[libimdb-commit] CVS: libimdb/parser Makefile,NONE,1.1 imdbDict.c,NONE,1.1 imdbParser.c,NONE,1.1 private.h,NONE,1.1

[Jerry V. <jve...@us...>]

Update of /cvsroot/libimdb/libimdb/parser
In directory sc8-pr-cvs1:/tmp/cvs-serv11812/parser

Added Files:
	Makefile imdbDict.c imdbParser.c private.h 
Log Message:
initial checkin

--- NEW FILE: Makefile ---
#
# $Id: Makefile,v 1.1 2003/06/16 04:18:38 jveldhuis Exp $
#
include ../config.mk
include ../constants.mk

CPPFLAGS += -I../include

LIB_BASE    = imdbParser
LIB_VERSION  = 1.0

build:
	$(MAKE) $(MYLIBRARY) LIBRARY=$(MYLIBRARY)
	$(MAKE) $(MYDYNAMIC) DYNAMIC=$(MYDYNAMIC) 

install:; @true;

include ../rules.mk

--- NEW FILE: imdbDict.c ---
/**
 * @file
 * @verbatim $Id: imdbDict.c,v 1.1 2003/06/16 04:18:38 jveldhuis Exp $ @endverbatim 
 */

static char rcs_id[] = "$Id: imdbDict.c,v 1.1 2003/06/16 04:18:38 jveldhuis Exp $";

#define _USE_STDIO
#define _USE_STDLIB
#define _USE_ERRNO
#define _USE_IO
#define _USE_ASSERT
#define _USE_STDARG
#define _USE_STRINGS
#define _USE_TYPES
#define _USE_CTYPE
#define _USE_MALLOC
#define _USE_LIMITS
#define _USE_TIME

#include "sysincludes.h"
#include "jlog/dicerr.h"
#include "libimdb/parser.h"
#include "libimdb/dict.h"
/*#include "private.h"*/

struct imdbDict_s
{
    int fd;
    char dir[1024];
    imdbParser_t *parser;
};

static inline void
writeIt(int fd, char *in)
{
    char *i=in;
    
    while(*i != '\0') {
	if ( isalnum(*i) || *i == '_' || *i=='.' || *i=='-') {
	    write(fd, i, 1);
	}
	else {
	    char buf[100];

	    sprintf(buf, "%%%02x", (unsigned int)*i);
	    write(fd, buf, strlen(buf));
	}
	i++;
    }
}

/**
 * move ending ', The', articles to front of title name.
 * articles include The,A,Une,Les,L\',Le,La,El,Das.
 */
static void
titleMoveArticlesToFront(const char *title, char *output)
{
    int len=strlen(title);
    const char *end=title+len;

    if ( strncmp(end-5, ", The", 5)==0 ||
	 strncmp(end-5, ", Une", 5)==0 ||
	 strncmp(end-5, ", Les", 5)==0 ||
	 strncmp(end-5, ", Das", 5)==0 ) {
	output[0]=*(end-3);
	output[1]=*(end-2);
	output[2]=*(end-1);
	output[3]=' ';
	strncpy(output+4, title, len-5);
	output[len-1]='\0';
    }
    else if ( strncmp(end-3, ", A", 3)==0 ) {
	output[0]='A';
	output[1]=' ';
	strncpy(output+2, title, len-3);
	output[len-1]='\0';
    }
    else if ( strncmp(end-4, ", L\'", 4)==0 ||
	      strncmp(end-4, ", Le", 4)==0 ||
	      strncmp(end-4, ", La", 4)==0 ||
	      strncmp(end-4, ", El", 4)==0 ) {
	output[0]=*(end-2);
	output[1]=*(end-1);
	output[2]=' ';
	strncpy(output+3, title, len-4);
	output[len-1]='\0';
    }
    else {
	strcpy(output, title);
    }
}

static int
listentry_cb(jlogHandle_t *logh,
	     imdbParser_t *imdbh,
	     const ListFile_e lf,
	     const void *entry)
{
    imdbDict_t *dict=imdbParser_getUserHandle(logh, imdbh);
    
    switch(lf) {
    case LIST_MOVIES:
	{
	    const struct MovieInList_s *data=entry;
	    
	    write(dict->fd, data->title, strlen(data->title));
	    write(dict->fd, "\t", 1);

	    switch(data->type) {
	    case isMovie:	write(dict->fd, "\t1", 2);
		break;
	    case isTVSeries:	write(dict->fd, "\t2", 2);
		break;
	    case isTVMovie:	write(dict->fd, "\t3", 2);
		break;
	    case isVideoMovie:	write(dict->fd, "\t4", 2);
		break;
	    case isVideoGame:	write(dict->fd, "\t5", 2);
		break;
	    case isTVMiniSeries:write(dict->fd, "\t6", 2);
		break;
	    }
	    write(dict->fd, "\n", 1);
	    if ( data->number > 1000 ) {
		/* stop */
		/*return(-1);*/
	    }
	}
	break;
    case LIST_DIRECTORS:
	{
	    const struct DirectorInList_s *data=entry;
	}
	break;
    case LIST_ACTORS:
	{
	    const struct ActorInList_s *data=entry;
	}
	break;
    case LIST_ACTRESSES:
	{
	    const struct ActorInList_s *data=entry;
	}
	break;
    }
    return(0);
}

imdbDict_t *
imdbDict_create(jlogHandle_t *logh,
		const char *dbsDir,
		const char *dictDir)
{
    imdbDict_t *dict;

    dict=malloc(sizeof(struct imdbDict_s));
    if ( dict == NULL ) {
	DICE_BAD_MEM_FAILURE(logh, imdbDict_t *, NULL);
    }

    dict->parser=imdbParser_open(logh, dbsDir, listentry_cb);
    if ( dict->parser == NULL ) {
	free(dict);
	DIC_FAILURE(logh, imdbDict_t *, NULL);
    }
    imdbParser_setUserHandle(logh, dict->parser, dict);
    dict->fd=-1;
    strcpy(dict->dir, dictDir);

    return(dict);
}

int
imdbDict_destroy(jlogHandle_t *logh,
		 imdbDict_t *dict)
{
    imdbParser_close(logh, dict->parser);
    free(dict);
    DIC_SUCCESS(logh, int, 0);
}

int
imdbDict_indexMovies(jlogHandle_t *logh,
		     imdbDict_t *dict)
{
    char file[1024];

    sprintf(file, "%s/moviedb.idx", dict->dir);

    dict->fd=open(file, O_CREAT|O_TRUNC|O_WRONLY, 0666);
    if ( dict->fd == -1 ) {
	DICE_FILE_OPEN_FAILURE(logh, file, strerror(errno), int, -1);
    }
    
    /* -1 means stop, -2 means abort */
    if ( imdbParser_moviesParse(logh, dict->parser) < -1 ) {
	close(dict->fd);
	unlink(file);
	DIC_FAILURE(logh, int, -1);
    }
    close(dict->fd);
    return(0);
}

--- NEW FILE: imdbParser.c ---
/**
 * @file
 * @verbatim $Id: imdbParser.c,v 1.1 2003/06/16 04:18:38 jveldhuis Exp $ @endverbatim 
 */

static char rcs_id[] = "$Id: imdbParser.c,v 1.1 2003/06/16 04:18:38 jveldhuis Exp $";

#define _USE_STDIO
#define _USE_STDLIB
#define _USE_ERRNO
#define _USE_IO
#define _USE_ASSERT
#define _USE_STDARG
#define _USE_STRINGS
#define _USE_TYPES
#define _USE_CTYPE
#define _USE_MALLOC
#define _USE_LIMITS
#define _USE_TIME
[...1333 lines suppressed...]
					   LIST_DIRECTORS));
}

int
imdbParser_actorsParse(jlogHandle_t *logh,
		       imdbParser_t *parser)
{
    return(imdbParser_castOrDirectorsParse(logh,
					   parser,
					   LIST_ACTORS));
}

int
imdbParser_actressesParse(jlogHandle_t *logh,
			  imdbParser_t *parser)
{
    return(imdbParser_castOrDirectorsParse(logh,
					   parser,
					   LIST_ACTRESSES));
}

--- NEW FILE: private.h ---
/**
 * @file
 * @verbatim $Id: private.h,v 1.1 2003/06/16 04:18:38 jveldhuis Exp $ @endverbatim 
 */

/*typedef void (FormatError_cb_t)(jlogHandle_t *logh,
				imdbParser_t *parser,
				const char *pathname,
				int lineNumber,
				const char *message);*/

typedef struct movieStats_s {
    int numTVMiniSeries;
    int numMovie;
    int numTVSeries;
    int numTVMovie;
    int numVideoMovie;
    int numVideoGame;
} movieStats_t;

struct imdbParser_s
{
    void *userHandle;
    char *dbsDir;
    
    /*FormatError_cb_t *formatError_cb;*/

    progressStarting_cb_t *progressStarting_cb;
    progressUpdate_cb_t *progressUpdate_cb;
    progressFinished_cb_t *progressFinished_cb;

    ListEntry_cb_t *listEntry_cb;

    struct {
	movieStats_t movieList;

	struct {
	    int numDirectors;
	    movieStats_t movies;
	} directorsList;

	struct {
	    int numActors;
	    movieStats_t movies;
	} actorsList;

	struct {
	    int numActresses;
	    movieStats_t movies;
	} actressesList;

    } stats;
};