improved rubbing out commands with modifiers

* This was actually broken if the command is preceded by `@` and `:` characters, which
  are __not__ modifiers.
  E.g. `Q:@I/foo^W` would have rubbed out the `:` register as well.
* Also, since it was all done in teco_state_process_edit_cmd(),
  it would also rub out modifier characters from within string arguments,
  E.g. `@I/::^EQ^W`
* Real commands now have their own ^W rubout implementation, while the generic
  fallback just rubs out until the start state is re-established.
  This fails to rub out modifiers as in `@I/^W`, though.
* Real command characters now use the common TECO_DEFINE_STATE_COMMAND().
* Added test cases for CTRL+W rub out.
  A few control characters are now portably available to tests
  via environment variables `$ESCAPE`, `$RUBOUT` and `$RUBOUT_WORD`.

2 files changed, 105 insertions, 57 deletions

diff --git a/src/cmdline.c b/src/cmdline.c
index 3fb7cb9..e367e9a 100644
--- a/src/cmdline.c
+++ b/src/cmdline.c
@@ -446,56 +446,22 @@ teco_state_process_edit_cmd(teco_machine_t *ctx, teco_machine_t *parent_ctx, gun
 		}
 		return TRUE;
 
-	case TECO_CTL_KEY('W'): /* rubout/reinsert command */
+	case TECO_CTL_KEY('W'): /* rubout/reinsert construct */
 		teco_interface_popup_clear();
 
-		/*
-		 * This mimics the behavior of the `Y` command,
-		 * so it also rubs out no-op commands.
-		 * See also teco_find_words().
-		 */
 		if (teco_cmdline.modifier_enabled) {
-			/* reinsert command */
-			/* @ and : are not separate states, but practically belong to the command */
-			while (ctx->current->is_start &&
-			       teco_cmdline.effective_len < teco_cmdline.str.len &&
-			       (teco_cmdline.str.data[teco_cmdline.effective_len] == ':' ||
-			        teco_cmdline.str.data[teco_cmdline.effective_len] == '@'))
-				if (!teco_cmdline_rubin(error))
-					return FALSE;
-
+			/* reinsert construct */
 			do {
 				if (!teco_cmdline_rubin(error))
 					return FALSE;
 			} while (!ctx->current->is_start &&
 			         teco_cmdline.effective_len < teco_cmdline.str.len);
-
-			while (ctx->current->is_start &&
-			       teco_cmdline.effective_len < teco_cmdline.str.len &&
-			       strchr(TECO_NOOPS, teco_cmdline.str.data[teco_cmdline.effective_len]))
-				if (!teco_cmdline_rubin(error))
-					return FALSE;
-
-			return TRUE;
+		} else {
+			/* rubout construct */
+			do
+				teco_cmdline_rubout();
+			while (!ctx->current->is_start);
 		}
-
-		/* rubout command */
-		while (ctx->current->is_start &&
-		       teco_cmdline.effective_len > 0 &&
-		       strchr(TECO_NOOPS, teco_cmdline.str.data[teco_cmdline.effective_len-1]))
-			teco_cmdline_rubout();
-
-		do
-			teco_cmdline_rubout();
-		while (!ctx->current->is_start);
-
-		/* @ and : are not separate states, but practically belong to the command */
-		while (ctx->current->is_start &&
-		       teco_cmdline.effective_len > 0 &&
-		       (teco_cmdline.str.data[teco_cmdline.effective_len-1] == ':' ||
-		        teco_cmdline.str.data[teco_cmdline.effective_len-1] == '@'))
-			teco_cmdline_rubout();
-
 		return TRUE;
 
 #if !defined(INTERFACE_GTK) && defined(SIGTSTP)
@@ -534,6 +500,69 @@ teco_state_caseinsensitive_process_edit_cmd(teco_machine_t *ctx, teco_machine_t
 }
 
 gboolean
+teco_state_command_process_edit_cmd(teco_machine_main_t *ctx, teco_machine_t *parent_ctx, gunichar key, GError **error)
+{
+	switch (key) {
+	case TECO_CTL_KEY('W'): /* rubout/reinsert command */
+		teco_interface_popup_clear();
+
+		/*
+		 * This mimics the behavior of the `Y` command,
+		 * so it also rubs out no-op commands.
+		 * See also teco_find_words().
+		 */
+		if (teco_cmdline.modifier_enabled) {
+			/* reinsert command */
+			/* @ and : are not separate states, but practically belong to the command */
+			while (ctx->parent.current->is_start &&
+			       teco_cmdline.effective_len < teco_cmdline.str.len &&
+			       (teco_cmdline.str.data[teco_cmdline.effective_len] == ':' ||
+			        teco_cmdline.str.data[teco_cmdline.effective_len] == '@'))
+				if (!teco_cmdline_rubin(error))
+					return FALSE;
+
+			do {
+				if (!teco_cmdline_rubin(error))
+					return FALSE;
+			} while (!ctx->parent.current->is_start &&
+			         teco_cmdline.effective_len < teco_cmdline.str.len);
+
+			while (ctx->parent.current->is_start &&
+			       teco_cmdline.effective_len < teco_cmdline.str.len &&
+			       strchr(TECO_NOOPS, teco_cmdline.str.data[teco_cmdline.effective_len]))
+				if (!teco_cmdline_rubin(error))
+					return FALSE;
+
+			return TRUE;
+		}
+
+		/* rubout command */
+		while (ctx->parent.current->is_start &&
+		       teco_cmdline.effective_len > 0 &&
+		       strchr(TECO_NOOPS, teco_cmdline.str.data[teco_cmdline.effective_len-1]))
+			teco_cmdline_rubout();
+
+		do
+			teco_cmdline_rubout();
+		while (!ctx->parent.current->is_start);
+
+		/*
+		 * @ and : are not separate states, but practically belong to the command.
+		 * We cannot rely on the last character though, since it might
+		 * be part of another command.
+		 */
+		while (ctx->parent.current->is_start &&
+		       (ctx->modifier_at || ctx->modifier_colon) &&
+		       teco_cmdline.effective_len > 0)
+			teco_cmdline_rubout();
+
+		return TRUE;
+	}
+
+	return teco_state_caseinsensitive_process_edit_cmd(&ctx->parent, parent_ctx, key, error);
+}
+
+gboolean
 teco_state_stringbuilding_start_process_edit_cmd(teco_machine_stringbuilding_t *ctx, teco_machine_t *parent_ctx,
                                                  gunichar key, GError **error)
 {
@@ -678,6 +707,10 @@ teco_state_stringbuilding_start_process_edit_cmd(teco_machine_stringbuilding_t *
 	 * Chaining to the parent (embedding) state machine's handler
 	 * makes sure that ^W at the beginning of the string argument
 	 * rubs out the entire string command.
+	 *
+	 * FIXME: This does not rub out modifiers in front of
+	 * string commands since this callback could be used in recursively
+	 * embedded string building machines as well.
 	 */
 	return teco_state_process_edit_cmd(parent_ctx, NULL, key, error);
 }
@@ -1050,6 +1083,10 @@ teco_state_qregspec_process_edit_cmd(teco_machine_qregspec_t *ctx, teco_machine_
 	 * the state machine. In particular ^W would crash.
 	 * This also makes sure that commands like <EQ> are completely
 	 * rub out via ^W.
+	 *
+	 * FIXME: This does not rub out modifiers in front of
+	 * Q-Reg commands since this callback could be used in recursively
+	 * embedded Q-Reg specification machines as well.
 	 */
 	return teco_state_process_edit_cmd(parent_ctx, NULL, key, error);
 }
diff --git a/src/core-commands.c b/src/core-commands.c
index 5d6e3b9..7845dc2 100644
--- a/src/core-commands.c
+++ b/src/core-commands.c
@@ -47,6 +47,25 @@
 #include "move-commands.h"
 #include "core-commands.h"
 
+gboolean teco_state_command_process_edit_cmd(teco_machine_main_t *ctx, teco_machine_t *parent_ctx,
+                                             gunichar key, GError **error);
+
+/**
+ * @class TECO_DEFINE_STATE_COMMAND
+ * @implements TECO_DEFINE_STATE_CASEINSENSITIVE
+ * @ingroup states
+ *
+ * Base state for everything that is the beginning of a one or two
+ * letter command.
+ */
+#define TECO_DEFINE_STATE_COMMAND(NAME, ...) \
+	TECO_DEFINE_STATE_CASEINSENSITIVE(NAME, \
+		.process_edit_cmd_cb = (teco_state_process_edit_cmd_cb_t) \
+		                       teco_state_command_process_edit_cmd, \
+		.style = SCE_SCITECO_COMMAND, \
+		##__VA_ARGS__ \
+	)
+
 static teco_state_t *teco_state_control_input(teco_machine_main_t *ctx, gunichar chr, GError **error);
 
 /*
@@ -822,11 +841,10 @@ teco_state_start_input(teco_machine_main_t *ctx, gunichar chr, GError **error)
 	                                          teco_ascii_toupper(chr), error);
 }
 
-TECO_DEFINE_STATE_CASEINSENSITIVE(teco_state_start,
+TECO_DEFINE_STATE_COMMAND(teco_state_start,
 	.end_of_macro_cb = NULL, /* Allowed at the end of a macro! */
 	.is_start = TRUE,
-	.keymacro_mask = TECO_KEYMACRO_MASK_START | TECO_KEYMACRO_MASK_CASEINSENSITIVE,
-	.style = SCE_SCITECO_COMMAND
+	.keymacro_mask = TECO_KEYMACRO_MASK_START | TECO_KEYMACRO_MASK_CASEINSENSITIVE
 );
 
 /*$ F<
@@ -983,9 +1001,7 @@ teco_state_fcommand_input(teco_machine_main_t *ctx, gunichar chr, GError **error
 	                                          teco_ascii_toupper(chr), error);
 }
 
-TECO_DEFINE_STATE_CASEINSENSITIVE(teco_state_fcommand,
-	.style = SCE_SCITECO_COMMAND
-);
+TECO_DEFINE_STATE_COMMAND(teco_state_fcommand);
 
 static void
 teco_undo_change_dir_action(gchar **dir, gboolean run)
@@ -1192,7 +1208,7 @@ teco_state_condcommand_input(teco_machine_main_t *ctx, gunichar chr, GError **er
 	return &teco_state_start;
 }
 
-TECO_DEFINE_STATE_CASEINSENSITIVE(teco_state_condcommand,
+TECO_DEFINE_STATE_COMMAND(teco_state_condcommand,
 	.style = SCE_SCITECO_OPERATOR
 );
 
@@ -1533,9 +1549,7 @@ teco_state_control_input(teco_machine_main_t *ctx, gunichar chr, GError **error)
 	                                          teco_ascii_toupper(chr), error);
 }
 
-TECO_DEFINE_STATE_CASEINSENSITIVE(teco_state_control,
-	.style = SCE_SCITECO_COMMAND
-);
+TECO_DEFINE_STATE_COMMAND(teco_state_control);
 
 static teco_state_t *
 teco_state_ascii_input(teco_machine_main_t *ctx, gunichar chr, GError **error)
@@ -1661,15 +1675,14 @@ teco_state_escape_end_of_macro(teco_machine_t *ctx, GError **error)
 	return teco_expressions_discard_args(error);
 }
 
-TECO_DEFINE_STATE_CASEINSENSITIVE(teco_state_escape,
+TECO_DEFINE_STATE_COMMAND(teco_state_escape,
 	.end_of_macro_cb = teco_state_escape_end_of_macro,
 	/*
 	 * The state should behave like teco_state_start
 	 * when it comes to function key macro masking.
 	 */
 	.is_start = TRUE,
-	.keymacro_mask = TECO_KEYMACRO_MASK_START | TECO_KEYMACRO_MASK_CASEINSENSITIVE,
-	.style = SCE_SCITECO_COMMAND
+	.keymacro_mask = TECO_KEYMACRO_MASK_START | TECO_KEYMACRO_MASK_CASEINSENSITIVE
 );
 
 /*$ ED flags
@@ -2464,9 +2477,7 @@ teco_state_ecommand_input(teco_machine_main_t *ctx, gunichar chr, GError **error
 	                                          teco_ascii_toupper(chr), error);
 }
 
-TECO_DEFINE_STATE_CASEINSENSITIVE(teco_state_ecommand,
-	.style = SCE_SCITECO_COMMAND
-);
+TECO_DEFINE_STATE_COMMAND(teco_state_ecommand);
 
 gboolean
 teco_state_insert_initial(teco_machine_main_t *ctx, GError **error)