copyout.cc

Go to the documentation of this file.

/*
 * Copyright (C) 1996-2023 The Squid Software Foundation and contributors
 *
 * Squid software is distributed under GPLv2+ license and includes
 * contributions from numerous individuals and organizations.
 * Please see the COPYING and CONTRIBUTORS files for details.
 */
 
// Author:  Jens-S. V?ckler <voeckler@rvs.uni-hannover.de>
//
// File:    copyout.cc
//          Tue Jun 15 1999
//
// (c) 1999 Lehrgebiet Rechnernetze und Verteilte Systeme
//          Universit?t Hannover, Germany
//
// Permission to use, copy, modify, distribute, and sell this software
// and its documentation for any purpose is hereby granted without fee,
// provided that (i) the above copyright notices and this permission
// notice appear in all copies of the software and related documentation,
// and (ii) the names of the Lehrgebiet Rechnernetze und Verteilte
// Systeme and the University of Hannover may not be used in any
// advertising or publicity relating to the software without the
// specific, prior written permission of Lehrgebiet Rechnernetze und
// Verteilte Systeme and the University of Hannover.
//
// THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF ANY KIND,
// EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY
// WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
//
// IN NO EVENT SHALL THE LEHRGEBIET RECHNERNETZE UND VERTEILTE SYSTEME OR
// THE UNIVERSITY OF HANNOVER BE LIABLE FOR ANY SPECIAL, INCIDENTAL,
// INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND, OR ANY DAMAGES
// WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER OR NOT
// ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF LIABILITY,
// ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
// SOFTWARE.
//
// Revision 1.2  1999/06/16 13:05:26  voeckler
// mmap file copying on Solaris.
//
// Revision 1.1  1999/06/15 21:10:47  voeckler
// Initial revision
//
//
#include "squid.h"
#include "copyout.hh"
 
#include <sys/stat.h>
#include <cerrno>
#include <cstring>
#include <fcntl.h>
#include <unistd.h>
#include <sys/mman.h>
 
#ifndef MAP_FILE
#define MAP_FILE 0
#endif // MAP_FILE
 
int
assert_copydir( const char* copydir )
// purpose: check, if copydir is a directory and that we can write into it.
// paramtr: copydir (IN): name of directory for copying bodies.
// returns: 0 if ok, -1 otherwise.
// further: errors are handled within. If the directory does not exist,
//          the assertion function will try to create it.
{
    struct stat st;
    int status = stat( copydir, &st );
 
    // check, if either "copydir" points to a valid directory,
    // or if copydir can be created.
    if ( status == 0 && ! S_ISDIR(st.st_mode) ) {
        // stat() returned true, but did not point to a directory
        fprintf( stderr, "copy dir \"%s\" is a file!\n", copydir );
        return -1;
    } else if ( S_ISDIR(st.st_mode) &&
                !( (st.st_uid == geteuid() && ( (st.st_mode & S_IWUSR) > 0 )) ||
                   (st.st_gid == getegid() && ( (st.st_mode & S_IWGRP) > 0 )) ||
                   ((st.st_mode & S_IWOTH) > 0) ) ) {
        fprintf( stderr, "copy dir \"%s\" is not accessible to me\n", copydir );
        return -1;
    }
    if ( status == -1 ) {
        // stat() returned with an error. 'File not found' is a legal error.
        if ( errno != ENOENT ) {
            // not a 'file not found' error, so this is hard error.
            fprintf( stderr, "accessing copy-out dir \"%s\": %s\n",
                     copydir, strerror(errno) );
            return -1;
        } else {
            // directory does not exist. Try to create it.
            if ( mkdir( copydir, 0750 ) == -1 ) {
                fprintf( stderr, "mkdir(%s): %s\n", copydir, strerror(errno) );
                return -1;
            }
        }
    }
 
    // postcondition: copydir exists and is a directory.
    return 0;
}
 
inline
unsigned
xlate( char ch )
{
    if ( ch == '\r' ) return 0u;
    else if ( ch == '\n' ) return 1u;
    else return 2u;
}
 
// shortcut for monotoneous typings...
#define BAUTZ(x) delete[] filename; close(input); close(out); return (x)
 
bool
copy_out( size_t filesize, size_t metasize, unsigned debug,
          const char* fn, const char* url, const char* copydir,
          bool copyHdr )
// purpose: copy content from squid disk file into separate file
// paramtr: metasize (IN): size of metadata to skip
//          fn (IN): current filename of squid disk file
//          url (IN): currently looked at URL to generate separate file
//          copydir (IN): base directory where to generate the file
//          copyHdr (IN): copy HTTP header, too, if set to true.
// returns: true, if successful, false otherwise.
{
    static const char* index = "index.html";
 
    // find hostname part after the scheme (okok, not counting port, etc.)
    const char* ptr = strstr( url, "://" );
    if ( ptr == nullptr || strlen(ptr) < 4 ) return false;
 
    // create filename to store contents into
    // NP: magic extra 5 bytes for the component delimiter and termination octets
    char *filename = new char[ strlen(ptr) + strlen(copydir) + strlen(index) +5 ];
    assert( filename != nullptr );
    strcpy( filename, copydir );
    strcat( filename, "/" );
    char* here = filename + strlen(filename);
    strcat( filename, ptr+3 );
 
    // handle server root (e.g. "http://www.focus.de" )
    if ( strchr( ptr+3, '/' ) == nullptr ) strcat( filename, "/" );
 
    // handle directories (e.g. "http://www.focus.de/A/" )
    if ( filename[strlen(filename)-1] == '/' ) strcat( filename, index );
 
    // create subdirectory structure
    for ( char* t = strchr(here,'/'); t; t = strchr(t,'/') ) {
        *t = 0;
        if ( mkdir( filename, 0775 ) == -1 && errno != EEXIST ) {
            fprintf( stderr, "mkdir(%s): %s\n", filename, strerror(errno) );
            delete[] filename;
            return false;
        } else if ( debug & 0x02 ) {
            fprintf( stderr, "# creating %s\n", filename );
        }
        *t = '/';
        ++t;
    }
 
    // create file
    int out = open( filename, O_CREAT | O_RDWR | O_TRUNC, 0664 );
    if ( out == -1 ) {
        fprintf( stderr, "open(%s,RDWR): %s\n", filename, strerror(errno) );
        delete[] filename;
        return false;
    } else if ( debug & 0x02 ) {
        fprintf( stderr, "# creating %s\n", filename );
    }
 
    // (re)open cache file
    int input = open( fn, O_RDONLY );
    if ( input == -1 ) {
        fprintf( stderr, "open(%s,RDONLY): %s\n", fn, strerror(errno) );
        delete[] filename;
        close(out);
        return false;
    }
 
    // find double CRLF sequence (actually, look at the FSM below)
    // XXX: this only looks at the already known buffer read previously,
    // which is globally passed (yuck)! As a limitation, the content data
    // *must* begin within the buffer size (that is: 16k)!
    if ( ! copyHdr ) {
        extern char*  linebuffer; // import from purge.cc
        extern size_t buffersize; // import from purge.cc
 
        unsigned state = 0;
        char* s = linebuffer + metasize;
        while ( s < linebuffer + buffersize && state < 4 ) {
            // state transition machine
            static unsigned table[4][3] = { {3,2,0}, {0,4,0}, {1,4,0}, {4,2,0} };
            //  old || \r | \n |else|
            // =====++====+====+====+
            //    0 ||  3 |  2 |  0 |
            //    1 ||  0 |  4 |  0 |
            //    2 ||  1 |  4 |  0 |
            //    3 ||  4 |  2 |  0 |
            state = table[ state ][ xlate(*s) ];
            ++s;
        }
 
        if ( state < 4 )
            // complain bitterly, if the HTTP header was too large ( > 16k ).
            fprintf( stderr, "WARNING: %s will contain partial HTTP header data!\n",
                     filename );
 
        // adjust to different seek size
        metasize = s - linebuffer;
    }
 
    // no need to copy zero content files
    if ( filesize - metasize <= 0 ) {
        BAUTZ( filesize-metasize == 0 );
    }
 
#ifdef USE_REGULAR_COPY
    // position input at start of server answer (contains HTTP headers)
    if ( lseek( input, metasize, SEEK_SET ) == -1 ) {
        fprintf( stderr, "lseek(%s,%lu): %s\n", fn, metasize, strerror(errno) );
        BAUTZ(false);
    }
 
    // file copy input into output via buffer (regular io)
    char buffer[32768];
    int rsize, wsize;
    while ( (rsize=read(input,buffer,sizeof(buffer))) > 0 ) {
        if ( (wsize=write(out,buffer,rsize)) <= 0 ) break;
    }
    if ( rsize < 0 || wsize < 0 ) perror( "while copying" );
#else // use mmap copy (compare: Stevens APUE 12.9)
    // precondition: filesize-metasize > 0
    // seek end of output file ...
    off_t position = lseek( out, filesize-metasize-1, SEEK_SET );
    if ( position == -1 ) {
        fprintf( stderr, "lseek(%s,%lu): %s\n", filename,
                 (unsigned long)filesize-metasize,
                 strerror(errno) );
        BAUTZ(false);
    } else if ( debug & 0x02 ) {
        fprintf( stderr, "# filesize=%lu, metasize=%lu, filepos=%ld\n",
                 (unsigned long)filesize, (unsigned long)metasize,
                 (long)position );
    }
 
    // ...and write 1 byte there (create a file that length)
    if ( write( out, "", 1 ) != 1 ) {
        perror( "write to output" );
        BAUTZ(false);
    }
 
    // create source mmap to copy from (mmap complete file)
    const auto src = static_cast<char *>(mmap(nullptr, filesize, PROT_READ, MAP_FILE | MAP_SHARED, input, 0));
    if (src == reinterpret_cast<const char *>(-1)) {
        perror( "mmap input" );
        BAUTZ(false);
    }
 
    // create destination mmap to copy into (mmap data portion)
    auto dst = static_cast<char *>(mmap(nullptr, filesize-metasize, PROT_READ | PROT_WRITE, MAP_FILE | MAP_SHARED, out, 0));
    if (dst == reinterpret_cast<char *>(-1)) {
        perror( "mmap output" );
        munmap( src, filesize );
        BAUTZ(false);
    }
 
    // copy file (beware of offset into wanted data, skip meta data)
    memcpy( dst, src+metasize, filesize-metasize );
 
    // clean up
    munmap( dst, filesize-metasize );
    munmap( src, filesize );
#endif // USE_REGULAR_COPY
 
    BAUTZ(true);
}