1 files changed, 99 insertions, 29 deletions

diff --git a/src/qreg-commands.c b/src/qreg-commands.c
index be0aada..cff4c84 100644
--- a/src/qreg-commands.c
+++ b/src/qreg-commands.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (C) 2012-2023 Robin Haberkorn
+ * Copyright (C) 2012-2024 Robin Haberkorn
  *
  * This program is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -50,7 +50,7 @@ teco_state_expectqreg_initial(teco_machine_main_t *ctx, GError **error)
 }
 
 teco_state_t *
-teco_state_expectqreg_input(teco_machine_main_t *ctx, gchar chr, GError **error)
+teco_state_expectqreg_input(teco_machine_main_t *ctx, gunichar chr, GError **error)
 {
 	teco_state_t *current = ctx->parent.current;
 
@@ -149,7 +149,7 @@ teco_state_loadqreg_done(teco_machine_main_t *ctx, const teco_string_t *str, GEr
 	if (str->len > 0) {
 		/* Load file into Q-Register */
 		g_autofree gchar *filename = teco_file_expand_path(str->data);
-		if (!teco_qreg_load(qreg, filename, error))
+		if (!qreg->vtable->load(qreg, filename, error))
 			return NULL;
 	} else {
 		/* Edit Q-Register */
@@ -202,7 +202,7 @@ teco_state_saveqreg_done(teco_machine_main_t *ctx, const teco_string_t *str, GEr
 		return &teco_state_start;
 
 	g_autofree gchar *filename = teco_file_expand_path(str->data);
-	return teco_qreg_save(qreg, filename, error) ? &teco_state_start : NULL;
+	return qreg->vtable->save(qreg, filename, error) ? &teco_state_start : NULL;
 }
 
 /*$ E% E%q
@@ -259,9 +259,9 @@ teco_state_queryqreg_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
 	if (teco_machine_main_eval_colon(ctx)) {
 		/* Query Q-Register's existence or string size */
 		if (qreg) {
-			gsize len;
-
-			if (!qreg->vtable->get_string(qreg, NULL, &len, error))
+			/* get_string() would return the size in bytes */
+			teco_int_t len = qreg->vtable->get_length(qreg, error);
+			if (len < 0)
 				return NULL;
 			teco_expressions_push(len);
 		} else {
@@ -281,10 +281,9 @@ teco_state_queryqreg_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
 			return NULL;
 		}
 
-		gint c = qreg->vtable->get_character(qreg, pos, error);
-		if (c < 0)
+		teco_int_t c;
+		if (!qreg->vtable->get_character(qreg, pos, &c, error))
 			return NULL;
-
 		teco_expressions_push(c);
 	} else {
 		/* Query integer */
@@ -311,6 +310,10 @@ teco_state_queryqreg_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
  * Positions are handled like buffer positions \(em they
  * begin at 0 up to the length of the string minus 1.
  * An error is thrown for invalid positions.
+ * If <q> is encoded as UTF-8 and there is
+ * an incomplete sequence at the requested position,
+ * -1 is returned.
+ * All other invalid Unicode sequences are returned as -2.
  * Both non-colon-modified forms of Q require register <q>
  * to be defined and fail otherwise.
  *
@@ -369,24 +372,50 @@ teco_state_setqregstring_nobuilding_done(teco_machine_main_t *ctx,
 	gint args = teco_expressions_args();
 
 	if (args > 0) {
-		g_autofree gchar *buffer = g_malloc(args);
+		guint codepage = teco_default_codepage();
+		if (colon_modified && !qreg->vtable->get_string(qreg, NULL, NULL, &codepage, error))
+			return NULL;
 
-		for (gint i = args; i > 0; i--) {
-			teco_int_t v;
-			if (!teco_expressions_pop_num_calc(&v, 0, error))
-				return NULL;
-			buffer[i-1] = (gchar)v;
+		g_autofree gchar *buffer = NULL;
+		gsize len = 0;
+
+		if (codepage == SC_CP_UTF8) {
+			/* the glib docs wrongly claim that one character can take 6 bytes */
+			buffer = g_malloc(4*args);
+			for (gint i = args; i > 0; i--) {
+				teco_int_t v;
+				if (!teco_expressions_pop_num_calc(&v, 0, error))
+					return NULL;
+				if (v < 0 || !g_unichar_validate(v)) {
+					teco_error_codepoint_set(error, "^U");
+					return NULL;
+				}
+				len += g_unichar_to_utf8(v, buffer+len);
+			}
+		} else {
+			buffer = g_malloc(args);
+			for (gint i = args; i > 0; i--) {
+				teco_int_t v;
+				if (!teco_expressions_pop_num_calc(&v, 0, error))
+					return NULL;
+				if (v < 0 || v > 0xFF) {
+					teco_error_codepoint_set(error, "^U");
+					return NULL;
+				}
+				buffer[len++] = v;
+			}
 		}
 
 		if (colon_modified) {
 			/* append to register */
 			if (!qreg->vtable->undo_append_string(qreg, error) ||
-			    !qreg->vtable->append_string(qreg, buffer, args, error))
+			    !qreg->vtable->append_string(qreg, buffer, len, error))
 				return NULL;
 		} else {
 			/* set register */
 			if (!qreg->vtable->undo_set_string(qreg, error) ||
-			    !qreg->vtable->set_string(qreg, buffer, args, error))
+			    !qreg->vtable->set_string(qreg, buffer, len,
+			                              codepage, error))
 				return NULL;
 		}
 	}
@@ -399,7 +428,8 @@ teco_state_setqregstring_nobuilding_done(teco_machine_main_t *ctx,
 	} else {
 		/* set register */
 		if (!qreg->vtable->undo_set_string(qreg, error) ||
-		    !qreg->vtable->set_string(qreg, str->data, str->len, error))
+		    !qreg->vtable->set_string(qreg, str->data, str->len,
+		                              teco_default_codepage(), error))
 			return NULL;
 	}
 
@@ -450,6 +480,26 @@ TECO_DEFINE_STATE_EXPECTQREG(teco_state_eucommand,
 	.expectqreg.type = TECO_QREG_OPTIONAL_INIT
 );
 
+static gboolean
+teco_state_setqregstring_building_initial(teco_machine_main_t *ctx, GError **error)
+{
+	if (ctx->mode > TECO_MODE_NORMAL)
+		return TRUE;
+
+	teco_qreg_t *qreg;
+	teco_machine_qregspec_get_results(ctx->expectqreg, &qreg, NULL);
+
+	/*
+	 * The expected codepage of string building constructs is determined
+	 * by the Q-Register.
+	 */
+	guint codepage;
+	if (!qreg->vtable->get_string(qreg, NULL, NULL, &codepage, error))
+		return FALSE;
+	teco_machine_stringbuilding_set_codepage(&ctx->expectstring.machine, codepage);
+	return TRUE;
+}
+
 static teco_state_t *
 teco_state_setqregstring_building_done(teco_machine_main_t *ctx, const teco_string_t *str, GError **error)
 {
@@ -467,6 +517,7 @@ teco_state_setqregstring_building_done(teco_machine_main_t *ctx, const teco_stri
  * characters \fBenabled\fP.
  */
 TECO_DEFINE_STATE_EXPECTSTRING(teco_state_setqregstring_building,
+	.initial_cb = (teco_state_initial_cb_t)teco_state_setqregstring_building_initial,
 	.expectstring.string_building = TRUE
 );
 
@@ -481,7 +532,7 @@ teco_state_getqregstring_got_register(teco_machine_main_t *ctx, teco_qreg_t *qre
 
 	g_auto(teco_string_t) str = {NULL, 0};
 
-	if (!qreg->vtable->get_string(qreg, &str.data, &str.len, error))
+	if (!qreg->vtable->get_string(qreg, &str.data, &str.len, NULL, error))
 		return NULL;
 
 	if (str.len > 0) {
@@ -604,8 +655,15 @@ teco_state_macro_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
 	} else {
 		g_auto(teco_qreg_table_t) table;
 		teco_qreg_table_init(&table, FALSE);
+
 		if (!teco_qreg_execute(qreg, &table, error))
 			return NULL;
+		if (teco_qreg_current && !teco_qreg_current->must_undo) {
+			/* currently editing local Q-Register */
+			teco_error_editinglocalqreg_set(error, teco_qreg_current->head.name.data,
+			                                teco_qreg_current->head.name.len);
+			return NULL;
+		}
 	}
 
 	return &teco_state_start;
@@ -632,6 +690,10 @@ teco_state_macro_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
  * Note that the string of <q> will be copied upon macro execution,
  * so subsequent changes to Q-Register <q> from inside the macro do
  * not modify the executed code.
+ *
+ * While \fBM\fP does not check the register's configured encoding
+ * (as reported by \fBEE\fP), its contents must be and are checked to be in
+ * valid UTF-8.
  */
 TECO_DEFINE_STATE_EXPECTQREG(teco_state_macro);
 
@@ -666,6 +728,9 @@ teco_state_macrofile_done(teco_machine_main_t *ctx, const teco_string_t *str, GE
  * It is otherwise similar to the \(lqM\(rq command.
  *
  * If <file> could not be read, the command yields an error.
+ *
+ * As all \*(ST code, the contents of <file> must be in valid UTF-8
+ * even if operating in the \(lqdefault ANSI\(rq mode as configured by \fBED\fP.
  */
 TECO_DEFINE_STATE_EXPECTFILE(teco_state_macrofile);
 
@@ -678,7 +743,7 @@ teco_state_copytoqreg_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
 	if (ctx->mode > TECO_MODE_NORMAL)
 		return &teco_state_start;
 
-	teco_int_t from, len;
+	gssize from, len; /* in bytes */
 
 	if (!teco_expressions_eval(FALSE, error))
 		return NULL;
@@ -702,32 +767,37 @@ teco_state_copytoqreg_got_register(teco_machine_main_t *ctx, teco_qreg_t *qreg,
 			len *= -1;
 		}
 	} else {
-		teco_int_t to = teco_expressions_pop_num(0);
-		from = teco_expressions_pop_num(0);
-
+		gssize to = teco_interface_glyphs2bytes(teco_expressions_pop_num(0));
+		from = teco_interface_glyphs2bytes(teco_expressions_pop_num(0));
 		len = to - from;
 
-		if (len < 0 || !teco_validate_pos(from) || !teco_validate_pos(to)) {
+		if (len < 0 || from < 0 || to < 0) {
 			teco_error_range_set(error, "X");
 			return NULL;
 		}
 	}
 
+	/*
+	 * NOTE: This does not use SCI_GETRANGEPOINTER+SCI_GETGAPPOSITION
+	 * since it may not be safe when copying from register to register.
+	 */
 	g_autofree gchar *str = g_malloc(len + 1);
 
-	struct Sci_TextRange text_range = {
-		.chrg = {.cpMin = from, .cpMax = from + len},
+	struct Sci_TextRangeFull range = {
+		.chrg = {from, from + len},
 		.lpstrText = str
 	};
-	teco_interface_ssm(SCI_GETTEXTRANGE, 0, (sptr_t)&text_range);
+	teco_interface_ssm(SCI_GETTEXTRANGEFULL, 0, (sptr_t)&range);
 
 	if (teco_machine_main_eval_colon(ctx)) {
 		if (!qreg->vtable->undo_append_string(qreg, error) ||
 		    !qreg->vtable->append_string(qreg, str, len, error))
 			return NULL;
 	} else {
+		guint cp = teco_interface_get_codepage();
+
 		if (!qreg->vtable->undo_set_string(qreg, error) ||
-		    !qreg->vtable->set_string(qreg, str, len, error))
+		    !qreg->vtable->set_string(qreg, str, len, cp, error))
 			return NULL;
 	}