Improve string extraction of ETR POT file

1 files changed, 153 insertions, 17 deletions

diff --git a/editor/editor_translation.cpp b/editor/editor_translation.cpp
index 194d78326d..77154ec344 100644
--- a/editor/editor_translation.cpp
+++ b/editor/editor_translation.cpp
@@ -156,30 +156,166 @@ void load_extractable_translations(const String &p_locale) {
 	}
 }
 
-List<StringName> get_extractable_message_list() {
+Vector<Vector<String>> get_extractable_message_list() {
 	ExtractableTranslationList *etl = _extractable_translations;
-	List<StringName> msgids;
-	while (etl->data) {
-		if (!strcmp(etl->lang, "source")) {
-			Vector<uint8_t> data;
-			data.resize(etl->uncomp_size);
-			int ret = Compression::decompress(data.ptrw(), etl->uncomp_size, etl->data, etl->comp_size, Compression::MODE_DEFLATE);
-			ERR_FAIL_COND_V_MSG(ret == -1, msgids, "Compressed file is corrupt.");
+	Vector<Vector<String>> list;
 
-			Ref<FileAccessMemory> fa;
-			fa.instantiate();
-			fa->open_custom(data.ptr(), data.size());
-
-			Ref<Translation> tr = TranslationLoaderPO::load_translation(fa);
+	while (etl->data) {
+		if (strcmp(etl->lang, "source")) {
+			etl++;
+			continue;
+		}
 
-			if (tr.is_valid()) {
-				tr->get_message_list(&msgids);
-				break;
+		Vector<uint8_t> data;
+		data.resize(etl->uncomp_size);
+		int ret = Compression::decompress(data.ptrw(), etl->uncomp_size, etl->data, etl->comp_size, Compression::MODE_DEFLATE);
+		ERR_FAIL_COND_V_MSG(ret == -1, list, "Compressed file is corrupt.");
+
+		Ref<FileAccessMemory> fa;
+		fa.instantiate();
+		fa->open_custom(data.ptr(), data.size());
+
+		// Taken from TranslationLoaderPO, modified to work specifically with POTs.
+		{
+			const String path = fa->get_path();
+
+			fa->seek(0);
+
+			enum Status {
+				STATUS_NONE,
+				STATUS_READING_ID,
+				STATUS_READING_STRING,
+				STATUS_READING_CONTEXT,
+				STATUS_READING_PLURAL,
+			};
+
+			Status status = STATUS_NONE;
+
+			String msg_id;
+			String msg_id_plural;
+			String msg_context;
+
+			int line = 1;
+			bool entered_context = false;
+			bool is_eof = false;
+
+			while (!is_eof) {
+				String l = fa->get_line().strip_edges();
+				is_eof = fa->eof_reached();
+
+				// If we reached last line and it's not a content line, break, otherwise let processing that last loop.
+				if (is_eof && l.is_empty()) {
+					if (status == STATUS_READING_ID || status == STATUS_READING_CONTEXT || status == STATUS_READING_PLURAL) {
+						ERR_FAIL_V_MSG(Vector<Vector<String>>(), "Unexpected EOF while reading POT file at: " + path + ":" + itos(line));
+					} else {
+						break;
+					}
+				}
+
+				if (l.begins_with("msgctxt")) {
+					ERR_FAIL_COND_V_MSG(status != STATUS_READING_STRING && status != STATUS_READING_PLURAL, Vector<Vector<String>>(),
+							"Unexpected 'msgctxt', was expecting 'msgid_plural' or 'msgstr' before 'msgctxt' while parsing: " + path + ":" + itos(line));
+
+					// In POT files, "msgctxt" appears before "msgid". If we encounter a "msgctxt", we add what we have read
+					// and set "entered_context" to true to prevent adding twice.
+					if (!msg_id.is_empty()) {
+						Vector<String> msgs;
+						msgs.push_back(msg_id);
+						msgs.push_back(msg_context);
+						msgs.push_back(msg_id_plural);
+						list.push_back(msgs);
+					}
+					msg_context = "";
+					l = l.substr(7, l.length()).strip_edges();
+					status = STATUS_READING_CONTEXT;
+					entered_context = true;
+				}
+
+				if (l.begins_with("msgid_plural")) {
+					if (status != STATUS_READING_ID) {
+						ERR_FAIL_V_MSG(Vector<Vector<String>>(), "Unexpected 'msgid_plural', was expecting 'msgid' before 'msgid_plural' while parsing: " + path + ":" + itos(line));
+					}
+					l = l.substr(12, l.length()).strip_edges();
+					status = STATUS_READING_PLURAL;
+				} else if (l.begins_with("msgid")) {
+					ERR_FAIL_COND_V_MSG(status == STATUS_READING_ID, Vector<Vector<String>>(), "Unexpected 'msgid', was expecting 'msgstr' while parsing: " + path + ":" + itos(line));
+
+					if (!msg_id.is_empty() && !entered_context) {
+						Vector<String> msgs;
+						msgs.push_back(msg_id);
+						msgs.push_back(msg_context);
+						msgs.push_back(msg_id_plural);
+						list.push_back(msgs);
+					}
+
+					l = l.substr(5, l.length()).strip_edges();
+					status = STATUS_READING_ID;
+					// If we did not encounter msgctxt, we reset context to empty to reset it.
+					if (!entered_context) {
+						msg_context = "";
+					}
+					msg_id = "";
+					msg_id_plural = "";
+					entered_context = false;
+				}
+
+				if (l.begins_with("msgstr[")) {
+					ERR_FAIL_COND_V_MSG(status != STATUS_READING_PLURAL, Vector<Vector<String>>(),
+							"Unexpected 'msgstr[]', was expecting 'msgid_plural' before 'msgstr[]' while parsing: " + path + ":" + itos(line));
+					l = l.substr(9, l.length()).strip_edges();
+				} else if (l.begins_with("msgstr")) {
+					ERR_FAIL_COND_V_MSG(status != STATUS_READING_ID, Vector<Vector<String>>(),
+							"Unexpected 'msgstr', was expecting 'msgid' before 'msgstr' while parsing: " + path + ":" + itos(line));
+					l = l.substr(6, l.length()).strip_edges();
+					status = STATUS_READING_STRING;
+				}
+
+				if (l.is_empty() || l.begins_with("#")) {
+					line++;
+					continue; // Nothing to read or comment.
+				}
+
+				ERR_FAIL_COND_V_MSG(!l.begins_with("\"") || status == STATUS_NONE, Vector<Vector<String>>(), "Invalid line '" + l + "' while parsing: " + path + ":" + itos(line));
+
+				l = l.substr(1, l.length());
+				// Find final quote, ignoring escaped ones (\").
+				// The escape_next logic is necessary to properly parse things like \\"
+				// where the backslash is the one being escaped, not the quote.
+				int end_pos = -1;
+				bool escape_next = false;
+				for (int i = 0; i < l.length(); i++) {
+					if (l[i] == '\\' && !escape_next) {
+						escape_next = true;
+						continue;
+					}
+
+					if (l[i] == '"' && !escape_next) {
+						end_pos = i;
+						break;
+					}
+
+					escape_next = false;
+				}
+
+				ERR_FAIL_COND_V_MSG(end_pos == -1, Vector<Vector<String>>(), "Expected '\"' at end of message while parsing: " + path + ":" + itos(line));
+
+				l = l.substr(0, end_pos);
+				l = l.c_unescape();
+
+				if (status == STATUS_READING_ID) {
+					msg_id += l;
+				} else if (status == STATUS_READING_CONTEXT) {
+					msg_context += l;
+				} else if (status == STATUS_READING_PLURAL) {
+					msg_id_plural += l;
+				}
+
+				line++;
 			}
 		}
 
 		etl++;
 	}
 
-	return msgids;
+	return list;
 }