#ifndef _RCLDOC_H_INCLUDED_

#ifndef _RCLDOC_H_INCLUDED_

#define _RCLDOC_H_INCLUDED_

#define _RCLDOC_H_INCLUDED_

#include <string>

#include <string>

#include <map>

#include <map>

#include "smallut.h"

#include "smallut.h"

namespace Rcl {

namespace Rcl {

    // We indicate the routine that sets them up during indexing

    // We indicate the routine that sets them up during indexing

    // Binary or url-encoded url. No transcoding: this is used to access files 

    // Binary or url-encoded url. No transcoding: this is used to access files 

    // Index: computed by Db::add caller. 

    // Index: computed by Db::add caller. 

    // Query: from doc data.

    // Query: from doc data.

    string url;

    // When we do path translation for documents from external indexes, we

    // When we do path translation for documents from external indexes, we

    // save the original path:

    // save the original path:

    string idxurl;

    // And the originating db. 0 is base, 1 first external etc.

    // And the originating db. 0 is base, 1 first external etc.

    int idxi;

    int idxi;

    // Internal path for multi-doc files. Ascii

    // Internal path for multi-doc files. Ascii

    // Set by FsIndexer::processone    

    // Set by FsIndexer::processone    

    string ipath;

    // Mime type. Set by FileInterner::internfile

    // Mime type. Set by FileInterner::internfile

    string mimetype;     

    // File modification time as decimal ascii unix time

    // File modification time as decimal ascii unix time

    // Set by FsIndexer::processone

    // Set by FsIndexer::processone

    string fmtime;

    // Data reference date (same format). Ie: mail date

    // Data reference date (same format). Ie: mail date

    // Possibly set by mimetype-specific handler

    // Possibly set by mimetype-specific handler

    // Filter::metaData["modificationdate"]

    // Filter::metaData["modificationdate"]

    string dmtime;

    // Charset we transcoded the 'text' field from (in case we want back)

    // Charset we transcoded the 'text' field from (in case we want back)

    // Possibly set by handler

    // Possibly set by handler

    string origcharset;  

    // A map for textual metadata like, author, keywords, abstract,

    // A map for textual metadata like, author, keywords, abstract,

    // title.  The entries are possibly set by the mimetype-specific

    // title.  The entries are possibly set by the mimetype-specific

    // handler. If a fieldname-to-prefix translation exists, the

    // handler. If a fieldname-to-prefix translation exists, the

    // terms in the value will be indexed with a prefix.

    // terms in the value will be indexed with a prefix.

    // Only some predefined fields are stored in the data record:

    // Only some predefined fields are stored in the data record:

    // "title", "keywords", "abstract", "author", but if a field name is

    // "title", "keywords", "abstract", "author", but if a field name is

    // in the "stored" configuration list, it will be stored too.

    // in the "stored" configuration list, it will be stored too.

    map<string, string> meta; 

    // Attribute for the "abstract" entry. true if it is just the top

    // Attribute for the "abstract" entry. true if it is just the top

    // of doc, not a native document attribute. Not stored directly, but

    // of doc, not a native document attribute. Not stored directly, but

    // as an indicative prefix at the beginning of the abstract (ugly hack)

    // as an indicative prefix at the beginning of the abstract (ugly hack)

    bool   syntabs;      

    bool   syntabs;      

    // File size. This is the size of the compressed file or of the

    // File size. This is the size of the compressed file or of the

    // external containing archive.

    // external containing archive.

    // Index: Set by caller prior to Db::Add. 

    // Index: Set by caller prior to Db::Add. 

    // Query: Set from data record

    // Query: Set from data record

    string pcbytes;       

    // Document size, ie, size of the .odt or .xls.

    // Document size, ie, size of the .odt or .xls.

    // Index: Set in internfile from the filter stack

    // Index: Set in internfile from the filter stack

    // Query: set from data record

    // Query: set from data record

    string fbytes;

    // Doc text size. 

    // Doc text size. 

    // Index: from text.length(). 

    // Index: from text.length(). 

    // Query: set by rcldb from index data record

    // Query: set by rcldb from index data record

    string dbytes;

    // Doc signature. Used for up to date checks. 

    // Doc signature. Used for up to date checks. 

    // Index: set by Db::Add caller. Query: set from doc data.

    // Index: set by Db::Add caller. Query: set from doc data.

    // This is opaque to rcldb, and could just as well be ctime, size,

    // This is opaque to rcldb, and could just as well be ctime, size,

    // ctime+size, md5, whatever.

    // ctime+size, md5, whatever.

    string sig;

    /////////////////////////////////////////////////

    /////////////////////////////////////////////////

    // The following fields don't go to the db record, so they can't

    // The following fields don't go to the db record, so they can't

    // be retrieved at query time

    // be retrieved at query time

    // Main document text. This is plaintext utf-8 text to be split

    // Main document text. This is plaintext utf-8 text to be split

    // and indexed

    // and indexed

    string text; 

    /////////////////////////////////////////////////

    /////////////////////////////////////////////////

    // Misc stuff

    // Misc stuff

    int pc; // relevancy percentage, used by sortseq, convenience

    int pc; // relevancy percentage, used by sortseq, convenience

    Doc()

    Doc()

    : idxi(0), syntabs(false), pc(0), xdocid(0),

    : idxi(0), syntabs(false), pc(0), xdocid(0),

      haspages(false), haschildren(false), onlyxattr(false) {

      haspages(false), haschildren(false), onlyxattr(false) {

    }

    }

    /** Get value for named field. If value pointer is 0, just test existence */

    /** Get value for named field. If value pointer is 0, just test existence */

    bool getmeta(const string& nm, string *value = 0) const

    {

    {

    if (it != meta.end()) {

    if (it != meta.end()) {

        if (value)

        if (value)

        *value = it->second;

        *value = it->second;

        return true;

        return true;

    } else {

    } else {

        return false;

        return false;

    }

    }

    }

    }

    /** Nocopy getvalue. sets pointer to entry value if exists */

    /** Nocopy getvalue. sets pointer to entry value if exists */

    bool peekmeta(const string& nm, const string **value = 0) const

    {

    {

    if (it != meta.end()) {

    if (it != meta.end()) {

        if (value)

        if (value)

        *value = &(it->second);

        *value = &(it->second);

        return true;

        return true;

    } else {

    } else {

        return false;

        return false;

    }

    }

    }

    }

    // Create entry or append text to existing entry.

    // Create entry or append text to existing entry.

    bool addmeta(const string& nm, const string& value) 

    {

    {

    if (mit == meta.end()) {

    if (mit == meta.end()) {

        meta[nm] = value;

        meta[nm] = value;

    } else if (mit->second.empty()) {

    } else if (mit->second.empty()) {

        mit->second = value;

        mit->second = value;

    } else {

    } else {

        // It may happen that the same attr exists several times

        // It may happen that the same attr exists several times

        // in the internfile stack. Avoid duplicating values.

        // in the internfile stack. Avoid duplicating values.

        if (mit->second != value)

        if (mit->second != value)

        mit->second += string(" - ") + value;

    }

    }

    return true;

    return true;

    }

    }

    /* Is this document stored as a regular filesystem file ?

    /* Is this document stored as a regular filesystem file ?

     * (as opposed to e.g. a webcache file), not a subdoc, 

     * (as opposed to e.g. a webcache file), not a subdoc, 

     */

     */

    bool isFsFile() {

    bool isFsFile() {

        string backend;

        getmeta(keybcknd, &backend);

        getmeta(keybcknd, &backend);

        if (!backend.empty() && backend.compare("FS"))

        if (!backend.empty() && backend.compare("FS"))

            return false;

            return false;

        return true;

        return true;

    }

    }

    // fields in the meta array, these are the names used). Defined in

    // fields in the meta array, these are the names used). Defined in

    // rcldoc.cpp. Fields stored in the meta[] array (ie, title,

    // rcldoc.cpp. Fields stored in the meta[] array (ie, title,

    // author), _must_ use these canonical values, not aliases. This is 

    // author), _must_ use these canonical values, not aliases. This is 

    // enforced in internfile.cpp and misc other bits of metadata-gathering 

    // enforced in internfile.cpp and misc other bits of metadata-gathering 

    // code

    // code

    static const string keyurl; // url

    // childurl. This is set when working with the parent of the result, to hold

    // childurl. This is set when working with the parent of the result, to hold

    // the child of interest url, typically to highlight a directory entry

    // the child of interest url, typically to highlight a directory entry

    static const string keychildurl; 

    // file name. This is set for filesystem-level containers or

    // file name. This is set for filesystem-level containers or

    // documents, and not inherited by subdocuments (which can get a

    // documents, and not inherited by subdocuments (which can get a

    // keyfn anyway from, e.g, an attachment filename value).  Subdocs

    // keyfn anyway from, e.g, an attachment filename value).  Subdocs

    // used to inherit the file name, but this was undesirable (you

    // used to inherit the file name, but this was undesirable (you

    // usually don't want to see all subdocs when searching for the

    // usually don't want to see all subdocs when searching for the

    // file name). Instead the container file name is now set in the

    // file name). Instead the container file name is now set in the

    // document record but not indexed (see next entry).

    // document record but not indexed (see next entry).

    static const string keyfn;  

    // Container file name. This is set for all subdocuments of a

    // Container file name. This is set for all subdocuments of a

    // given top level container. It is not indexed by default but

    // given top level container. It is not indexed by default but

    // stored in the document record keyfn field if this is still

    // stored in the document record keyfn field if this is still

    // empty when we create it, for display purposes.

    // empty when we create it, for display purposes.

    static const string keytcfn;

    static const string keyipt; // ipath

    static const string keytp;  // mime type

    static const string keyfmt; // file mtime

    static const string keydmt; // document mtime

    static const string keymt;  // mtime dmtime if set else fmtime

    static const string keyoc;  // original charset

    static const string keypcs;  // document outer container size

    static const string keyfs;  // document size

    static const string keyds;  // document text size

    static const string keysz;  // dbytes if set else fbytes else pcbytes

    static const string keysig; // sig

    static const string keyrr;  // relevancy rating

    static const string keycc;  // Collapse count

    static const string keyabs; // abstract

    static const string keyau;  // author

    static const string keytt;  // title

    static const string keykw;  // keywords

    static const string keymd5; // file md5 checksum

    static const string keybcknd; // backend type for data not from the filesys

    // udi back from index. Only set by Rcl::Query::getdoc().

    // udi back from index. Only set by Rcl::Query::getdoc().

    static const string keyudi;

    static const string keyapptg; // apptag. Set from localfields (fsindexer)

    static const string keybght;  // beagle hit type ("beagleHitType")

};

};

}

}

#endif /* _RCLDOC_H_INCLUDED_ */

#endif /* _RCLDOC_H_INCLUDED_ */