update mupdf

mupdf/Makefile

@@ -454,6 +454,7 @@ incdir ?= $(prefix)/include
 mandir ?= $(prefix)/share/man
 docdir ?= $(prefix)/share/doc/mupdf
 pydir ?= $(shell python3 -c "import sysconfig; print(sysconfig.get_path('platlib'))")
+SO_INSTALL_MODE ?= 644
 
 third: $(THIRD_LIB)
 extra-libs: $(THIRD_GLUT_LIB)
@@ -612,13 +613,17 @@ $(error OUT=$(OUT) does not contain shared)
 endif
 
 # C++, Python and C# shared libraries.
+#
+# To disable automatic use of a venv, use `make VENV_FLAG= ...` or `VENV_FLAG=
+# make ...`.
+#
+VENV_FLAG ?= --venv
 c++-%: shared-%
-	./scripts/mupdfwrap.py --venv -d $(OUT) -b 01
+	./scripts/mupdfwrap.py $(VENV_FLAG) -d $(OUT) -b 01
 python-%: c++-%
-	./scripts/mupdfwrap.py --venv -d $(OUT) -b 23
+	./scripts/mupdfwrap.py $(VENV_FLAG) -d $(OUT) -b 23
 csharp-%: c++-%
-	./scripts/mupdfwrap.py --venv -d $(OUT) -b --csharp 23
-
+	./scripts/mupdfwrap.py $(VENV_FLAG) -d $(OUT) -b --csharp 23
 
 # Installs of C, C++, Python and C# shared libraries
 #
@@ -632,21 +637,21 @@ endif
 
 install-shared-c: install-shared-check shared install-headers
 	install -d $(DESTDIR)$(libdir)
-	install -m 644 $(OUT)/libmupdf.$(SO)$(SO_VERSION) $(DESTDIR)$(libdir)/
+	install -m $(SO_INSTALL_MODE) $(OUT)/libmupdf.$(SO)$(SO_VERSION) $(DESTDIR)$(libdir)/
 ifneq ($(OS),OpenBSD)
 	ln -sf libmupdf.$(SO)$(SO_VERSION) $(DESTDIR)$(libdir)/libmupdf.$(SO)
 endif
 
 install-shared-c++: install-shared-c c++
 	install -m 644 platform/c++/include/mupdf/*.h $(DESTDIR)$(incdir)/mupdf
-	install -m 644 $(OUT)/libmupdfcpp.$(SO)$(SO_VERSION) $(DESTDIR)$(libdir)/
+	install -m $(SO_INSTALL_MODE) $(OUT)/libmupdfcpp.$(SO)$(SO_VERSION) $(DESTDIR)$(libdir)/
 ifneq ($(OS),OpenBSD)
 	ln -sf libmupdfcpp.$(SO)$(SO_VERSION) $(DESTDIR)$(libdir)/libmupdfcpp.$(SO)
 endif
 
 install-shared-python: install-shared-c++ python
 	install -d $(DESTDIR)$(pydir)/mupdf
-	install -m 644 $(OUT)/_mupdf.$(SO) $(DESTDIR)$(pydir)/mupdf
+	install -m $(SO_INSTALL_MODE) $(OUT)/_mupdf.$(SO) $(DESTDIR)$(pydir)/mupdf
 	install -m 644 $(OUT)/mupdf.py $(DESTDIR)$(pydir)/mupdf/__init__.py
 
 else

mupdf/Makelists

@@ -586,6 +586,7 @@ TESSERACT_SRC += thirdparty/tesseract/src/api/pdfrenderer.cpp
 TESSERACT_SRC += thirdparty/tesseract/src/api/renderer.cpp
 TESSERACT_SRC += thirdparty/tesseract/src/api/wordstrboxrenderer.cpp
 TESSERACT_SRC += thirdparty/tesseract/src/arch/dotproduct.cpp
+TESSERACT_SRC += thirdparty/tesseract/src/arch/dotproductneon.cpp
 TESSERACT_SRC += thirdparty/tesseract/src/arch/intsimdmatrix.cpp
 TESSERACT_SRC += thirdparty/tesseract/src/arch/simddetect.cpp
 TESSERACT_SRC += thirdparty/tesseract/src/ccmain/applybox.cpp

mupdf/include/mupdf/fitz/string-util.h

@@ -249,6 +249,12 @@ const char *fz_runeptr(const char *str, int idx);
 */
 int fz_utflen(const char *s);
 
+/*
+	Convert a wchar string into a new heap allocated utf8 one.
+*/
+char *fz_utf8_from_wchar(fz_context *ctx, const wchar_t *s);
+
+
 /**
 	Locale-independent decimal to binary conversion. On overflow
 	return (-)INFINITY and set errno to ERANGE. On underflow return

mupdf/include/mupdf/fitz/system.h

@@ -165,9 +165,6 @@ static __inline int signbit(double x)
 
 #ifdef _WIN32
 
-char *fz_utf8_from_wchar(const wchar_t *s);
-wchar_t *fz_wchar_from_utf8(const char *s);
-
 /* really a FILE* but we don't want to include stdio.h here */
 void *fz_fopen_utf8(const char *name, const char *mode);
 int fz_remove_utf8(const char *name);
@@ -184,6 +181,8 @@ void fz_free_argv(int argc, char **argv);
 
 int64_t fz_stat_ctime(const char *path);
 int64_t fz_stat_mtime(const char *path);
+int fz_mkdir(char *path);
+
 
 /* inline is standard in C++. For some compilers we can enable it within
  * C too. Some compilers think they know better than we do about when

mupdf/platform/gl/gl-main.c

@@ -499,25 +499,6 @@ static void save_history(void)
 	js_freestate(J);
 }
 
-static int
-fz_mkdir(char *path)
-{
-#ifdef _WIN32
-	int ret;
-	wchar_t *wpath = fz_wchar_from_utf8(path);
-
-	if (wpath == NULL)
-		return -1;
-
-	ret = _wmkdir(wpath);
-
-	free(wpath);
-
-	return ret;
-#else
-	return mkdir(path, S_IRWXU | S_IRWXG | S_IRWXO);
-#endif
-}
 
 static int create_accel_path(char outname[], size_t len, int create, const char *absname, ...)
 {

mupdf/platform/win32/libtesseract.vcxproj

@@ -565,6 +565,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <PrecompiledHeaderFile>pch.h</PrecompiledHeaderFile>
       <AdditionalIncludeDirectories>..\..\include;..\..\thirdparty\leptonica\src;..\..\thirdparty\tesseract\include;..\..\thirdparty\tesseract\src\api;..\..\thirdparty\tesseract\src\arch;..\..\thirdparty\tesseract\src\ccmain;..\..\thirdparty\tesseract\src\ccstruct;..\..\thirdparty\tesseract\src\ccutil;..\..\thirdparty\tesseract\src\classify;..\..\thirdparty\tesseract\src\dict;..\..\thirdparty\tesseract\src\lstm;..\..\thirdparty\tesseract\src\textord;..\..\thirdparty\tesseract\src\viewer;..\..\thirdparty\tesseract\src\wordrec;..\..\thirdparty\tesseract\src\cutil;..\..\scripts\tesseract</AdditionalIncludeDirectories>
+      <LanguageStandard>stdcpp17</LanguageStandard>
       <UndefinePreprocessorDefinitions>
       </UndefinePreprocessorDefinitions>
       <RuntimeLibrary>MultiThreadedDLL</RuntimeLibrary>
@@ -650,4 +651,4 @@
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
   <ImportGroup Label="ExtensionTargets">
   </ImportGroup>
-</Project>
+</Project>

mupdf/platform/x11/pdfapp.c

@@ -25,26 +25,6 @@
 #define MAX(a,b) ((a) > (b) ? (a) : (b))
 #endif
 
-static int
-fz_mkdir(char *path)
-{
-#ifdef _WIN32
-	int ret;
-	wchar_t *wpath = fz_wchar_from_utf8(path);
-
-	if (wpath == NULL)
-		return -1;
-
-	ret = _wmkdir(wpath);
-
-	free(wpath);
-
-	return ret;
-#else
-	return mkdir(path, S_IRWXU | S_IRWXG | S_IRWXO);
-#endif
-}
-
 static int create_accel_path(fz_context *ctx, char outname[], size_t len, int create, const char *absname, ...)
 {
 	va_list args;

mupdf/scripts/pipcl.py

@@ -348,8 +348,8 @@ def __init__(self,
                 assumed to be relative to `root`.
 
                 `to_` identifies what the file should be called within a wheel
-                or when installing. If `to_` ends with `/`, the leaf of `from_`
-                is appended to it.
+                or when installing. If `to_` is '' or ends with `/`, the leaf
+                of `from_` is appended to it.
 
                 Initial `$dist-info/` in `_to` is replaced by
                 `{name}-{version}.dist-info/`; this is useful for license files
@@ -1541,7 +1541,7 @@ def build_extension(
         if command_was_run and darwin():
             # We need to patch up references to shared libraries in `libs`.
             sublibraries = list()
-            for lib in libs:
+            for lib in () if libs is None else libs:
                 for libpath in libpaths:
                     found = list()
                     for suffix in '.so', '.dylib':
@@ -1745,6 +1745,7 @@ def run( command, capture=False, check=1):
     else:
         return (cp.returncode, cp.stdout) if capture else cp.returncode
 
+
 def darwin():
     return sys.platform.startswith( 'darwin')
 
@@ -1760,6 +1761,9 @@ def pyodide():
 def linux():
     return platform.system() == 'Linux'
 
+def openbsd():
+    return platform.system() == 'OpenBSD'
+
 class PythonFlags:
     '''
     Compile/link flags for the current python, for example the include path
@@ -1875,6 +1879,8 @@ def macos_patch( library, *sublibraries):
     log2( f'macos_patch(): library={library}  sublibraries={sublibraries}')
     if not darwin():
         return
+    if not sublibraries:
+        return
     subprocess.run( f'otool -L {library}', shell=1, check=1)
     command = 'install_name_tool'
     names = []
@@ -2172,6 +2178,36 @@ def _so_suffix():
     return sysconfig.get_config_var('EXT_SUFFIX')
 
 
+def get_soname(path):
+    '''
+    If we are on Linux and `path` is softlink and points to a shared library
+    for which `objdump -p` contains 'SONAME', return the pointee. Otherwise
+    return `path`. Useful if Linux shared libraries have been created with
+    `-Wl,-soname,...`, where we need to embed the versioned library.
+    '''
+    log1(f'{path=} {os.path.abspath(path)=}.')
+    if linux() and os.path.islink(path):
+        path2 = os.path.realpath(path)
+        log1(f'Is link: {path} -> {path2}.')
+        if subprocess.run(f'objdump -p {path2}|grep SONAME', shell=1, check=0).returncode == 0:
+            log1(f'SONAME, returning {path2=}.')
+            return path2
+        log1(f'Not SONAME')
+    elif openbsd():
+        # Return newest .so with version suffix.
+        sos = glob.glob(f'{path}.*')
+        log1(f'{sos=}')
+        sos2 = list()
+        for so in sos:
+            suffix = so[len(path):]
+            if not suffix or re.match('^[.][0-9.]*[0-9]$', suffix):
+                sos2.append(so)
+        sos2.sort(key=lambda p: os.path.getmtime(p))
+        log1(f'{sos2=}')
+        return sos2[-1]
+    return path
+
+
 def install_dir(root=None):
     '''
     Returns install directory used by `install()`.

mupdf/scripts/wrap/__main__.py

@@ -1211,6 +1211,7 @@ def _get_m_command( build_dirs, j=None, make=None, m_target=None, m_vars=None):
         make_args += f' {m_vars}'
     suffix = None
     build_prefix = ''
+    build_suffix = ''
     in_prefix = True
     for i, flag in enumerate( flags):
         if flag in ('x32', 'x64') or re.match('py[0-9]', flag):
@@ -1219,6 +1220,7 @@ def _get_m_command( build_dirs, j=None, make=None, m_target=None, m_vars=None):
             # when creating wheels; we need to ignore
             # them.
             jlib.log('Ignoring {flag=}')
+            build_suffix += f'-{flag}'
         else:
             if 0: pass  # lgtm [py/unreachable-statement]
             elif flag == 'debug':
@@ -1259,6 +1261,8 @@ def _get_m_command( build_dirs, j=None, make=None, m_target=None, m_vars=None):
     assert suffix, f'Leaf must contain "shared-" or "fpic-": build_dirs.dir_so={build_dirs.dir_so}'
     if build_prefix:
         make_args += f' build_prefix={build_prefix}'
+    if build_suffix:
+        make_args += f' build_suffix={build_suffix}'
     if m_target:
         make_args += f' {m_target}'
     command = f'cd {build_dirs.dir_mupdf} &&'
@@ -1456,7 +1460,11 @@ def link_l_flags(sos):
     if state.state_.pyodide:
         # Don't add '-Wl,-rpath*' etc if building for Pyodide.
         ld_origin = False
-    return jlib.link_l_flags( sos, ld_origin)
+    ret = jlib.link_l_flags( sos, ld_origin)
+    r = os.environ.get('LDFLAGS')
+    if r:
+        ret += f' {r}'
+    return ret
 
 
 def build( build_dirs, swig_command, args, vs_upgrade, make_command):

mupdf/setup.py

@@ -309,30 +309,30 @@ def build():
     if windows():
         infix = '' if sys.maxsize == 2**31 - 1 else '64'
         names = [
-                f'mupdfcpp{infix}.dll', # C and C++.
-                '_mupdf.pyd',           # Python internals.
-                'mupdf.py',             # Python.
+                f'{build_dir()}/mupdfcpp{infix}.dll',   # C and C++.
+                f'{build_dir()}/_mupdf.pyd',            # Python internals.
+                f'{build_dir()}/mupdf.py',              # Python.
                 ]
     elif macos():
         log( f'Contents of {build_dir()} are:')
         for leaf in os.listdir(build_dir()):
             log( f'    {leaf}')
         names = [
-                'libmupdf.dylib',   # C.
-                'libmupdfcpp.so',   # C++.
-                '_mupdf.so',        # Python internals.
-                'mupdf.py',         # Python.
+                f'{build_dir()}/libmupdf.dylib',    # C.
+                f'{build_dir()}/libmupdfcpp.so',    # C++.
+                f'{build_dir()}/_mupdf.so',         # Python internals.
+                f'{build_dir()}/mupdf.py',          # Python.
                 ]
     else:
         names = [
-                'libmupdf.so',      # C.
-                'libmupdfcpp.so',   # C++.
-                '_mupdf.so',        # Python internals.
-                'mupdf.py',         # Python.
+                pipcl.get_soname(f'{build_dir()}/libmupdf.so'),     # C.
+                pipcl.get_soname(f'{build_dir()}/libmupdfcpp.so'),  # C++.
+                f'{build_dir()}/_mupdf.so',                         # Python internals.
+                f'{build_dir()}/mupdf.py',                          # Python.
                 ]
     paths = []
     for name in names:
-        paths.append((f'{build_dir()}/{name}', name))
+        paths.append((name, ''))
 
     log(f'build(): returning: {paths}')
     return paths

mupdf/source/fitz/string.c

@@ -938,3 +938,28 @@ void *fz_memmem(const void *h0, size_t k, const void *n0, size_t l)
 
 	return twoway_memmem(h, h+k, n, l);
 }
+
+char *
+fz_utf8_from_wchar(fz_context *ctx, const wchar_t *s)
+{
+	const wchar_t *src = s;
+	char *d;
+	char *dst;
+	int len = 1;
+
+	while (*src)
+	{
+		len += fz_runelen(*src++);
+	}
+
+	d = Memento_label(fz_malloc(ctx, len), "utf8_from_wchar");
+	dst = d;
+	src = s;
+	while (*src)
+	{
+		dst += fz_runetochar(dst, *src++);
+	}
+	*dst = 0;
+
+	return d;
+}