a better API for loading PGF & NGF files

src/runtime/c/data.h

@@ -121,7 +121,8 @@ struct PGF_INTERNAL_DECL PgfPGFRoot {
 #pragma GCC diagnostic ignored "-Wattributes"
 
 struct PgfPGF : DB {
-    PGF_INTERNAL_DECL PgfPGF(const char* fpath) : DB(fpath) {};
+    PGF_INTERNAL_DECL PgfPGF(const char* fpath, int flags, int mode)
+                         : DB(fpath, flags, mode) {};
     PGF_INTERNAL_DECL ~PgfPGF() {};
 };
 

src/runtime/c/db.cxx

@@ -270,26 +270,35 @@ struct malloc_state
     moffset root_offset;
 };
 
-DB::DB(const char* pathname) {
-    fd = open(pathname, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR);
-    if (fd < 0)
-        throw std::system_error(errno, std::generic_category());
-
-    size_t file_size = lseek(fd, 0, SEEK_END);
-
-    if (file_size == ((off_t) -1))
-        throw std::system_error(errno, std::generic_category());
-
+DB::DB(const char* pathname, int flags, int mode) {
+    size_t file_size;
     bool is_new = false;
-    if (file_size == 0) {
+
+    if (pathname == NULL) {
+        fd = -1;
         file_size = getpagesize();
-        if (ftruncate(fd, file_size) < 0)
+        is_new    = true;
+    } else {
+        fd = open(pathname, flags, mode);
+        if (fd < 0)
             throw std::system_error(errno, std::generic_category());
-        is_new = true;
+
+        file_size = lseek(fd, 0, SEEK_END);
+        if (file_size == ((off_t) -1))
+            throw std::system_error(errno, std::generic_category());
+
+        is_new = false;
+        if (file_size == 0) {
+            file_size = getpagesize();
+            if (ftruncate(fd, file_size) < 0)
+                throw std::system_error(errno, std::generic_category());
+            is_new = true;
+        }
     }
 
+    int mflags = (fd < 0) ? (MAP_PRIVATE | MAP_ANONYMOUS) : MAP_SHARED;
     ms = (malloc_state*)
-        mmap(NULL, file_size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0);
+        mmap(NULL, file_size, PROT_READ | PROT_WRITE, mflags, fd, 0);
     if (ms == MAP_FAILED)
         throw std::system_error(errno, std::generic_category());
 
@@ -306,7 +315,9 @@ DB::~DB() {
         ms->top + size + sizeof(size_t);
 
     munmap(ms,size);
-    close(fd);
+
+    if (fd >= 0)
+        close(fd);
 }
 
 void DB::sync()
@@ -803,8 +814,10 @@ DB::malloc_internal(size_t bytes)
             size_t new_size =
                 old_size + alloc_size;
 
-            if (ftruncate(fd, new_size) < 0)
-                throw std::system_error(errno, std::generic_category());
+            if (fd >= 0) {
+                if (ftruncate(fd, new_size) < 0)
+                    throw std::system_error(errno, std::generic_category());
+            }
 
             malloc_state* new_ms =
                 (malloc_state*) mremap(ms, old_size, new_size, MREMAP_MAYMOVE);

src/runtime/c/db.h

@@ -63,7 +63,7 @@ private:
     friend class PgfReader;
 
 public:
-    DB(const char* pathname);
+    DB(const char* pathname, int flags, int mode);
     ~DB();
 
     template<class A>

src/runtime/c/pgf.cxx

@@ -1,3 +1,4 @@
+#include <fcntl.h>
 #include "data.h"
 #include "reader.h"
 
@@ -10,21 +11,14 @@ pgf_exn_clear(PgfExn* err)
 }
 
 PGF_API
-PgfPGF *pgf_read(const char* fpath, PgfExn* err)
+PgfPGF *pgf_read_pgf(const char* fpath, PgfExn* err)
 {
     PgfPGF *pgf = NULL;
 
     pgf_exn_clear(err);
 
     try {
-        std::string fpath_n = fpath;
-        size_t len = fpath_n.length();
-        if (len > 4 && fpath_n.substr(len-4) == ".pgf")
-            fpath_n[len-3] = 'n';
-        else if (!(len > 4 && fpath_n.substr(len-4) == ".ngf"))
-            fpath_n += ".ngf";
-
-        pgf = new PgfPGF(fpath_n.c_str());
+        pgf = new PgfPGF(NULL, 0, 0);
 
         if (DB::get_root<PgfPGFRoot>() == 0) {
             std::ifstream in(fpath, std::ios::binary);
@@ -55,6 +49,80 @@ PgfPGF *pgf_read(const char* fpath, PgfExn* err)
     return NULL;
 }
 
+PGF_API
+PgfPGF *pgf_boot_ngf(const char* pgf_path, const char* ngf_path, PgfExn* err)
+{
+    PgfPGF *pgf = NULL;
+
+    pgf_exn_clear(err);
+
+    try {
+        pgf = new PgfPGF(ngf_path, O_CREAT | O_EXCL | O_RDWR, S_IRUSR | S_IWUSR);
+
+        std::ifstream in(pgf_path, std::ios::binary);
+        if (in.fail()) {
+            throw std::system_error(errno, std::generic_category());
+        }
+
+        PgfReader rdr(&in);
+        ref<PgfPGFRoot> pgf_root = rdr.read_pgf();
+
+        pgf->set_root(pgf_root);
+
+        DB::sync();
+
+        return pgf;
+    } catch (std::system_error& e) {
+        err->type = PGF_EXN_SYSTEM_ERROR;
+        err->code = e.code().value();
+    } catch (pgf_error& e) {
+        err->type = PGF_EXN_PGF_ERROR;
+        err->msg  = strdup(e.what());
+    }
+
+    if (pgf != NULL)
+        delete pgf;
+
+    return NULL;
+}
+
+PGF_API
+PgfPGF *pgf_read_ngf(const char *fpath, PgfExn* err)
+{
+    PgfPGF *pgf = NULL;
+
+    pgf_exn_clear(err);
+
+    try {
+        pgf = new PgfPGF(fpath, O_CREAT | O_RDWR, S_IRUSR | S_IWUSR);
+
+        if (DB::get_root<PgfPGFRoot>() == 0) {
+            ref<PgfPGFRoot> pgf = DB::malloc<PgfPGFRoot>();
+            pgf->major_version = 2;
+            pgf->minor_version = 0;
+            pgf->gflags = 0;
+            pgf->abstract.name = DB::malloc<PgfText>();
+            pgf->abstract.name->size = 0;
+            pgf->abstract.aflags = 0;
+            pgf->abstract.funs = 0;
+            pgf->abstract.cats = 0;
+        }
+
+        return pgf;
+    } catch (std::system_error& e) {
+        err->type = PGF_EXN_SYSTEM_ERROR;
+        err->code = e.code().value();
+    } catch (pgf_error& e) {
+        err->type = PGF_EXN_PGF_ERROR;
+        err->msg  = strdup(e.what());
+    }
+
+    if (pgf != NULL)
+        delete pgf;
+
+    return NULL;
+}
+
 PGF_API
 void pgf_free(PgfPGF *pgf)
 {

src/runtime/c/pgf.h

@@ -80,9 +80,25 @@ typedef struct {
     const char *msg;
 } PgfExn;
 
+/* Reads a PGF file and keeps it in memory. */
 PGF_API_DECL
-PgfPGF *pgf_read(const char* fpath, PgfExn* err);
+PgfPGF *pgf_read_pgf(const char* fpath, PgfExn* err);
 
+/* Reads a PGF file and stores the unpacked data in an NGF file
+ * ready to be shared with other process, or used for quick startup.
+ * The NGF file is platform dependent and should not be copied
+ * between machines. */
+PGF_API_DECL
+PgfPGF *pgf_boot_ngf(const char* pgf_path, const char* ngf_path, PgfExn* err);
+
+/* Tries to read the grammar from an already booted NGF file.
+ * If the file does not exist then a new one is created, and the
+ * grammar is set to be empty. It can later be populated with
+ * rules dynamically. */
+PGF_API_DECL
+PgfPGF *pgf_read_ngf(const char* fpath, PgfExn* err);
+
+/* Release the grammar when it is no longer needed. */
 PGF_API_DECL
 void pgf_free(PgfPGF *pgf);
 

src/runtime/haskell/PGF2.hsc

@@ -14,7 +14,7 @@
 -------------------------------------------------
 
 module PGF2 (-- * PGF
-             PGF,readPGF,
+             PGF,readPGF,bootNGF,readNGF,
 
              -- * Abstract syntax
              AbsName,abstractName,
@@ -48,7 +48,44 @@ readPGF fpath =
   withCString fpath $ \c_fpath ->
   allocaBytes (#size PgfExn) $ \c_exn ->
   mask_ $ do
-    c_pgf <- pgf_read c_fpath c_exn
+    c_pgf <- pgf_read_pgf c_fpath c_exn
+    ex_type <- (#peek PgfExn, type) c_exn :: IO (#type PgfExnType)
+    if ex_type == (#const PGF_EXN_NONE)
+      then do fptr <- newForeignPtr pgf_free_fptr c_pgf
+              return (PGF fptr Map.empty)
+      else if ex_type == (#const PGF_EXN_SYSTEM_ERROR)
+             then do errno <- (#peek PgfExn, code) c_exn
+                     ioError (errnoToIOError "readPGF" (Errno errno) Nothing (Just fpath))
+             else do c_msg <- (#peek PgfExn, msg) c_exn
+                     msg <- peekCString c_msg
+                     free c_msg
+                     throwIO (PGFError msg)
+
+bootNGF :: FilePath -> FilePath -> IO PGF
+bootNGF pgf_path ngf_path =
+  withCString pgf_path $ \c_pgf_path ->
+  withCString ngf_path $ \c_ngf_path ->
+  allocaBytes (#size PgfExn) $ \c_exn ->
+  mask_ $ do
+    c_pgf <- pgf_boot_ngf c_pgf_path c_ngf_path c_exn
+    ex_type <- (#peek PgfExn, type) c_exn :: IO (#type PgfExnType)
+    if ex_type == (#const PGF_EXN_NONE)
+      then do fptr <- newForeignPtr pgf_free_fptr c_pgf
+              return (PGF fptr Map.empty)
+      else if ex_type == (#const PGF_EXN_SYSTEM_ERROR)
+             then do errno <- (#peek PgfExn, code) c_exn
+                     ioError (errnoToIOError "bootNGF" (Errno errno) Nothing (Just pgf_path))
+             else do c_msg <- (#peek PgfExn, msg) c_exn
+                     msg <- peekCString c_msg
+                     free c_msg
+                     throwIO (PGFError msg)
+
+readNGF :: FilePath -> IO PGF
+readNGF fpath =
+  withCString fpath $ \c_fpath ->
+  allocaBytes (#size PgfExn) $ \c_exn ->
+  mask_ $ do
+    c_pgf <- pgf_read_ngf c_fpath c_exn
     ex_type <- (#peek PgfExn, type) c_exn :: IO (#type PgfExnType)
     if ex_type == (#const PGF_EXN_NONE)
       then do fptr <- newForeignPtr pgf_free_fptr c_pgf

src/runtime/haskell/PGF2/FFI.hsc

@@ -31,8 +31,14 @@ foreign import ccall unsafe "pgf_utf8_decode"
 foreign import ccall unsafe "pgf_utf8_encode"
   pgf_utf8_encode :: Word32 -> Ptr CString -> IO ()
 
-foreign import ccall "pgf_read"
-  pgf_read :: CString -> Ptr PgfExn -> IO (Ptr PgfPGF)
+foreign import ccall "pgf_read_pgf"
+  pgf_read_pgf :: CString -> Ptr PgfExn -> IO (Ptr PgfPGF)
+
+foreign import ccall "pgf_boot_ngf"
+  pgf_boot_ngf :: CString -> CString -> Ptr PgfExn -> IO (Ptr PgfPGF)
+
+foreign import ccall "pgf_read_ngf"
+  pgf_read_ngf :: CString -> Ptr PgfExn -> IO (Ptr PgfPGF)
 
 foreign import ccall "&pgf_free"
   pgf_free_fptr :: FinalizerPtr PgfPGF