Changeset 34509

Timestamp:

2020-10-22T17:42:08+13:00 (4 years ago)

Author:

ak19

Message:

Related to previous commits 34506-34508. Storing both the Win 8.3 Short filename of gsdlsourcefilename and its long filename version.

File:

• main/trunk/gli/src/org/greenstone/gatherer/metadata/DocXMLFile.java (modified) (4 diffs)

main/trunk/gli/src/org/greenstone/gatherer/metadata/DocXMLFile.java

@@ -77 +77 @@
             ///System.err.println("Unable to find meta for file path form " + file_relative_path);
             return metadata_values; // we're done
-    } /// else { System.err.println("@@@ file rel path: " + file_relative_path + " matched" ); }
+    } ///else { System.err.println("@@@ file rel path: " + file_relative_path + " matched" ); }
 
     MetadataSet extracted_metadata_set = MetadataSetManager.getMetadataSet(MetadataSetManager.EXTRACTED_METADATA_NAMESPACE);
@@ -293 +293 @@
         buffered_reader.close();
 
+        // Work out if is_unix_path
+        int import_index = gsdlsourcefilename_value.indexOf("import");
+        if (import_index != -1) {
+            String tempStr = gsdlsourcefilename_value.substring(import_index + "import".length());
+            is_unix_path = tempStr.startsWith("/");
+        }
+        // We're only interested in the path relative to the import folder
+        // Lop off "import" folder prefix
+        gsdlsourcefilename_value = adjustForRelativeToImportDir(gsdlsourcefilename_value);
 
         // Now that we're done skimming, we actually need to decode gsdlsourcefilename
@@ -303 +312 @@
             fileRenameMethod = FILE_RENAME_METHOD_URL; // default for building
         }
-        
-        // If gsdlsourcefilename was encoded, we remove it from the map under its encoded
-        // filename, decode it and add it back into map using its decoded filename.
         if(!fileRenameMethod.equals(FILE_RENAME_METHOD_NONE)) {         
             gsdlsourcefilename_value = decodeSourceFilename(gsdlsourcefilename_value, fileRenameMethod, is_unix_path);          
-        }       
+        }
+
+        // Now we can finally put the gsdlsourcefilename path relative to import dir into the hashmap
+        ///System.err.println("@@@ into map: " + gsdlsourcefilename_value);
+        if (source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value) == null) {
+            source_file_name_to_description_elements_mapping.put(gsdlsourcefilename_value, new ArrayList());
+        }
+        ((ArrayList) source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value)).add(new Integer(description_element_start_gsdlsourcefilename_value));      
         
         // Next, if Windows, check if dealing with Win 8.3 Short Filename
-        // In that case, convert short file name to full name - works only if the file exists       
+        // In that case, convert short file name to long filename - works only if the file exists       
         if(isWin && gsdlsourcefilename_value.indexOf("~") != -1) {
+            
+            String long_gsdlsourcefilename = gsdlsourcefilename_value;
+            
             // gsdlsourcefilename is stored from import folder onwards: import/opt_subdir/filename.ext
             // This may contain Win 8.3 shortening. To get Win Long filename, prefix current collection dir
             // and if resulting file exists, getCanonicalPath() which produces Win Long filename.
             File currentCollectionFolder = Gatherer.c_man.getCollection().getCollectionDirectory();
-            File f = new File(currentCollectionFolder, /*"import" + File.separator +*/ gsdlsourcefilename_value);
+            File f = new File(currentCollectionFolder, "import" + File.separator + gsdlsourcefilename_value); // should work even if linux style slashes in gsdlsourcefilename_value
             ///System.err.println("### file: " + f.getAbsolutePath());
             
             if(f.exists()) {
-                gsdlsourcefilename_value = f.getCanonicalPath();
-                ///System.err.println("### canon: " + gsdlsourcefilename_value);
-                
-            } // else couldn't find a version of the filename stored in doc.xml that exists, giving up, leave gsdlsourcefilename_value as is    
+                long_gsdlsourcefilename = f.getCanonicalPath();
+                ///System.err.println("### canon: " + long_gsdlsourcefilename);             
+            } // else couldn't find a version of the filename stored in doc.xml that exists, giving up, leave gsdlsourcefilename_value as is
+            
+            // Again, we're only interested in the path relative to the import folder
+            long_gsdlsourcefilename = adjustForRelativeToImportDir(long_gsdlsourcefilename);
+            if(!gsdlsourcefilename_value.equals(long_gsdlsourcefilename)) { // truly distinct Win long and short file names
+                // Put a copy of the ref to gsdlsourcefilename's metadata list under the long filename as well
+                ///System.err.println("@@@ long filename into map: " + long_gsdlsourcefilename);        
+                Object arrList = source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value);
+                source_file_name_to_description_elements_mapping.put(long_gsdlsourcefilename, arrList);
+            }
         }
         
-        // We're only interested in the path relative to the import folder
-        int import_index = gsdlsourcefilename_value.indexOf("import");
+    }
+    catch (FileNotFoundException exception) {
+        DebugStream.printStackTrace(exception);
+    }
+    catch (IOException exception) {
+        DebugStream.printStackTrace(exception);
+    } catch (Exception exception) { // e.g. exception decoding gsdlsourcefilename
+        DebugStream.printStackTrace(exception);
+    }
+    }
+    
+    private String adjustForRelativeToImportDir(String gsdlsourcefilename_value) {
+        int import_index = gsdlsourcefilename_value.indexOf("import");
         if (import_index != -1) {
             gsdlsourcefilename_value = gsdlsourcefilename_value.substring(import_index + "import".length());
 
-            is_unix_path = gsdlsourcefilename_value.startsWith("/");
+            boolean is_unix_path = gsdlsourcefilename_value.startsWith("/");
             gsdlsourcefilename_value = gsdlsourcefilename_value.substring(1);
 
@@ -347 +382 @@
                 gsdlsourcefilename_value = gsdlsourcefilename_value.replaceAll("\\\\", "/");
             }
-        
-            ///System.err.println("@@@ into map: " + gsdlsourcefilename_value);
-        
-            if (source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value) == null) {
-                source_file_name_to_description_elements_mapping.put(gsdlsourcefilename_value, new ArrayList());
-            }
-            
-            ((ArrayList) source_file_name_to_description_elements_mapping.get(gsdlsourcefilename_value)).add(new Integer(description_element_start_gsdlsourcefilename_value));
-        }
-    }
-    catch (FileNotFoundException exception) {
-        DebugStream.printStackTrace(exception);
-    }
-    catch (IOException exception) {
-        DebugStream.printStackTrace(exception);
-    } catch (Exception exception) { // e.g. exception decoding gsdlsourcefilename
-        DebugStream.printStackTrace(exception);
-    }
-    }
-    
+        }
+        return gsdlsourcefilename_value;
+    }
     
     protected String decodeSourceFilename(String relative_sourcefile_path,