readtags.c

/*
*   Copyright (c) 1996-2003, Darren Hiebert
*
*   This source code is released into the public domain.
*
*   This module contains functions for reading tag files.
*/

/*
*   INCLUDE FILES
*/
#ifdef HAVE_CTAGS_INLINE_H
/* the build script of ctags can insert thee macro
   definition for READTAGS_INLINE. */
#include "inline.h"
#else
/* Makefile generated by configure defines inline. */
#define READTAGS_INLINE static inline
#endif

#include <stdlib.h>
#include <string.h>
#include <ctype.h>
#include <stdio.h>
#include <errno.h>
#include <sys/types.h>  /* to declare off_t */

#include "readtags.h"

/*
*   MACROS
*/
#define TAB '\t'


/*
*   DATA DECLARATIONS
*/
typedef struct {
	size_t size;
	char *buffer;
} vstring;

/* Define readtags' own off_t. */
#ifdef _WIN32
typedef long long rt_off_t;
#else
typedef off_t rt_off_t;
#endif

/* Information about current tag file */
struct sTagFile {
		/* has the file been opened and this structure initialized? */
	unsigned char initialized;
		/* format of tag file */
	unsigned char format;
		/* 1 "u-ctags" is set to !_TAG_OUTPUT_MODE pseudo tag
		 * and "slash" is set to !_TAG_OUTPUT_FILESEP
		 * pseudo tag in the tags file. */
	unsigned char inputUCtagsMode;
		/* how is the tag file sorted? */
	tagSortType sortMethod;
		/* pointer to file structure */
	FILE* fp;
		/* file position of first character of `line' */
	rt_off_t pos;
		/* size of tag file in seekable positions */
	rt_off_t size;
		/* last line read */
	vstring line;
		/* name of tag in last line read */
	vstring name;
		/* defines tag search state */
	struct {
				/* file position of last match for tag */
			rt_off_t pos;
				/* name of tag last searched for */
			char *name;
				/* length of name for partial matches */
			size_t nameLength;
				/* performing partial match */
			short partial;
				/* ignoring case */
			short ignorecase;
	} search;
		/* miscellaneous extension fields */
	struct {
				/* number of entries in `list' */
			unsigned short max;
				/* list of key value pairs */
			tagExtensionField *list;
	} fields;
		/* buffers to be freed at close */
	struct {
			/* name of program author */
		char *author;
			/* name of program */
		char *name;
			/* URL of distribution */
		char *url;
			/* program version */
		char *version;
	} program;
		/* 0 (initial state set by calloc), errno value,
		 * or tagErrno typed value */
	int err;
};

/*
*   DATA DEFINITIONS
*/
static const char *const EmptyString = "";
static const char *const PseudoTagPrefix = "!_";
static const size_t PseudoTagPrefixLength = 2;

/*
*   FUNCTION DEFINITIONS
*/

static rt_off_t readtags_ftell(FILE *fp)
{
	rt_off_t pos;

#ifdef _WIN32
	pos = _ftelli64(fp);
#else
	pos = ftell(fp);
#endif

	return pos;
}

static int readtags_fseek(FILE *fp, rt_off_t pos, int whence)
{
	int ret;

#ifdef _WIN32
	ret = _fseeki64(fp, pos, whence);
#else
	ret = fseek(fp, pos, whence);
#endif

	return ret;
}

/* Converts a hexadecimal digit to its value */
static int xdigitValue (unsigned char digit)
{
	if (digit >= '0' && digit <= '9')
		return digit - '0';
	else if (digit >= 'a' && digit <= 'f')
		return 10 + digit - 'a';
	else if (digit >= 'A' && digit <= 'F')
		return 10 + digit - 'A';
	else
		return 0;
}

static unsigned char altchars[0x100] = {
	['\\'] = '\\',
	['a']  = '\a', /* u-ctags extension */
	['b']  = '\b', /* u-ctags extension */
	['f']  = '\f', /* u-ctags extension */
	['n']  = '\n',
	['r']  = '\r',
	['t']  = '\t',
	['v']  = '\v', /* u-ctags extension */
};

/*
 * Reads the first character from the string, possibly un-escaping it, and
 * advances *s to the start of the next character.
 */

READTAGS_INLINE
int readTagCharacter (const char **const s)
{
	const unsigned char *p = (const unsigned char *) *s;
	int c = *p;

	p++;

	if (c == '\\')
	{
		unsigned char a;
		switch (*p)
		{
		default:
			a = altchars[*p];
			if (a) {
				c = a;
				p++;
			}
			break;
		case 'x':
			if (isxdigit (p[1]) && isxdigit (p[2]))
			{
				int val = (xdigitValue (p[1]) << 4) | xdigitValue (p[2]);
				if (val < 0x80)
				{
					p += 3;
					c = val;
				}
			}
			break;
		}
	}

	*s = (const char *) p;

	return c;
}

/*
 * Compare two strings, ignoring case.
 * Return 0 for match, < 0 for smaller, > 0 for bigger
 * Make sure case is folded to uppercase in comparison (like for 'sort -f')
 * This makes a difference when one of the chars lies between upper and lower
 * ie. one of the chars [ \ ] ^ _ ` for ascii. (The '_' in particular !)
 */
static int taguppercmp (const char *s1, const char *s2)
{
	int result;
	int c1, c2;
	do
	{
		c1 = (unsigned char)*s1++;
		c2 = readTagCharacter (&s2);

		result = toupper (c1) - toupper (c2);
	} while (result == 0  &&  c1 != '\0'  &&  c2 != '\0');
	return result;
}

static int tagnuppercmp (const char *s1, const char *s2, size_t n)
{
	int result;
	int c1, c2;
	do
	{
		c1 = (unsigned char)*s1++;
		c2 = readTagCharacter (&s2);

		result = toupper (c1) - toupper (c2);
	} while (result == 0  &&  --n > 0  &&  c1 != '\0'  &&  c2 != '\0');
	return result;
}

static int tagcmp (const char *s1, const char *s2)
{
	int result;
	int c1, c2;
	do
	{
		c1 = (unsigned char)*s1++;
		c2 = readTagCharacter (&s2);

		result = c1 - c2;
	} while (result == 0  &&  c1 != '\0'  &&  c2 != '\0');
	return result;
}

static int tagncmp (const char *s1, const char *s2, size_t n)
{
	int result;
	int c1, c2;
	do
	{
		c1 = *s1++;
		c2 = readTagCharacter (&s2);

		result = c1 - c2;
	} while (result == 0  &&  --n > 0  &&  c1 != '\0'  &&  c2 != '\0');
	return result;
}

static tagResult growString (vstring *s)
{
	tagResult result = TagFailure;
	size_t newLength;
	char *newLine;
	if (s->size == 0)
	{
		newLength = 128;
		newLine = (char*) malloc (newLength);
		if (newLine)
			*newLine = '\0';
	}
	else
	{
		newLength = 2 * s->size;
		newLine = (char*) realloc (s->buffer, newLength);
	}
	if (newLine == NULL)
		perror ("string too large");
	else
	{
		s->buffer = newLine;
		s->size = newLength;
		result = TagSuccess;
	}
	return result;
}

/* Copy name of tag out of tag line */
static tagResult copyName (tagFile *const file)
{
	size_t length;
	const char *end = strchr (file->line.buffer, '\t');
	if (end == NULL)
	{
		end = strchr (file->line.buffer, '\n');
		if (end == NULL)
			end = strchr (file->line.buffer, '\r');
	}
	if (end != NULL)
		length = end - file->line.buffer;
	else
		length = strlen (file->line.buffer);
	while (length >= file->name.size)
	{
		if (growString (&file->name) != TagSuccess)
			return TagFailure;
	}
	strncpy (file->name.buffer, file->line.buffer, length);
	file->name.buffer [length] = '\0';
	return TagSuccess;
}

/* Return 1 on success.
 * Return 0 on failure or EOF.
 * errno is set to *err unless EOF.
 */
static int readTagLineRaw (tagFile *const file, int *err)
{
	int result = 1;
	int reReadLine;

	/*  If reading the line places any character other than a null or a
	 *  newline at the last character position in the buffer (one less than
	 *  the buffer size), then we must resize the buffer and reattempt to read
	 *  the line.
	 */
	do
	{
		char *const pLastChar = file->line.buffer + file->line.size - 2;
		char *line;

		file->pos = readtags_ftell (file->fp);
		if (file->pos < 0)
		{
			*err = errno;
			result = 0;
			break;
		}
		reReadLine = 0;
		*pLastChar = '\0';
		line = fgets (file->line.buffer, (int) file->line.size, file->fp);
		if (line == NULL)
		{
			/* read error */
			*err = 0;
			if (! feof (file->fp))
				*err = errno;
			result = 0;
		}
		else if (*pLastChar != '\0'  &&
					*pLastChar != '\n'  &&  *pLastChar != '\r')
		{
			/*  buffer overflow */
			if (growString (&file->line) != TagSuccess)
			{
				*err = ENOMEM;
				result = 0;
			}

			if (readtags_fseek (file->fp, file->pos, SEEK_SET) < 0)
			{
				*err = errno;
				result = 0;
			}
			reReadLine = 1;
		}
		else
		{
			size_t i = strlen (file->line.buffer);
			while (i > 0  &&
				   (file->line.buffer [i - 1] == '\n' || file->line.buffer [i - 1] == '\r'))
			{
				file->line.buffer [i - 1] = '\0';
				--i;
			}
		}
	} while (reReadLine  &&  result);
	if (result)
	{
		if (copyName (file) != TagSuccess)
		{
			*err = ENOMEM;
			result = 0;
		}
	}
	return result;
}

/* Return 1 on success.
 * Return 0 on failure or EOF.
 * errno is set to *err unless EOF.
 */
static int readTagLine (tagFile *const file, int *err)
{
	int result;
	do
	{
		result = readTagLineRaw (file, err);
	} while (result && *file->name.buffer == '\0');
	return result;
}

static tagResult growFields (tagFile *const file)
{
	tagResult result = TagFailure;
	unsigned short newCount = (unsigned short) 2 * file->fields.max;
	tagExtensionField *newFields;

	if (!(newCount > file->fields.max)) {
		errno = EOVERFLOW;
		perror ("too many extension fields");
		goto out;				/* integer overflow */
	}

	newFields = (tagExtensionField*)
			realloc (file->fields.list, newCount * sizeof (tagExtensionField));
	if (newFields == NULL)
		perror ("too many extension fields");
	else
	{
		file->fields.list = newFields;
		file->fields.max = newCount;
		result = TagSuccess;
	}

 out:
	return result;
}

static tagResult parseExtensionFields (tagFile *const file, tagEntry *const entry,
									   char *const string, int *err)
{
	char *p = string;
	char *tail = string + (string? strlen(string):0);
	size_t q_len;

	while (p != NULL  &&  *p != '\0')
	{
		while (*p == TAB)
			*p++ = '\0';
		if (*p != '\0')
		{
			char *colon;
			char *field = p;
			p = strchr (p, TAB);
			if (p != NULL)
				*p++ = '\0';
			colon = strchr (field, ':');
			if (colon == NULL)
				entry->kind = field;
			else
			{
				const char *key = field;
				char *q = colon + 1;
				const char *value = q;
				const int key_len = colon - key;
				*colon = '\0';

				q_len = tail - q;

				/* Unescaping */
				while (*q != '\0')
				{
					const char *next = q;
					int ch = readTagCharacter (&next);
					size_t skip = next - q;

					*q = (char) ch;
					q++;
					q_len -= skip;
					if (skip > 1)
					{
						/* + 1 is for moving the area including the last '\0'. */
						memmove (q, next, q_len + 1);
						if (p)
							p -= skip - 1;
						if (tail != string)
							tail -= skip - 1;
					}
				}

				if (key_len == 4)
				{
					if (memcmp (key, "kind", 4) == 0)
						entry->kind = value;
					else if (memcmp (key, "file", 4) == 0)
						entry->fileScope = 1;
					else if (memcmp (key, "line", 4) == 0)
					{
						char *endptr = NULL;
						long m = strtol (value, &endptr, 10);
						if (*endptr != '\0' || m < 0)
						{
							*err = TagErrnoUnexpectedLineno;
							return TagFailure;
						}
						entry->address.lineNumber = m;
					}
					else
						goto normalField;
				}
				else
				{
				normalField:
					if (entry->fields.count == file->fields.max)
					{
						if (growFields (file) != TagSuccess)
						{
							*err = errno;
							return TagFailure;
						}
					}
					file->fields.list [entry->fields.count].key = key;
					file->fields.list [entry->fields.count].value = value;
					++entry->fields.count;
				}
			}
		}
	}
	return TagSuccess;
}

static int isOdd (unsigned int i)
{
	return  (i % 2);
}

static unsigned int countContinuousBackslashesBackward(const char *from,
						     const char *till)
{
	unsigned int counter = 0;

	for (; from > till; from--)
	{
		if (*from == '\\')
			counter++;
		else
			break;
	}
	return counter;
}

/* When unescaping, the input string becomes shorter.
 * e.g. \t occupies two bytes on the tag file.
 * It is converted to 0x9 and occupies one byte.
 * memmove called here for shortening the line
 * buffer. */
READTAGS_INLINE
char *unescapeInPlace (char *q, char **tab, size_t *p_len)
{
	char *p = q;

	while (*p != '\0')
	{
		const char *next = p;
		int ch = readTagCharacter (&next);
		size_t skip = next - p;

		*p = (char) ch;
		p++;
		*p_len -= skip;
		if (skip > 1)
		{
			/* + 1 is for moving the area including the last '\0'. */
			memmove (p, next, *p_len + 1);
			if (*tab)
				*tab -= skip - 1;
		}
	}

	return p;
}

static tagResult parseTagLine (tagFile *file, tagEntry *const entry, int *err)
{
	int i;
	char *p = file->line.buffer;
	size_t p_len = strlen (p);
	char *tab = strchr (p, TAB);

	memset(entry, 0, sizeof(*entry));

	entry->name = p;
	if (tab != NULL)
	{
		*tab = '\0';
	}

	p = unescapeInPlace (p, &tab, &p_len);

	if (tab != NULL)
	{
		p = tab + 1;
		entry->file = p;
		tab = strchr (p, TAB);
		if (file->inputUCtagsMode)
		{
			if (tab != NULL)
			{
				*tab = '\0';
			}
			p = unescapeInPlace (p, &tab, &p_len);
		}

		if (tab != NULL)
		{
			int fieldsPresent;
			int combinedPattern;
			*tab = '\0';
			p = tab + 1;
			if (*p == '/'  ||  *p == '?')
			{
				/* parse pattern */
				int delimiter = *(unsigned char*) p;
				entry->address.lineNumber = 0;
				entry->address.pattern = p;
				do
				{
					p = strchr (p + 1, delimiter);
				} while (p != NULL
					 &&  isOdd (countContinuousBackslashesBackward (p - 1,
											entry->address.pattern)));

				if (p == NULL)
				{
					/* TODO: invalid pattern */
				}
				else
					++p;
			}
			else if (isdigit (*(unsigned char*) p))
			{
				/* parse line number */
				entry->address.pattern = p;
				entry->address.lineNumber = atol (p);
				while (isdigit (*(unsigned char*) p))
					++p;
				if (p)
				{
					combinedPattern = (strncmp (p, ";/", 2) == 0) ||
											(strncmp (p, ";?", 2) == 0);
					if (combinedPattern)
					{
						++p;
						/* parse pattern */
						int delimiter = *(unsigned char*) p;
						do
						{
							p = strchr (p + 1, delimiter);
						} while (p != NULL
							 &&  isOdd (countContinuousBackslashesBackward (p - 1,
													entry->address.pattern)));

						if (p == NULL)
						{
							/* TODO: invalid pattern */
						}
						else
							++p;
					}
				}
			}
			else
			{
				/* TODO: invalid pattern */
			}

			if (p)
			{
				fieldsPresent = (strncmp (p, ";\"", 2) == 0);
				*p = '\0';
				if (fieldsPresent)
				{
					if (parseExtensionFields (file, entry, p + 2, err) != TagSuccess)
						return TagFailure;
				}
			}
		}
	}
	if (entry->fields.count > 0)
		entry->fields.list = file->fields.list;
	for (i = entry->fields.count  ;  i < file->fields.max  ;  ++i)
	{
		file->fields.list [i].key = NULL;
		file->fields.list [i].value = NULL;
	}
	return TagSuccess;
}

static char *duplicate (const char *str)
{
	char *result = NULL;
	if (str != NULL)
	{
		result = strdup (str);
		if (result == NULL)
			perror (NULL);
	}
	return result;
}

static int isPseudoTagLine (const char *buffer)
{
	return (strncmp (buffer, PseudoTagPrefix, PseudoTagPrefixLength) == 0);
}

static tagResult readPseudoTags (tagFile *const file, tagFileInfo *const info)
{
	fpos_t startOfLine;
	int err = 0;
	tagResult result = TagSuccess;
	const size_t prefixLength = strlen (PseudoTagPrefix);
	int tag_output_mode_u_ctags = 0;
	int tag_output_filesep_slash = 0;

	info->file.format     = 1;
	info->file.sort       = TAG_UNSORTED;
	info->program.author  = NULL;
	info->program.name    = NULL;
	info->program.url     = NULL;
	info->program.version = NULL;

	while (1)
	{
		if (fgetpos (file->fp, &startOfLine) < 0)
		{
			err = errno;
			break;
		}
		if (! readTagLine (file, &err))
			break;
		if (!isPseudoTagLine (file->line.buffer))
			break;
		else
		{
			tagEntry entry;
			const char *key, *value;
			if (parseTagLine (file, &entry, &err) != TagSuccess)
				break;
			key = entry.name + prefixLength;
			value = entry.file;
			if (strcmp (key, "TAG_FILE_SORTED") == 0)
			{
				char *endptr = NULL;
				long m = strtol (value, &endptr, 10);
				if (*endptr != '\0' || m < 0 || m > 2)
				{
					err = TagErrnoUnexpectedSortedMethod;
					break;
				}
				file->sortMethod = (tagSortType) m;
			}
			else if (strcmp (key, "TAG_FILE_FORMAT") == 0)
			{
				char *endptr = NULL;
				long m = strtol (value, &endptr, 10);
				if (*endptr != '\0' || m < 1 || m > 2)
				{
					err = TagErrnoUnexpectedFormat;
					break;
				}
				file->format = (unsigned char) m;
			}
			else if (strcmp (key, "TAG_PROGRAM_AUTHOR") == 0)
			{
				file->program.author = duplicate (value);
				if (value && file->program.author == NULL)
				{
					err = ENOMEM;
					break;
				}
			}
			else if (strcmp (key, "TAG_PROGRAM_NAME") == 0)
			{
				file->program.name = duplicate (value);
				if (value && file->program.name == NULL)
				{
					err = ENOMEM;
					break;
				}
			}
			else if (strcmp (key, "TAG_PROGRAM_URL") == 0)
			{
				file->program.url = duplicate (value);
				if (value && file->program.url == NULL)
				{
					err = ENOMEM;
					break;
				}
			}
			else if (strcmp (key, "TAG_PROGRAM_VERSION") == 0)
			{
				file->program.version = duplicate (value);
				if (value && file->program.version == NULL)
				{
					err = ENOMEM;
					break;
				}
			}
			else if (strcmp (key, "TAG_OUTPUT_MODE") == 0)
			{
				if (strcmp (value, "u-ctags") == 0)
					tag_output_mode_u_ctags = 1;
			}
			else if (strcmp (key, "TAG_OUTPUT_FILESEP") == 0)
			{
				if (strcmp (value, "slash") == 0)
					tag_output_filesep_slash = 1;
			}

			info->file.format     = file->format;
			info->file.sort       = file->sortMethod;
			info->program.author  = file->program.author;
			info->program.name    = file->program.name;
			info->program.url     = file->program.url;
			info->program.version = file->program.version;
		}
	}

	if (tag_output_mode_u_ctags && tag_output_filesep_slash)
		file->inputUCtagsMode = 1;

	if (fsetpos (file->fp, &startOfLine) < 0)
		err = errno;

	info->status.error_number = err;
	if (err)
		result = TagFailure;
	return result;
}

static int doesFilePointPseudoTag (tagFile *const file, void *unused)
{
	return isPseudoTagLine (file->name.buffer);
}

static tagResult gotoFirstLogicalTag (tagFile *const file)
{
	fpos_t startOfLine;

	if (readtags_fseek(file->fp, 0, SEEK_SET) == -1)
	{
		file->err = errno;
		return TagFailure;
	}

	while (1)
	{
		if (fgetpos (file->fp, &startOfLine) < 0)
		{
			file->err = errno;
			return TagFailure;
		}
		if (! readTagLine (file, &file->err))
		{
			if (file->err)
				return TagFailure;
			break;
		}
		if (!isPseudoTagLine (file->line.buffer))
			break;
	}
	if (fsetpos (file->fp, &startOfLine) < 0)
	{
		file->err = errno;
		return TagFailure;
	}
	return TagSuccess;
}

static tagFile *initialize (const char *const filePath, tagFileInfo *const info)
{
	tagFile *result = (tagFile*) calloc ((size_t) 1, sizeof (tagFile));

	if (result == NULL)
	{
		info->status.opened = 0;
		info->status.error_number = ENOMEM;
		return NULL;
	}

	if (growString (&result->line) != TagSuccess)
		goto mem_error;
	if (growString (&result->name) != TagSuccess)
		goto mem_error;
	result->fields.max = 20;
	result->fields.list = (tagExtensionField*) calloc (
		result->fields.max, sizeof (tagExtensionField));
	if (result->fields.list == NULL)
		goto mem_error;

#if defined(__GLIBC__) && (__GLIBC__ >= 2) \
	&& defined(__GLIBC_MINOR__) && (__GLIBC_MINOR__ >= 3)
	result->fp = fopen (filePath, "rbm");
#endif
	if (result->fp == NULL)
	{
		errno = 0;
		result->fp = fopen (filePath, "rb");
	}
	if (result->fp == NULL)
	{
		info->status.error_number = errno;
		goto file_error;
	}

	/* Record the size of the tags file to `size` field of result. */
	if (readtags_fseek (result->fp, 0, SEEK_END) == -1)
	{
		info->status.error_number = errno;
		goto file_error;
	}
	result->size = readtags_ftell (result->fp);
	if (result->size == -1)
	{
		/* fseek() retruns an int value.
		 * We observed following behavior on Windows;
		 * if sizeof(int) of the platform is too small for
		 * representing the size of the tags file, fseek()
		 * returns -1 and it doesn't set errno.
		 */
		info->status.error_number = errno;
		if (info->status.error_number == 0)
			info->status.error_number = TagErrnoFileMaybeTooBig;

		goto file_error;
	}
	if (readtags_fseek(result->fp, 0, SEEK_SET) == -1)
	{
		info->status.error_number = errno;
		goto file_error;
	}

	if (readPseudoTags (result, info) == TagFailure)
		goto file_error;

	info->status.opened = 1;
	result->initialized = 1;

	return result;
 mem_error:
	info->status.error_number = ENOMEM;
 file_error:
	free (result->line.buffer);
	free (result->name.buffer);
	free (result->fields.list);
	if (result->fp)
		fclose (result->fp);
	free (result);
	info->status.opened = 0;
	return NULL;
}

static void terminate (tagFile *const file)
{
	fclose (file->fp);

	free (file->line.buffer);
	free (file->name.buffer);
	free (file->fields.list);

	if (file->program.author != NULL)
		free (file->program.author);
	if (file->program.name != NULL)
		free (file->program.name);
	if (file->program.url != NULL)
		free (file->program.url);
	if (file->program.version != NULL)
		free (file->program.version);
	if (file->search.name != NULL)
		free (file->search.name);

	memset (file, 0, sizeof (tagFile));

	free (file);
}

static tagResult readNext (tagFile *const file, tagEntry *const entry)
{
	tagResult result;

	if (file == NULL)
		return TagFailure;

	if (! file->initialized)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	if (! readTagLine (file, &file->err))
		return TagFailure;

	result = (entry != NULL)
		? parseTagLine (file, entry, &file->err)
		: TagSuccess;

	return result;
}

static const char *readFieldValue (
	const tagEntry *const entry, const char *const key)
{
	const char *result = NULL;
	int i;
	if (strcmp (key, "kind") == 0)
		result = entry->kind;
	else if (strcmp (key, "file") == 0)
		result = EmptyString;
	else for (i = 0  ;  i < entry->fields.count  &&  result == NULL  ;  ++i)
		if (strcmp (entry->fields.list [i].key, key) == 0)
			result = entry->fields.list [i].value;
	return result;
}

static int readTagLineSeek (tagFile *const file, const rt_off_t pos)
{
	if (readtags_fseek (file->fp, pos, SEEK_SET) < 0)
	{
		file->err = errno;
		return 0;
	}

	/* read probable partial line */
	if (!readTagLine (file, &file->err))
		return 0;

	/* read complete line */
	if (pos > 0)
		return readTagLine (file, &file->err);

	return 1;
}

static int nameComparison (tagFile *const file)
{
	int result;
	if (file->search.ignorecase)
	{
		if (file->search.partial)
			result = tagnuppercmp (file->search.name, file->name.buffer,
					file->search.nameLength);
		else
			result = taguppercmp (file->search.name, file->name.buffer);
	}
	else
	{
		if (file->search.partial)
			result = tagncmp (file->search.name, file->name.buffer,
					file->search.nameLength);
		else
			result = tagcmp (file->search.name, file->name.buffer);
	}
	return result;
}

static tagResult findFirstNonMatchBefore (tagFile *const file)
{
#define JUMP_BACK 512
	int more_lines;
	int comp;
	rt_off_t start = file->pos;
	rt_off_t pos = start;
	do
	{
		if (pos < (rt_off_t) JUMP_BACK)
			pos = 0;
		else
			pos = pos - JUMP_BACK;
		more_lines = readTagLineSeek (file, pos);
		if (more_lines == 0 && file->err)
			return TagFailure;
		comp = nameComparison (file);
	} while (more_lines  &&  comp == 0  &&  pos > 0  &&  pos < start);
	return TagSuccess;
}

static tagResult findFirstMatchBefore (tagFile *const file)
{
	tagResult result = TagFailure;
	int more_lines;
	rt_off_t start = file->pos;
	if (findFirstNonMatchBefore (file) != TagSuccess)
		return TagFailure;
	do
	{
		more_lines = readTagLine (file, &file->err);
		if (more_lines == 0 && file->err)
			return TagFailure;
		if (nameComparison (file) == 0)
			result = TagSuccess;
	} while (more_lines  &&  result != TagSuccess  &&  file->pos < start);
	return result;
}

static tagResult findBinary (tagFile *const file)
{
	tagResult result = TagFailure;
	rt_off_t lower_limit = 0;
	rt_off_t upper_limit = file->size;
	rt_off_t last_pos = 0;
	rt_off_t pos = upper_limit / 2;
	while (result != TagSuccess)
	{
		if (! readTagLineSeek (file, pos))
		{
			if (file->err)
				break;
			/* in case we fell off end of file */
			result = findFirstMatchBefore (file);
			break;
		}
		else if (pos == last_pos)
		{
			/* prevent infinite loop if we backed up to beginning of file */
			break;
		}
		else
		{
			const int comp = nameComparison (file);
			last_pos = pos;
			if (comp < 0)
			{
				upper_limit = pos;
				pos = lower_limit + ((upper_limit - lower_limit) / 2);
			}
			else if (comp > 0)
			{
				lower_limit = pos;
				pos = lower_limit + ((upper_limit - lower_limit) / 2);
			}
			else if (pos == 0)
				result = TagSuccess;
			else
			{
				result = findFirstMatchBefore (file);
				if (result != TagSuccess && file->err)
					break;
			}
		}
	}
	return result;
}

static tagResult findSequentialFull (tagFile *const file,
									 int (* isAcceptable) (tagFile *const, void *),
									 void *data)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	tagResult result = TagFailure;
	while (result == TagFailure)
	{
		if (! readTagLine (file, &file->err))
			break;
		if (isAcceptable (file, data))
			result = TagSuccess;
	}
	return result;
}

static int nameAcceptable (tagFile *const file, void *unused)
{
	return (nameComparison (file) == 0);
}

static tagResult findSequential (tagFile *const file)
{
	return findSequentialFull (file, nameAcceptable, NULL);
}

static tagResult find (tagFile *const file, tagEntry *const entry,
					   const char *const name, const int options)
{
	tagResult result;
	if (file->search.name != NULL)
		free (file->search.name);
	file->search.name = duplicate (name);
	if (file->search.name == NULL)
	{
		file->err = ENOMEM;
		return TagFailure;
	}
	file->search.nameLength = strlen (name);
	file->search.partial = (options & TAG_PARTIALMATCH) != 0;
	file->search.ignorecase = (options & TAG_IGNORECASE) != 0;
	if (readtags_fseek (file->fp, 0, SEEK_END) < 0)
	{
		file->err = errno;
		return TagFailure;
	}
	file->size = readtags_ftell (file->fp);
	if (file->size == -1)
	{
		file->err = errno;
		return TagFailure;
	}
	if (readtags_fseek(file->fp, 0, SEEK_SET) == -1)
	{
		file->err = errno;
		return TagFailure;
	}
	if ((file->sortMethod == TAG_SORTED      && !file->search.ignorecase) ||
		(file->sortMethod == TAG_FOLDSORTED  &&  file->search.ignorecase))
	{
		result = findBinary (file);
		if (result == TagFailure && file->err)
			return TagFailure;
	}
	else
	{
		result = findSequential (file);
		if (result == TagFailure && file->err)
			return TagFailure;
	}

	if (result != TagSuccess)
		file->search.pos = file->size;
	else
	{
		file->search.pos = file->pos;
		result = (entry != NULL)
			? parseTagLine (file, entry, &file->err)
			: TagSuccess;
	}
	return result;
}

static tagResult findNextFull (tagFile *const file, tagEntry *const entry,
							   int sorted,
							   int (* isAcceptable) (tagFile *const, void *),
							   void *data)
{
	tagResult result;
	if (sorted)
	{
		result = tagsNext (file, entry);
		if (result == TagSuccess  && !isAcceptable (file, data))
			result = TagFailure;
	}
	else
	{
		result = findSequentialFull (file, isAcceptable, data);
		if (result == TagSuccess  &&  entry != NULL)
			result = parseTagLine (file, entry, &file->err);
	}
	return result;
}

static tagResult findNext (tagFile *const file, tagEntry *const entry)
{
	return findNextFull (file, entry,
						 (file->sortMethod == TAG_SORTED      && !file->search.ignorecase) ||
						 (file->sortMethod == TAG_FOLDSORTED  &&  file->search.ignorecase),
						 nameAcceptable, NULL);
}

static tagResult findPseudoTag (tagFile *const file, int rewindBeforeFinding, tagEntry *const entry)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	if (rewindBeforeFinding)
	{
		if (readtags_fseek(file->fp, 0, SEEK_SET) == -1)
		{
			file->err = errno;
			return TagFailure;
		}
	}
	return findNextFull (file, entry,
						 (file->sortMethod == TAG_SORTED || file->sortMethod == TAG_FOLDSORTED),
						 doesFilePointPseudoTag,
						 NULL);
}


/*
*  EXTERNAL INTERFACE
*/

extern tagFile *tagsOpen (const char *const filePath, tagFileInfo *const info)
{
	tagFileInfo infoDummy;
	return initialize (filePath, info? info: &infoDummy);
}

extern tagResult tagsSetSortType (tagFile *const file, const tagSortType type)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	switch (type)
	{
	case TAG_UNSORTED:
	case TAG_SORTED:
	case TAG_FOLDSORTED:
		file->sortMethod = type;
		return TagSuccess;
	default:
		file->err = TagErrnoUnexpectedSortedMethod;
		return TagFailure;
	}
}

extern tagResult tagsFirst (tagFile *const file, tagEntry *const entry)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	if (gotoFirstLogicalTag (file) != TagSuccess)
		return TagFailure;
	return readNext (file, entry);
}

extern tagResult tagsNext (tagFile *const file, tagEntry *const entry)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	return readNext (file, entry);
}

extern const char *tagsField (const tagEntry *const entry, const char *const key)
{
	const char *result = NULL;
	if (entry != NULL)
		result = readFieldValue (entry, key);
	return result;
}

extern tagResult tagsFind (tagFile *const file, tagEntry *const entry,
						   const char *const name, const int options)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	return find (file, entry, name, options);
}

extern tagResult tagsFindNext (tagFile *const file, tagEntry *const entry)
{
	if (file == NULL)
		return TagFailure;

	if (!file->initialized || file->err)
	{
		file->err = TagErrnoInvalidArgument;
		return TagFailure;
	}

	return findNext (file, entry);
}

extern tagResult tagsFirstPseudoTag (tagFile *const file, tagEntry *const entry)
{
	return findPseudoTag (file, 1, entry);
}

extern tagResult tagsNextPseudoTag (tagFile *const file, tagEntry *const entry)
{
	return findPseudoTag (file, 0, entry);
}

extern tagResult tagsFindPseudoTag (tagFile *const file, tagEntry *const entry,
									const char *const name, const int match)
{
	size_t len;
	tagEntry entry0;
	tagEntry *entryp = entry? entry: &entry0;

	tagResult r = tagsFirstPseudoTag (file, entryp);
	if (r != TagSuccess)
		return r;

	if (match & TAG_PARTIALMATCH)
		len = strlen (name);

	do
	{
		if (match & TAG_PARTIALMATCH)
		{
			if (strncmp (entryp->name, name, len) == 0)
				return TagSuccess;
		}
		else
		{
			if (strcmp (entryp->name, name) == 0)
				return TagSuccess;
		}
		r = tagsNextPseudoTag (file, entryp);
	}
	while (r == TagSuccess);

	return r;
}

extern tagResult tagsClose (tagFile *const file)
{
	tagResult result = TagFailure;
	if (file != NULL  &&  file->initialized)
	{
		terminate (file);
		result = TagSuccess;
	}
	return result;
}

extern int tagsGetErrno (tagFile *const file)
{
	if (file == NULL)
		return TagErrnoInvalidArgument;
	return file->err;
}