Changeset 15591 for gsdl/trunk/src/colservr/source.cpp

Timestamp:

2008-05-20T13:12:52+12:00 (16 years ago)

Author:

mdewsnip

Message:

(Adding new DB support) Moved all the code in gdbmsource into source, and changed mggdbmsource and lucenegdbmsource to inherit from source instead of gdbmsource. This is part of removing GDBM from these files altogether.

File:

• gsdl/trunk/src/colservr/source.cpp (modified) (3 diffs)

gsdl/trunk/src/colservr/source.cpp

@@ -25 +25 @@
 
 #include "source.h"
+#include "fileutil.h"
+#include "OIDtools.h"
+#include "gsdltools.h"
 #include <assert.h>
 
 
-// default constructor does nothing
-sourceclass::sourceclass () {
-}
-
-// default destructor does nothing
-sourceclass::~sourceclass () {
+sourceclass::sourceclass ()
+{
+  db_ptr = NULL;
+  textsearchptr = NULL;
+  classname = "source";
+}
+
+sourceclass::~sourceclass ()
+{
+  if (db_ptr != NULL) delete db_ptr;
+  if (textsearchptr != NULL) delete textsearchptr;
 }
 
 // configure should be called once for each configuration line
-// default does nothing
-void sourceclass::configure (const text_t &/*key*/, const text_tarray &/*cfgline*/) {
+void sourceclass::configure (const text_t &key, const text_tarray &cfgline)
+{
+  if (cfgline.size() >= 1) {
+    const text_t &value = cfgline[0];
+
+    if (key == "collection") collection = value;
+    else if (key == "collectdir") collectdir = value;
+    else if (key == "gsdlhome") gsdlhome = value;
+    else if (key == "gdbmhome") dbhome = value;
+  }
+
+  if (key == "indexmap") {
+    indexmap.importmap (cfgline);
+
+  } else if (key == "defaultindex") {
+    indexmap.from2to (cfgline[0], defaultindex);
+
+  } else if (key == "subcollectionmap") {
+    subcollectionmap.importmap (cfgline);
+
+  } else if (key == "defaultsubcollection") {
+    subcollectionmap.from2to (cfgline[0], defaultsubcollection);
+
+  } else if (key == "languagemap") {
+    languagemap.importmap (cfgline);
+
+  } else if (key == "defaultlanguage") {
+    languagemap.from2to (cfgline[0], defaultlanguage);
+  } else if (key == "indexstem") {
+    indexstem = cfgline[0];
+  }
 }
 
 // init should be called after all the configuration is done but
 // before any other methods are called
-// default does nothing
-bool sourceclass::init (ostream &/*logout*/) {
+bool sourceclass::init (ostream &logout)
+{
+  outconvertclass text_t2ascii;
+  
+  if (dbhome.empty()) dbhome = gsdlhome;
+
+  if (!sourceclass::init (logout)) return false;
+
+  if (defaultindex.empty()) {
+    // use first index in map as default if no default is set explicitly
+    text_tarray toarray;
+    indexmap.gettoarray(toarray);
+    if (toarray.size()) {
+      defaultindex = toarray[0];
+    }
+  }
+
+  if (defaultsubcollection.empty()) {
+    // use first subcollection in map as default if no default is set explicitly
+    text_tarray toarray;
+    subcollectionmap.gettoarray(toarray);
+    if (toarray.size()) {
+      defaultsubcollection = toarray[0];
+    }
+  }
+
+  if (defaultlanguage.empty()) {
+    // use first language in map as default if no default is set explicitly
+    text_tarray toarray;
+    languagemap.gettoarray(toarray);
+    if (toarray.size()) {
+      defaultlanguage = toarray[0];
+    }
+  }
+  
+  // get the collection directory name
+  if (collectdir.empty()) {
+    collectdir = filename_cat (gsdlhome, "collect", collection);
+  }
+  
+  // get the filename for the database and make sure it exists
+  if (indexstem.empty()) {
+    indexstem = collection;
+  }
+  db_filename = filename_cat(dbhome, "collect", collection, "index", "text", indexstem);
+  if (littleEndian()) db_filename += ".ldb";
+  else db_filename += ".bdb";
+
+  if (!file_exists(db_filename)) {
+    logout << text_t2ascii
+       << "warning: database \"" << db_filename << "\" does not exist\n\n";
+    // return false;
+  }
+  
   return true;
 }
@@ -50 +139 @@
 
 // translate_OID translates OIDs using ".pr", ."fc" etc.
-bool sourceclass::translate_OID (const text_t &/*OIDin*/, text_t &/*OIDout*/, 
-                 comerror_t &err, ostream &/*logout*/) {
+bool sourceclass::translate_OID (const text_t &OIDin, text_t &OIDout, 
+                 comerror_t &err, ostream &logout)
+{
+  outconvertclass text_t2ascii;
+
   err = noError;
-
-  return false;
+  if (db_ptr == NULL) {
+    // most likely a configuration problem
+    logout << text_t2ascii 
+       << "configuration error: " << classname << " contains a null dbclass\n\n";
+    err = configurationError;
+    return true;
+  }
+
+  // open the database
+  db_ptr->setlogout(&logout);
+  if (!db_ptr->opendatabase (db_filename, DB_READER, 100, false)) {
+    // most likely a system problem (we have already checked that the database exists)
+    logout << text_t2ascii
+       << "system problem: open on database \"" << db_filename << "\" failed\n\n";
+    err = systemProblem;
+    return true;
+  }
+
+  infodbclass info;
+  OIDout = db_ptr->translate_OID (OIDin, info);
+  db_ptr->closedatabase();  // Important that local library doesn't leave any files open
+  return true;
 }
 
@@ -60 +172 @@
 // get_metadata fills out the metadata if possible, if it is not responsable
 // for the given OID then it will return false.
-bool sourceclass::get_metadata (const text_t &/*requestParams*/, const text_t &/*refParams*/, 
-                bool /*getParents*/, const text_tset &/*fields*/, 
-                const text_t &/*OID*/, MetadataInfo_tmap &metadata, 
-                comerror_t &err, ostream &/*logout*/) {
+bool sourceclass::get_metadata (const text_t &requestParams, const text_t &refParams, 
+                bool getParents, const text_tset &fields, 
+                const text_t &OID, MetadataInfo_tmap &metadata, 
+                comerror_t &err, ostream &logout)
+{
+  outconvertclass text_t2ascii;
+
   metadata.erase(metadata.begin(), metadata.end());
+
   err = noError;
-
-  return false;
-}
-
-bool sourceclass::get_document (const text_t &/*OID*/, text_t &/*doc*/, 
-                comerror_t &err, ostream &/*logout*/) {
+  if (db_ptr == NULL) {
+    // most likely a configuration problem
+    logout << text_t2ascii 
+       << "configuration error: " << classname <<" contains a null dbclass\n\n";
+    err = configurationError;
+    return true;
+  }
+
+  // open the database
+  db_ptr->setlogout(&logout);
+  if (!db_ptr->opendatabase (db_filename, DB_READER, 100, false)) {
+    // most likely a system problem (we have already checked that the database exists)
+    logout << text_t2ascii
+       << "system problem: open on database \"" << db_filename << "\" failed\n\n";
+    err = systemProblem;
+    return true;
+  }
+
+  // get the metadata - if getParents is set we need to get 
+  // info for all parents of OID as well as OID
+  vector<infodbclass> info_array;
+  text_tarray OIDs;
+  if (getParents) get_parents_array (OID, OIDs);
+  OIDs.push_back (OID);
+
+  text_tarray::const_iterator this_OID = OIDs.begin();
+  text_tarray::const_iterator end_OID = OIDs.end();
+
+  while (this_OID != end_OID) {
+    infodbclass info;
+    if (!db_ptr->getinfo(*this_OID, info)) return false;
+
+    // adjust the metadata
+    text_t &contains = info["contains"];
+    if (contains.empty()) info["haschildren"] = 0;
+    else info["haschildren"] = 1;
+    //contains.clear();
+
+    info_array.push_back(info);
+    ++this_OID;
+  }
+
+  // if fields set is empty we want to get all available metadata
+  text_tset tfields = fields;
+  if (tfields.empty() && !info_array.empty()) {
+    infodbclass::iterator t_info = info_array[0].begin(); 
+    infodbclass::iterator e_info = info_array[0].end(); 
+    while (t_info != e_info) {
+      if ((*t_info).first != "contains")
+    tfields.insert ((*t_info).first);
+      ++t_info;
+    }
+    tfields.insert ("hasnext");
+    tfields.insert ("hasprevious");
+  }
+    
+  // collect together the metadata
+  bool donenextprevtest = false;
+  bool hasnext=false, hasprevious=false;
+  MetadataInfo_t this_metadata;
+  text_tarray *pos_metadata;
+  text_tset::const_iterator fields_here = tfields.begin();
+  text_tset::const_iterator fields_end = tfields.end();
+
+  while (fields_here != fields_end) {
+    this_metadata.clear();
+    this_metadata.isRef = false;
+
+    vector<infodbclass>::reverse_iterator this_info = info_array.rbegin(); 
+    vector<infodbclass>::reverse_iterator end_info = info_array.rend(); 
+    MetadataInfo_t *tmetaptr = &this_metadata;
+    while (this_info != end_info) {
+
+      pos_metadata = (*this_info).getmultinfo(*fields_here);
+      if ((*fields_here == "hasnext" || *fields_here == "hasprevious")) {
+    
+    // collect metadata
+    if (!donenextprevtest) {
+      donenextprevtest = true;
+      
+      // cache parent contents array
+      text_t thisparent = get_parent (OID);
+      if (!thisparent.empty()) {
+        if (thisparent != parentOID) {
+          parentOID = thisparent;
+          parentcontents.erase(parentcontents.begin(), parentcontents.end());
+          if (db_ptr->getinfo(parentOID, parentinfo)) {
+        text_t &parentinfocontains = parentinfo["contains"];
+        if (!parentinfocontains.empty())
+          splitchar (parentinfocontains.begin(), parentinfocontains.end(), 
+                 ';', parentcontents);
+          }
+        }
+        
+        // do tests
+        text_tarray::const_iterator parentcontents_here = parentcontents.begin();
+        text_tarray::const_iterator parentcontents_end = parentcontents.end();
+        text_t shrunk_OID = OID;
+        shrink_parent (shrunk_OID);
+        while (parentcontents_here != parentcontents_end) {
+          if (*parentcontents_here == shrunk_OID) {
+        if (parentcontents_here == parentcontents.begin()) hasprevious = false;
+        else hasprevious = true;
+        
+        ++parentcontents_here;
+        
+        if (parentcontents_here == parentcontents.end()) hasnext = false;
+        else hasnext = true;
+        
+        break;
+          }  
+          
+          ++parentcontents_here;
+        }
+     
+        // fill in metadata
+        if ((*fields_here == "hasnext" && hasnext) ||
+        (*fields_here == "hasprevious" && hasprevious))
+          tmetaptr->values.push_back("1");
+        else
+          tmetaptr->values.push_back("0");
+      } else
+        tmetaptr->values.push_back("0");
+    }
+      }
+      //else if (pos_metadata != NULL && *fields_here != "contains") {
+      else if (pos_metadata != NULL) {
+    tmetaptr->values = *pos_metadata;
+      }
+      else
+    tmetaptr->values.push_back("");
+
+      ++this_info;
+      if (this_info != end_info) {
+    tmetaptr->parent = new MetadataInfo_t();
+    tmetaptr = tmetaptr->parent;
+      }
+    }
+    metadata[*fields_here] = this_metadata;
+    ++fields_here;
+  }
+
+  db_ptr->closedatabase();  // Important that local library doesn't leave any files open
+  return true;
+}
+
+bool sourceclass::get_document (const text_t &OID, text_t &doc, 
+                comerror_t &err, ostream &logout)
+{
+  outconvertclass text_t2ascii;
+
   err = noError;
-
-  return false;
-}
-
-bool sourceclass::is_searchable(bool &issearchable, comerror_t &err, ostream &logout) {
-
-  // currently defaults to claiming the collection is searchable!!
+  if (db_ptr == NULL) {
+    // most likely a configuration problem
+    logout << text_t2ascii 
+       << "configuration error: " << classname << " contains a null dbclass\n\n";
+    err = configurationError;
+    return true;
+  }
+
+  // open the database
+  db_ptr->setlogout(&logout);
+  if (!db_ptr->opendatabase (db_filename, DB_READER, 100, false)) {
+    // most likely a system problem (we have already checked that the database exists)
+    logout << text_t2ascii
+       << "system problem: open on database \"" << db_filename << "\" failed\n\n";
+    err = systemProblem;
+    return true;
+  }
+
+  text_t tOID = OID;
+  if (needs_translating (OID)) 
+    translate_OID (OID, tOID, err, logout);
+  infodbclass info;
+  if (!db_ptr->getinfo(tOID, info)) {
+    db_ptr->closedatabase();  // Important that local library doesn't leave any files open
+    return false;
+  }
+  
+  if (info["hastxt"].getint() == 1) {
+    int docnum = info["docnum"].getint();
+    
+    // set the collection directory
+    textsearchptr->setcollectdir (collectdir);
+    
+    // get the text
+    textsearchptr->docTargetDocument(defaultindex, defaultsubcollection,
+                   defaultlanguage, collection, docnum, doc);
+  }
+
+  db_ptr->closedatabase();  // Important that local library doesn't leave any files open
+  return true;
+}
+
+bool sourceclass::is_searchable(bool &issearchable, comerror_t &err, ostream &logout)
+{
+  err = noError;
+  issearchable = false;
+
+  text_tarray fromarray;
+  indexmap.getfromarray(fromarray);
+  if (fromarray.size() == 0) {
+    return true;
+  } else if (fromarray.size() == 1) {
+    if (fromarray[0] == "dummy:text") {
+      // always return true - issearchable is false here though
+      return true;
+    }
+  }
   issearchable = true;
-
-  err = noError;
-  
-  return false;
-}
+  return true;
+}
+
 
 bool operator==(const sourceptr &x, const sourceptr &y) {