stdio: ungetc(), update (+) and append (a) modes (DONE)

CMakeLists.txt

@@ -154,6 +154,7 @@ set(SOURCES
   src/libc/stdio/setvbuf.c
   src/libc/stdio/snprintf.c
   src/libc/stdio/sprintf.c
+  src/libc/stdio/ungetc.c
   src/libc/stdio/vasprintf.c
   src/libc/stdio/vdprintf.c
   src/libc/stdio/vfprintf.c

STATUS

@@ -85,6 +85,7 @@ TEST:       Function/symbol/macro needs to be tested
 
 7.19 <stdio.h>
      7.19.1    Introduction                  TEST (no wide-oriented streams *)
+     7.19.1    stdin, stdout, stderr         TODO
 
      7.19.4.1  remove                        TEST
      7.19.4.2  rename                        TODO
@@ -93,22 +94,22 @@ TEST:       Function/symbol/macro needs to be tested
 
      7.19.5.1  fclose                        -
      7.19.5.2  fflush                        -
-     7.19.5.3  fopen                         TEST ("a" and "+" modes remain)
+     7.19.5.3  fopen                         -
      (EXT)     fdopen                        TEST
-     7.19.5.4  freopen                       TEST
+     7.19.5.4  freopen                       TEST (mode changes untested)
      7.19.5.5  setbuf                        -
      7.19.5.6  setvbuf                       -
 
-     7.19.6.1  fprintf                       LDEPS(fwrite)
+     7.19.6.1  fprintf                       TEST
      7.19.6.2  fscanf                        TODO
-     7.19.6.3  printf                        LDEPS(fwrite, stdout)
+     7.19.6.3  printf                        LDEPS(stdout)
      7.19.6.4  scanf                         TODO
      7.19.6.5  snprintf                      -
      7.19.6.6  sprintf                       -
      7.19.6.7  sscanf                        TODO
-     7.19.6.8  vfprintf                      LDEPS(fwrite)
+     7.19.6.8  vfprintf                      TEST
      7.19.6.9  vfscanf                       TODO
-     7.19.6.10 vprintf                       LDEPS(fwrite, stdout)
+     7.19.6.10 vprintf                       LDEPS(stdout)
      7.19.6.11 vscanf                        TODO
      7.19.6.12 vsnprintf                     -
      7.19.6.13 vsprintf                      -
@@ -128,10 +129,10 @@ TEST:       Function/symbol/macro needs to be tested
      7.19.7.8  putc                          LDPES(fputc)
      7.19.7.9  putchar                       LDEPS(fputc)
      7.19.7.10 puts                          LDEPS(fputs)
-     7.19.7.11 ungetc                        TODO
+     7.19.7.11 ungetc                        -
 
-     7.19.8.1  fread                         TEST ("a" and "+" modes remain)
-     7.19.8.2  fwrite                        TEST ("a" and "+" modes remain)
+     7.19.8.1  fread                         -
+     7.19.8.2  fwrite                        -
 
      7.19.9.1  fgetpos                       -
      7.19.9.2  fseek                         -
@@ -200,7 +201,7 @@ TEST:       Function/symbol/macro needs to be tested
      (EXT)     strncasecmp                   -
      (EXT)     strdup                        -
      (EXT)     strndup                       -
-     (EXT)     memrchr                       -
+     (EXT)     memrchr                       - (Unoptimized: byte-by-byte)
 
 7.22 <tgmath.h> => GCC
 

include/stdio.h

@@ -259,6 +259,9 @@ extern int feof(FILE *__fp);
 /* Test the error flag. */
 extern int ferror(FILE *__fp);
 
+/* Print a message followed by strerror(errno) to stdout. */
+extern void perror(char const *__s);
+
 #ifdef __cplusplus
 }
 #endif

include/target/gint/bits/types/FILE.h

@@ -41,6 +41,12 @@
   but not yet written to the file descriptor. ftell() reports fdpos + bufpos.
   The rest of the buffer is undefined.
 
+  The ungetc() function pushes back characters into the buffer; if the FILE is
+  unbuffered, then it's made buffered temporarily to hold the characters and
+  cleared at the next fflush() or read. The buffer is put in reading mode. For
+  this reason, reading functions should test [fp->buf] to check whether there
+  is a buffer instead of [fp->bufmode != _IONBF].
+
   Many fields in the FILE structure are abstracted away by API calls in layers:
 
   1. [fd], [fdpos] and [error] are updated by the primitive functions of
@@ -67,6 +73,8 @@ typedef struct {
 	size_t bufpos;
 	size_t bufread;
 	size_t bufsize;
+	/* Number of ungetc()'d characters at the start of buffer data */
+	int bufungetc;
 	/* Buffering mode; one of _IOFBF, _IOLBF, or _IONBF */
 	uint8_t bufmode   :2;
 	/* We own the buffer and it needs to be freed */

src/libc/stdio/fflush.c

@@ -5,30 +5,38 @@
 
 int fflush(FILE *fp)
 {
-	// TODO: fflush(NULL) should flush "all" files (do we track them?)
+	/* TODO: fflush(NULL) should flush "all" files (do we track them?) */
 	if(!fp) {
 		errno = EINVAL;
 		return EOF;
 	}
+
 	if(!fp->buf)
 		return 0;
 
+	int rc = 0;
+
 	/* In reading mode, reset the file offset */
-	if(fp->bufdir == __FILE_BUF_READ && fp->bufpos < fp->bufread) {
+	if(__fp_hasbuf_read(fp)) {
 		fp->fdpos = fp->fdpos - fp->bufread + fp->bufpos;
 		lseek(fp->fd, fp->fdpos, SEEK_SET);
-		fp->bufpos = 0;
-		fp->bufread = 0;
-		return 0;
 	}
 
 	/* In writing mode, write pending data */
-	if(fp->bufdir == __FILE_BUF_WRITE && fp->bufpos > 0) {
+	else if(__fp_hasbuf_write(fp)) {
 		ssize_t written = __fp_write(fp, fp->buf, fp->bufpos);
-		int rc = (written == (ssize_t)fp->bufpos ? 0 : EOF);
-		fp->bufpos = 0;
-		return rc;
+		rc = (written == (ssize_t)fp->bufpos ? 0 : EOF);
+
+		/* TODO: fflush(): Keep data that couldn't be written */
 	}
 
-	return 0;
+	fp->bufpos = 0;
+	fp->bufread = 0;
+	fp->bufungetc = 0;
+
+	/* Clear buffering for unbuffered streams that used ungetc() */
+	if(fp->bufmode == _IONBF)
+		__fp_remove_buffer(fp);
+
+	return rc;
 }

src/libc/stdio/fileutil.c

@@ -1,4 +1,5 @@
 #include "fileutil.h"
+#include <gint/defs/util.h>
 #include <unistd.h>
 #include <stdlib.h>
 #include <string.h>
@@ -34,20 +35,74 @@ void __fp_close(FILE *fp, bool free_fp)
 		free(fp);
 }
 
+void __fp_remove_buffer(FILE *fp)
+{
+	if(fp->bufowned)
+		free(fp->buf);
+
+	fp->buf = NULL;
+	fp->bufowned = false;
+	fp->bufsize = 0;
+}
+
+bool __fp_set_buffer(FILE *fp, void *buf, size_t size)
+{
+	bool owned = false;
+
+	if(!buf) {
+		owned = true;
+		buf = malloc(size);
+		if(!buf)
+			return false;
+	}
+
+	fp->buf = buf;
+	fp->bufowned = owned;
+	fp->bufsize = size;
+	return true;
+}
+
 void __fp_buffer_mode_read(FILE *fp)
 {
-	if(fp->bufmode != _IONBF && fp->bufdir == __FILE_BUF_WRITE &&
-	   fp->bufpos > 0)
+	if(__fp_hasbuf_write(fp)) {
 		fflush(fp);
-	fp->bufdir = __FILE_BUF_READ;
+	}
+	if(fp->buf)
+		fp->bufdir = __FILE_BUF_READ;
 }
 
 void __fp_buffer_mode_write(FILE *fp)
 {
-	if(fp->bufmode != _IONBF && fp->bufdir == __FILE_BUF_READ &&
-	   fp->bufpos < fp->bufread)
+	if(__fp_hasbuf_read(fp))
 		fflush(fp);
-	fp->bufdir = __FILE_BUF_WRITE;
+	if(fp->buf)
+		fp->bufdir = __FILE_BUF_WRITE;
+}
+
+ssize_t __fp_buffered_read(FILE *fp, void *data, size_t request_size)
+{
+	if(!fp->buf || __fp_hasbuf_write(fp))
+		return -1;
+
+	int read_size = min((int)request_size, fp->bufread - fp->bufpos);
+	if(read_size <= 0)
+		return 0;
+
+	memcpy(data, fp->buf + fp->bufpos, read_size);
+	fp->bufpos += read_size;
+	fp->bufungetc = max(fp->bufungetc - read_size, 0);
+
+	/* Rewind the buffer if we read it fully */
+	if(fp->bufpos >= fp->bufread) {
+		fp->bufread = 0;
+		fp->bufpos = 0;
+
+		/* Clear temporary ungetc() buffers of _IONBF streams */
+		if(fp->bufmode == _IONBF)
+			__fp_remove_buffer(fp);
+	}
+
+	return read_size;
 }
 
 ssize_t __fp_read(FILE *fp, void *data, size_t size)
@@ -77,12 +132,6 @@ ssize_t __fp_write(FILE *fp, void const *data, size_t size)
 {
 	size_t written = 0;
 
-	if(fp->append) {
-		int rc = fseek(fp, 0, SEEK_END);
-		if(rc < 0)
-			return EOF;
-	}
-
 	while(written < size) {
 		ssize_t rc = write(fp->fd, data + written, size - written);
 

src/libc/stdio/fileutil.h

@@ -9,18 +9,38 @@ extern "C" {
 #include <stdbool.h>
 #include <fcntl.h>
 
+/* Check whether a buffer has written data to flush. */
+#define __fp_hasbuf_write(fp) \
+   (fp->buf && fp->bufdir == __FILE_BUF_WRITE && fp->bufpos > 0)
+/* Check whether a buffer has read data to yield. */
+#define __fp_hasbuf_read(fp) \
+   (fp->buf && fp->bufdir == __FILE_BUF_READ && fp->bufpos < fp->bufread)
+
 /* Open a file descriptor in a pre-allocated FILE. */
 int __fp_open(FILE *fp, int fd, bool use_buffering);
 
 /* Close fp and free all of its resources. */
 void __fp_close(FILE *fp, bool free_fp);
 
+/* Remove the buffer on fp. */
+void __fp_remove_buffer(FILE *fp);
+
+/* Set the specified buffer on fp (can be NULL), in which case malloc().
+   Returns true on success, false on error. */
+bool __fp_set_buffer(FILE *fp, void *buf, size_t size);
+
 /* Set reading mode on the buffer. */
 void __fp_buffer_mode_read(FILE *fp);
 
 /* Set writing mode on the buffer. */
 void __fp_buffer_mode_write(FILE *fp);
 
+/* Reads data from the buffer. Always reads as much as possible, up to size.
+   Returns amount of data read; if >= 0 but < size, the buffer should be
+   refilled. Returns -1 to indicate that unbuffered access should be used.
+   Allows reading from temporary ungetc() buffers and cleans them. */
+ssize_t __fp_buffered_read(FILE *fp, void *data, size_t size);
+
 /* Reads data from a file descriptor; updates the fdpos and sets the error
    indicator. Returns 0 on success, EOF on error. */
 ssize_t __fp_read(FILE *fp, void *data, size_t size);

src/libc/stdio/fread.c

@@ -9,43 +9,44 @@ size_t fread(void *data, size_t membsize, size_t nmemb, FILE *fp)
 		return 0;
 	}
 
-	size_t size;
-	if(__builtin_umul_overflow(membsize, nmemb, &size)) {
+	size_t request_size;
+	if(__builtin_umul_overflow(membsize, nmemb, &request_size)) {
 		fp->error = 1;
 		return 0;
 	}
 
-	if(!fp->buf) {
-		ssize_t rc = __fp_read(fp, data, size);
-		return (rc == EOF) ? 0 : rc;
-	}
+	/* If the stream if unbuffered, we might have no buffer for the reads.
+	   If it's buffered, we always have one. It's also possible that fp is
+	   unbuffered (_IONBF) but has a buffer temporarily because ungetc()
+	   has been used, in which case we have to transition from buffered
+	   reads into direct reads midway. We use __fp_buffered_read() to
+	   handle this. */
 
+	size_t read_size = 0;
 	__fp_buffer_mode_read(fp);
 
-	size_t size_read = 0;
-	while(size_read < size) {
-		/* Take what's available in the buffer (may be 0, but only
-		   during the first iteration) */
-		size_t size_frag = fp->bufread - fp->bufpos;
-		if(size_frag > size - size_read)
-			size_frag = size - size_read;
-		memcpy(data + size_read, fp->buf + fp->bufpos, size_frag);
-		size_read += size_frag;
-		fp->bufpos += size_frag;
+	while(read_size < request_size) {
+		int remaining = request_size - read_size;
+		int chunk = __fp_buffered_read(fp, data+read_size, remaining);
 
-		if(fp->bufpos >= fp->bufread) {
-			fp->bufread = 0;
-			fp->bufpos = 0;
+		/* Stream is not/no longer buffered, finish unbuffered */
+		if(chunk < 0) {
+			ssize_t rc = __fp_read(fp, data+read_size, remaining);
+			return read_size + (rc == EOF ? 0 : rc);
 		}
-		if(size_read >= size)
+
+		read_size += chunk;
+		if(read_size >= request_size)
 			break;
 
-		/* Get more data */
-		ssize_t rc = __fp_read(fp, fp->buf, fp->bufsize);
-		if(rc <= 0) /* EOF or error */
-			break;
-		fp->bufread = rc;
+		/* Get more data from the file descriptor into the buffer */
+		if(fp->buf) {
+			ssize_t rc = __fp_read(fp, fp->buf, fp->bufsize);
+			if(rc <= 0) /* EOF or error */
+				break;
+			fp->bufread = rc;
+		}
 	}
 
-	return size_read;
+	return read_size;
 }

src/libc/stdio/fwrite.c

@@ -15,6 +15,11 @@ size_t fwrite(void const *data, size_t membsize, size_t nmemb, FILE *fp)
 		return 0;
 	}
 
+	if(fp->append && fseek(fp, 0, SEEK_END) == EOF) {
+		fp->error = 1;
+		return 0;
+	}
+
 	if(!fp->buf) {
 		ssize_t rc = __fp_write(fp, data, size);
 		return (rc == EOF) ? 0 : rc;

src/libc/stdio/setvbuf.c

@@ -1,36 +1,23 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <stdbool.h>
+#include "fileutil.h"
 
 int setvbuf(FILE * restrict fp, char * restrict buf, int mode, size_t size)
 {
-	if(fp->bufmode != _IONBF) {
+	if(fp->bufmode != _IONBF || fp->buf) {
 		fflush(fp);
-		if(fp->bufowned)
-			free(fp->buf);
 	}
 
-	fp->buf = NULL;
-	fp->bufowned = false;
+	__fp_remove_buffer(fp);
 	fp->bufmode = _IONBF;
-	fp->bufsize = 0;
 
 	if(mode == _IONBF)
 		return 0;
 
-	if(buf) {
-		fp->buf = buf;
-		fp->bufsize = size;
-		fp->bufmode = mode;
-	}
-	else {
-		fp->buf = malloc(size);
-		if(!fp->buf)
-			return -1;
-		fp->bufsize = size;
-		fp->bufowned = true;
-		fp->bufmode = mode;
-	}
+	if(!__fp_set_buffer(fp, buf, size))
+		return -1;
 
+	fp->bufmode = mode;
 	return 0;
 }

src/libc/stdio/ungetc.c

@@ -0,0 +1,53 @@
+#include <stdio.h>
+#include <errno.h>
+#include <string.h>
+#include "fileutil.h"
+
+int ungetc(int c, FILE *fp)
+{
+	if(c == EOF)
+		return c;
+	if(!fp->readable) {
+		errno = EINVAL;
+		return EOF;
+	}
+
+	/* If there is no buffer, create a temporary one */
+	if(!fp->buf)
+		__fp_set_buffer(fp, NULL, 16);
+
+	__fp_buffer_mode_read(fp);
+
+	/* If there is a spot available in the buffer, use it */
+	if(fp->bufpos > 0) {
+		fp->buf[--fp->bufpos] = c;
+		fp->bufungetc++;
+		fp->eof = 0;
+		return c;
+	}
+
+	/* Otherwise, make some room. If the buffer is full of ungetc()
+	   characters then we can't preserve the semantics, so we fail. */
+	if(fp->bufungetc >= (int)fp->bufsize)
+		return EOF;
+
+	/* If the buffer is full, discard read data that isn't ungetc()'d. */
+	if(fp->bufread >= fp->bufsize) {
+		int discarded = fp->bufread - fp->bufungetc;
+		if(fseek(fp, -discarded, SEEK_CUR))
+			return EOF;
+		fp->bufread -= discarded;
+	}
+
+	/* Then move whichever type of read data remains to the very end of the
+	   buffer, so we have as much space as possible for more ungetc(). */
+	memmove(fp->buf + fp->bufsize - fp->bufread, fp->buf, fp->bufread);
+	fp->bufpos = fp->bufsize - fp->bufread;
+	fp->bufread = fp->bufsize;
+
+	/* Because we've made space, conclude like the first case */
+	fp->buf[--fp->bufpos] = c;
+	fp->bufungetc++;
+	fp->eof = 0;
+	return c;
+}