hcache.c #1

/*
 * This file has been donated to Jam.
 */

# include "jam.h"
# include "lists.h"
# include "parse.h"
# include "rules.h"
# include "regexp.h"
# include "headers.h"
# include "newstr.h"
# include "hash.h"
# include "hcache.h"
# include "variable.h"
# include "search.h"

#ifdef OPT_HEADER_CACHE_EXT

/*
 * Craig W. McPheeters, Alias|Wavefront.  Nov/2001.
 * Jan/2002.  Extensions by Matt Armstrong.
 *    See the file README.header_scan_cache for details on the extensions.
 * Jan/2002.  Modification to extensions by Craig.
 *
 * hcache.c, hcache.h - handle cacheing of #includes in source files
 *
 * Create a cache of files scanned for headers.	 When starting jam,
 * look for the cache file and load it if present.  When finished the
 * binding phase, create a new header cache.  The cache contains
 * files, their timestamps and the header files found in their scan.
 * During the binding phase of jam, look in the header cache first for
 * the headers contained in a file.  If the cache is present and
 * valid, use its contents.  This can result in dramatic speedups on
 * large projects (eg. 3min -> 1min startup on one project.)
 *
 * External routines:
 *    hcache_init() - read and parse the local .jamdeps file.
 *    hcache_done() - write a new .jamdeps file
 *    hcache() - return list of headers on target.  Use cache or do a scan.
 *    
 * The dependency file format is an ascii file with 1 line per target.
 * Each line has the following fields:
 * @boundname@ timestamp age num @file@ @file@ num @hdrscan@ @hdrscan@ ... \n
 *   where the first number is the number of headers, and the second is the
 *   number of elements in the hdrscan list.  
 *
 * Filenames may contain any ascii or non-ascii characters.  If they
 * contain the '@' or '#' characters, they are quoted on output and
 * their quoting is handled on input.  Often the '\' character is used
 * for quoting, but as that is so common in NT pathnames, the '#'
 * character is used instead. Both '@' and '#' are characters
 * disallowed in Perforce filenames - and they should be rare in other
 * SCM systems hopefully.  CWM.
 */

struct hcachedata {
    const char		*boundname;
    time_t		time;
    LIST		*includes;
    LIST		*hdrscan; /* the HDRSCAN value for this target */
    int			age;	  /* if too old, we'll remove it from cache */
    struct hcachedata	*next;
} ;

typedef struct hcachedata HCACHEDATA ;


static struct hash *hcachehash = 0;
static HCACHEDATA  *hcachelist = 0; 

static int queries = 0;
static int hits = 0;

#define CACHE_FILE_VERSION "version 1"

/*
 * Return the name of the header cache file.  May return NULL.
 *
 * The user sets this by setting the HCACHEFILE variable in a Jamfile.
 * We cache the result so the user can't change the cache file during
 * header scanning.
 */
static const char*
cache_name(void)
{
    static const char *name = 0;
    if( !name ) {
	LIST *hcachevar = var_get( "HCACHEFILE" );

	if( hcachevar ) {
	    TARGET *t = bindtarget( hcachevar->string );

	    pushsettings( t->settings );
	    t->boundname = search( t->name, &t->time );
	    popsettings( t->settings ); 
	    name = copystr( t->boundname );
	}
    }
    return name;
}

/*
 * Return the maximum age a cache entry can have before it is purged
 * from the cache.
 *
 * A maxage of 0 indicates that the cache entries should never be
 * purged, in effect disabling the aging of cache entries.
 */
static int cache_maxage(void) {
    int age = 100;
    LIST *var = var_get( "HCACHEMAXAGE" ); 
    if( var ) {
	age = atoi( var->string );
	if( age < 0 ) age = 0;
    } 
    return age;
}

/*
 * Read any spaces we're on.  Return the first non-space character
 */
static int skip_spaces( FILE *f ) {
    int ch = fgetc( f ); 
    while( ch == ' ' ) ch = fgetc( f ); 
    return ch;
}

/*
 * Read a string from the file.	 Handle quoted characters.  The
 * returned value is as returned by newstr(), so it need not be freed.	
 */
static const char * read_string( FILE *f ) {
    int ch, i = 0;
    char filename[ MAXJPATH ]; 
    ch = skip_spaces( f );
    if( ch != '@' ) return 0; 
    ch = fgetc( f );
    while( ch != '@' && ch != EOF && i < MAXJPATH ) {
	if( ch == '#' ) /* Quote */
	     filename[ i++ ] = fgetc( f );
	else filename[ i++ ] = ch;
	ch = fgetc( f );
    } 
    if( ch != '@' ) return 0; 
    filename[ i ] = 0;
    return newstr( filename );
}

static int read_int( FILE *f ) {
    int	 ch, i = 0;
    char num[ 30 ]; 
    ch = skip_spaces( f );
    while( ch >= '0' && ch <= '9' ) {
	num[ i++ ] = ch;
	ch = fgetc( f );
    }
    num[ i ] = 0; 
    return atoi( num );
}

static void write_string( FILE *f, const char *s ) {
    int i = 0; 
    fputc( '@', f );
    while( s[ i ] != 0 ) {
	if( s[ i ] == '@' || s[ i ] == '#' ) fputc( '#', f ); /* Quote */
	fputc( s[ i++ ], f );
    }
    fputs( "@ ", f );
}

static void write_int( FILE *f, int i ) { fprintf( f, "%d ", i ); }

void hcache_init(void) {
    HCACHEDATA	cachedata, *c, *last = 0;
    FILE	*f;
    int		bad_cache = 1, ch;
    const char	*version, *hcachename; 
    hcachehash = hashinit( sizeof( HCACHEDATA ), "hcache" ); 
    if( ! (hcachename = cache_name()) ) return; 
    if( ! (f = fopen( hcachename, "rb" )) ) return; 
    version = read_string( f );
    ch = fgetc( f );
    if (!version || strcmp( version, CACHE_FILE_VERSION ) || ch != '\n' ) { goto bail;
    } 
    for(;;) {
	int i, count, ch;
	LIST *l; 
	c = &cachedata; 
	c->boundname = read_string( f );
	if( !c->boundname ) /* Test for eof */ break; 
	c->time = read_int( f );
	c->age = read_int( f ) + 1; /* we're getting older... */ 
	if( !c->boundname ) goto bail; 
	/* headers */
	count = read_int( f );
	for( l = 0, i = 0; i < count; ++i ) {
	    const char *s = read_string( f );
	    if( !s ) goto bail;
	    l = list_new( l, s, 0 );
	}
	c->includes = l; 
	/* hdrscan */
	count = read_int( f );
	for( l = 0, i = 0; i < count; ++i ) {
	    const char *s = read_string( f );
	    if( !s ) goto bail;
	    l = list_new( l, s, 0 );
	}
	c->hdrscan = l; 
	/* Read the newline */
	ch = skip_spaces( f );
	if( ch != '\n' ) goto bail; 
	if( !hashenter( hcachehash, (HASHDATA **)&c ) ) {
	    printf( "jam: can't insert header cache item, bailing on %s\n", hcachename );
	    goto bail;
	} 
	c->next = 0;
	if( last ) last->next = c;
	else hcachelist = c;
	last = c;
    } 
    bad_cache = 0; 
    if( DEBUG_HEADER ) printf( "hcache read from file %s\n", hcachename ); 
 bail:
    /* If its bad, no worries, it'll be overwritten in hcache_done() */
    if( bad_cache ) printf( "jam: warning: the cache was invalid: %s\n", hcachename ); 
    fclose( f );
}

void hcache_done(void) {
    FILE	*f;
    HCACHEDATA	*c;
    int		header_count = 0;
    const char *hcachename;
    int		maxage; 
    if( !hcachehash ) return; 
    if( ! (hcachename = cache_name()) ) return; 
    if( ! (f = fopen( hcachename, "wb" ) ) ) return; 
    maxage = cache_maxage(); 
    /* print out the version */
    fprintf( f, "@%s@\n", CACHE_FILE_VERSION ); 
    c = hcachelist;
    for( c = hcachelist; c; c = c->next ) {
	LIST	*l; 
	if( maxage == 0 ) c->age = 0;
	else if( c->age > maxage ) continue; 
	write_string( f, c->boundname );
	write_int( f, c->time );
	write_int( f, c->age ); 
	write_int( f, list_length( c->includes ) );
	for( l = c->includes; l; l = list_next( l ) ) {
	    write_string( f, l->string );
	} 
	write_int( f, list_length( c->hdrscan ) );
	for( l = c->hdrscan; l; l = list_next( l ) ) {
	    write_string( f, l->string );
	} 
	fputc( '\n', f );
	++header_count;
    } 
    if( DEBUG_HEADER )
	printf( "hcache written to %s.	 %d dependencies, %.0f%% hit rate\n",
	       hcachename, header_count,
	       queries ? 100.0 * hits / queries : 0 ); 
    fclose( f );
}

LIST * hcache( TARGET *t, LIST *hdrscan ) {
    HCACHEDATA	cachedata, *c = &cachedata;
    LIST	*l = 0;
    int		use_cache = 1; 
    ++queries; 
    c->boundname = t->boundname; 
    if( hashcheck( hcachehash, (HASHDATA **) &c ) ) {
	if( c->time == t->time ) {
	    LIST *l1 = hdrscan, *l2 = c->hdrscan;
	    while( l1 && l2 ) {
		if( l1->string != l2->string ) {
		    l1 = 0;
		} else {
		    l1 = list_next( l1 );
		    l2 = list_next( l2 );
		}
	    }
	    if( l1 || l2 ) use_cache = 0;
	} else use_cache = 0; 
	if( use_cache ) {
	    if( DEBUG_HEADER ) printf( "using header cache for %s\n", t->boundname );
	    c->age = 0; /* The entry has been used, its young again */
	    ++hits;
	    l = list_copy( 0, c->includes );
	    return l;
	} else {
	    if( DEBUG_HEADER ) printf( "header cache out of date for %s\n", t->boundname );
	    list_free( c->includes );
	    list_free( c->hdrscan );
	    c->includes = 0;
	    c->hdrscan = 0;
	}
    } else {
	if( hashenter( hcachehash, (HASHDATA **)&c ) ) {
	    c->boundname = newstr( c->boundname );
	    c->next = hcachelist;
	    hcachelist = c;
	}
    } 
    /* 'c' points at the cache entry.  Its out of date. */ 
    l = headers1( t->boundname, hdrscan ); 
    c->time = t->time;
    c->age = 0;
    c->includes = list_copy( 0, l );
    c->hdrscan = list_copy( 0, hdrscan ); 
    return l;
} 
#endif