support +line[,column] and filename:line:column syntaxes when opening files

* This is done via the new opener.tes in the standard library.
* Some programs that use $EDITOR expect the +line syntax to work.
* You can copy filename:line:column directly from GCC error messages
  and filename:line from grep output.
* Since there may be safe file names beginning with "+" or containing colons,
  there needs to be a way to turn this off, especially for scripts that don't
  know anything about the filenames to open.
  This is done with "--".
  Unfortunately, the first "--", that stops parameter processing,
  is always removed from the command line and not passed down into TECO land.
  This is not a problem for stand-alone scripts,
  since the script filename is already stopping option processing, so "--"
  would get passed down.
  But when calling the profile via `sciteco -- ...`, you could prevent leading
  minus signs to cause problems but since the `--` is removed, opener.tes cannot
  use it as a hint.
  Therefore, we introduced `-S` as a new alternative to `--`, that's always passed
  down as `--` (i.e. it is equivalent to "-- --").
  In other words, `sciteco -S *` will always open exactly the specified files
  without any danger of misinterpreting certain file names.
  Should we ever switch to a custom option parsing algorithm, we might preserve
  "--" (unless after --mung) and thus get rid of "-S".
* This advanced behavior can be tweaked by the user relatively easily.
  In the easiest case, we could replace M[opener] with
  <:L;R 0X.f [* @EB/^EN.f/ ]* L>
  in ~/.teco_ini to completely disable the special syntax.

7 files changed, 94 insertions, 35 deletions

diff --git a/TODO b/TODO
index adaa334..dc85057 100644
--- a/TODO
+++ b/TODO
@@ -466,14 +466,6 @@ Features:
  * Get into mentors.debian.net. First step to being adopted
    into the Debian repositories.
  * Get meta-rhaberkorn into https://layers.openembedded.org
- * sample.teco_ini: Support opening files on certain lines
-   (filename:line).
-   Theoretically, this could also be added to the <EB> syntax,
-   although the colon character is allowed in filenames under Windows.
-   In principe there is little need for that in interactive mode,
-   but it would ease opening filenames copied from compiler errors
-   or grep -n results.
-   Other editors use the +line[,col] syntax (see nano).
  * <:EF> for saving and closing a buffer, similar to <:EX>.
  * Bash completions.
  * FreeBSD: rctl(8) theoretically allows setting up per-process actions
diff --git a/doc/sciteco.1.in b/doc/sciteco.1.in
index 2b3d3bf..cc17237 100644
--- a/doc/sciteco.1.in
+++ b/doc/sciteco.1.in
@@ -21,7 +21,7 @@ Scintilla-based \fBT\fPext \fBE\fPditor and \fBCO\fPrrector
 .OP "--no-profile"
 .OP "-8|--8bit"
 .RI [ "UI option .\|.\|." ]
-.OP "--"
+.OP "--|-S"
 .RI [ script ]
 .RI [ "argument .\|.\|." ]
 .YS
@@ -86,7 +86,9 @@ In any case the current buffer position (called
 .IR dot )
 is left at the beginning of the buffer.
 Optionally \(lq\-\-\(rq might be used to explicitly separate \*(ST options and
-macro arguments.
+macro arguments, but is never passed down as a macro argument.
+Since it's sometimes useful to pass down \(rq\-\-\(rq to the profile macro,
+you can use \(lq\-S\(rq, which is equivalent to \(lq\-\- \-\-\(rq.
 .
 .LP
 If the munged macro does not request program termination using the
@@ -155,7 +157,14 @@ the first non-option argument.
 The profile will usually set up various Scintilla and \*(ST options,
 configure syntax highlighting,
 define commonly used macros and open files specified as arguments to \*(ST.
-It will usually leave the editor in interactive mode.
+The recommended command-line processor from \(lqopener.tes\(rq allows
+opening files on specific lines or lines and columns
+(\fB+\fIline\fR[\fB,\fIcolumn\fR] \fIfilename\fP or
+\fIfilename\fB:\fIline\fR[\fB:\fIcolumn\fR][\fB:\fR]).
+This special syntax can be inhibited by prefixing the files names with the
+special \(lq\-S\(rq separator.
+.LP
+The profile macro will usually leave the editor in interactive mode.
 Profile execution can be suppressed with the
 .B \-\-no\-profile
 option.
@@ -289,12 +298,6 @@ on Windows.
 On all other platforms (including UNIX/Linux) this variable
 defaults to the standard library installation path at
 .BR "@scitecolibdir@" .
-.TP
-.SCITECO_TOPIC "$SCITECO_SCINTILLUA_LEXERS"
-.B SCITECO_SCINTILLUA_LEXERS
-The Scintillua \(lqlexers/\(rq directory.
-This is passed as the \(lqscintillua.lexers\(rq library property when
-loading a Scintillua lexer via the \fBSCI_SETILEXER\fP Scintilla message.
 .
 .LP
 The \fBHOME\fP, \fBSCITECOCONFIG\fP and \fBSCITECOPATH\fP environment
@@ -321,9 +324,16 @@ environment before initializing Curses, so these variables
 can be modified in the profile macro.
 .
 .LP
+.SCITECO_TOPIC "$GTK_CSD"
 On GTK+, you may turn off the infamous client-side window decorations
 by setting the environment variable \fBGTK_CSD\fP to \(lq0\(rq.
 .
+.LP
+.SCITECO_TOPIC "$SCITECO_SCINTILLUA_LEXERS"
+The \fBSCITECO_SCINTILLUA_LEXERS\fP environment variable specifies
+the Scintillua \(lqlexers/\(rq directory.
+This is passed as the \(lqscintillua.lexers\(rq library property when
+loading a Scintillua lexer via the \fBSCI_SETILEXER\fP Scintilla message.
 .
 .SH SIGNALS
 .
diff --git a/freebsd/pkg-plist b/freebsd/pkg-plist
index c129643..48a4b61 100644
--- a/freebsd/pkg-plist
+++ b/freebsd/pkg-plist
@@ -91,6 +91,7 @@ share/man/man7/%%PROGRAM_PREFIX%%sciteco.7.gz
 %%LEXILLA%%%%DATADIR%%/lib/lexers/xml.tes
 %%LEXILLA%%%%DATADIR%%/lib/lexers/yaml.tes
 %%DATADIR%%/lib/lexers/woman.tes
+%%DATADIR%%/lib/opener.tes
 %%DATADIR%%/lib/session.tes
 %%DATADIR%%/lib/string.tes
 %%DATADIR%%/lib/women/grosciteco.tes.1.woman
diff --git a/lib/Makefile.am b/lib/Makefile.am
index 8bf80be..70f39b7 100644
--- a/lib/Makefile.am
+++ b/lib/Makefile.am
@@ -1,6 +1,6 @@
 
-dist_scitecolib_DATA = color.tes lexer.tes session.tes fnkeys.tes
-dist_scitecolib_DATA += string.tes getopt.tes
+dist_scitecolib_DATA = color.tes lexer.tes session.tes opener.tes \
+                       fnkeys.tes string.tes getopt.tes
 
 # Helper script for creating lexer definitions
 EXTRA_DIST = scite2co.lua
diff --git a/lib/opener.tes b/lib/opener.tes
new file mode 100644
index 0000000..6a57317
--- /dev/null
+++ b/lib/opener.tes
@@ -0,0 +1,37 @@
+!*$
+ * M[opener] -- Open a number of files from the current buffer
+ *
+ * This is usually the unnamed buffer as initialized from the command line.
+ * It supports both the +line[,column] and filename:line[:column] syntaxes.
+ * Since this may make it hard to open files with certain file names, all
+ * filenames after "--" are interpreted verbatim.
+ *!
+@[opener]{
+  <.-Z;
+    !* --/-S stops processing of special arguments *!
+    0A--"= 1A--"= 2A-10"=
+      L <:L;R 0X.f [* EBN.f ]* L> 1;
+    '''
+
+    1U.l 1U.c
+    !* +line[,column] *!
+    0A-+"=
+      C 0A"D \U.l W 0A-,"= C \U.c ' 0A-10"=L' '
+    '
+
+    !* filename:line[:column][:] *!
+    LR -A-:"=R' <-A"DR|1;'> -A-:"=
+      \U.a R <-A"DR|1;'> -A-:"=
+        \U.l Q.aU.c R
+      |
+        Q.aU.l 1U.c
+      '
+    |
+      LR
+    '
+
+    0X.f [*
+      EBN.f Q.c-1,Q.l-1ESFINDCOLUMN:J
+    ]*
+  L>
+}
diff --git a/sample.teco_ini b/sample.teco_ini
index 0fbf40d..a9caf26 100644
--- a/sample.teco_ini
+++ b/sample.teco_ini
@@ -7,6 +7,7 @@ EMQ[$SCITECOPATH]/color.tes
 !* Load lexer and buffer session libraries *!
 EMQ[$SCITECOPATH]/lexer.tes
 EMQ[$SCITECOPATH]/session.tes
+EMQ[$SCITECOPATH]/opener.tes
 
 !* Automatic lexing and session management using ED hooks *!
 @#ED{
@@ -81,8 +82,5 @@ Z"=
   M[session.load]
 |
   [session.path] !* disables session saving *!
-  [.f
-    <:L;R 0X.f [* EBN.f ]* L>
-  ].f
-  -EF
+  M[opener] -EF
 '
diff --git a/src/main.c b/src/main.c
index e93a3de..fb542e7 100644
--- a/src/main.c
+++ b/src/main.c
@@ -141,7 +141,7 @@ teco_process_options(gchar ***argv)
 
 	g_autoptr(GError) error = NULL;
 
-	g_autoptr(GOptionContext) options = g_option_context_new("[--] [SCRIPT] [ARGUMENT...]");
+	g_autoptr(GOptionContext) options = g_option_context_new("[--|-S] [SCRIPT] [ARGUMENT...]");
 
 	g_option_context_set_summary(
 		options,
@@ -168,27 +168,48 @@ teco_process_options(gchar ***argv)
 	 * in many situations.
 	 * It is also strictly required to make hash-bang lines like
 	 * #!/usr/bin/sciteco -m
-	 * work.
+	 * work (without additional --).
 	 */
 	g_option_context_set_strict_posix(options, TRUE);
 
+	/*
+	 * The first unknown parameter will be left in argv and
+	 * terminates option parsing (see above).
+	 * This means we can use "-S" as an alternative to "--",
+	 * that is always preserved and passed down to the macro.
+	 */
+	g_option_context_set_ignore_unknown_options(options, TRUE);
+
 	if (!g_option_context_parse_strv(options, argv, &error)) {
 		g_fprintf(stderr, "Option parsing failed: %s\n",
 			  error->message);
 		exit(EXIT_FAILURE);
 	}
 
-	/*
-	 * GOption will NOT remove "--" if followed by an
-	 * option-argument, which may interfer with scripts
-	 * doing their own option handling and interpreting "--".
-	 *
-	 * NOTE: This is still true if we're parsing in GNU-mode
-	 * and "--" is not the first non-option argument as in
-	 * sciteco foo -- -C bar.
-	 */
-	if ((*argv)[0] && !g_strcmp0((*argv)[1], "--"))
+	if ((*argv)[0] && !g_strcmp0((*argv)[1], "-S")) {
+		/* translate -S to --, this is always passed down */
+		(*argv)[1][1] = '-';
+	} else if ((*argv)[0] && !g_strcmp0((*argv)[1], "--")) {
+		/*
+		 * GOption will NOT remove "--" if followed by an
+		 * option-argument, which may interfer with scripts
+		 * doing their own option handling and interpreting "--".
+		 * Otherwise, GOption will always remove "--".
+		 *
+		 * NOTE: This is still true if we're parsing in GNU-mode
+		 * and "--" is not the first non-option argument as in
+		 * sciteco foo -- -C bar.
+		 */
 		g_free(teco_strv_remove(*argv, 1));
+	} else if ((*argv)[0] && (*argv)[1] && *(*argv)[1] == '-') {
+		/*
+		 * GOption does not remove "--" if it is followed by "-",
+		 * so if the first parameter starts with "-", we know it's
+		 * not a known built-in parameter.
+		 */
+		g_fprintf(stderr, "Unknown option \"%s\"\n", (*argv)[1]);
+		exit(EXIT_FAILURE);
+	}
 
 	gchar *mung_filename = NULL;