Consolidate text buffer functions

There are many scattered functions that look into the contents of
buffers to do various text manipulations (such as escaping or
unescaping data, calculating text stats, guessing if content is
binary, etc).  This groups all those functions together into a
new file and converts the code to use that.

This has two enhancements to existing functionality.  The old
text stats function is significantly rewritten and the BOM
detection code was extended (although largely we can't deal with
anything other than a UTF8 BOM).

15 files changed, 388 insertions, 289 deletions

diff --git a/src/buf_text.c b/src/buf_text.c
new file mode 100644
index 000000000..3c5024e6c
--- /dev/null
+++ b/src/buf_text.c
@@ -0,0 +1,208 @@
+/*
+ * Copyright (C) 2009-2012 the libgit2 contributors
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#include "buf_text.h"
+
+int git_buf_text_puts_escaped(
+	git_buf *buf,
+	const char *string,
+	const char *esc_chars,
+	const char *esc_with)
+{
+	const char *scan;
+	size_t total = 0, esc_len = strlen(esc_with), count;
+
+	if (!string)
+		return 0;
+
+	for (scan = string; *scan; ) {
+		/* count run of non-escaped characters */
+		count = strcspn(scan, esc_chars);
+		total += count;
+		scan += count;
+		/* count run of escaped characters */
+		count = strspn(scan, esc_chars);
+		total += count * (esc_len + 1);
+		scan += count;
+	}
+
+	if (git_buf_grow(buf, buf->size + total + 1) < 0)
+		return -1;
+
+	for (scan = string; *scan; ) {
+		count = strcspn(scan, esc_chars);
+
+		memmove(buf->ptr + buf->size, scan, count);
+		scan += count;
+		buf->size += count;
+
+		for (count = strspn(scan, esc_chars); count > 0; --count) {
+			/* copy escape sequence */
+			memmove(buf->ptr + buf->size, esc_with, esc_len);
+			buf->size += esc_len;
+			/* copy character to be escaped */
+			buf->ptr[buf->size] = *scan;
+			buf->size++;
+			scan++;
+		}
+	}
+
+	buf->ptr[buf->size] = '\0';
+
+	return 0;
+}
+
+void git_buf_text_unescape(git_buf *buf)
+{
+	buf->size = git__unescape(buf->ptr);
+}
+
+int git_buf_text_common_prefix(git_buf *buf, const git_strarray *strings)
+{
+	size_t i;
+	const char *str, *pfx;
+
+	git_buf_clear(buf);
+
+	if (!strings || !strings->count)
+		return 0;
+
+	/* initialize common prefix to first string */
+	if (git_buf_sets(buf, strings->strings[0]) < 0)
+		return -1;
+
+	/* go through the rest of the strings, truncating to shared prefix */
+	for (i = 1; i < strings->count; ++i) {
+
+		for (str = strings->strings[i], pfx = buf->ptr;
+			 *str && *str == *pfx; str++, pfx++)
+			/* scanning */;
+
+		git_buf_truncate(buf, pfx - buf->ptr);
+
+		if (!buf->size)
+			break;
+	}
+
+	return 0;
+}
+
+bool git_buf_text_is_binary(const git_buf *buf)
+{
+	const char *scan = buf->ptr, *end = buf->ptr + buf->size;
+	int printable = 0, nonprintable = 0;
+
+	while (scan < end) {
+		unsigned char c = *scan++;
+
+		if (c > 0x1F && c < 0x7F)
+			printable++;
+		else if (c == '\0')
+			return true;
+		else if (!git__isspace(c))
+			nonprintable++;
+	}
+
+	return ((printable >> 7) < nonprintable);
+}
+
+int git_buf_text_detect_bom(git_bom_t *bom, const git_buf *buf, size_t offset)
+{
+	const char *ptr;
+	size_t len;
+
+	/* need at least 2 bytes after offset to look for any BOM */
+	if (buf->size < offset + 2)
+		return 0;
+
+	ptr = buf->ptr + offset;
+	len = buf->size - offset;
+
+	switch (*ptr++) {
+	case 0:
+		if (len >= 4 && ptr[0] == 0 && ptr[1] == '\xFE' && ptr[2] == '\xFF') {
+			*bom = GIT_BOM_UTF32_BE;
+			return 4;
+		}
+		break;
+	case '\xEF':
+		if (len >= 3 && ptr[0] == '\xBB' && ptr[1] == '\xBF') {
+			*bom = GIT_BOM_UTF8;
+			return 3;
+		}
+		break;
+	case '\xFE':
+		if (*ptr == '\xFF') {
+			*bom = GIT_BOM_UTF16_BE;
+			return 2;
+		}
+		break;
+	case '\xFF':
+		if (*ptr != '\xFE')
+			break;
+		if (len >= 4 && ptr[1] == 0 && ptr[2] == 0) {
+			*bom = GIT_BOM_UTF32_LE;
+			return 4;
+		} else {
+			*bom = GIT_BOM_UTF16_LE;
+			return 2;
+		}
+		break;
+	default:
+		break;
+	}
+
+	return 0;
+}
+
+bool git_buf_text_gather_stats(
+	git_buf_text_stats *stats, const git_buf *buf, bool skip_bom)
+{
+	const char *scan = buf->ptr, *end = buf->ptr + buf->size;
+	int skip;
+
+	memset(stats, 0, sizeof(*stats));
+
+	/* BOM detection */
+	skip = git_buf_text_detect_bom(&stats->bom, buf, 0);
+	if (skip_bom)
+		scan += skip;
+
+	/* Ignore EOF character */
+	if (buf->size > 0 && end[-1] == '\032')
+		end--;
+
+	/* Counting loop */
+	while (scan < end) {
+		unsigned char c = *scan++;
+
+		if ((c > 0x1F && c < 0x7F) || c > 0x9f)
+			stats->printable++;
+		else switch (c) {
+			case '\0':
+				stats->nul++;
+				stats->nonprintable++;
+				break;
+			case '\n':
+				stats->lf++;
+				break;
+			case '\r':
+				stats->cr++;
+				if (scan < end && *scan == '\n')
+					stats->crlf++;
+				break;
+			case '\t': case '\f': case '\v': case '\b': case 0x1b: /*ESC*/
+				stats->printable++;
+				break;
+			default:
+				stats->nonprintable++;
+				break;
+			}
+	}
+
+	return (stats->nul > 0 ||
+		((stats->printable >> 7) < stats->nonprintable));
+}
diff --git a/src/buf_text.h b/src/buf_text.h
new file mode 100644
index 000000000..c22499bfe
--- /dev/null
+++ b/src/buf_text.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright (C) 2009-2012 the libgit2 contributors
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#ifndef INCLUDE_buf_text_h__
+#define INCLUDE_buf_text_h__
+
+#include "buffer.h"
+
+typedef enum {
+	GIT_BOM_NONE = 0,
+	GIT_BOM_UTF8 = 1,
+	GIT_BOM_UTF16_LE = 2,
+	GIT_BOM_UTF16_BE = 3,
+	GIT_BOM_UTF32_LE = 4,
+	GIT_BOM_UTF32_BE = 5
+} git_bom_t;
+
+typedef struct {
+	git_bom_t bom; /* BOM found at head of text */
+	unsigned int nul, cr, lf, crlf; /* NUL, CR, LF and CRLF counts */
+	unsigned int printable, nonprintable; /* These are just approximations! */
+} git_buf_text_stats;
+
+/**
+ * Append string to buffer, prefixing each character from `esc_chars` with
+ * `esc_with` string.
+ *
+ * @param buf Buffer to append data to
+ * @param string String to escape and append
+ * @param esc_chars Characters to be escaped
+ * @param esc_with String to insert in from of each found character
+ * @return 0 on success, <0 on failure (probably allocation problem)
+ */
+extern int git_buf_text_puts_escaped(
+	git_buf *buf,
+	const char *string,
+	const char *esc_chars,
+	const char *esc_with);
+
+/**
+ * Append string escaping characters that are regex special
+ */
+GIT_INLINE(int) git_buf_text_puts_escape_regex(git_buf *buf, const char *string)
+{
+	return git_buf_text_puts_escaped(buf, string, "^.[]$()|*+?{}\\", "\\");
+}
+
+/**
+ * Unescape all characters in a buffer in place
+ *
+ * I.e. remove backslashes
+ */
+extern void git_buf_text_unescape(git_buf *buf);
+
+/**
+ * Fill buffer with the common prefix of a array of strings
+ *
+ * Buffer will be set to empty if there is no common prefix
+ */
+extern int git_buf_text_common_prefix(git_buf *buf, const git_strarray *strs);
+
+/**
+ * Check quickly if buffer looks like it contains binary data
+ *
+ * @param buf Buffer to check
+ * @return true if buffer looks like non-text data
+ */
+extern bool git_buf_text_is_binary(const git_buf *buf);
+
+/**
+ * Check if a buffer begins with a UTF BOM
+ *
+ * @param bom Set to the type of BOM detected or GIT_BOM_NONE
+ * @param buf Buffer in which to check the first bytes for a BOM
+ * @param offset Offset into buffer to look for BOM
+ * @return Number of bytes of BOM data (or 0 if no BOM found)
+ */
+extern int git_buf_text_detect_bom(
+	git_bom_t *bom, const git_buf *buf, size_t offset);
+
+/**
+ * Gather stats for a piece of text
+ *
+ * Fill the `stats` structure with counts of unreadable characters, carriage
+ * returns, etc, so it can be used in heuristics.  This automatically skips
+ * a trailing EOF (\032 character).  Also it will look for a BOM at the
+ * start of the text and can be told to skip that as well.
+ *
+ * @param stats Structure to be filled in
+ * @param buf Text to process
+ * @param skip_bom Exclude leading BOM from stats if true
+ * @return Does the buffer heuristically look like binary data
+ */
+extern bool git_buf_text_gather_stats(
+	git_buf_text_stats *stats, const git_buf *buf, bool skip_bom);
+
+#endif
diff --git a/src/buffer.c b/src/buffer.c
index e55b0a230..0e2c005df 100644
--- a/src/buffer.c
+++ b/src/buffer.c
@@ -31,15 +31,7 @@ void git_buf_init(git_buf *buf, size_t initial_size)
 		git_buf_grow(buf, initial_size);
 }
 
-int git_buf_grow(git_buf *buf, size_t target_size)
-{
-	int error = git_buf_try_grow(buf, target_size);
-	if (error != 0)
-		buf->ptr = git_buf__oom;
-	return error;
-}
-
-int git_buf_try_grow(git_buf *buf, size_t target_size)
+int git_buf_try_grow(git_buf *buf, size_t target_size, bool mark_oom)
 {
 	char *new_ptr;
 	size_t new_size;
@@ -67,8 +59,12 @@ int git_buf_try_grow(git_buf *buf, size_t target_size)
 	new_size = (new_size + 7) & ~7;
 
 	new_ptr = git__realloc(new_ptr, new_size);
-	if (!new_ptr)
+
+	if (!new_ptr) {
+		if (mark_oom)
+			buf->ptr = git_buf__oom;
 		return -1;
+	}
 
 	buf->asize = new_size;
 	buf->ptr   = new_ptr;
@@ -141,51 +137,6 @@ int git_buf_puts(git_buf *buf, const char *string)
 	return git_buf_put(buf, string, strlen(string));
 }
 
-int git_buf_puts_escaped(
-	git_buf *buf, const char *string, const char *esc_chars, const char *esc_with)
-{
-	const char *scan;
-	size_t total = 0, esc_len = strlen(esc_with), count;
-
-	if (!string)
-		return 0;
-
-	for (scan = string; *scan; ) {
-		/* count run of non-escaped characters */
-		count = strcspn(scan, esc_chars);
-		total += count;
-		scan += count;
-		/* count run of escaped characters */
-		count = strspn(scan, esc_chars);
-		total += count * (esc_len + 1);
-		scan += count;
-	}
-
-	ENSURE_SIZE(buf, buf->size + total + 1);
-
-	for (scan = string; *scan; ) {
-		count = strcspn(scan, esc_chars);
-
-		memmove(buf->ptr + buf->size, scan, count);
-		scan += count;
-		buf->size += count;
-
-		for (count = strspn(scan, esc_chars); count > 0; --count) {
-			/* copy escape sequence */
-			memmove(buf->ptr + buf->size, esc_with, esc_len);
-			buf->size += esc_len;
-			/* copy character to be escaped */
-			buf->ptr[buf->size] = *scan;
-			buf->size++;
-			scan++;
-		}
-	}
-
-	buf->ptr[buf->size] = '\0';
-
-	return 0;
-}
-
 static const char b64str[64] =
 	"ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
 
@@ -497,59 +448,6 @@ int git_buf_cmp(const git_buf *a, const git_buf *b)
 		(a->size < b->size) ? -1 : (a->size > b->size) ? 1 : 0;
 }
 
-int git_buf_common_prefix(git_buf *buf, const git_strarray *strings)
-{
-	size_t i;
-	const char *str, *pfx;
-
-	git_buf_clear(buf);
-
-	if (!strings || !strings->count)
-		return 0;
-
-	/* initialize common prefix to first string */
-	if (git_buf_sets(buf, strings->strings[0]) < 0)
-		return -1;
-
-	/* go through the rest of the strings, truncating to shared prefix */
-	for (i = 1; i < strings->count; ++i) {
-
-		for (str = strings->strings[i], pfx = buf->ptr;
-			 *str && *str == *pfx; str++, pfx++)
-			/* scanning */;
-
-		git_buf_truncate(buf, pfx - buf->ptr);
-
-		if (!buf->size)
-			break;
-	}
-
-	return 0;
-}
-
-bool git_buf_is_binary(const git_buf *buf)
-{
-	size_t i;
-	int printable = 0, nonprintable = 0;
-
-	for (i = 0; i < buf->size; i++) {
-		unsigned char c = buf->ptr[i];
-		if (c > 0x1F && c < 0x7F)
-			printable++;
-		else if (c == '\0')
-			return true;
-		else if (!git__isspace(c))
-			nonprintable++;
-	}
-
-	return ((printable >> 7) < nonprintable);
-}
-
-void git_buf_unescape(git_buf *buf)
-{
-	buf->size = git__unescape(buf->ptr);
-}
-
 int git_buf_splice(
 	git_buf *buf,
 	size_t where,
diff --git a/src/buffer.h b/src/buffer.h
index a2896d486..379216bfc 100644
--- a/src/buffer.h
+++ b/src/buffer.h
@@ -27,30 +27,35 @@ extern char git_buf__oom[];
  * For the cases where GIT_BUF_INIT cannot be used to do static
  * initialization.
  */
-void git_buf_init(git_buf *buf, size_t initial_size);
+extern void git_buf_init(git_buf *buf, size_t initial_size);
 
 /**
- * Grow the buffer to hold at least `target_size` bytes.
+ * Attempt to grow the buffer to hold at least `target_size` bytes.
  *
- * If the allocation fails, this will return an error and the buffer
- * will be marked as invalid for future operations.  The existing
- * contents of the buffer will be preserved however.
- * @return 0 on success or -1 on failure
+ * If the allocation fails, this will return an error.  If mark_oom is true,
+ * this will mark the buffer as invalid for future operations; if false,
+ * existing buffer content will be preserved, but calling code must handle
+ * that buffer was not expanded.
  */
-int git_buf_grow(git_buf *buf, size_t target_size);
+extern int git_buf_try_grow(git_buf *buf, size_t target_size, bool mark_oom);
 
 /**
- * Attempt to grow the buffer to hold at least `target_size` bytes.
+ * Grow the buffer to hold at least `target_size` bytes.
  *
- * This is just like `git_buf_grow` except that even if the allocation
- * fails, the git_buf will still be left in a valid state.
+ * If the allocation fails, this will return an error and the buffer will be
+ * marked as invalid for future operations, invaliding contents.
+ *
+ * @return 0 on success or -1 on failure
  */
-int git_buf_try_grow(git_buf *buf, size_t target_size);
+GIT_INLINE(int) git_buf_grow(git_buf *buf, size_t target_size)
+{
+	return git_buf_try_grow(buf, target_size, true);
+}
 
-void git_buf_free(git_buf *buf);
-void git_buf_swap(git_buf *buf_a, git_buf *buf_b);
-char *git_buf_detach(git_buf *buf);
-void git_buf_attach(git_buf *buf, char *ptr, size_t asize);
+extern void git_buf_free(git_buf *buf);
+extern void git_buf_swap(git_buf *buf_a, git_buf *buf_b);
+extern char *git_buf_detach(git_buf *buf);
+extern void git_buf_attach(git_buf *buf, char *ptr, size_t asize);
 
 /**
  * Test if there have been any reallocation failures with this git_buf.
@@ -92,18 +97,6 @@ int git_buf_join_n(git_buf *buf, char separator, int nbuf, ...);
 int git_buf_join(git_buf *buf, char separator, const char *str_a, const char *str_b);
 
 /**
- * Copy string into buf prefixing every character that is contained in the
- * esc_chars string with the esc_with string.
- */
-int git_buf_puts_escaped(
-	git_buf *buf, const char *string, const char *esc_chars, const char *esc_with);
-
-GIT_INLINE(int) git_buf_puts_escape_regex(git_buf *buf, const char *string)
-{
-	return git_buf_puts_escaped(buf, string, "^.[]$()|*+?{}\\", "\\");
-}
-
-/**
  * Join two strings as paths, inserting a slash between as needed.
  * @return 0 on success, -1 on failure
  */
@@ -146,15 +139,6 @@ void git_buf_rtrim(git_buf *buf);
 
 int git_buf_cmp(const git_buf *a, const git_buf *b);
 
-/* Fill buf with the common prefix of a array of strings */
-int git_buf_common_prefix(git_buf *buf, const git_strarray *strings);
-
-/* Check if buffer looks like it contains binary data */
-bool git_buf_is_binary(const git_buf *buf);
-
-/* Unescape all characters in a buffer */
-void git_buf_unescape(git_buf *buf);
-
 /* Write data as base64 encoded in buffer */
 int git_buf_put_base64(git_buf *buf, const char *data, size_t len);
 
diff --git a/src/config.c b/src/config.c
index 5ee0d39ff..6347f7df7 100644
--- a/src/config.c
+++ b/src/config.c
@@ -10,6 +10,7 @@
 #include "config.h"
 #include "git2/config.h"
 #include "vector.h"
+#include "buf_text.h"
 #if GIT_WIN32
 # include <windows.h>
 #endif
@@ -803,7 +804,7 @@ int git_config_rename_section(
 	int error = -1;
 	struct rename_data data;
 
-	git_buf_puts_escape_regex(&pattern,  old_section_name);
+	git_buf_text_puts_escape_regex(&pattern,  old_section_name);
 	git_buf_puts(&pattern, "\\..+");
 	if (git_buf_oom(&pattern))
 		goto cleanup;
diff --git a/src/config_file.c b/src/config_file.c
index 7cc812aa4..354a91986 100644
--- a/src/config_file.c
+++ b/src/config_file.c
@@ -10,6 +10,7 @@
 #include "fileops.h"
 #include "filebuf.h"
 #include "buffer.h"
+#include "buf_text.h"
 #include "git2/config.h"
 #include "git2/types.h"
 #include "strmap.h"
@@ -854,17 +855,14 @@ fail_parse:
 
 static int skip_bom(diskfile_backend *cfg)
 {
-	static const char utf8_bom[] = { '\xef', '\xbb', '\xbf' };
+	git_bom_t bom;
+	int bom_offset = git_buf_text_detect_bom(&bom,
+		&cfg->reader.buffer, cfg->reader.read_ptr - cfg->reader.buffer.ptr);
 
-	if (cfg->reader.buffer.size < sizeof(utf8_bom))
-		return 0;
-
-	if (memcmp(cfg->reader.read_ptr, utf8_bom, sizeof(utf8_bom)) == 0)
-		cfg->reader.read_ptr += sizeof(utf8_bom);
+	if (bom == GIT_BOM_UTF8)
+		cfg->reader.read_ptr += bom_offset;
 
-	/* TODO: the reference implementation does pretty stupid
-		stuff with the BoM
-	*/
+	/* TODO: reference implementation is pretty stupid with BoM */
 
 	return 0;
 }
diff --git a/src/crlf.c b/src/crlf.c
index 5e86b4eb6..80204ebf0 100644
--- a/src/crlf.c
+++ b/src/crlf.c
@@ -148,8 +148,11 @@ static int crlf_apply_to_odb(git_filter *self, git_buf *dest, const git_buf *sou
 	if (filter->attrs.crlf_action == GIT_CRLF_AUTO ||
 		filter->attrs.crlf_action == GIT_CRLF_GUESS) {
 
-		git_text_stats stats;
-		git_text_gather_stats(&stats, source);
+		git_buf_text_stats stats;
+
+		/* Check heuristics for binary vs text... */
+		if (git_buf_text_gather_stats(&stats, source, false))
+			return -1;
 
 		/*
 		 * We're currently not going to even try to convert stuff
@@ -159,12 +162,6 @@ static int crlf_apply_to_odb(git_filter *self, git_buf *dest, const git_buf *sou
 		if (stats.cr != stats.crlf)
 			return -1;
 
-		/*
-		 * And add some heuristics for binary vs text, of course...
-		 */
-		if (git_text_is_binary(&stats))
-			return -1;
-
 #if 0
 		if (crlf_action == CRLF_GUESS) {
 			/*
diff --git a/src/diff_output.c b/src/diff_output.c
index 3d5e03a29..e137fd0f2 100644
--- a/src/diff_output.c
+++ b/src/diff_output.c
@@ -142,7 +142,7 @@ static int diff_delta_is_binary_by_content(
 		search.ptr  = map->data;
 		search.size = min(map->len, 4000);
 
-		if (git_buf_is_binary(&search))
+		if (git_buf_text_is_binary(&search))
 			file->flags |= GIT_DIFF_FILE_BINARY;
 		else
 			file->flags |= GIT_DIFF_FILE_NOT_BINARY;
diff --git a/src/filter.c b/src/filter.c
index f2ab1b85a..6d27c0c46 100644
--- a/src/filter.c
+++ b/src/filter.c
@@ -13,75 +13,6 @@
 #include "git2/config.h"
 #include "blob.h"
 
-/* Tweaked from Core Git. I wonder what we could use this for... */
-void git_text_gather_stats(git_text_stats *stats, const git_buf *text)
-{
-	size_t i;
-
-	memset(stats, 0, sizeof(*stats));
-
-	for (i = 0; i < git_buf_len(text); i++) {
-		unsigned char c = text->ptr[i];
-
-		if (c == '\r') {
-			stats->cr++;
-
-			if (i + 1 < git_buf_len(text) && text->ptr[i + 1] == '\n')
-				stats->crlf++;
-		}
-
-		else if (c == '\n')
-			stats->lf++;
-
-		else if (c == 127)
-			/* DEL */
-			stats->nonprintable++;
-
-		else if (c <= 0x1F || (c >= 0x80 && c <= 0x9F)) {
-			switch (c) {
-				/* BS, HT, ESC and FF */
-			case '\b': case '\t': case '\033': case '\014':
-				stats->printable++;
-				break;
-			case 0:
-				stats->nul++;
-				/* fall through */
-			default:
-				stats->nonprintable++;
-			}
-		}
-
-		else
-			stats->printable++;
-	}
-
-	/* If file ends with EOF then don't count this EOF as non-printable. */
-	if (git_buf_len(text) >= 1 && text->ptr[text->size - 1] == '\032')
-		stats->nonprintable--;
-}
-
-/*
- * Fresh from Core Git
- */
-int git_text_is_binary(git_text_stats *stats)
-{
-	if (stats->nul)
-		return 1;
-
-	if ((stats->printable >> 7) < stats->nonprintable)
-		return 1;
-	/*
-	 * Other heuristics? Average line length might be relevant,
-	 * as might LF vs CR vs CRLF counts..
-	 *
-	 * NOTE! It might be normal to have a low ratio of CRLF to LF
-	 * (somebody starts with a LF-only file and edits it with an editor
-	 * that adds CRLF only to lines that are added..). But do  we
-	 * want to support CR-only? Probably not.
-	 */
-	return 0;
-}
-
 int git_filters_load(git_vector *filters, git_repository *repo, const char *path, int mode)
 {
 	int error;
diff --git a/src/filter.h b/src/filter.h
index b9beb4942..2b5051c69 100644
--- a/src/filter.h
+++ b/src/filter.h
@@ -9,6 +9,7 @@
 
 #include "common.h"
 #include "buffer.h"
+#include "buf_text.h"
 #include "git2/odb.h"
 #include "git2/repository.h"
 
@@ -31,14 +32,6 @@ typedef enum {
 	GIT_CRLF_AUTO,
 } git_crlf_t;
 
-typedef struct {
-	/* NUL, CR, LF and CRLF counts */
-	unsigned int nul, cr, lf, crlf;
-
-	/* These are just approximations! */
-	unsigned int printable, nonprintable;
-} git_text_stats;
-
 /*
  * FILTER API
  */
@@ -99,24 +92,4 @@ extern int git_filter_add__crlf_to_odb(git_vector *filters, git_repository *repo
 /* Add CRLF, from ODB to worktree */
 extern int git_filter_add__crlf_to_workdir(git_vector *filters, git_repository *repo, const char *path);
 
-
-/*
- * PLAINTEXT API
- */
-
-/*
- * Gather stats for a piece of text
- *
- * Fill the `stats` structure with information on the number of
- * unreadable characters, carriage returns, etc, so it can be
- * used in heuristics.
- */
-extern void git_text_gather_stats(git_text_stats *stats, const git_buf *text);
-
-/*
- * Process `git_text_stats` data generated by `git_text_stat` to see
- * if it qualifies as a binary file
- */
-extern int git_text_is_binary(git_text_stats *stats);
-
 #endif
diff --git a/src/path.c b/src/path.c
index 98351bec3..87eded3c4 100644
--- a/src/path.c
+++ b/src/path.c
@@ -511,7 +511,7 @@ static bool _check_dir_contents(
 	size_t sub_size = strlen(sub);
 
 	/* leave base valid even if we could not make space for subdir */
-	if (git_buf_try_grow(dir, dir_size + sub_size + 2) < 0)
+	if (git_buf_try_grow(dir, dir_size + sub_size + 2, false) < 0)
 		return false;
 
 	/* save excursion */
diff --git a/src/pathspec.c b/src/pathspec.c
index fc6547afe..993b44667 100644
--- a/src/pathspec.c
+++ b/src/pathspec.c
@@ -6,6 +6,7 @@
  */
 
 #include "pathspec.h"
+#include "buf_text.h"
 #include "attr_file.h"
 
 /* what is the common non-wildcard prefix for all items in the pathspec */
@@ -15,7 +16,7 @@ char *git_pathspec_prefix(const git_strarray *pathspec)
 	const char *scan;
 
 	if (!pathspec || !pathspec->count ||
-		git_buf_common_prefix(&prefix, pathspec) < 0)
+		git_buf_text_common_prefix(&prefix, pathspec) < 0)
 		return NULL;
 
 	/* diff prefix will only be leading non-wildcards */
@@ -31,7 +32,7 @@ char *git_pathspec_prefix(const git_strarray *pathspec)
 		return NULL;
 	}
 
-	git_buf_unescape(&prefix);
+	git_buf_text_unescape(&prefix);
 
 	return git_buf_detach(&prefix);
 }
diff --git a/src/submodule.c b/src/submodule.c
index 15158f0d8..c117255d4 100644
--- a/src/submodule.c
+++ b/src/submodule.c
@@ -12,6 +12,7 @@
 #include "git2/index.h"
 #include "git2/submodule.h"
 #include "buffer.h"
+#include "buf_text.h"
 #include "vector.h"
 #include "posix.h"
 #include "config_file.h"
@@ -782,7 +783,7 @@ int git_submodule_reload(git_submodule *submodule)
 		git_buf path = GIT_BUF_INIT;
 
 		git_buf_sets(&path, "submodule\\.");
-		git_buf_puts_escape_regex(&path, submodule->name);
+		git_buf_text_puts_escape_regex(&path, submodule->name);
 		git_buf_puts(&path, ".*");
 
 		if (git_buf_oom(&path))
diff --git a/tests-clar/core/buffer.c b/tests-clar/core/buffer.c
index 236bf39da..40fc4c571 100644
--- a/tests-clar/core/buffer.c
+++ b/tests-clar/core/buffer.c
@@ -1,5 +1,6 @@
 #include "clar_libgit2.h"
 #include "buffer.h"
+#include "buf_text.h"
 
 #define TESTSTR "Have you seen that? Have you seeeen that??"
 const char *test_string = TESTSTR;
@@ -576,37 +577,37 @@ void test_core_buffer__11(void)
 
 	t.strings = t1;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "");
 
 	t.strings = t2;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "some");
 
 	t.strings = t3;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "");
 
 	t.strings = t4;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "happ");
 
 	t.strings = t5;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "happ");
 
 	t.strings = t6;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "");
 
 	t.strings = t7;
 	t.count = 3;
-	cl_git_pass(git_buf_common_prefix(&a, &t));
+	cl_git_pass(git_buf_text_common_prefix(&a, &t));
 	cl_assert_equal_s(a.ptr, "");
 
 	git_buf_free(&a);
@@ -641,19 +642,19 @@ void test_core_buffer__puts_escaped(void)
 	git_buf a = GIT_BUF_INIT;
 
 	git_buf_clear(&a);
-	cl_git_pass(git_buf_puts_escaped(&a, "this is a test", "", ""));
+	cl_git_pass(git_buf_text_puts_escaped(&a, "this is a test", "", ""));
 	cl_assert_equal_s("this is a test", a.ptr);
 
 	git_buf_clear(&a);
-	cl_git_pass(git_buf_puts_escaped(&a, "this is a test", "t", "\\"));
+	cl_git_pass(git_buf_text_puts_escaped(&a, "this is a test", "t", "\\"));
 	cl_assert_equal_s("\\this is a \\tes\\t", a.ptr);
 
 	git_buf_clear(&a);
-	cl_git_pass(git_buf_puts_escaped(&a, "this is a test", "i ", "__"));
+	cl_git_pass(git_buf_text_puts_escaped(&a, "this is a test", "i ", "__"));
 	cl_assert_equal_s("th__is__ __is__ a__ test", a.ptr);
 
 	git_buf_clear(&a);
-	cl_git_pass(git_buf_puts_escape_regex(&a, "^match\\s*[A-Z]+.*"));
+	cl_git_pass(git_buf_text_puts_escape_regex(&a, "^match\\s*[A-Z]+.*"));
 	cl_assert_equal_s("\\^match\\\\s\\*\\[A-Z\\]\\+\\.\\*", a.ptr);
 
 	git_buf_free(&a);
@@ -663,7 +664,7 @@ static void assert_unescape(char *expected, char *to_unescape) {
 	git_buf buf = GIT_BUF_INIT;
 
 	cl_git_pass(git_buf_sets(&buf, to_unescape));
-	git_buf_unescape(&buf);
+	git_buf_text_unescape(&buf);
 	cl_assert_equal_s(expected, buf.ptr);
 	cl_assert_equal_sz(strlen(expected), buf.size);
 
diff --git a/tests-clar/object/blob/filter.c b/tests-clar/object/blob/filter.c
index 785489849..b9bbfff0c 100644
--- a/tests-clar/object/blob/filter.c
+++ b/tests-clar/object/blob/filter.c
@@ -4,7 +4,7 @@
 #include "filter.h"
 
 static git_repository *g_repo = NULL;
-#define NUM_TEST_OBJECTS 6
+#define NUM_TEST_OBJECTS 8
 static git_oid g_oids[NUM_TEST_OBJECTS];
 static const char *g_raw[NUM_TEST_OBJECTS] = {
 	"",
@@ -12,16 +12,20 @@ static const char *g_raw[NUM_TEST_OBJECTS] = {
 	"foo\rbar\r",
 	"foo\r\nbar\r\n",
 	"foo\nbar\rboth\r\nreversed\n\ragain\nproblems\r",
-	"123\n\000\001\002\003\004abc\255\254\253\r\n"
+	"123\n\000\001\002\003\004abc\255\254\253\r\n",
+	"\xEF\xBB\xBFThis is UTF-8\n",
+	"\xFE\xFF\x00T\x00h\x00i\x00s\x00!"
 };
-static git_off_t g_len[NUM_TEST_OBJECTS] = { -1, -1, -1, -1, -1, 17 };
-static git_text_stats g_stats[NUM_TEST_OBJECTS] = {
-	{ 0, 0, 0, 0, 0, 0 },
-	{ 0, 0, 2, 0, 6, 0 },
-	{ 0, 2, 0, 0, 6, 0 },
-	{ 0, 2, 2, 2, 6, 0 },
-	{ 0, 4, 4, 1, 31, 0 },
-	{ 1, 1, 2, 1, 9, 5 }
+static git_off_t g_len[NUM_TEST_OBJECTS] = { -1, -1, -1, -1, -1, 17, -1, 12 };
+static git_buf_text_stats g_stats[NUM_TEST_OBJECTS] = {
+	{ 0, 0, 0, 0, 0, 0, 0 },
+	{ 0, 0, 0, 2, 0, 6, 0 },
+	{ 0, 0, 2, 0, 0, 6, 0 },
+	{ 0, 0, 2, 2, 2, 6, 0 },
+	{ 0, 0, 4, 4, 1, 31, 0 },
+	{ 0, 1, 1, 2, 1, 9, 5 },
+	{ GIT_BOM_UTF8, 0, 0, 1, 0, 16, 0 },
+	{ GIT_BOM_UTF16_BE, 5, 0, 0, 0, 7, 5 },
 };
 static git_buf g_crlf_filtered[NUM_TEST_OBJECTS] = {
 	{ "", 0, 0 },
@@ -29,7 +33,9 @@ static git_buf g_crlf_filtered[NUM_TEST_OBJECTS] = {
 	{ "foo\rbar\r", 0, 8 },
 	{ "foo\nbar\n", 0, 8 },
 	{ "foo\nbar\rboth\nreversed\n\ragain\nproblems\r", 0, 38 },
-	{ "123\n\000\001\002\003\004abc\255\254\253\n", 0, 16 }
+	{ "123\n\000\001\002\003\004abc\255\254\253\n", 0, 16 },
+	{ "\xEF\xBB\xBFThis is UTF-8\n", 0, 17 },
+	{ "\xFE\xFF\x00T\x00h\x00i\x00s\x00!", 0, 12 }
 };
 
 void test_object_blob_filter__initialize(void)
@@ -76,12 +82,12 @@ void test_object_blob_filter__stats(void)
 	int i;
 	git_blob *blob;
 	git_buf buf = GIT_BUF_INIT;
-	git_text_stats stats;
+	git_buf_text_stats stats;
 
 	for (i = 0; i < NUM_TEST_OBJECTS; i++) {
 		cl_git_pass(git_blob_lookup(&blob, g_repo, &g_oids[i]));
 		cl_git_pass(git_blob__getbuf(&buf, blob));
-		git_text_gather_stats(&stats, &buf);
+		git_buf_text_gather_stats(&stats, &buf, false);
 		cl_assert(memcmp(&g_stats[i], &stats, sizeof(stats)) == 0);
 		git_blob_free(blob);
 	}