Support emscripten build with PBRST support, also via cmake

CMakeLists.txt

@@ -1,30 +1,39 @@
+# This cmake configuration file has been successfully tested on the following platforms:
+# Linux, macOS, MSYS2/MINGW64, emscripten.
+
+# When building for emscripten, build a native Linux version first, create the addbooks cache,
+# and just then start the build for emscripten, providing the bibref-addbooks-cache folder name
+# as an option when starting cmake. See the variable BIBREF_ADDBOOKS_CACHE_FOLDER below.
+
 cmake_minimum_required(VERSION 3.16)
 project(bibref)
 
-option(PBRST "Include the brst parser" OFF)
+set(BIBREF_ADDBOOKS_CACHE_FOLDER ../bibref-addbooks-cache CACHE STRING
+ "Set the previously created bibref-addbooks-cache folder for the Emscripten build")
 
 set(CMAKE_CXX_STANDARD 11)
 set(CMAKE_CXX_FLAGS "-pedantic-errors")
 
 find_package(PkgConfig REQUIRED)
 pkg_check_modules(sword REQUIRED sword)
 
+if(EMSCRIPTEN)
+find_library(SWORD_LIB NAMES libsword.a)
+endif(EMSCRIPTEN)
+
 set(Boost_USE_STATIC_LIBS OFF)
 set(Boost_USE_MULTITHREADED ON)
 set(Boost_USE_STATIC_RUNTIME OFF)
 find_package(Boost CONFIG COMPONENTS system filesystem)
 
-if(APPLE) # readline seems to have trouble on MacOS
+if(APPLE OR EMSCRIPTEN) # readline seems to have trouble on macOS, unsupported on emscripten
 else()
 # One needs readline >= 8.0 to make this work:
 pkg_check_modules (readline REQUIRED readline)
 endif()
 
 include_directories(${sword_INCLUDE_DIRS})
 
-if(PBRST)
-add_compile_definitions(WITH_PBRST)
-
 set(PARSER_DIR "${CMAKE_CURRENT_BINARY_DIR}")
 find_package(FLEX 2.6 REQUIRED)
 find_package(BISON 3.0 REQUIRED)
@@ -38,7 +47,7 @@ bison_target(PARSER "${CMAKE_CURRENT_SOURCE_DIR}/statements/pbrst.y" "${PARSER_O
 add_flex_bison_dependency(LEXER PARSER)
 
 include_directories(${PARSER_DIR} ${CMAKE_CURRENT_SOURCE_DIR})
-set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DYYDEBUG=1 -DIN_BIBREF")
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DYYDEBUG=1")
 
 #add_executable(pbrst-cli statements/pbrst-cli.c ${PARSER_DIR}/pbrst.tab.h "${LEXER_OUT}" "${PARSER_OUT}")
 
@@ -47,7 +56,6 @@ SET(PBRST_SRCS
  fingerprint-wrapper.cpp fingerprint-wrapper.h
  ${PARSER_DIR}/pbrst.tab.h "${LEXER_OUT}" "${PARSER_OUT}"
 )
-endif(PBRST)
 
 add_executable(bibref main.cpp book.cpp book.h books.cpp books.h fingerprint.cpp fingerprint.h cli.cpp cli.h psalmsinfo.cpp psalmsinfo.h
  ${PBRST_SRCS}
@@ -63,16 +71,25 @@ if(APPLE)
 set(sword_LIBRARIES_folder "-L/usr/local/lib") # this is hardcoded (via configure/make install), FIXME
 endif()
 
-target_link_libraries(bibref
-    ${sword_LIBRARIES} ${sword_LIBRARIES_folder} ${Boost_LIBRARIES} ${readline_LIBRARIES}
-)
-
+if(EMSCRIPTEN)
+target_link_libraries(bibref ${SWORD_LIB} ${Boost_LIBRARIES})
+target_link_options(bibref PUBLIC -sSINGLE_FILE -sUSE_BOOST_HEADERS -sALLOW_MEMORY_GROWTH -O3 -fexceptions)
+target_link_options(bibref PUBLIC "SHELL:
+  --preload-file '$ENV{HOME}/.sword@/'")
+target_link_options(bibref PUBLIC "SHELL:
+  --preload-file '${BIBREF_ADDBOOKS_CACHE_FOLDER}@/bibref-addbooks-cache'")
+target_link_options(bibref PUBLIC -sEXTRA_EXPORTED_RUNTIME_METHODS=['cwrap'] -sMODULARIZE -sEXPORT_NAME=bibref)
+else()
+target_link_libraries(bibref ${sword_LIBRARIES} ${sword_LIBRARIES_folder} ${Boost_LIBRARIES} ${readline_LIBRARIES})
+endif(EMSCRIPTEN)
 
 install(
  CODE " set(ENV{SWORD_PATH} \"$ENV{HOME}/.sword\") "
 # Download and install LXX and SBLGNT via CrossWire...
 ## Now we disable this because LXX was updated to version 3.0 and its versification ("LXX") differs from version 2.5 ("KJV"),
 ## so we want to avoid installing the new version for the moment. Instead, we use the bundled versions...
+## In fact, now (February 2025) the bundled version contains LXX version 3.0, but
+## it is simpler to maintain a ZIP bundle than managing the modules via installmgr.
 ## CODE " execute_process (COMMAND bash -c \"test -d \\\"$SWORD_PATH/mods.d\\\" || (mkdir -p \\\"$SWORD_PATH/mods.d\\\"; echo yes | installmgr -init)\") "
 ## CODE " execute_process (COMMAND bash -c \"installmgr -s | grep \\\"\\\\[CrossWire\\\\]\\\" || echo yes | installmgr -sc\") "
 ## CODE " execute_process (COMMAND bash -c \"echo yes | installmgr -r CrossWire\") "

Makefile

@@ -23,13 +23,16 @@
 #    in the cmake configuration are important, otherwise the SWORD utilities
 #    and examples will also be built, but some of them will eventually fail
 #    and the library will not be copied to the official emsdk folder.
-#    (In that case you need to do that manually.)
+#    (In that case you need to do that manually.) There will be no shared library created.
 #
-# 3. Step 2 will make sure that libsword.a is visible for the rest of the process,
-#    but the first LDFLAGS setting makes sure that one can extend the search path
-#    by the scope of the ".." folder.
+# 3. Step 2 will make sure that libsword.a is visible for the rest of the process.
+#    Unfortunately, since static libraries are not available published properly by
+#    the SWORD library setup, we still have to work around manually that
+#    the first LDFLAGS setting extends the scope to find the library also under the ".." folder.
 #
-# 4. Issue "emmake make" in the current directory.
+# 4. Create symlinks in this folder to the files statements/{pbrst.c,pbrst.tab.c,pbrst.tab.h}.
+#    These files will be present only if you successfully compile the native program above.
+#    Issue then "emmake make" in the current directory.
 #
 # See also the GitHub actions in the .github/ folder for more information on the
 # steps above. They may be somewhat different, but hopefully still working.
@@ -67,12 +70,16 @@ TARGET_HTML ?= bibref.html
 
 BUILD_DIR ?= ./wasm-build
 
-SRCS := book.cpp books.cpp cli.cpp fingerprint.cpp main.cpp psalmsinfo.cpp
-OBJS := $(SRCS:%=$(BUILD_DIR)/%.o)
+CXX_SRCS := book.cpp books.cpp cli.cpp fingerprint.cpp main.cpp psalmsinfo.cpp \
+	books-wrapper.cpp fingerprint-wrapper.cpp
+CXX_OBJS := $(CXX_SRCS:%=$(BUILD_DIR)/%.o)
+C_SRCS := pbrst.tab.c pbrst.c
+C_OBJS := $(C_SRCS:%=$(BUILD_DIR)/%.o)
 DOXS := book.dox books.dox cli.dox fingerprint.dox main.dox psalmsinfo.dox
 
-CPPFLAGS += -I/usr/include/sword -s USE_BOOST_HEADERS=1
+CPPFLAGS += -I/usr/include/sword -s USE_BOOST_HEADERS=1 -DYYDEBUG=1
 
+CFLAGS := -s EXPORTED_FUNCTIONS=[_bibref_wasm,_bibref_wasm_vocabulary] -s EXTRA_EXPORTED_RUNTIME_METHODS=['cwrap'] -fexceptions
 CXXFLAGS := -s EXPORTED_FUNCTIONS=[_bibref_wasm,_bibref_wasm_vocabulary] -s EXTRA_EXPORTED_RUNTIME_METHODS=['cwrap'] -fexceptions
 
 BIBREF_NATIVE_FOLDER := $(shell find . -name bibref -executable -printf "%h\n" | sort | head -1)
@@ -87,15 +94,19 @@ ifeq ($(TARGET_HTML),bibref.js)
 LDFLAGS += -s MODULARIZE=1 -s EXPORT_NAME=bibref
 endif
 
-$(BUILD_DIR)/$(TARGET_HTML): $(OBJS)
+$(BUILD_DIR)/$(TARGET_HTML): $(CXX_OBJS) $(C_OBJS)
 	$(MKDIR_P) $(dir $@)
 	echo addbooks | $(BIBREF_NATIVE_FOLDER)/bibref # Make sure bibref generates its cache.
-	$(CXX) $(OBJS) -o $@ $(LDFLAGS)
+	$(LD) $(CXX_OBJS) $(C_OBJS) -o $@ $(LDFLAGS)
 
 $(BUILD_DIR)/%.cpp.o: %.cpp
 	$(MKDIR_P) $(dir $@)
 	$(CXX) $(CPPFLAGS) $(CXXFLAGS) -c $< -o $@
 
+$(BUILD_DIR)/%.c.o: %.c
+	$(MKDIR_P) $(dir $@)
+	$(CC) $(CPPFLAGS) $(CFLAGS) -c $< -o $@
+
 #######################################################################################################
 # Documentation related code
 

cli.cpp

@@ -7,12 +7,9 @@
 #include <algorithm>
 #include <vector>
 
-#ifdef WITH_PBRST
 extern "C" char* brst_scan_string(char *string, int correct_raw, int correct_differ,
   int correct_cover, int correct_versification, int show_dump);
-// #include "statements/pbrst.tab.h" // use flex/bison parser for bibref statements (brst)
 #include "pbrst.tab.h" // the statements folder must be included among the folders
-#endif
 
 using namespace std;
 
@@ -75,10 +72,8 @@ string tokensCmd = "tokens";
 string searchCmd = "search";
 string quitCmd = "quit";
 string helpCmd = "help";
-#ifdef WITH_PBRST
 string statementCmd = "statement";
 string statementCmd2 = "Statement";
-#endif
 
 string errorNotRecognized = "Sorry, the command you entered was not recognized or its syntax is invalid.";
 string errorTextIncomplete = "Either " + textCmd + "1 or " + textCmd + "2 must be used.";
@@ -111,11 +106,8 @@ vector<string> commands {addbooksCmd, compareCmd + "12", jaccardCmd + "12",
       minuniqueCmd + "1", latintextCmd + "1", latintextCmd + "2",
       extendCmd, getrefsCmd, lookupCmd, maxresultsCmd, sqlCmd, psalminfoCmd,
       rawCmd, rawCmd + "1", rawCmd + "2",
-      printCmd + "1", printCmd + "2", colorsCmd, tokensCmd, searchCmd
-#ifdef WITH_PBRST
-      , statementCmd
-#endif
-                          };
+      printCmd + "1", printCmd + "2", colorsCmd, tokensCmd, searchCmd, statementCmd
+};
 vector<string> vocabulary = commands;
 
 void add_vocabulary_item(string item) {
@@ -261,10 +253,8 @@ string getHelp(const string &key)
       "Tokenization is done via Strong's numbers.",
       "* `help` *command*: Show some hints on usage of *command*, or get general help if no "
       "parameter is given.",
-#ifdef WITH_PBRST
       "* `statement` ...: Analyze the given statement, see https://matek.hu/zoltan/blog-20250102.php "
       "for some explanations.",
-#endif
       "* `quit`: Exit program."};
   string retval;
   for (int i = 0; i < helpStr.size(); i++) {
@@ -851,7 +841,6 @@ void processGetrefsCmd(string input) {
   info("Finished"); // Success!
 }
 
-#ifdef WITH_PBRST
 void processStatementCmd(string input) {
   char *output = brst_scan_string((char*)input.c_str(), 0, 0, 0, 0, 0);
   string output_s(output);
@@ -865,7 +854,6 @@ void processStatementCmd(string input) {
     }
   info("Finished"); // Success!
 }
-#endif
 
 string cli_process(char *buf) {
   string rawinput(buf); // Convert the input to string.
@@ -925,10 +913,8 @@ string cli_process(char *buf) {
     processExtendCmd(input);
   else if (boost::starts_with(input, getrefsCmd))
     processGetrefsCmd(input);
-#ifdef WITH_PBRST
   else if (boost::starts_with(input, statementCmd) || boost::starts_with(input, statementCmd2))
     processStatementCmd(input);
-#endif
   // If the input is not recognized, we show an error...
   else if (input.length() != 0) // unless there was no input at all
     error(errorNotRecognized);
@@ -999,14 +985,12 @@ void cli(const char *input_prepend, const char *output_prepend, bool addbooks, b
     line = string(bufline);
     // Handle multiline inputs (for the statement command)...
     boost::algorithm::trim(line); // Trim the input.
-#ifdef WITH_PBRST
     if (!multiline && (boost::starts_with(line, statementCmd) || boost::starts_with(line, statementCmd2))) {
       multiline = true;
     }
     if (multiline && boost::ends_with(line, ".")) {
       multiline = false;
     }
-#endif // WITH_PBRST
     strcat(buf, bufline);
     if (!multiline) {
       cli_process(buf); // Process the input.

qt/CMakeLists.txt

@@ -1,8 +1,6 @@
 cmake_minimum_required(VERSION 3.16)
 project(bibref-qt LANGUAGES CXX C)
 
-option(PBRST "Include the brst parser" OFF)
-
 find_package(Qt6 REQUIRED COMPONENTS Core Gui Widgets Concurrent SvgWidgets)
 
 find_package(PkgConfig REQUIRED)
@@ -50,9 +48,6 @@ set(CMAKE_AUTOUIC ON)
 add_definitions(-DPROJECT_SOURCE_DIR="${PROJECT_SOURCE_DIR}"
     -DINSTALL_PREFIX="${CMAKE_INSTALL_PREFIX}")
 
-if(PBRST)
-add_compile_definitions(WITH_PBRST)
-
 set(PARSER_DIR "${CMAKE_CURRENT_BINARY_DIR}")
 find_package(FLEX 2.6 REQUIRED)
 find_package(BISON 3.0 REQUIRED)
@@ -66,16 +61,14 @@ bison_target(PARSER "${CMAKE_CURRENT_SOURCE_DIR}/../statements/pbrst.y" "${PARSE
 add_flex_bison_dependency(LEXER PARSER)
 
 include_directories(${PARSER_DIR} "${CMAKE_CURRENT_SOURCE_DIR}/..")
-set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DYYDEBUG=1 -DIN_BIBREF")
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DYYDEBUG=1")
 
 SET(PBRST_SRCS
  books-wrapper.cpp books-wrapper.h
  fingerprint-wrapper.cpp fingerprint-wrapper.h
  "${LEXER_OUT}" "${PARSER_OUT}"
 )
 
-endif(PBRST)
-
 # set the Application icon, the first line is the property added to Info.plist
 set(MACOSX_BUNDLE_ICON_FILE bibref-qt.icns)
 set(bibref_qt_ICON ${CMAKE_CURRENT_SOURCE_DIR}/logo-Psalm40-256.icns)

qt/statementwindow.cpp

@@ -4,7 +4,6 @@
 #include <QtWidgets>
 #include <iostream>
 
-#ifdef WITH_PBRST
 #include <string>
 #include <boost/algorithm/string/trim.hpp>
 #include <boost/algorithm/string/predicate.hpp>
@@ -15,7 +14,6 @@ extern "C" char* brst_scan_string(char *string, int correct_raw, int correct_dif
 #include "pbrst.tab.h" // the statements folder must be included among the folders
 
 using namespace std;
-#endif
 
 StatementWindow::StatementWindow(QWidget *parent)
     : QMainWindow(parent)
@@ -123,7 +121,6 @@ void StatementWindow::setupFileMenu()
 
 void StatementWindow::parse()
 {
-#ifdef WITH_PBRST
     char* output = brst_scan_string((char*)editor->toPlainText().toStdString().c_str(), 0, 0, 0, 0, 0);
     string output_s(output);
     vector<string> statementAnalysis;
@@ -195,30 +192,24 @@ void StatementWindow::parse()
         link += QUrl::toPercentEncoding(QString::fromStdString(graphviz_input));
         QDesktopServices::openUrl(QUrl(link));
     }
-
-#endif
 }
 
 void StatementWindow::showSvg()
 {
-#ifdef WITH_PBRST
     auto vwindow = new VisualizeWindow(this, graphviz_input);
     // vwindow->resize(600, 400);
     vwindow->show();
     vwindow->setWindowIcon(QIcon::fromTheme("emblem-photos"));
     vwindow->setWindowTitle(tr("Visualize"));
-#endif // WITH_PBRST
 }
 
 void StatementWindow::setupProveMenu()
 {
-#ifdef WITH_PBRST
     QMenu *proveMenu = new QMenu(tr("&Prove"), this);
     menuBar()->addMenu(proveMenu);
 
     proveMenu->addAction(QIcon::fromTheme("tools-check-spelling"), tr("&Parse"), QKeySequence::Forward,
                         this, &StatementWindow::parse);
-#endif // WITH_PBRST
 }
 
 void StatementWindow::setupHelpMenu()