<#\\1\\&lt;#g'" \
-                " -e 's#([^\\\\]|^)>#\\1\\&gt;#g'" \
-                " -e 's#\\\\([&<>])#\\1#g'";
-        ## Prelude
-        par = "#!/bin/sh\n### kernel\n" \
-                "preface()(sed \"s/^/$1/\")\n" \
-                "unpreface()(sed \"s/^$1//\")\n" \
-                "shexpand()(eval \"$( (echo 'cat<<.';preface +;echo .)" \
-                " | unpreface + )\")\n" \
-                "### library\n" \
+        kernel = "quote()(sed \"s/^/$1/\")\n" \
+                "unquote()(sed \"s/^$1//\")\n" \
+                "shexpand()(eval " \
+                "\"$(echo 'cat<<" bodyend "';cat;echo " bodyend ")\")\n" \
                 "shellfix()(" shellfix ")\n" \
-                "htmlfix()(" htmlfix ")\n" \
-                "### variables\n" \
-                "ST_POSTPROC=" postproc "\n" \
-                "ST_BODYFUNC=" bodyfunc "\n" \
-                "ST_SOPATH=" sopath "\n" \
-                "### header\n"
+                "ST_SOPATH=" sopath
 }
 
-### End a block
-
-end[endn] && $0 == end[endn] {
-        pushpar(end[endn--] "\n)", true)
-        printpar()
-        subdocp = false
-        next
-}
-
-### Line continuation
-
-/\\$/ {
+/\\$/ { # line continuation
         getline nl
-        sub(/\\$/,"")
+        sub(/\\$/, "")
         $0 = $0 " " nl
 }
 
-### Special commands
-## These call subtext-internal functions
-
-/^#so/ {                        # Insert $2 verbatim (if in sopath), else error
-        pushpar($0)
+/^#so/ { # source a file
+        pushpar()
         source($2)
         next
 }
 
-/^###$/ {                       # Delimit document
-        printpar()
-        docp = !docp
-        if (docp) {
-                print pretext "unpreface ':'<<'_'|eval \"$ST_POSTPROC\""
-        } else {
-                end_text()
-        }
+/^#/ { # head lines
+        sub(/^#+[       ]*/,"")
+        head = head (head?"\n":"") $0
         next
 }
 
-### Escape sequences
-
-/^\.\./ && docp {               # Begin a heredoc
-        # ..[<command>] [<options>] [<<delim]
-        ## ends with DELIM or '..'
-        subdocp = true
-        end[++endn] = (match($0, "<<") ? substr($0, RSTART + RLENGTH) : "..")
-        command = substr($0, 3, RSTART ? RSTART - 2 : length($0))
-        $0 = "$$(" (command ? command : "cat") "<<" end[endn]
+/^\.\./ { # block
+        body = body (body?"\n":"") \
+                "$$(unquote + << .." \
+                (length>2 ? " | " substr($1,3) " " quote(2) : "") \
+                slurp("..") \
+                "..\n)"
+        next
 }
 
-/^\./ && docp {                 # One-line command
-        # .<command> [<parameters>]
-        ## wraps the line in $( ... ), basically (also quotes)
-        specialp = 2
-        gsub(/"/, "\\\\&", $0)
-        ln = "$$(" substr($1, 2)
-        for (f=2; f<=NF; f++) {
-                ln = ln " \"" $f "\""
-        }
-        ln = ln ")"
-        $0 = ln
+/^\./ { # line
+        body = body (body?"\n":"") \
+                "$$(" substr($1, 2) " " quote(2) ")"
+        next
 }
 
-/^\\/ && docp { # \ at the beginning of a line escapes the next character
-        $0 = substr($0, 2)
+/^$/ { # line break
+        if (!pushpar()) next
 }
 
-### Book-keeping
-
-/^$/ {
-        if (!par)
-                next
-        printpar()
-}
-
-{
-        pushpar($0)
-        if (--specialp < 0)
-                specialp = 0
+{ # regular text
+        par = par (par?"\n":"") $0
 }
 
 END {
-        if (dead)
-                exit dead
-        if (par)
-                printpar()
-        while (endn > 0)
-                print "\n" end[endn--] "\n)"
-        if (docp)
-                end_text()
-        if (runbody)
-                print bodyfunc
-}
-
-function end_text() {
-        print "_\n" posttext
+        if (dead) exit dead
+
+        print "#!/bin/sh"
+        print "### generated with subtext"
+        print kernel
+        print "### head"
+        print head
+        print "### body"
+        print "body(){ shexpand << \\_"
+        pushpar()
+        print body | shellfix
+        close(shellfix)
+        print "_"
+        print "}"
+        if (!dryrun) print "body"
+}
+
+function quote(begin, end,      out) {
+        if (!begin) begin = 1
+        if (!end) end = NF
+        for (i=begin; i<=NF; i++) {
+                gsub(/"/, "\\\"", $i)
+                out = out (out?" ":"") "\"" $i "\""
+        }
+        return out
 }
 
-function pushpar(text, force_newline) {
-        par = par ((par || force_newline) ? "\n" : "") text
+function slurp(to,      out, nl) {
+        while (nl != to) {
+                getline nl
+                out = out "\n+" nl
+        }
+        sub("\\+"to"$", "", out)
+        return out
 }
 
-function printpar() {
-        specialp = specialp || (match(par, /^[  ]*</))
-        if (docp) {
-                if (!subdocp && !specialp)
-                        par = "<p>" par "</p>"
-                shx = shellfix shxwrap
-                print par | shx
-                close(shx)
-        } else
-                print par
+function pushpar() {
+        if (!par) return 0
+        if (!match(par, /^[     ]*</))
+                par = "<p>" par "</p>"
+        body = body (body?"\n":"") par
         par = ""
+        return 1
 }
 
-function get_value(var, env_var, default) {
-        if (var)
-                return var
-        else
-                return default
-}
-
-function bool(var, default) {
-        if (var=="false" || var=="no" || var=="0")
-                return false
-        else if (var=="true" || var=="yes" || var=="1")
-                return true
-        else
-                return var || default
-}
-
-function source(name) {
-        found = false
-        sp = ""
+function source(name,   found, sp) {
         split(sopath, asopath, ":")
         for (dir in asopath) {
                 fn = asopath[dir] "/" name
                 sp = asopath[dir] "\n\t" sp
                 while ((getline ln < fn) > 0) {
-                        found = true
-                        pushpar(ln)
+                        found = 1
+                        head = head (head?"\n":"") ln
                 }
-                if (found)
-                        break
+                if (found) break
+        }
+        if (!found) {
+                printf "Couldn't source %s; looked in:\n\t%s", name, sp
+                dead = 127
+                exit
         }
-        if (!found)
-                die("Couldn't source " name "; looked in:\n\t" sp, 9)
-}
-
-function die(message, code) {
-        dead = code
-        print "!!" FILENAME ":" NR ": " message > "/dev/stderr"
-        exit
 }