Rewrote the handling of protoarticles

yujitach · Aug 11, 2015 · 5172cd2 · 5172cd2
1 parent f0019ce
commit 5172cd2
Show file tree

Hide file tree

Showing 11 changed files with 322 additions and 93 deletions.
diff --git a/BatchImportOperation.h b/BatchImportOperation.h
@@ -13,7 +13,7 @@
 @class Article;
 @class ArticleList;
 @interface BatchImportOperation : NSOperation
--(BatchImportOperation*)initWithXMLData:(NSData*)d
+-(BatchImportOperation*)initWithProtoArticles:(NSArray*)d
                           originalQuery:(NSString*)q;
 @property(readonly) NSMutableSet*generated;
 @end
diff --git a/BatchImportOperation.m b/BatchImportOperation.m
@@ -15,24 +15,23 @@
 #import "AppDelegate.h"
 #import "MOC.h"
 #import "NSString+magic.h"
-#import "InspireXMLArticle.h"
+#import "ProtoArticle.h"
 
 @implementation BatchImportOperation
 {
-    NSData*xmlData;
+    NSArray*elements;
     NSString*query;
     NSManagedObjectContext*secondMOC;
     NSMutableSet*generated;
     dispatch_group_t group;
 }
 @synthesize generated;
--(BatchImportOperation*)initWithXMLData:(NSData*)d
+-(BatchImportOperation*)initWithProtoArticles:(NSArray *)d
                           originalQuery:(NSString*)q;
 {
     self=[super init];
-    xmlData=d;
+    elements=d;
     query=[q copy];
-    [xmlData writeToFile:@"/tmp/spiresTemporary.xml" atomically:YES];
 /*    NSInteger cap=[[NSUserDefaults standardUserDefaults] integerForKey:@"batchImportCap"];
     if(cap<100)cap=100;
     if([elements count]>cap){
@@ -55,13 +54,11 @@ -(NSString*)description
 
 #pragma mark setters from XML
 
--(void)populatePropertiesOfArticle:(Article*)o fromProtoArticle:(ProtoArticle*)element
+-(void)populatePropertiesOfArticle:(Article*)o fromProtoArticle:(NSObject<ProtoArticle>*)element
 {
-    o.spiresKey=@([element.spiresKey integerValue]);
-    o.inspireKey=@([element.inspireKey integerValue]);
     // Here I'm cheating: -setAuthorNames: puts the collaboration name in the author list,
     // so "collaboration" needs to be set up before that
-    for(NSString*key in [@"eprint,title,collaboration,doi,abstract,comments,citecount,pages,date" componentsSeparatedByString:@","]){
+    for(NSString*key in [@"spiresKey,inspireKey,eprint,title,collaboration,doi,abstract,comments,citecount,pages,date" componentsSeparatedByString:@","]){
         NSObject*x=[element valueForKey:key];
         if(x){
             [o setValue:x forKey:key];
@@ -93,8 +90,8 @@ -(void)treatElements:(NSMutableArray*)a withKey:(NSString*)key
     if([a count]==0)
         return ;
     NSMutableDictionary*dict=[NSMutableDictionary dictionary];
-    for(ProtoArticle*e in a){
-        NSString*v=[e valueForKey:key];
+    for(NSObject<ProtoArticle>*e in a){
+        NSObject<NSCopying>*v=[e valueForKey:key];
         dict[v] = e;
     }
 
@@ -116,17 +113,14 @@ -(void)treatElements:(NSMutableArray*)a withKey:(NSString*)key
             [secondMOC deleteObject:data];
             continue;
         }
-        NSString*v=[data valueForKey:key];
-        if([v isKindOfClass:[NSNumber class]]){
-            v=[(NSNumber*)v stringValue];
-        }
-        ProtoArticle*e=dict[v];
+        NSObject<NSCopying>*v=[data valueForKey:key];
+        NSObject<ProtoArticle>*e=dict[v];
         [self populatePropertiesOfArticle:data.article fromProtoArticle:e];
         [generated addObject:data.article];
         [a removeObject:e];
     }
     NSEntityDescription*articleEntity=[NSEntityDescription entityForName:@"Article" inManagedObjectContext:secondMOC];
-    for(ProtoArticle*e in a){
+    for(NSObject<ProtoArticle>*e in a){
         Article*article=(Article*)[[NSManagedObject alloc] initWithEntity:articleEntity insertIntoManagedObjectContext:secondMOC];
         [self populatePropertiesOfArticle:article fromProtoArticle:e];
         [generated addObject:article];
@@ -138,7 +132,7 @@ -(void)batchAddEntriesOfSPIRES:(NSArray*)a
     NSMutableArray*lookForSpiresKey=[NSMutableArray array];
     NSMutableArray*lookForDOI=[NSMutableArray array];
     NSMutableArray*lookForTitle=[NSMutableArray array];
-    for(ProtoArticle*element in a){
+    for(NSObject<ProtoArticle>*element in a){
         if(element.eprint){
             [lookForEprint addObject:element];
         }else if(element.spiresKey){
@@ -192,8 +186,6 @@ -(void)main
 {
 
     [secondMOC performBlockAndWait:^{
-        NSArray*elements=[InspireXMLArticle articlesFromXMLData:xmlData];
-        NSLog(@"spires returned %d entries",(int)[elements count]);
             [self batchAddEntriesOfSPIRES:elements];
             [secondMOC save:NULL];
     }];

diff --git a/InspireXMLArticle.h → InspireXMLParser.h b/InspireXMLArticle.h → InspireXMLParser.h
@@ -8,7 +8,7 @@
 
 #import "ProtoArticle.h"
 
-@interface InspireXMLArticle : ProtoArticle
+@interface InspireXMLParser : NSObject<NSXMLParserDelegate>
 +(NSString*)usedTags;
 +(NSArray*)articlesFromXMLData:(NSData*)data;
 @end
diff --git a/InspireXMLArticle.m → InspireXMLParser.m b/InspireXMLArticle.m → InspireXMLParser.m
@@ -6,28 +6,31 @@
 //
 //
 
-#import "InspireXMLArticle.h"
+#import "InspireXMLParser.h"
+#import "LightweightArticle.h"
 
-@interface InspireXMLArticle ()
--(void)addAuthor:(NSString*)name;
+@interface InspireXMLParser ()
+@property NSMutableArray*articles;
 @end
 
-
-@interface InspireXMLParser: NSObject<NSXMLParserDelegate>
--(instancetype)initWithXMLData:(NSData*)data;
-@property (readonly) NSMutableArray*articles;
-@end
-
-
 @implementation InspireXMLParser
 {
     NSMutableString*currentString;
     NSString*currentTag;
     NSString*currentCode;
     NSMutableDictionary*subfieldDic;
-    InspireXMLArticle*currentArticle;
+    LightweightArticle*currentArticle;
 }
 @synthesize articles;
++(NSString*)usedTags
+{
+    return @"001,970,100,700,710,520,037,245,300,773,961,024";
+}
++(NSArray*)articlesFromXMLData:(NSData*)data
+{
+    InspireXMLParser*parser=[[InspireXMLParser alloc] initWithXMLData:data];
+    return parser.articles;
+}
 -(instancetype)initWithXMLData:(NSData*)data
 {
     self=[super init];
@@ -46,7 +49,7 @@ -(void)parser:(NSXMLParser *)parser didStartElement:(NSString *)elementName name
 {
     currentString=[NSMutableString string];
     if([elementName isEqualToString:@"record"]){
-        currentArticle=[[InspireXMLArticle alloc] init];
+        currentArticle=[[LightweightArticle alloc] init];
     }else if([elementName isEqualToString:@"controlfield"]){
 
     }else if([elementName isEqualToString:@"datafield"]){
@@ -62,36 +65,37 @@ -(void)parser:(NSXMLParser *)parser didEndElement:(NSString *)elementName namesp
         [articles addObject:currentArticle];
         currentArticle=nil;
     }else if([elementName isEqualToString:@"controlfield"]){
-        [currentArticle setValue:currentString forKey:@"inspireKey"];
+        currentArticle.inspireKey=@([currentString integerValue]);
     }else if([elementName isEqualToString:@"datafield"]){
         if([currentTag isEqualToString:@"970"]){
             NSString*s=subfieldDic[@"a"];
-            [currentArticle setValue:[s substringFromIndex:[@"SPIRES-" length]] forKey:@"spiresKey"];
+            NSString*spiresKey=[s substringFromIndex:[@"SPIRES-" length]];
+            currentArticle.spiresKey=@([spiresKey integerValue]);
         }else if([currentTag isEqualToString:@"100"]){
             [currentArticle addAuthor:subfieldDic[@"a"]];
         }else if([currentTag isEqualToString:@"700"]){
             [currentArticle addAuthor:subfieldDic[@"a"]];
         }else if([currentTag isEqualToString:@"710"]){
-            [currentArticle setValue:subfieldDic[@"g"] forKey:@"collaboration"];
+            currentArticle.collaboration=subfieldDic[@"g"];
         }else if([currentTag isEqualToString:@"520"]){
             if([subfieldDic[@"9"] isEqualToString:@"arXiv"]){
-                [currentArticle setValue:subfieldDic[@"a"] forKey:@"abstract"];
+                currentArticle.abstract=subfieldDic[@"a"];
             }
         }else if([currentTag isEqualToString:@"037"]){
             if([subfieldDic[@"9"] isEqualToString:@"arXiv"]){
-                [currentArticle setValue:subfieldDic[@"a"] forKey:@"eprint"];
+                currentArticle.eprint=subfieldDic[@"a"];
             }
         }else if([currentTag isEqualToString:@"245"]){
-            [currentArticle setValue:subfieldDic[@"a"] forKey:@"title"];
+            currentArticle.title=subfieldDic[@"a"];
         }else if([currentTag isEqualToString:@"300"]){
-            [currentArticle setValue:@([subfieldDic[@"a"] integerValue]) forKey:@"pages"];
+            currentArticle.pages=@([subfieldDic[@"a"] integerValue]);
         }else if([currentTag isEqualToString:@"773"]){
             NSString*title=subfieldDic[@"p"];
             if(title && ![title isEqualToString:@""]){
-                [currentArticle setValue:title forKey:@"journalTitle"];
-                [currentArticle setValue:subfieldDic[@"v"] forKey:@"journalVolume"];
-                [currentArticle setValue:subfieldDic[@"c"] forKey:@"journalPage"];
-                [currentArticle setValue:@([subfieldDic[@"y"] integerValue]) forKey:@"journalYear"];
+                currentArticle.journalTitle=title;
+                currentArticle.journalVolume=subfieldDic[@"v"];
+                currentArticle.journalPage=subfieldDic[@"c"];
+                currentArticle.journalYear=@([subfieldDic[@"y"] integerValue]);
             }
         }else if([currentTag isEqualToString:@"961"]){
             NSString*dateString=subfieldDic[@"x"];
@@ -100,11 +104,11 @@ -(void)parser:(NSXMLParser *)parser didEndElement:(NSString *)elementName namesp
                     dateString=[dateString stringByAppendingString:@"-00"];
                 }
                 NSDate*date=[NSDate dateWithString:[NSString stringWithFormat:@"%@ 00:00:00 +0000",dateString]];
-                [currentArticle setValue:date forKey:@"date"];
+                currentArticle.date=date;
             }
         }else if([currentTag isEqualToString:@"024"]){
             if([subfieldDic[@"2"] isEqualToString:@"DOI"]){
-                [currentArticle setValue:subfieldDic[@"a" ] forKey:@"doi"];
+                currentArticle.doi=subfieldDic[@"a"];
             }
         }
         subfieldDic=nil;
@@ -119,31 +123,3 @@ -(void)parser:(NSXMLParser *)parser foundCharacters:(NSString *)string
 }
 @end
 
-@implementation InspireXMLArticle
-{
-    NSMutableArray*authorArray;
-}
-+(NSString*)usedTags
-{
-    return @"001,970,100,700,710,520,037,245,300,773,961,024";
-}
-+(NSArray*)articlesFromXMLData:(NSData*)data
-{
-    InspireXMLParser*parser=[[InspireXMLParser alloc] initWithXMLData:data];
-    return parser.articles;
-}
--(instancetype)init
-{
-    self=[super init];
-    authorArray=[NSMutableArray array];
-    return self;
-}
--(NSArray*)authors
-{
-    return authorArray;
-}
--(void)addAuthor:(NSString*)name
-{
-    [authorArray addObject:name];
-}
-@end
diff --git a/LightweightArticle.h b/LightweightArticle.h
@@ -0,0 +1,30 @@
+//
+//  LightweightArticle.h
+//  inspire
+//
+//  Created by Yuji on 2015/08/11.
+//
+//
+
+#import <Foundation/Foundation.h>
+#import "ProtoArticle.h"
+
+@interface LightweightArticle : NSObject<ProtoArticle>
+@property (strong) NSString*title;
+@property (strong) NSNumber*inspireKey;
+@property (strong) NSNumber*spiresKey;
+@property (strong) NSString*eprint;
+@property (strong) NSArray*authors;
+@property (strong) NSString*abstract;
+@property (strong) NSString*collaboration;
+@property (strong) NSNumber*pages;
+@property (strong) NSNumber*citecount;
+@property (strong) NSDate*date;
+@property (strong) NSString*doi;
+@property (strong) NSString*comments;
+@property (strong) NSString*journalTitle;
+@property (strong) NSString*journalVolume;
+@property (strong) NSString*journalPage;
+@property (strong) NSNumber*journalYear;
+-(void)addAuthor:(NSString*)author;
+@end