buf: remove internal `git_buf_text` namespace

The `git_buf_text` namespace is unnecessary and strange.  Remove it,
just keep the functions prefixed with `git_buf`.

14 files changed, 414 insertions, 473 deletions

diff --git a/src/attr_file.c b/src/attr_file.c
index d1b90c5ba..6575e5a49 100644
--- a/src/attr_file.c
+++ b/src/attr_file.c
@@ -10,7 +10,6 @@
 #include "repository.h"
 #include "filebuf.h"
 #include "attrcache.h"
-#include "buf_text.h"
 #include "git2/blob.h"
 #include "git2/tree.h"
 #include "blob.h"
@@ -192,7 +191,7 @@ int git_attr_file__load(
 
 	/* advance over a UTF8 BOM */
 	content_str = git_buf_cstr(&content);
-	bom_offset = git_buf_text_detect_bom(&bom, &content);
+	bom_offset = git_buf_detect_bom(&bom, &content);
 
 	if (bom == GIT_BOM_UTF8)
 		content_str += bom_offset;
diff --git a/src/blob.c b/src/blob.c
index ddc2494c8..169e34503 100644
--- a/src/blob.c
+++ b/src/blob.c
@@ -14,7 +14,6 @@
 
 #include "filebuf.h"
 #include "filter.h"
-#include "buf_text.h"
 
 const void *git_blob_rawcontent(const git_blob *blob)
 {
@@ -401,7 +400,7 @@ int git_blob_is_binary(const git_blob *blob)
 
 	git_buf_attach_notowned(&content, git_blob_rawcontent(blob),
 		(size_t)min(size, GIT_FILTER_BYTES_TO_CHECK_NUL));
-	return git_buf_text_is_binary(&content);
+	return git_buf_is_binary(&content);
 }
 
 int git_blob_filter_options_init(
diff --git a/src/buf_text.c b/src/buf_text.c
deleted file mode 100644
index 0fd22319f..000000000
--- a/src/buf_text.c
+++ /dev/null
@@ -1,316 +0,0 @@
-/*
- * Copyright (C) the libgit2 contributors. All rights reserved.
- *
- * 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, alloclen;
-
-	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;
-	}
-
-	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, total, 1);
-	if (git_buf_grow_by(buf, alloclen) < 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_crlf_to_lf(git_buf *tgt, const git_buf *src)
-{
-	const char *scan = src->ptr;
-	const char *scan_end = src->ptr + src->size;
-	const char *next = memchr(scan, '\r', src->size);
-	size_t new_size;
-	char *out;
-
-	GIT_ASSERT(tgt != src);
-
-	if (!next)
-		return git_buf_set(tgt, src->ptr, src->size);
-
-	/* reduce reallocs while in the loop */
-	GIT_ERROR_CHECK_ALLOC_ADD(&new_size, src->size, 1);
-	if (git_buf_grow(tgt, new_size) < 0)
-		return -1;
-
-	out = tgt->ptr;
-	tgt->size = 0;
-
-	/* Find the next \r and copy whole chunk up to there to tgt */
-	for (; next; scan = next + 1, next = memchr(scan, '\r', scan_end - scan)) {
-		if (next > scan) {
-			size_t copylen = (size_t)(next - scan);
-			memcpy(out, scan, copylen);
-			out += copylen;
-		}
-
-		/* Do not drop \r unless it is followed by \n */
-		if (next + 1 == scan_end || next[1] != '\n')
-			*out++ = '\r';
-	}
-
-	/* Copy remaining input into dest */
-	if (scan < scan_end) {
-		size_t remaining = (size_t)(scan_end - scan);
-		memcpy(out, scan, remaining);
-		out += remaining;
-	}
-
-	tgt->size = (size_t)(out - tgt->ptr);
-	tgt->ptr[tgt->size] = '\0';
-
-	return 0;
-}
-
-int git_buf_text_lf_to_crlf(git_buf *tgt, const git_buf *src)
-{
-	const char *start = src->ptr;
-	const char *end = start + src->size;
-	const char *scan = start;
-	const char *next = memchr(scan, '\n', src->size);
-	size_t alloclen;
-
-	GIT_ASSERT(tgt != src);
-
-	if (!next)
-		return git_buf_set(tgt, src->ptr, src->size);
-
-	/* attempt to reduce reallocs while in the loop */
-	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, src->size, src->size >> 4);
-	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);
-	if (git_buf_grow(tgt, alloclen) < 0)
-		return -1;
-	tgt->size = 0;
-
-	for (; next; scan = next + 1, next = memchr(scan, '\n', end - scan)) {
-		size_t copylen = next - scan;
-
-		/* if we find mixed line endings, carry on */
-		if (copylen && next[-1] == '\r')
-			copylen--;
-
-		GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, copylen, 3);
-		if (git_buf_grow_by(tgt, alloclen) < 0)
-			return -1;
-
-		if (copylen) {
-			memcpy(tgt->ptr + tgt->size, scan, copylen);
-			tgt->size += copylen;
-		}
-
-		tgt->ptr[tgt->size++] = '\r';
-		tgt->ptr[tgt->size++] = '\n';
-	}
-
-	tgt->ptr[tgt->size] = '\0';
-	return git_buf_put(tgt, scan, end - scan);
-}
-
-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;
-	git_bom_t bom;
-	int printable = 0, nonprintable = 0;
-
-	scan += git_buf_text_detect_bom(&bom, buf);
-
-	if (bom > GIT_BOM_UTF8)
-		return 1;
-
-	while (scan < end) {
-		unsigned char c = *scan++;
-
-		/* Printable characters are those above SPACE (0x1F) excluding DEL,
-		 * and including BS, ESC and FF.
-		 */
-		if ((c > 0x1F && c != 127) || c == '\b' || c == '\033' || c == '\014')
-			printable++;
-		else if (c == '\0')
-			return true;
-		else if (!git__isspace(c))
-			nonprintable++;
-	}
-
-	return ((printable >> 7) < nonprintable);
-}
-
-bool git_buf_text_contains_nul(const git_buf *buf)
-{
-	return (memchr(buf->ptr, '\0', buf->size) != NULL);
-}
-
-int git_buf_text_detect_bom(git_bom_t *bom, const git_buf *buf)
-{
-	const char *ptr;
-	size_t len;
-
-	*bom = GIT_BOM_NONE;
-	/* need at least 2 bytes to look for any BOM */
-	if (buf->size < 2)
-		return 0;
-
-	ptr = buf->ptr;
-	len = buf->size;
-
-	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);
-	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)
-			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;
-			}
-	}
-
-	/* Treat files with a bare CR as binary */
-	return (stats->cr != stats->crlf || stats->nul > 0 ||
-		((stats->printable >> 7) < stats->nonprintable));
-}
diff --git a/src/buf_text.h b/src/buf_text.h
deleted file mode 100644
index 726b0ae7b..000000000
--- a/src/buf_text.h
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * Copyright (C) the libgit2 contributors. All rights reserved.
- *
- * 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 "common.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);
-
-/**
- * Replace all \r\n with \n.
- *
- * @return 0 on success, -1 on memory error
- */
-extern int git_buf_text_crlf_to_lf(git_buf *tgt, const git_buf *src);
-
-/**
- * Replace all \n with \r\n. Does not modify existing \r\n.
- *
- * @return 0 on success, -1 on memory error
- */
-extern int git_buf_text_lf_to_crlf(git_buf *tgt, const git_buf *src);
-
-/**
- * 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 quickly if buffer contains a NUL byte
- *
- * @param buf Buffer to check
- * @return true if buffer contains a NUL byte
- */
-extern bool git_buf_text_contains_nul(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
- * @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);
-
-/**
- * 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 f6ecc6e41..ffce73c68 100644
--- a/src/buffer.c
+++ b/src/buffer.c
@@ -7,7 +7,6 @@
 #include "buffer.h"
 #include "posix.h"
 #include "git2/buffer.h"
-#include "buf_text.h"
 #include <ctype.h>
 
 /* Used as default value for git_buf->ptr so that people can always
@@ -187,16 +186,6 @@ int git_buf_set(git_buf *buf, const void *data, size_t len)
 	return 0;
 }
 
-int git_buf_is_binary(const git_buf *buf)
-{
-	return git_buf_text_is_binary(buf);
-}
-
-int git_buf_contains_nul(const git_buf *buf)
-{
-	return git_buf_text_contains_nul(buf);
-}
-
 int git_buf_sets(git_buf *buf, const char *string)
 {
 	return git_buf_set(buf, string, string ? strlen(string) : 0);
@@ -1058,3 +1047,312 @@ invalid:
 	git_error_set(GIT_ERROR_INVALID, "invalid quoted line");
 	return -1;
 }
+
+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, alloclen;
+
+	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;
+	}
+
+	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, total, 1);
+	if (git_buf_grow_by(buf, alloclen) < 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_unescape(git_buf *buf)
+{
+	buf->size = git__unescape(buf->ptr);
+}
+
+int git_buf_crlf_to_lf(git_buf *tgt, const git_buf *src)
+{
+	const char *scan = src->ptr;
+	const char *scan_end = src->ptr + src->size;
+	const char *next = memchr(scan, '\r', src->size);
+	size_t new_size;
+	char *out;
+
+	GIT_ASSERT(tgt != src);
+
+	if (!next)
+		return git_buf_set(tgt, src->ptr, src->size);
+
+	/* reduce reallocs while in the loop */
+	GIT_ERROR_CHECK_ALLOC_ADD(&new_size, src->size, 1);
+	if (git_buf_grow(tgt, new_size) < 0)
+		return -1;
+
+	out = tgt->ptr;
+	tgt->size = 0;
+
+	/* Find the next \r and copy whole chunk up to there to tgt */
+	for (; next; scan = next + 1, next = memchr(scan, '\r', scan_end - scan)) {
+		if (next > scan) {
+			size_t copylen = (size_t)(next - scan);
+			memcpy(out, scan, copylen);
+			out += copylen;
+		}
+
+		/* Do not drop \r unless it is followed by \n */
+		if (next + 1 == scan_end || next[1] != '\n')
+			*out++ = '\r';
+	}
+
+	/* Copy remaining input into dest */
+	if (scan < scan_end) {
+		size_t remaining = (size_t)(scan_end - scan);
+		memcpy(out, scan, remaining);
+		out += remaining;
+	}
+
+	tgt->size = (size_t)(out - tgt->ptr);
+	tgt->ptr[tgt->size] = '\0';
+
+	return 0;
+}
+
+int git_buf_lf_to_crlf(git_buf *tgt, const git_buf *src)
+{
+	const char *start = src->ptr;
+	const char *end = start + src->size;
+	const char *scan = start;
+	const char *next = memchr(scan, '\n', src->size);
+	size_t alloclen;
+
+	GIT_ASSERT(tgt != src);
+
+	if (!next)
+		return git_buf_set(tgt, src->ptr, src->size);
+
+	/* attempt to reduce reallocs while in the loop */
+	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, src->size, src->size >> 4);
+	GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, alloclen, 1);
+	if (git_buf_grow(tgt, alloclen) < 0)
+		return -1;
+	tgt->size = 0;
+
+	for (; next; scan = next + 1, next = memchr(scan, '\n', end - scan)) {
+		size_t copylen = next - scan;
+
+		/* if we find mixed line endings, carry on */
+		if (copylen && next[-1] == '\r')
+			copylen--;
+
+		GIT_ERROR_CHECK_ALLOC_ADD(&alloclen, copylen, 3);
+		if (git_buf_grow_by(tgt, alloclen) < 0)
+			return -1;
+
+		if (copylen) {
+			memcpy(tgt->ptr + tgt->size, scan, copylen);
+			tgt->size += copylen;
+		}
+
+		tgt->ptr[tgt->size++] = '\r';
+		tgt->ptr[tgt->size++] = '\n';
+	}
+
+	tgt->ptr[tgt->size] = '\0';
+	return git_buf_put(tgt, scan, end - scan);
+}
+
+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;
+}
+
+int git_buf_is_binary(const git_buf *buf)
+{
+	const char *scan = buf->ptr, *end = buf->ptr + buf->size;
+	git_bom_t bom;
+	int printable = 0, nonprintable = 0;
+
+	scan += git_buf_detect_bom(&bom, buf);
+
+	if (bom > GIT_BOM_UTF8)
+		return 1;
+
+	while (scan < end) {
+		unsigned char c = *scan++;
+
+		/* Printable characters are those above SPACE (0x1F) excluding DEL,
+		 * and including BS, ESC and FF.
+		 */
+		if ((c > 0x1F && c != 127) || c == '\b' || c == '\033' || c == '\014')
+			printable++;
+		else if (c == '\0')
+			return true;
+		else if (!git__isspace(c))
+			nonprintable++;
+	}
+
+	return ((printable >> 7) < nonprintable);
+}
+
+int git_buf_contains_nul(const git_buf *buf)
+{
+	return (memchr(buf->ptr, '\0', buf->size) != NULL);
+}
+
+int git_buf_detect_bom(git_bom_t *bom, const git_buf *buf)
+{
+	const char *ptr;
+	size_t len;
+
+	*bom = GIT_BOM_NONE;
+	/* need at least 2 bytes to look for any BOM */
+	if (buf->size < 2)
+		return 0;
+
+	ptr = buf->ptr;
+	len = buf->size;
+
+	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_gather_text_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_detect_bom(&stats->bom, buf);
+	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)
+			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;
+			}
+	}
+
+	/* Treat files with a bare CR as binary */
+	return (stats->cr != stats->crlf || stats->nul > 0 ||
+		((stats->printable >> 7) < stats->nonprintable));
+}
diff --git a/src/buffer.h b/src/buffer.h
index 8c2096bce..e75ecc15c 100644
--- a/src/buffer.h
+++ b/src/buffer.h
@@ -17,6 +17,21 @@
  * } git_buf;
  */
 
+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;
+
 extern char git_buf__initbuf[];
 extern char git_buf__oom[];
 
@@ -219,4 +234,81 @@ int git_buf_splice(
 	const char *data,
 	size_t nb_to_insert);
 
+/**
+ * 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_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_puts_escape_regex(git_buf *buf, const char *string)
+{
+	return git_buf_puts_escaped(buf, string, "^.[]$()|*+?{}\\", "\\");
+}
+
+/**
+ * Unescape all characters in a buffer in place
+ *
+ * I.e. remove backslashes
+ */
+extern void git_buf_unescape(git_buf *buf);
+
+/**
+ * Replace all \r\n with \n.
+ *
+ * @return 0 on success, -1 on memory error
+ */
+extern int git_buf_crlf_to_lf(git_buf *tgt, const git_buf *src);
+
+/**
+ * Replace all \n with \r\n. Does not modify existing \r\n.
+ *
+ * @return 0 on success, -1 on memory error
+ */
+extern int git_buf_lf_to_crlf(git_buf *tgt, const git_buf *src);
+
+/**
+ * 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_common_prefix(git_buf *buf, const git_strarray *strs);
+
+/**
+ * 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
+ * @return Number of bytes of BOM data (or 0 if no BOM found)
+ */
+extern int git_buf_detect_bom(git_bom_t *bom, const git_buf *buf);
+
+/**
+ * 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_gather_text_stats(
+	git_buf_text_stats *stats, const git_buf *buf, bool skip_bom);
+
 #endif
diff --git a/src/checkout.c b/src/checkout.c
index cadc4c82d..27bf6fe00 100644
--- a/src/checkout.c
+++ b/src/checkout.c
@@ -26,7 +26,6 @@
 #include "diff.h"
 #include "diff_generate.h"
 #include "pathspec.h"
-#include "buf_text.h"
 #include "diff_xdiff.h"
 #include "path.h"
 #include "attr.h"
diff --git a/src/config.c b/src/config.c
index a6a8cb246..7a5dadd87 100644
--- a/src/config.c
+++ b/src/config.c
@@ -10,7 +10,6 @@
 #include "git2/config.h"
 #include "git2/sys/config.h"
 
-#include "buf_text.h"
 #include "config_backend.h"
 #include "regexp.h"
 #include "sysdir.h"
@@ -1497,7 +1496,7 @@ int git_config_rename_section(
 	int error = 0;
 	struct rename_data data;
 
-	git_buf_text_puts_escape_regex(&pattern, old_section_name);
+	git_buf_puts_escape_regex(&pattern, old_section_name);
 
 	if ((error = git_buf_puts(&pattern, "\\..+")) < 0)
 		goto cleanup;
diff --git a/src/config_parse.c b/src/config_parse.c
index 7da92d3e7..ea32c36ba 100644
--- a/src/config_parse.c
+++ b/src/config_parse.c
@@ -7,8 +7,6 @@
 
 #include "config_parse.h"
 
-#include "buf_text.h"
-
 #include <ctype.h>
 
 const char *git_config_escapes = "ntb\"\\";
@@ -231,7 +229,7 @@ static int skip_bom(git_parse_ctx *parser)
 {
 	git_buf buf = GIT_BUF_INIT_CONST(parser->content, parser->content_len);
 	git_bom_t bom;
-	int bom_offset = git_buf_text_detect_bom(&bom, &buf);
+	int bom_offset = git_buf_detect_bom(&bom, &buf);
 
 	if (bom == GIT_BOM_UTF8)
 		git_parse_advance_chars(parser, bom_offset);
diff --git a/src/crlf.c b/src/crlf.c
index 81b5216bc..1de9d8c3b 100644
--- a/src/crlf.c
+++ b/src/crlf.c
@@ -15,7 +15,6 @@
 #include "futils.h"
 #include "hash.h"
 #include "filter.h"
-#include "buf_text.h"
 #include "repository.h"
 
 typedef enum {
@@ -219,7 +218,7 @@ static int crlf_apply_to_odb(
 	if (ca->crlf_action == GIT_CRLF_BINARY || !git_buf_len(from))
 		return GIT_PASSTHROUGH;
 
-	is_binary = git_buf_text_gather_stats(&stats, from, false);
+	is_binary = git_buf_gather_text_stats(&stats, from, false);
 
 	/* Heuristics to see if we can skip the conversion.
 	 * Straight from Core Git.
@@ -247,7 +246,7 @@ static int crlf_apply_to_odb(
 		return GIT_PASSTHROUGH;
 
 	/* Actually drop the carriage returns */
-	return git_buf_text_crlf_to_lf(to, from);
+	return git_buf_crlf_to_lf(to, from);
 }
 
 static int crlf_apply_to_workdir(
@@ -262,7 +261,7 @@ static int crlf_apply_to_workdir(
 	if (git_buf_len(from) == 0 || output_eol(ca) != GIT_EOL_CRLF)
 		return GIT_PASSTHROUGH;
 
-	is_binary = git_buf_text_gather_stats(&stats, from, false);
+	is_binary = git_buf_gather_text_stats(&stats, from, false);
 
 	/* If there are no LFs, or all LFs are part of a CRLF, nothing to do */
 	if (stats.lf == 0 || stats.lf == stats.crlf)
@@ -281,7 +280,7 @@ static int crlf_apply_to_workdir(
 			return GIT_PASSTHROUGH;
 	}
 
-	return git_buf_text_lf_to_crlf(to, from);
+	return git_buf_lf_to_crlf(to, from);
 }
 
 static int convert_attrs(
diff --git a/src/diff_driver.c b/src/diff_driver.c
index 6b19012c7..e9f63cb17 100644
--- a/src/diff_driver.c
+++ b/src/diff_driver.c
@@ -13,7 +13,6 @@
 #include "diff.h"
 #include "strmap.h"
 #include "map.h"
-#include "buf_text.h"
 #include "config.h"
 #include "regexp.h"
 #include "repository.h"
@@ -428,8 +427,8 @@ int git_diff_driver_content_is_binary(
 	 * let's just use the simple NUL-byte detection that core git uses.
 	 */
 
-	/* previously was: if (git_buf_text_is_binary(&search)) */
-	if (git_buf_text_contains_nul(&search))
+	/* previously was: if (git_buf_is_binary(&search)) */
+	if (git_buf_contains_nul(&search))
 		return 1;
 
 	return 0;
diff --git a/src/ident.c b/src/ident.c
index 7eccf9a43..ae3ef1b45 100644
--- a/src/ident.c
+++ b/src/ident.c
@@ -10,7 +10,6 @@
 #include "git2/sys/filter.h"
 #include "filter.h"
 #include "buffer.h"
-#include "buf_text.h"
 
 static int ident_find_id(
 	const char **id_start, const char **id_end, const char *start, size_t len)
@@ -105,7 +104,7 @@ static int ident_apply(
 	GIT_UNUSED(self); GIT_UNUSED(payload);
 
 	/* Don't filter binary files */
-	if (git_buf_text_is_binary(from))
+	if (git_buf_is_binary(from))
 		return GIT_PASSTHROUGH;
 
 	if (git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
diff --git a/src/pathspec.c b/src/pathspec.c
index 8b17a6e27..8f1bdf0fa 100644
--- a/src/pathspec.c
+++ b/src/pathspec.c
@@ -9,7 +9,6 @@
 
 #include "git2/pathspec.h"
 #include "git2/diff.h"
-#include "buf_text.h"
 #include "attr_file.h"
 #include "iterator.h"
 #include "repository.h"
@@ -25,7 +24,7 @@ char *git_pathspec_prefix(const git_strarray *pathspec)
 	const char *scan;
 
 	if (!pathspec || !pathspec->count ||
-		git_buf_text_common_prefix(&prefix, pathspec) < 0)
+		git_buf_common_prefix(&prefix, pathspec) < 0)
 		return NULL;
 
 	/* diff prefix will only be leading non-wildcards */
@@ -41,7 +40,7 @@ char *git_pathspec_prefix(const git_strarray *pathspec)
 		return NULL;
 	}
 
-	git_buf_text_unescape(&prefix);
+	git_buf_unescape(&prefix);
 
 	return git_buf_detach(&prefix);
 }
diff --git a/src/submodule.c b/src/submodule.c
index 1d059a043..14869452f 100644
--- a/src/submodule.c
+++ b/src/submodule.c
@@ -12,7 +12,6 @@
 #include "git2/types.h"
 #include "git2/index.h"
 #include "buffer.h"
-#include "buf_text.h"
 #include "vector.h"
 #include "posix.h"
 #include "config_backend.h"