Converted all strings to char; fixed resulting issues. (resolves #31 …

…and #32)

CHANGELOG.md

@@ -7,6 +7,15 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
 
+### Changed
+- Converted unsigned char strings to char strings, which broken iso8859-1
+  processing.  Updated iso8859-1 processing so that it works as before.
+  Updated CFLAGS to catch any issues that arise in the future. [#31]
+
+### Fixed
+- src/Makefile.am no longer deletes src/config_file.h when `make
+  maintainer-clean` is run.
+
 ## [1.4.0] - 2021-02-11
 ### Added
 - Regression tests for basic functionality, based on old custom scripts.
@@ -163,6 +172,7 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 [1.2.1]: https://github.com/dharple/detox/compare/v1.2.0...v1.2.1
 [1.2.0]: https://github.com/dharple/detox/releases/tag/v1.2.0
 
+[#31]: https://github.com/dharple/detox/issues/31
 [#30]: https://github.com/dharple/detox/issues/30
 [#19]: https://github.com/dharple/detox/issues/19
 [#17]: https://github.com/dharple/detox/issues/17

src/Makefile.am

@@ -1,8 +1,6 @@
-BUILT_SOURCES = config_file.h
+AM_CFLAGS = -g -O2 -Wall -DYY_NO_INPUT -DYY_NO_UNPUT -DDATADIR=\"$(datadir)\" -DSYSCONFDIR=\"$(sysconfdir)\"
 AM_YFLAGS = -d
 
-AM_CPPFLAGS = -DDATADIR=\"$(datadir)\" -DSYSCONFDIR=\"$(sysconfdir)\"
-
 bin_PROGRAMS = detox inline-detox
 detox_SOURCES = detox.c file.c clean_string.c table.c parse_table.c config_file_yacc.y config_file_lex.l config_file_spoof.c config_file_dump.c parse_options.c
 inline_detox_SOURCES = $(detox_SOURCES)

src/clean_string.c

@@ -26,14 +26,16 @@
 #include "parse_table.h"
 #include "table.h"
 
+#define ISO8859_1_UPPER_BIT 0x80
+#define ISO8859_LOWER_BITS  0x7f
 
 /*
  * Translates ISO8859.1 characters (Latin-1) into lower ASCII characters.
  */
-unsigned char *clean_iso8859_1_basic(unsigned char *s, void *opts)
+char *clean_iso8859_1_basic(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk, *replace_walk;
-	int replace_pos;
+	char *output, *input_walk, *output_walk, *replace_walk;
+	int new_value;
 
 	if (s == NULL) {
 		return NULL;
@@ -49,18 +51,20 @@ unsigned char *clean_iso8859_1_basic(unsigned char *s, void *opts)
 	output_walk = output;
 
 	while (*input_walk != '\0') {
-		if (*input_walk >= ISO8859_1_OFFSET) {
-			replace_pos = *input_walk - ISO8859_1_OFFSET;
-			replace_walk = (unsigned char *)&iso8859_1_trans[replace_pos];
-
-			while (*replace_walk != '\0') {
-				*output_walk++ = *replace_walk++;
-			}
-			input_walk++;
-		}
-		else {
+		if ((*input_walk & ISO8859_1_UPPER_BIT) == 0) {
 			*output_walk++ = *input_walk++;
+			continue;
+		}
+
+		new_value = *input_walk & ISO8859_LOWER_BITS;
+
+		replace_walk = (char *)&iso8859_1_trans[new_value];
+
+		while (*replace_walk != '\0') {
+			*output_walk++ = *replace_walk++;
 		}
+
+		input_walk++;
 	}
 
 	*output_walk = '\0';
@@ -71,9 +75,10 @@ unsigned char *clean_iso8859_1_basic(unsigned char *s, void *opts)
 /*
  * Translates ISO8859.1 characters (Latin-1) into lower ASCII characters.
  */
-unsigned char *clean_iso8859_1(unsigned char *s, void *opts)
+char *clean_iso8859_1(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk, *replace_walk;
+	char *output, *input_walk, *output_walk, *replace_walk;
+	int new_value;
 
 	struct translation_table *table = NULL;
 	struct clean_string_options *options = NULL;
@@ -100,30 +105,33 @@ unsigned char *clean_iso8859_1(unsigned char *s, void *opts)
 	output_walk = output;
 
 	while (*input_walk != '\0') {
-		if (*input_walk >= ISO8859_1_OFFSET) {
-			replace_walk = table_get(table, *input_walk);
-			if (replace_walk == NULL) {
-				if (table->default_translation == NULL) {
-					/*
-					 * Null translation == leave it alone
-					 */
-					*output_walk++ = *input_walk++;
-					continue;
-				}
-				else {
-					replace_walk = table->default_translation;
-				}
-			}
+		if ((*input_walk & ISO8859_1_UPPER_BIT) == 0) {
+			*output_walk++ = *input_walk++;
+			continue;
+		}
 
-			while (*replace_walk != '\0') {
-				*output_walk++ = *replace_walk++;
-			}
+		new_value = (unsigned char) *input_walk;
 
-			input_walk++;
+		replace_walk = table_get(table, new_value);
+
+		if (replace_walk == NULL) {
+			if (table->default_translation == NULL) {
+				/*
+				 * Null translation == leave it alone
+				 */
+				*output_walk++ = *input_walk++;
+				continue;
+			}
+			else {
+				replace_walk = table->default_translation;
+			}
 		}
-		else {
-			*output_walk++ = *input_walk++;
+
+		while (*replace_walk != '\0') {
+			*output_walk++ = *replace_walk++;
 		}
+
+		input_walk++;
 	}
 
 	*output_walk = '\0';
@@ -149,9 +157,9 @@ unsigned char *clean_iso8859_1(unsigned char *s, void *opts)
  *     ( ) [ ] { }
  *
  */
-unsigned char *clean_safe_basic(unsigned char *s, void *opts)
+char *clean_safe_basic(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk;
+	char *output, *input_walk, *output_walk;
 
 	if (s == NULL) {
 		return NULL;
@@ -238,9 +246,9 @@ unsigned char *clean_safe_basic(unsigned char *s, void *opts)
 /*
  * Translates unsafe characters
  */
-unsigned char *clean_safe(unsigned char *s, void *opts)
+char *clean_safe(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk, *replace_walk;
+	char *output, *input_walk, *output_walk, *replace_walk;
 
 	struct translation_table *table = NULL;
 	struct clean_string_options *options = NULL;
@@ -300,10 +308,10 @@ unsigned char *clean_safe(unsigned char *s, void *opts)
  * Cleans up any CGI encoded characters, in the form "%" followed by 2 hex
  * digits.
  */
-unsigned char *clean_uncgi(unsigned char *s, void *opts)
+char *clean_uncgi(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk;
-	unsigned char conv[3];
+	char *output, *input_walk, *output_walk;
+	char conv[3];
 
 	if (s == NULL) {
 		return NULL;
@@ -323,7 +331,7 @@ unsigned char *clean_uncgi(unsigned char *s, void *opts)
 			conv[0] = input_walk[1];
 			conv[1] = input_walk[2];
 			conv[2] = 0;
-			*output_walk++ = (unsigned char)strtol(conv, NULL, 16);
+			*output_walk++ = (char) strtol(conv, NULL, 16);
 			input_walk += 3;
 		}
 		else {
@@ -348,9 +356,9 @@ unsigned char *clean_uncgi(unsigned char *s, void *opts)
  * Strips any "-", "_" or "#" from the beginning of a string.
  *
  */
-unsigned char *clean_wipeup(unsigned char *s, void *opts)
+char *clean_wipeup(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk;
+	char *output, *input_walk, *output_walk;
 	int matched;
 	int remove_trailing;
 
@@ -436,9 +444,9 @@ unsigned char *clean_wipeup(unsigned char *s, void *opts)
  * Translates UTF-8 characters (Unicode Translation Format - 8 Bit) into
  * Unicode and then lower ASCII characters.
  */
-unsigned char *clean_utf_8_basic(unsigned char *s, void *opts)
+char *clean_utf_8_basic(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk, *replace_walk;
+	char *output, *input_walk, *output_walk, *replace_walk;
 	int new_value, expected_chars;
 
 	if (s == NULL) {
@@ -529,7 +537,7 @@ unsigned char *clean_utf_8_basic(unsigned char *s, void *opts)
 			continue;
 		}
 
-		replace_walk = (unsigned char *)&unicode_trans[new_value];
+		replace_walk = (char *)&unicode_trans[new_value];
 
 		while (*replace_walk != '\0') {
 			*output_walk++ = *replace_walk++;
@@ -545,9 +553,9 @@ unsigned char *clean_utf_8_basic(unsigned char *s, void *opts)
  * Translates UTF-8 characters (Unicode Translation Format - 8 Bit) into
  * Unicode and then runs the translation table.
  */
-unsigned char *clean_utf_8(unsigned char *s, void *opts)
+char *clean_utf_8(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk, *replace_walk;
+	char *output, *input_walk, *output_walk, *replace_walk;
 	int new_value, expected_chars;
 
 	struct translation_table *table = NULL;
@@ -687,9 +695,9 @@ unsigned char *clean_utf_8(unsigned char *s, void *opts)
 /*
  * Trims a file down to specified length.
  */
-unsigned char *clean_max_length(unsigned char *s, void *opts)
+char *clean_max_length(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk;
+	char *output, *input_walk, *output_walk;
 	size_t max_length;
 	size_t s_length;
 	size_t ext_length;
@@ -741,9 +749,9 @@ unsigned char *clean_max_length(unsigned char *s, void *opts)
 /*
  * Converts all characters to lowercase.
  */
-unsigned char *clean_lower(unsigned char *s, void *opts)
+char *clean_lower(char *s, void *opts)
 {
-	unsigned char *output, *input_walk, *output_walk;
+	char *output, *input_walk, *output_walk;
 
 	if (s == NULL) {
 		return NULL;

src/clean_string.h

@@ -19,15 +19,15 @@ struct clean_string_options {
 	void *translation_table;
 };
 
-extern unsigned char *clean_iso8859_1_basic(unsigned char *s, void *options);
-extern unsigned char *clean_iso8859_1(unsigned char *s, void *options);
-extern unsigned char *clean_safe_basic(unsigned char *s, void *options);
-extern unsigned char *clean_safe(unsigned char *s, void *options);
-extern unsigned char *clean_uncgi(unsigned char *s, void *options);
-extern unsigned char *clean_wipeup(unsigned char *s, void *options);
-extern unsigned char *clean_utf_8_basic(unsigned char *s, void *options);
-extern unsigned char *clean_utf_8(unsigned char *s, void *options);
-extern unsigned char *clean_max_length(unsigned char *s, void *opts);
-extern unsigned char *clean_lower(unsigned char *s, void *opts);
+extern char *clean_iso8859_1_basic(char *s, void *options);
+extern char *clean_iso8859_1(char *s, void *options);
+extern char *clean_safe_basic(char *s, void *options);
+extern char *clean_safe(char *s, void *options);
+extern char *clean_uncgi(char *s, void *options);
+extern char *clean_wipeup(char *s, void *options);
+extern char *clean_utf_8_basic(char *s, void *options);
+extern char *clean_utf_8(char *s, void *options);
+extern char *clean_max_length(char *s, void *opts);
+extern char *clean_lower(char *s, void *opts);
 
 #endif				/* __CLEAN_STRING_H */

src/detox.h

@@ -32,7 +32,7 @@ struct detox_sequence_list {
 struct detox_sequence_entry {
 	struct detox_sequence_entry *next;
 
-	unsigned char *(*cleaner) (unsigned char *str, void *options);
+	char *(*cleaner) (char *str, void *options);
 	void *options;
 };
 
@@ -42,7 +42,7 @@ struct detox_sequence_entry {
 struct detox_ignore_entry {
 	struct detox_ignore_entry *next;
 
-	unsigned char *filename;
+	char *filename;
 };
 
 /*

src/file.c

@@ -32,15 +32,15 @@ static char badfiles[3][30] = {
 /*
  * Internal function declarations
  */
-static int ignore_file(unsigned char *filename, struct detox_options *options);
+static int ignore_file(char *filename, struct detox_options *options);
 
 /*
  * Renames file to a safe filename.
  */
-unsigned char *parse_file(unsigned char *filename, struct detox_options *options)
+char *parse_file(char *filename, struct detox_options *options)
 {
-	unsigned char *old_filename, *old_filename_ptr, *new_filename;
-	unsigned char *work, *hold;
+	char *old_filename, *old_filename_ptr, *new_filename;
+	char *work, *hold;
 
 	struct stat stat_info_old;
 	struct stat stat_info_new;
@@ -157,9 +157,9 @@ unsigned char *parse_file(unsigned char *filename, struct detox_options *options
 /*
  * Handles directory.
  */
-void parse_dir(unsigned char *indir, struct detox_options *options)
+void parse_dir(char *indir, struct detox_options *options)
 {
-	unsigned char *new_file, *work;
+	char *new_file, *work;
 	DIR *dir_handle;
 	struct dirent *dir_entry;
 	struct stat stat_info;
@@ -177,7 +177,7 @@ void parse_dir(unsigned char *indir, struct detox_options *options)
 	}
 
 	new_file_length = strlen(indir) + 1024;
-	new_file = (char *)malloc(new_file_length);
+	new_file = malloc(new_file_length);
 	if (new_file == NULL) {
 		fprintf(stderr, "out of memory: %s\n", strerror(errno));
 		return;
@@ -230,7 +230,7 @@ void parse_dir(unsigned char *indir, struct detox_options *options)
 /*
  * Handles a special file.
  */
-void parse_special(unsigned char *in, struct detox_options *options)
+void parse_special(char *in, struct detox_options *options)
 {
 	struct stat stat_info;
 	char *new_file, *work;
@@ -287,7 +287,7 @@ void parse_special(unsigned char *in, struct detox_options *options)
 /*
  * Determines if the file should be ignored
  */
-static int ignore_file(unsigned char *filename, struct detox_options *options)
+static int ignore_file(char *filename, struct detox_options *options)
 {
 	struct detox_ignore_entry *ignore_walk;
 	int x;
@@ -312,11 +312,11 @@ static int ignore_file(unsigned char *filename, struct detox_options *options)
 /*
  * Renames file to a safe filename.
  */
-void parse_inline(unsigned char *filename, struct detox_options *options)
+void parse_inline(char *filename, struct detox_options *options)
 {
 	struct detox_sequence_entry *sequence;
 	FILE *fp;
-	unsigned char *base, *work, *hold;
+	char *base, *work, *hold;
 	size_t buf_size;
 
 	if (filename != NULL) {