Make Reader always read from a ByteSource

9 files changed, 203 insertions, 190 deletions

diff --git a/src/byte_source.c b/src/byte_source.c
index 2e4f66cb..ef6bf3bb 100644
--- a/src/byte_source.c
+++ b/src/byte_source.c
@@ -17,12 +17,21 @@
 #include "byte_source.h"
 
 #include "caret.h"
+#include "serd_config.h"
 #include "system.h"
 
 #include "serd/serd.h"
 
+#include <sys/stat.h>
+
+#if USE_POSIX_FADVISE && USE_FILENO
+#  include <fcntl.h>
+#endif
+
 #include <stdbool.h>
 #include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
 
 SerdStatus
@@ -50,28 +59,34 @@ serd_byte_source_page(SerdByteSource* const source)
   return SERD_SUCCESS;
 }
 
-SerdStatus
-serd_byte_source_open_source(SerdByteSource* const     source,
-                             const SerdReadFunc        read_func,
-                             const SerdStreamErrorFunc error_func,
-                             const SerdStreamCloseFunc close_func,
-                             void* const               stream,
-                             const SerdNode* const     name,
-                             const size_t              page_size)
+SerdByteSource*
+serd_byte_source_new_function(const SerdReadFunc        read_func,
+                              const SerdStreamErrorFunc error_func,
+                              const SerdStreamCloseFunc close_func,
+                              void* const               stream,
+                              const SerdNode* const     name,
+                              const size_t              page_size)
 {
-  assert(page_size > 0);
-  memset(source, '\0', sizeof(*source));
-  source->read_func   = read_func;
-  source->error_func  = error_func;
-  source->close_func  = close_func;
-  source->stream      = stream;
-  source->page_size   = page_size;
-  source->buf_size    = page_size;
-  source->name        = serd_node_copy(name);
-  source->caret.file  = source->name;
-  source->caret.line  = 1u;
-  source->caret.col   = 1u;
-  source->from_stream = true;
+  if (!page_size) {
+    return NULL;
+  }
+
+  SerdByteSource* source = (SerdByteSource*)calloc(1, sizeof(SerdByteSource));
+
+  source->read_func  = read_func;
+  source->error_func = error_func;
+  source->close_func = close_func;
+  source->stream     = stream;
+  source->page_size  = page_size;
+  source->buf_size   = page_size;
+  source->type       = FROM_FUNCTION;
+
+  source->name =
+    name ? serd_node_copy(name) : serd_new_string(SERD_STRING("func"));
+
+  source->caret.file = source->name;
+  source->caret.line = 1u;
+  source->caret.col  = 1u;
 
   if (page_size > 1) {
     source->file_buf = (uint8_t*)serd_allocate_buffer(page_size);
@@ -81,58 +96,112 @@ serd_byte_source_open_source(SerdByteSource* const     source,
     source->read_buf = &source->read_byte;
   }
 
-  return SERD_SUCCESS;
+  return source;
 }
 
-SerdStatus
-serd_byte_source_prepare(SerdByteSource* const source)
+static bool
+is_directory(const char* const path)
 {
-  if (source->page_size == 0) {
-    return SERD_FAILURE;
+#ifdef _MSC_VER
+  struct stat st;
+  return !stat(path, &st) && (st.st_mode & _S_IFDIR);
+#else
+  struct stat st;
+  return !stat(path, &st) && S_ISDIR(st.st_mode);
+#endif
+}
+
+SerdByteSource*
+serd_byte_source_new_filename(const char* const path, const size_t page_size)
+{
+  if (page_size == 0 || is_directory(path)) {
+    return NULL;
   }
 
-  source->prepared = true;
+  FILE* const fd = fopen(path, "rb");
+  if (!fd) {
+    return NULL;
+  }
 
-  if (source->from_stream) {
-    return (source->page_size > 1 ? serd_byte_source_page(source)
-                                  : serd_byte_source_advance(source));
+  SerdByteSource* source = (SerdByteSource*)calloc(1, sizeof(SerdByteSource));
+
+  source->read_func  = (SerdReadFunc)fread;
+  source->error_func = (SerdStreamErrorFunc)ferror;
+  source->close_func = (SerdStreamCloseFunc)fclose;
+  source->stream     = fd;
+  source->page_size  = page_size;
+  source->buf_size   = page_size;
+
+  source->name = serd_new_file_uri(SERD_STRING(path), SERD_EMPTY_STRING());
+  source->type = FROM_FILENAME;
+
+  source->caret.file = source->name;
+  source->caret.line = 1u;
+  source->caret.col  = 1u;
+
+  if (page_size > 1) {
+    source->file_buf = (uint8_t*)serd_allocate_buffer(page_size);
+    source->read_buf = source->file_buf;
+    memset(source->file_buf, '\0', page_size);
+  } else {
+    source->read_buf = &source->read_byte;
   }
 
-  return SERD_SUCCESS;
+#if USE_POSIX_FADVISE && USE_FILENO
+  posix_fadvise(fileno(fd), 0, 0, POSIX_FADV_SEQUENTIAL);
+#endif
+
+  return source;
 }
 
-SerdStatus
-serd_byte_source_open_string(SerdByteSource* const source,
-                             const char* const     utf8,
-                             const SerdNode* const name)
+SerdByteSource*
+serd_byte_source_new_string(const char* const     string,
+                            const SerdNode* const name)
 {
-  memset(source, '\0', sizeof(*source));
+  SerdByteSource* source = (SerdByteSource*)calloc(1, sizeof(SerdByteSource));
 
   source->page_size = 1;
+  source->read_buf  = (const uint8_t*)string;
+  source->type      = FROM_STRING;
+
   source->name =
     name ? serd_node_copy(name) : serd_new_string(SERD_STRING("string"));
 
-  source->read_buf   = (const uint8_t*)utf8;
   source->caret.file = source->name;
   source->caret.line = 1u;
   source->caret.col  = 1u;
 
-  return SERD_SUCCESS;
+  return source;
 }
 
 SerdStatus
-serd_byte_source_close(SerdByteSource* const source)
+serd_byte_source_prepare(SerdByteSource* const source)
 {
-  SerdStatus st = SERD_SUCCESS;
-  if (source->close_func) {
-    st = source->close_func(source->stream) ? SERD_ERR_UNKNOWN : SERD_SUCCESS;
-  }
+  source->prepared = true;
+  if (source->type != FROM_STRING) {
+    if (source->page_size > 1) {
+      return serd_byte_source_page(source);
+    }
 
-  if (source->page_size > 1) {
-    serd_free_aligned(source->file_buf);
+    return serd_byte_source_advance(source);
   }
 
-  serd_node_free(source->name);
-  memset(source, '\0', sizeof(*source));
-  return st;
+  return SERD_SUCCESS;
+}
+
+void
+serd_byte_source_free(SerdByteSource* const source)
+{
+  if (source) {
+    if (source->close_func) {
+      source->close_func(source->stream);
+    }
+
+    if (source->page_size > 1) {
+      serd_free_aligned(source->file_buf);
+    }
+
+    serd_node_free(source->name);
+    free(source);
+  }
 }
diff --git a/src/byte_source.h b/src/byte_source.h
index 1c9fbd1f..d054e156 100644
--- a/src/byte_source.h
+++ b/src/byte_source.h
@@ -26,42 +26,29 @@
 #include <stddef.h>
 #include <stdint.h>
 
-typedef int (*SerdStreamCloseFunc)(void*);
-
-typedef struct {
-  SerdReadFunc        read_func;   ///< Read function (e.g. fread)
-  SerdStreamErrorFunc error_func;  ///< Error function (e.g. ferror)
-  SerdStreamCloseFunc close_func;  ///< Function for closing stream
-  void*               stream;      ///< Stream (e.g. FILE)
-  size_t              page_size;   ///< Number of bytes to read at a time
-  size_t              buf_size;    ///< Number of bytes in file_buf
-  SerdNode*           name;        ///< Name of stream (referenced by cur)
-  SerdCaret           caret;       ///< Caret for error reporting
-  uint8_t*            file_buf;    ///< Buffer iff reading pages from a file
-  const uint8_t*      read_buf;    ///< Pointer to file_buf or read_byte
-  size_t              read_head;   ///< Offset into read_buf
-  uint8_t             read_byte;   ///< 1-byte 'buffer' used when not paging
-  bool                from_stream; ///< True iff reading from `stream`
-  bool                prepared;    ///< True iff prepared for reading
-  bool                eof;         ///< True iff end of file reached
-} SerdByteSource;
-
-SerdStatus
-serd_byte_source_open_string(SerdByteSource* source,
-                             const char*     utf8,
-                             const SerdNode* name);
-
-SerdStatus
-serd_byte_source_open_source(SerdByteSource*     source,
-                             SerdReadFunc        read_func,
-                             SerdStreamErrorFunc error_func,
-                             SerdStreamCloseFunc close_func,
-                             void*               stream,
-                             const SerdNode*     name,
-                             size_t              page_size);
-
-SerdStatus
-serd_byte_source_close(SerdByteSource* source);
+typedef enum {
+  FROM_STRING,   ///< Reading from a user-provided buffer
+  FROM_FILENAME, ///< Reading from a file we opened
+  FROM_FUNCTION, ///< Reading from a user-provided function
+} SerdByteSourceType;
+
+struct SerdByteSourceImpl {
+  SerdReadFunc        read_func;  ///< Read function (e.g. fread)
+  SerdStreamErrorFunc error_func; ///< Error function (e.g. ferror)
+  SerdStreamCloseFunc close_func; ///< Function for closing stream
+  void*               stream;     ///< Stream (e.g. FILE)
+  size_t              page_size;  ///< Number of bytes to read at a time
+  size_t              buf_size;   ///< Number of bytes in file_buf
+  SerdNode*           name;       ///< Name of stream (referenced by cur)
+  SerdCaret           caret;      ///< File position for error reporting
+  uint8_t*            file_buf;   ///< Buffer iff reading pages from a file
+  const uint8_t*      read_buf;   ///< Pointer to file_buf or read_byte
+  size_t              read_head;  ///< Offset into read_buf
+  SerdByteSourceType  type;       ///< Type of input
+  uint8_t             read_byte;  ///< 1-byte 'buffer' used when not paging
+  bool                prepared;   ///< True iff prepared for reading
+  bool                eof;        ///< True iff end of file reached
+};
 
 SerdStatus
 serd_byte_source_prepare(SerdByteSource* source);
@@ -92,7 +79,7 @@ serd_byte_source_advance(SerdByteSource* source)
     ++source->caret.col;
   }
 
-  if (source->from_stream) {
+  if (source->type != FROM_STRING) {
     if (++source->read_head >= source->buf_size) {
       st = serd_byte_source_page(source);
     }
diff --git a/src/n3.c b/src/n3.c
index 46fe281b..a6aa71ba 100644
--- a/src/n3.c
+++ b/src/n3.c
@@ -1678,7 +1678,7 @@ skip_until(SerdReader* const reader, const uint8_t byte)
 SerdStatus
 read_turtleTrigDoc(SerdReader* const reader)
 {
-  while (!reader->source.eof) {
+  while (!reader->source->eof) {
     const size_t     orig_stack_size = reader->stack.size;
     const SerdStatus st              = read_n3_statement(reader);
     if (st > SERD_FAILURE) {
@@ -1699,7 +1699,7 @@ SerdStatus
 read_nquadsDoc(SerdReader* const reader)
 {
   SerdStatus st = SERD_SUCCESS;
-  while (!st && !reader->source.eof) {
+  while (!st && !reader->source->eof) {
     const size_t orig_stack_size = reader->stack.size;
 
     SerdStatementFlags flags   = 0;
diff --git a/src/reader.c b/src/reader.c
index ed6caafd..fe88ee14 100644
--- a/src/reader.c
+++ b/src/reader.c
@@ -37,7 +37,7 @@ r_err(SerdReader* const reader, const SerdStatus st, const char* const fmt, ...)
 {
   va_list args;
   va_start(args, fmt);
-  const SerdError e = {st, &reader->source.caret, fmt, &args};
+  const SerdError e = {st, &reader->source->caret, fmt, &args};
   serd_world_error(reader->world, &e);
   va_end(args);
   return st;
@@ -69,7 +69,8 @@ tolerate_status(const SerdReader* const reader, const SerdStatus status)
   }
 
   if (status == SERD_ERR_INTERNAL || status == SERD_ERR_OVERFLOW ||
-      status == SERD_ERR_BAD_WRITE || status == SERD_ERR_NO_DATA) {
+      status == SERD_ERR_BAD_WRITE || status == SERD_ERR_NO_DATA ||
+      status == SERD_ERR_BAD_CALL) {
     return false;
   }
 
@@ -145,7 +146,7 @@ emit_statement(SerdReader* const reader,
   serd_node_zero_pad(o);
 
   const SerdStatement statement = {{ctx.subject, ctx.predicate, o, ctx.graph},
-                                   &reader->source.caret};
+                                   &reader->source->caret};
 
   const SerdStatus st =
     serd_sink_write_statement(reader->sink, *ctx.flags, &statement);
@@ -163,7 +164,11 @@ read_statement(SerdReader* const reader)
 SerdStatus
 serd_reader_read_document(SerdReader* const reader)
 {
-  if (!reader->source.prepared) {
+  if (!reader->source) {
+    return SERD_ERR_BAD_CALL;
+  }
+
+  if (!reader->source->prepared) {
     SerdStatus st = serd_reader_prepare(reader);
     if (st) {
       return st;
@@ -241,12 +246,12 @@ serd_reader_add_blank_prefix(SerdReader* const reader, const char* const prefix)
 static SerdStatus
 skip_bom(SerdReader* const me)
 {
-  if (serd_byte_source_peek(&me->source) == 0xEF) {
-    serd_byte_source_advance(&me->source);
-    if (serd_byte_source_peek(&me->source) != 0xBB ||
-        serd_byte_source_advance(&me->source) ||
-        serd_byte_source_peek(&me->source) != 0xBF ||
-        serd_byte_source_advance(&me->source)) {
+  if (serd_byte_source_peek(me->source) == 0xEF) {
+    serd_byte_source_advance(me->source);
+    if (serd_byte_source_peek(me->source) != 0xBB ||
+        serd_byte_source_advance(me->source) ||
+        serd_byte_source_peek(me->source) != 0xBF ||
+        serd_byte_source_advance(me->source)) {
       r_err(me, SERD_ERR_BAD_SYNTAX, "corrupt byte order mark\n");
       return SERD_ERR_BAD_SYNTAX;
     }
@@ -256,60 +261,23 @@ skip_bom(SerdReader* const me)
 }
 
 SerdStatus
-serd_reader_start_stream(SerdReader* const         reader,
-                         const SerdReadFunc        read_func,
-                         const SerdStreamErrorFunc error_func,
-                         void* const               stream,
-                         const SerdNode* const     name,
-                         const size_t              page_size)
+serd_reader_start(SerdReader* const reader, SerdByteSource* const byte_source)
 {
-  return serd_byte_source_open_source(
-    &reader->source, read_func, error_func, NULL, stream, name, page_size);
-}
-
-SerdStatus
-serd_reader_start_file(SerdReader* reader, const char* uri, bool bulk)
-{
-  char* const path = serd_parse_file_uri(uri, NULL);
-  if (!path) {
-    return SERD_ERR_BAD_ARG;
-  }
-
-  FILE* fd = serd_world_fopen(reader->world, path, "rb");
-  free(path);
-  if (!fd) {
-    return SERD_ERR_UNKNOWN;
-  }
+  serd_reader_finish(reader);
 
-  SerdNode* const  name = serd_new_uri(SERD_STRING(uri));
-  const SerdStatus st   = serd_byte_source_open_source(
-    &reader->source,
-    bulk ? (SerdReadFunc)fread : serd_file_read_byte,
-    (SerdStreamErrorFunc)ferror,
-    (SerdStreamCloseFunc)fclose,
-    fd,
-    name,
-    bulk ? SERD_PAGE_SIZE : 1u);
-  serd_node_free(name);
-  return st;
-}
+  reader->source = byte_source;
 
-SerdStatus
-serd_reader_start_string(SerdReader* const     reader,
-                         const char* const     utf8,
-                         const SerdNode* const name)
-{
-  return serd_byte_source_open_string(&reader->source, utf8, name);
+  return reader->source ? SERD_SUCCESS : SERD_ERR_BAD_ARG;
 }
 
 static SerdStatus
 serd_reader_prepare(SerdReader* const reader)
 {
-  SerdStatus st = serd_byte_source_prepare(&reader->source);
+  SerdStatus st = serd_byte_source_prepare(reader->source);
   if (st == SERD_SUCCESS) {
     st = skip_bom(reader);
   } else if (st == SERD_FAILURE) {
-    reader->source.eof = true;
+    reader->source->eof = true;
   }
   return st;
 }
@@ -318,10 +286,14 @@ SerdStatus
 serd_reader_read_chunk(SerdReader* const reader)
 {
   SerdStatus st = SERD_SUCCESS;
-  if (!reader->source.prepared) {
+  if (!reader->source) {
+    return SERD_ERR_BAD_CALL;
+  }
+
+  if (!reader->source->prepared) {
     st = serd_reader_prepare(reader);
-  } else if (reader->source.eof) {
-    st = serd_byte_source_advance(&reader->source);
+  } else if (reader->source->eof) {
+    st = serd_byte_source_advance(reader->source);
   }
 
   return st ? st : read_statement(reader);
@@ -330,5 +302,6 @@ serd_reader_read_chunk(SerdReader* const reader)
 SerdStatus
 serd_reader_finish(SerdReader* const reader)
 {
-  return serd_byte_source_close(&reader->source);
+  reader->source = NULL;
+  return SERD_SUCCESS;
 }
diff --git a/src/reader.h b/src/reader.h
index a8c154dd..76f46506 100644
--- a/src/reader.h
+++ b/src/reader.h
@@ -50,7 +50,7 @@ struct SerdReaderImpl {
   SerdNode*       rdf_first;
   SerdNode*       rdf_rest;
   SerdNode*       rdf_nil;
-  SerdByteSource  source;
+  SerdByteSource* source;
   SerdStack       stack;
   SerdSyntax      syntax;
   unsigned        next_id;
@@ -107,7 +107,7 @@ read_turtleTrigDoc(SerdReader* reader);
 static inline int
 peek_byte(SerdReader* reader)
 {
-  SerdByteSource* source = &reader->source;
+  SerdByteSource* source = reader->source;
 
   return source->eof ? EOF : (int)source->read_buf[source->read_head];
 }
@@ -120,7 +120,7 @@ eat_byte_safe(SerdReader* reader, const int byte)
   const int c = peek_byte(reader);
   assert(c == byte);
 
-  serd_byte_source_advance(&reader->source);
+  serd_byte_source_advance(reader->source);
   return c;
 }
 
diff --git a/src/serdi.c b/src/serdi.c
index 2e04ae5a..3025b494 100644
--- a/src/serdi.c
+++ b/src/serdi.c
@@ -27,6 +27,7 @@
 #  include <io.h>
 #endif
 
+#include <errno.h>
 #include <limits.h>
 #include <stdbool.h>
 #include <stdio.h>
@@ -104,25 +105,36 @@ read_file(SerdWorld* const      world,
   syntax = syntax ? syntax : serd_guess_syntax(filename);
   syntax = syntax ? syntax : SERD_TRIG;
 
-  SerdStatus  st     = SERD_SUCCESS;
-  SerdReader* reader = serd_reader_new(world, syntax, flags, sink, stack_size);
-
-  serd_reader_add_blank_prefix(reader, add_prefix);
-
+  SerdByteSource* byte_source = NULL;
   if (!strcmp(filename, "-")) {
     SerdNode* name = serd_new_string(SERD_STRING("stdin"));
 
-    st = serd_reader_start_stream(
-      reader, serd_file_read_byte, (SerdStreamErrorFunc)ferror, stdin, name, 1);
+    byte_source = serd_byte_source_new_function(
+      serd_file_read_byte, (SerdStreamErrorFunc)ferror, NULL, stdin, name, 1);
 
     serd_node_free(name);
   } else {
-    st = serd_reader_start_file(reader, filename, bulk_read);
+    byte_source =
+      serd_byte_source_new_filename(filename, bulk_read ? SERD_PAGE_SIZE : 1u);
+  }
+
+  if (!byte_source) {
+    SERDI_ERRORF(
+      "failed to open input file `%s' (%s)\n", filename, strerror(errno));
+
+    return SERD_ERR_UNKNOWN;
   }
 
+  SerdReader* reader = serd_reader_new(world, syntax, flags, sink, stack_size);
+
+  serd_reader_add_blank_prefix(reader, add_prefix);
+
+  SerdStatus st = serd_reader_start(reader, byte_source);
+
   st = st ? st : serd_reader_read_document(reader);
 
   serd_reader_free(reader);
+  serd_byte_source_free(byte_source);
 
   return st;
 }
@@ -332,6 +344,9 @@ main(int argc, char** argv)
   SerdStatus st         = SERD_SUCCESS;
   SerdNode*  input_name = NULL;
   if (input_string) {
+    SerdByteSource* const byte_source =
+      serd_byte_source_new_string(input_string, NULL);
+
     SerdReader* const reader =
       serd_reader_new(world,
                       input_syntax ? input_syntax : SERD_TRIG,
@@ -341,13 +356,12 @@ main(int argc, char** argv)
 
     serd_reader_add_blank_prefix(reader, add_prefix);
 
-    SerdNode* name = serd_new_string(SERD_STRING("string"));
-    if (!(st = serd_reader_start_string(reader, input_string, name))) {
+    if (!(st = serd_reader_start(reader, byte_source))) {
       st = serd_reader_read_document(reader);
     }
 
-    serd_node_free(name);
     serd_reader_free(reader);
+    serd_byte_source_free(byte_source);
   }
 
   size_t prefix_len = 0;
diff --git a/src/string.c b/src/string.c
index 097e73ef..13fb9263 100644
--- a/src/string.c
+++ b/src/string.c
@@ -58,6 +58,8 @@ serd_strerror(const SerdStatus status)
     return "Error writing to file";
   case SERD_ERR_NO_DATA:
     return "Unexpected end of input";
+  case SERD_ERR_BAD_CALL:
+    return "Invalid call";
   }
 
   return "Unknown error";
diff --git a/src/world.c b/src/world.c
index e0ce9201..bd70d615 100644
--- a/src/world.c
+++ b/src/world.c
@@ -19,14 +19,7 @@
 #include "caret.h"
 #include "namespaces.h"
 #include "node.h"
-#include "serd_config.h"
-#include "system.h"
 
-#if defined(USE_POSIX_FADVISE)
-#  include <fcntl.h>
-#endif
-
-#include <errno.h>
 #include <stdarg.h>
 #include <stdio.h>
 #include <stdlib.h>
@@ -34,26 +27,6 @@
 
 #define BLANK_CHARS 12
 
-FILE*
-serd_world_fopen(SerdWorld* world, const char* path, const char* mode)
-{
-  FILE* fd = fopen(path, mode);
-  if (!fd) {
-    char message[1024] = {0};
-    serd_system_strerror(errno, message, sizeof(message));
-
-    serd_world_errorf(
-      world, SERD_ERR_INTERNAL, "failed to open file %s (%s)\n", path, message);
-    return NULL;
-  }
-
-#if USE_POSIX_FADVISE && USE_FILENO
-  posix_fadvise(fileno(fd), 0, 0, POSIX_FADV_SEQUENTIAL);
-#endif
-
-  return fd;
-}
-
 SerdStatus
 serd_world_error(const SerdWorld* const world, const SerdError* const e)
 {
diff --git a/src/world.h b/src/world.h
index 44e21166..a70a6e28 100644
--- a/src/world.h
+++ b/src/world.h
@@ -20,7 +20,6 @@
 #include "serd/serd.h"
 
 #include <stdint.h>
-#include <stdio.h>
 
 struct SerdWorldImpl {
   SerdNodes*      nodes;
@@ -37,10 +36,6 @@ struct SerdWorldImpl {
   uint32_t        next_blank_id;
 };
 
-/// Open a file configured for fast sequential reading
-FILE*
-serd_world_fopen(SerdWorld* world, const char* path, const char* mode);
-
 SerdStatus
 serd_world_error(const SerdWorld* world, const SerdError* e);