internal/rosa/azalea: implement semicolon and comma insertion

internal/rosa/azalea/azalea.go

@@ -4,8 +4,11 @@ package azalea
 
 import (
 	"errors"
+	"fmt"
 	"io"
+	"os"
 	"strconv"
+	"strings"
 	"text/scanner"
 )
 
@@ -302,6 +305,7 @@ func (ScanError) Error() string {
 
 // Parse parses expressions from r.
 func Parse(r io.Reader) (e []any, err error) {
+	r, err = insertSemicolons(r)
 	var p parser
 	p.s.Init(r)
 
@@ -311,12 +315,7 @@ func Parse(r io.Reader) (e []any, err error) {
 		scanner.ScanRawStrings |
 		scanner.ScanComments |
 		scanner.SkipComments
-	p.s.IsIdentRune = func(ch rune, i int) bool {
-		if i == 0 && ch >= '0' && ch <= '9' {
-			return false
-		}
-		return ch > 0 && ch < rune(len(idents)) && idents[ch]
-	}
+	p.s.IsIdentRune = isIdentRune
 
 	defer func() {
 		v := recover()
@@ -346,3 +345,89 @@ func Parse(r io.Reader) (e []any, err error) {
 	}
 	return
 }
+func isIdentRune(ch rune, i int) bool {
+	if i == 0 && ch >= '0' && ch <= '9' {
+		return false
+	}
+	return ch > 0 && ch < rune(len(idents)) && idents[ch]
+}
+func insertSemicolons(r io.Reader) (io.Reader, error) {
+	var s scanner.Scanner
+	s.Init(r)
+	s.Mode = scanner.ScanIdents |
+		scanner.ScanInts |
+		scanner.ScanStrings |
+		scanner.ScanRawStrings |
+		scanner.ScanComments |
+		scanner.SkipComments
+	s.Whitespace = 0
+	s.IsIdentRune = isIdentRune
+
+	var buf strings.Builder
+	nlsemi := false
+	nlcomma := false
+	b := ""
+
+	tok := s.Scan()
+	for tok != scanner.EOF {
+		switch tok {
+		case '\n':
+			if b[len(b)-1] == '{' && nlsemi {
+				buf.WriteRune(';')
+				nlsemi = false
+			}
+			if b[len(b)-1] == '[' && nlcomma {
+				buf.WriteRune(',')
+				nlcomma = false
+			}
+		case ';':
+			buf.WriteRune(tok)
+			nlsemi = false
+		case ',':
+			buf.WriteRune(tok)
+			nlcomma = false
+		case '{', '[':
+			nlsemi = false
+			buf.WriteRune(tok)
+			b += string(tok)
+		case '}', ']':
+			if len(b) > 0 && b[len(b)-1] == '{' && nlsemi {
+				buf.WriteRune(';')
+				nlsemi = false
+			}
+			if len(b) > 0 && b[len(b)-1] == '[' && nlcomma {
+				buf.WriteRune(',')
+				nlcomma = false
+			}
+			if len(b) > 1 && b[len(b)-1] == '{' {
+				nlsemi = true
+			}
+			if len(b) > 1 && b[len(b)-1] == '[' {
+				nlcomma = true
+				nlsemi = true
+			}
+
+			if len(b) > 1 {
+				b = b[:len(b)-1]
+			}
+			buf.WriteRune(tok)
+		case '+', ':', '=':
+			buf.WriteRune(tok)
+			nlsemi = false
+			nlcomma = false
+		case scanner.String, scanner.RawString, scanner.Int, scanner.Ident:
+			buf.WriteString(s.TokenText())
+			if len(b) > 0 && b[len(b)-1] == '{' {
+				nlsemi = true
+			} else if len(b) > 0 && b[len(b)-1] == '[' {
+				nlcomma = true
+			}
+		default:
+			buf.WriteString(s.TokenText())
+		}
+
+		tok = s.Scan()
+	}
+	fmt.Fprintln(os.Stderr, buf.String())
+	return strings.NewReader(buf.String()), nil
+}

internal/rosa/azalea/testdata/gcc.az

@@ -1,19 +1,19 @@
 package gcc {
-	description = "The GNU Compiler Collection";
-	website = "https://www.gnu.org/software/gcc";
-	anitya = 6502;
+	description = "The GNU Compiler Collection"
+	website = "https://www.gnu.org/software/gcc"
+	anitya = 6502
 
-	version# = "16.1.0";
+	version# = "16.1.0"
 	source = remoteTar {
 		url = "https://ftp.tsukuba.wide.ad.jp/software/gcc/releases/"+
-			"gcc-"+version+"/gcc-"+version+".tar.gz";
-		checksum = "4ASoWbxaA2FW7PAB0zzHDPC5XnNhyaAyjtDPpGzceSLeYnEIXsNYZR3PA_Zu5P0K";
-		compress = gzip;
-	};
+			"gcc-"+version+"/gcc-"+version+".tar.gz"
+		checksum = "4ASoWbxaA2FW7PAB0zzHDPC5XnNhyaAyjtDPpGzceSLeYnEIXsNYZR3PA_Zu5P0K"
+		compress = gzip
+	}
 	patches = [
-		"musl-off64_t-loff_t.patch",
-		"musl-legacy-lfs.patch",
-	];
+		"musl-off64_t-loff_t.patch"
+		"musl-legacy-lfs.patch"
+	]
 
 	// GCC spends most of its time in its many configure scripts, however
 	// it also saturates the CPU for a consequential amount of time.
@@ -21,37 +21,37 @@ package gcc {
 
 	exec = make {
 		configure = {
-			"disable-multilib";
-			"enable-default-pie";
-			"disable-nls";
-			"with-gnu-as";
-			"with-gnu-ld";
-			"with-system-zlib";
-			"enable-languages": "c,c++,go";
-			"with-native-system-header-dir": "/system/include";
+			"disable-multilib"
+			"enable-default-pie"
+			"disable-nls"
+			"with-gnu-as"
+			"with-gnu-ld"
+			"with-system-zlib"
+			"enable-languages": "c,c++,go"
+			"with-native-system-header-dir": "/system/include"
 			"with-multilib-list": arch {
-				amd64, arm64 = "''";
-				default = unset;
-			};
-		};
+				amd64, arm64 = "''"
+				default = unset
+			}
+		}
 		make = [
-			"BOOT_CFLAGS='-O2 -g'",
-			noop { key = value; } + "\x00",
-			"bootstrap",
-		];
+			"BOOT_CFLAGS='-O2 -g'"
+			noop { key = value } + "\x00"
+			"bootstrap"
+		]
 
 		// This toolchain is hacked to pieces, it is not expected to ever work
 		// well in its current state. That does not matter as long as the
 		// toolchain it produces passes its own test suite.
-		skip-check = true;
-	};
+		skip-check = true
+	}
 
 	inputs = [
-		binutils,
+		binutils
 
-		mpc,
-		zlib,
-		libucontext,
-		kernel-headers,
-	];
+		mpc
+		zlib
+		libucontext
+		kernel-headers
+	]
 }