fix: bom detection on source files moved into cjit core

cleaned up a bit more main.c by moving bom detection which is only
needed inside cjit_add_source

eventually this needs a bit more cleanup to normalize the way files
are loaded

src/cjit.c

@@ -28,6 +28,9 @@
 #include <stdbool.h>
 #include <errno.h>
 #include <unistd.h> // getpid/write
+#include <fcntl.h> // open(2)
+#include <inttypes.h>
+#include <sys/stat.h> // fstat(2)
 
 #define MAX_PATH 260 // rather short paths
 #define MAX_STRING 20480 // max 20KiB strings
@@ -251,9 +254,54 @@ static int has_source_extension(const char *path) {
 	return (is_source? 1 : -1);
 }
 
+static int detect_bom(const char *filename,size_t *filesize) {
+	uint8_t bom[3];
+	int res;
+	int fd = open(filename, O_RDONLY | O_BINARY);
+	if(fd<0) {
+		_err("%s: error opening file: %s",__func__,filename);
+		_err("%s",strerror(errno));
+		return -1;
+	}
+	struct stat st;
+	if (fstat(fd, &st) == -1) {
+		_err("%s: error analyzing file: %s",__func__,filename);
+		_err("%s",strerror(errno));
+		close(fd);
+		return -1;
+	}
+	*filesize = st.st_size;
+	res = read(fd,bom,3);
+	close(fd);
+	if (res!=3) {
+		_err("%s: error reading file: %s",__func__,filename);
+		_err("%s",strerror(errno));
+		return -1;
+	}
+	// _err("%s bom: %x %x %x",filename,bom[0],bom[1],bom[2]);
+	if (bom[0] == 0xFF && bom[1] == 0xFE) {
+		return 1; // UTF-16 LE
+	} else if (bom[0] == 0xFE && bom[1] == 0xFF) {
+		return 2; // UTF-16 BE
+	} else if (bom[0] == 0xEF && bom[1] == 0xBB && bom[2] == 0xBF) {
+		return 3; // UTF-8
+	} else {
+		return 0; // No BOM
+	}
+}
+
 static bool cjit_add_source(CJITState *cjit, const char *path) {
-	long length = file_size(path);
-	if (length == -1) return false;
+	size_t length;
+	int res = detect_bom(path,&length);
+	if(res<0) {
+		_err("Cannot open file: %s",path);
+		_err("Execution aborted.");
+		return false;
+	} else if(res>0) {
+		_err("UTF BOM detected in file: %s",path);
+		_err("Encoding is not yet supported, execution aborted.");
+		return false;
+	}
 	FILE *file = fopen(path, "rb");
 	if (!file) {
 		_err("%s: fopen error: ", __func__, strerror(errno));

src/cjit.h

@@ -61,8 +61,6 @@ extern bool extract_assets(CJITState *CJIT);
 
 /////////////
 // from file.c
-extern int   detect_bom(const char *filename);
-extern long  file_size(const char *filename);
 extern char* file_load(const char *filename, unsigned int *len);
 extern char *load_stdin();
 extern char* dir_load(const char *path);

src/file.c

@@ -31,82 +31,8 @@
 
 #include <ftw.h> // _GNU_SOURCE
 
-int detect_bom(const char *filename) {
-	uint8_t bom[3];
-	int res;
-	int fd = open(filename, O_RDONLY | O_BINARY);
-	res = read(fd,bom,3);
-	if (res!=3) {
-		_err("read error: %s",strerror(errno));
-		return -1;
-	}
-	close(fd);
-	// _err("%s bom: %x %x %x",filename,bom[0],bom[1],bom[2]);
-	if (bom[0] == 0xFF && bom[1] == 0xFE) {
-		return 1; // UTF-16 LE
-	} else if (bom[0] == 0xFE && bom[1] == 0xFF) {
-		return 2; // UTF-16 BE
-	} else if (bom[0] == 0xEF && bom[1] == 0xBB && bom[2] == 0xBF) {
-		return 3; // UTF-8
-	} else {
-		return 0; // No BOM
-	}
-}
-
-bool append_path(char **stored_path, const char *new_path) {
-  // TODO: sanitize input checking only path chars are there
-  // support both / and \ for windows
-  if (*stored_path == NULL) {
-    // If stored_path is NULL, allocate memory and copy new_path
-    *stored_path = malloc(strlen(new_path) + 1);
-    if (*stored_path == NULL) {
-      _err("Memory allocation failed");
-      return(false);
-    }
-    strcpy(*stored_path, new_path);
-  } else {
-    // If stored_path is not NULL, append new_path separated by ':'
-    size_t new_length = strlen(*stored_path) + strlen(new_path) + 2;
-    char *temp = realloc(*stored_path, new_length);
-    if (temp == NULL) {
-      _err("Memory allocation failed");
-      return(false);
-    }
-    *stored_path = temp;
-    strcat(*stored_path, ":");
-    strcat(*stored_path, new_path);
-  }
-  return(true);
-}
-
-bool prepend_path(char **stored_path, const char *new_path) {
-  if (*stored_path == NULL) {
-    // If stored_path is NULL, allocate memory and copy new_path
-    *stored_path = malloc(strlen(new_path) + 1);
-    if (*stored_path == NULL) {
-      _err("Memory allocation failed");
-      return(false);
-    }
-    strcpy(*stored_path, new_path);
-  } else {
-    // If stored_path is not NULL, prepend new_path separated by ':'
-    size_t new_length = strlen(*stored_path) + strlen(new_path) + 2;
-    char *temp = malloc(new_length);
-    if (temp == NULL) {
-      _err("Memory allocation failed");
-      return(false);
-    }
-    strcpy(temp, new_path);
-    strcat(temp, ":");
-    strcat(temp, *stored_path);
-    free(*stored_path);
-    *stored_path = temp;
-  }
-  return(true);
-}
-
 // Function to get the length of a file in bytes
-long file_size(const char *filename) {
+static long file_size(const char *filename) {
     FILE *file = fopen(filename, "rb");
     if (file == NULL) {
         _err("%s: fopen error: %s",__func__,strerror(errno));

src/main.c

@@ -269,24 +269,7 @@ int main(int argc, char **argv) {
 				  goto endgame;
 			  } else free(stdin_code);
 		  } else { // load any file path
-			  int res = detect_bom(code_path);
-			  // returned values:
-			  // 0  : no BOM, all OK
-			  // <0 : file not found
-			  // 1  : BOM found, UTF16-LE
-			  // 2  : BOM found, UTF16-BE
-			  // 3  : BOM found, UTF8
-			  if(res ==0) {
-				  cjit_add_file(CJIT, code_path);
-			  } else if(res<0) {
-				  _err("Cannot open file: %s",code_path);
-				  _err("Execution aborted.");
-				  goto endgame;
-			  } else {
-				  _err("UTF BOM detected in file: %s",code_path);
-				  _err("Encoding is not yet supported, execution aborted.");
-				  goto endgame;
-			  }
+			  cjit_add_file(CJIT, code_path);
 		  }
 	  }
   }