Revision: 28113
http://sourceforge.net/p/bibdesk/svn/28113
Author: hofman
Date: 2022-12-14 16:15:52 +0000 (Wed, 14 Dec 2022)
Log Message:
-----------
Get items for ACM from doi, either from meta or input elements. Add URL for
item in ACM site.
Modified Paths:
--------------
trunk/bibdesk/BDSKACMDLParser.m
Modified: trunk/bibdesk/BDSKACMDLParser.m
===================================================================
--- trunk/bibdesk/BDSKACMDLParser.m 2022-12-14 14:56:00 UTC (rev 28112)
+++ trunk/bibdesk/BDSKACMDLParser.m 2022-12-14 16:15:52 UTC (rev 28113)
@@ -39,44 +39,70 @@
#import "BDSKACMDLParser.h"
#import <AGRegex/AGRegex.h>
#import "DOMNode_BDSKExtensions.h"
+#import "NSURL_BDSKExtensions.h"
+#import "NSString_BDSKExtensions.h"
+#import "BibItem.h"
@implementation BDSKACMDLParser
-+ (NSString *)citationNodeXPath { return
@"./head/meta[@name='citation_abstract_html_url']/@content"; }
-
-+ (NSString *)citationURLStringFromNode:(DOMNode *)node {
- NSString *nodeString = [[NSURL URLWithString:[node stringValue]] query];
-
- AGRegex *doiRegex = [AGRegex regexWithPattern:@"^id=([0-9]+)\\.([0-9]+)$"];
- AGRegexMatch *match = [doiRegex findInString:nodeString];
++ (BOOL)canParseDocument:(DOMDocument *)domDocument fromURL:(NSURL *)url {
+ if ([url hasDomain:@"dl.acm.org"] == NO)
+ return NO;
- if ([match count] != 3)
+ NSString *doiNodeXPath = nil;
+ if ([url hasFirstPathComponent:@"doi"])
+ doiNodeXPath = @"./head/meta[@name='dc.Identifier' and
@scheme='doi']/@content";
+ else if ([url hasFirstPathComponent:@"action"])
+ doiNodeXPath = @"./body//input[@name='doiVal']/@value";
+ else
return nil;
+ DOMNode *node = [[domDocument documentElement]
singleNodeForXPath:doiNodeXPath];
- NSString *parentNumber = [match groupAtIndex:1];
- NSString *articleNumber = [match groupAtIndex:2];
+ if (node == nil)
+ return NO;
- return [NSString
stringWithFormat:@"/downformats.cfm?id=%@&parent_id=%@&expformat=bibtex",
articleNumber, parentNumber];
+ return nil != [self citationURLStringFromNode:node];
}
-- (NSString *)bibTeXStringFromDownload:(BDSKCitationDownload *)download {
- NSString *string = [download string];
+- (NSArray *)itemsReturningError:(NSError **)outError {
+ NSString *doiNodeXPath = nil;
+ if ([[self URL] hasFirstPathComponent:@"doi"])
+ doiNodeXPath = @"./head/meta[@name='dc.Identifier' and
@scheme='doi']/@content";
+ else
+ doiNodeXPath = @"./body//input[@name='doiVal']/@value";
+
+ NSArray *doiNodes = [[[self domDocument] documentElement]
nodesForXPath:doiNodeXPath];
+ NSMutableSet *dois = [NSMutableSet set];
- // remove characters before the first @ symbol
- NSRange range = [string rangeOfString:@"@"];
- if (range.location == NSNotFound)
- return nil;
+ for (DOMNode *doiNode in doiNodes) {
+ NSString *doi = [doiNode stringValue];
+
+ if ([dois containsObject:doi]) continue;
+ [dois addObject:doi];
+
+ NSURL *doiURL = [NSURL URLWithString:[@"https://doi.org/";
stringByAppendingString:doi]];
+ NSMutableURLRequest *request = [NSMutableURLRequest
requestWithURL:doiURL];
+ [request setValue:@"application/x-bibtex,
text/bibliography;style=bibtex;q=0.5" forHTTPHeaderField:@"Accept"];
+
+ NSString *urlString = [@"https://dl.acm.org/doi/";
stringByAppendingString:doi];
+
+ [self addDownloadWithRequest:request contextInfo:urlString];
+ }
- string = [string substringFromIndex:range.location];
+ return nil;
+}
+
+- (NSString *)bibTeXStringFromDownload:(BDSKCitationDownload *)download {
+ return [[download string] stringByRemovingSurroundingWhitespace];
+}
+
+- (NSArray *)itemsFromDownload:(BDSKCitationDownload *)download error:(NSError
**)outError {
+ NSArray *items = [super itemsFromDownload:download error:outError];
- // remove spaces in cite key (for some reason, ACM will use author names
with spaces in the cite key
- // but btparse chokes on these)
- range = [string rangeOfString:@","];
- if (range.location == NSNotFound)
- return nil;
+ [[items firstObject] addURLString:[download contextInfo]];
- return [string stringByReplacingOccurrencesOfString:@" " withString:@""
options:0 range:NSMakeRange(0, range.location)];
+ return items;
}
+ (NSString *)name {return @"ACM"; }
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
_______________________________________________
Bibdesk-commit mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bibdesk-commit
