/* unpack.c -- David Bremner, 1991;
 * Changes to this file Copyright 1991, 1994 Liam R. Quin.
 * All Rights Reserved.
 * This code is NOT in the public domain.
 * See the file COPYRIGHT for full details.
 *
 * $Id: unpack.c,v 1.14 2001/05/30 20:42:16 liam Exp $
 *
 * Uncompress or dearchive a file and return a an open file descriptor to a 
 * copy of the file in /tmp
 *
 * If a filename contains a ':' it is assumed to be of the form 
 * archive-path:component
 *
 * Compression is determined via magic number
 *
 * Original by David Bremner (bremner@cs.sfu.ca).
 * Id: LQT_UnpackAndOpen.c,v 1.1 1991/06/05 18:39:47 bremner Exp 
 *
 */

#include "error.h"
#include <stdio.h>
#include <sys/types.h>
#include "globals.h"

#ifdef HAVE_STRING_H
# include <string.h>
#else
# include <strings.h>
#endif

#ifdef HAVE_STDLIB_H
# include <stdlib.h>
#endif

#ifdef HAVE_FCNTL_H
# ifdef HAVE_SYSV_FCNTL_H
#  include <sys/stat.h>
# endif
# include <fcntl.h>
#endif

#ifdef HAVE_UNISTD_H
# include <unistd.h>
#endif

#include "emalloc.h"
#include "fileinfo.h"
#include "liblqtext.h"
#include "lqtrace.h"

#define COMPRESS_MAGIC  "\037\235\220"
#define COMPRESS_LEN (sizeof(COMPRESS_MAGIC) - 1)
#define UNCOMPRESS_COMMAND "compress -d < '%s' > '%s'"
#define UNGZIP_COMMAND "gzip -d -c < '%s' > '%s'"
#define BUNZIP_COMMAND "bzip2 -d -c < '%s' > '%s'"
#define GZIP_MAGIC "\037\213"
#define GZIP_MAGIC_LEN (sizeof(GZIP_MAGIC) - 1)
#define AR_COMMAND "ar p '%s' '%s' > '%s'"
/* Archives are accessed as archivename(component), where the ( and )
 * characters are START_CHAR and END_CHAR respectively.  Define END_CHAR to
 * be \0 to disable it.
 * TODO: combine with LQT_GetFilterType and make dynamically configured.
 */
#define START_CHAR '('
#define END_CHAR ')'

PRIVATE int DoCommand(
#ifdef HAVE_PROTO
    t_LQTEXT_Database *db,
    char *Command,
    char *InputFile
#endif
);

PRIVATE int UnpackArchive(
#ifdef HAVE_PROTO
    t_LQTEXT_Database *,
    char *,
    char *
#endif
);

extern char *mktemp(
#ifdef HAVE_PROTO
    char *pattern
#endif
);

/* <Function>
 *   <Name>LQT_UnpackAndOpen
 *   <Class>Database/Documents
 *   <Purpose>
 *      Tries to open the named file, using compress or gunzip as
 *	necessary.  Can append a .Z or .gz to the file name.
 *	Currently, LQT_UnpackAndOpen makes a copy of a file if necessary;
 *	a future version may create a pipe, and the interface will change.
 *   <Returns>
 *      <LIST>
 *        <LI>an open file descriptor on success;
 *        <LI>-1 if the file couldn't be opened.
 *	</LIST>
 *   <SeeAlso>
 *	LQT_FindFile
 *	LQT_MakeInput
 * </Function>
 */
API int
LQT_UnpackAndOpen(db, FileName)
    t_LQTEXT_Database *db;
    char *FileName;
{
    int fd;
    char magic_buffer[4];
    char *Component;
  
    if ((fd = open(FileName, O_RDONLY, 0)) < 0) {
	char *CompressedName = emalloc(
	    "LQT_UnpackAndOpen filename",
	    strlen(FileName) + 5 /* for .bz2 + \0 */
	);

	/* gzip? */
	(void) sprintf(CompressedName, "%s.gz", FileName);
	if ((fd = open(CompressedName, O_RDONLY, 0)) >= 0) {
	    (void) close(fd);
	    return DoCommand(db, UNGZIP_COMMAND, CompressedName);
	}

	/* compressed */
	(void) sprintf(CompressedName, "%s.Z", FileName);
	if ((fd = open(CompressedName, O_RDONLY, 0)) >= 0) {
	    (void) close(fd);
	    return DoCommand(db, UNCOMPRESS_COMMAND, CompressedName);
	}

	/* or with bzip2? */
	(void) sprintf(CompressedName, "%s.bz2", FileName);
	if ((fd = open(CompressedName, O_RDONLY, 0)) >= 0) {
	    (void) close(fd);
	    return DoCommand(db, BUNZIP_COMMAND, CompressedName);
	}

	/* Check to see if it could be a name of the form
	 * archivename(component);
	 * Archive Component
	 */
	if ((Component = strchr(FileName, START_CHAR)) != (char *) 0) {
	    char *ArchiveName = emalloc(
		"LQT_UnpackAndOpen archive name",
		Component - FileName + 1
	    );
	    char *ArchiveLocation = 0; /* search filepath */
	    char *p;
	    int Newfd;

	    if (!ArchiveName) return -1;
	    if (END_CHAR && FileName[strlen(FileName) - 1] != END_CHAR) {
		(void) efree(ArchiveName);
		return -1;
	    }

	    (void) strncpy(ArchiveName, FileName, Component - FileName);
	    ArchiveName[Component - FileName] = '\0';
	    ArchiveLocation = LQT_FindFile(db, ArchiveName);

	    if (!ArchiveLocation) {
		(void) efree(ArchiveName);
		return -1;
	    }
	    ++Component; /* skip over FIRST_CHAR */
	    if (END_CHAR) {
		char *q;
		p = emalloc(
		    "LQT_UnpackAndOpen ar component",
		    strlen(Component) + 1
		);
		(void) strcpy(p, Component);
		q = strrchr(p, END_CHAR);
		if (q && q[1] == '\0') {
		    *q = '\0';
		}
	    } else {
		p = emalloc("LQT_UnpackAndOpen component", strlen(Component) + 1);
		(void) strncpy(p, Component, strlen(Component));
	    }
	    Component = p;
#ifdef ASCIITRACE
	    LQT_Trace(LQTRACE_FINDFILE,
		"archive %s; Component %s",
		ArchiveLocation,
		Component
	    );
#endif
	    (void) close(fd);
	    Newfd = UnpackArchive(db, ArchiveLocation, Component);
	    (void) efree(ArchiveName);
	    (void) efree(Component);
	    return Newfd;
	}
	return -1;
    }

    if (read(fd, magic_buffer, COMPRESS_LEN) == COMPRESS_LEN) {

	if (strncmp(magic_buffer, COMPRESS_MAGIC, COMPRESS_LEN) == 0) {
	    (void) close(fd);
	    return DoCommand(db, UNCOMPRESS_COMMAND, FileName);
	} else if (strncmp(magic_buffer, GZIP_MAGIC, GZIP_MAGIC_LEN) == 0) {
	    (void) close(fd);
	    return DoCommand(db, UNGZIP_COMMAND, FileName);
	}
    }


    /* If we fall thorough to here, it is just a regular file */

    /* reset file pointer */
    if (lseek(fd,0L,SEEK_SET) < 0 ) {
	(void) close(fd);
	return -1;
    } else {
	return fd;
    }
/*NOTREACHED*/
}

PRIVATE int
UnpackArchive(db, ArchiveLocation, Component)
    t_LQTEXT_Database *db;
    char *ArchiveLocation; /* a full pathname */
    char *Component;
{
    int fd;
    char *Tmp = emalloc(
	"ArchiveLocation",
	strlen(ArchiveLocation) + strlen(AR_COMMAND)
    );

    (void) sprintf(Tmp, AR_COMMAND, ArchiveLocation, "%s", "%s");
    fd = DoCommand(db, Tmp, Component);
    (void) efree(Tmp);
    return fd;
}

PRIVATE int
DoCommand(db, Command, InputFile)
    t_LQTEXT_Database *db;
    char *Command;
    char *InputFile;
{
    char *template;
    char *Buffer;
    int fd;
    char *tmpdir = getenv("TMPDIR");

    if (!tmpdir) {
	tmpdir = "/tmp";
    }

#define PATTERN "/lqtextZZXXXXXX"

    template = emalloc("DoComamnd template", strlen(tmpdir) + sizeof(PATTERN));

    (void) strcpy(template, tmpdir);
    (void) strcat(template, PATTERN);
    (void) mktemp(template);

    Buffer = emalloc(
	"DoCommand buffer",
	strlen(Command) + strlen(InputFile) + strlen(template)
    );

    (void) sprintf(Buffer, Command, InputFile, template);
#ifdef ASCIITRACE
    LQT_Trace(LQTRACE_FINDFILE,
	"findfile/Unpack command: %s",
	Buffer
    );
#endif

    /* System() will create the file, so we have not to
     * call Error(E_FATAL...) until we have ensured that
     * we have removed the temporary file.
     * An interrupt would also leave the file lying around.
     * However, we need to keep the file around for long enough to
     * unpack it, so there's not much we can do.
     */
    system(Buffer);
    (void) efree(Buffer);

    fd = LQT_UnpackAndOpen(db, template);
    (void) unlink(template); 
    efree(template);
    return fd;
}