Mirrors/yacy_search_server
1
0
Fork 0
mirror of https://github.com/yacy/yacy_search_server.git synced 2025-07-22 09:14:38 -04:00
Code Issues Packages Projects Releases Wiki Activity

fix (todo) remove redundant holding of email link nameproperty in parser document

Browse Source
This commit is contained in:
reger
2015-12-18 02:35:44 +01:00
parent 71c416f383
commit 0c5548a7ff
2 changed files with 13 additions and 10 deletions
htroot
ViewFile.java
source/net/yacy/document
Document.java

10
htroot/ViewFile.java

@ -328,17 +328,17 @@ public class ViewFile {
int i = 0;
if (document.getEmaillinks() != null) {
Iterator<Entry<String, String>> emailit = document.getEmaillinks().entrySet().iterator();
Iterator<AnchorURL> emailit = document.getEmaillinks().iterator();
while (emailit.hasNext()) {
Entry<String, String> eentry = emailit.next();
AnchorURL eentry = emailit.next();
prop.put("viewMode_links_" + i + "_nr", i);
prop.put("viewMode_links_" + i + "_dark", dark ? "1" : "0");
prop.put("viewMode_links_" + i + "_type", "email");
prop.put("viewMode_links_" + i + "_text", (eentry.getValue().isEmpty()) ? "&nbsp;" : eentry.getValue());
prop.put("viewMode_links_" + i + "_text", (eentry.getTextProperty().isEmpty()) ? "&nbsp;" : eentry.getTextProperty());
prop.put("viewMode_links_" + i + "_url", "#");
prop.put("viewMode_links_" + i + "_link", eentry.getKey());
prop.put("viewMode_links_" + i + "_link", eentry.toNormalform(true));
prop.put("viewMode_links_" + i + "_rel", "");
prop.put("viewMode_links_" + i + "_name", "");
prop.put("viewMode_links_" + i + "_name", eentry.getNameProperty());
dark = !dark;
i++;
}

13
source/net/yacy/document/Document.java

@ -83,9 +83,9 @@ public class Document {
// the anchors and images - Maps are URL-to-EntityDescription mappings.
// The EntityDescription appear either as visible text in anchors or as alternative
// text in image tags.
private LinkedHashMap<AnchorURL, String> audiolinks, videolinks, applinks, hyperlinks;
private LinkedHashMap<AnchorURL, String> audiolinks, videolinks, applinks, hyperlinks; // TODO: check if redundant value (set to key.getNameProperty()) is needed
private LinkedHashMap<DigestURL, String> inboundlinks, outboundlinks;
private Map<String, String> emaillinks;
private Set<AnchorURL> emaillinks; // mailto: links
private MultiProtocolURL favicon;
private boolean resorted;
private final Set<String> languages;
@ -473,7 +473,10 @@ dc_rights
return this.applinks;
}
public Map<String, String> getEmaillinks() {
/**
* @return mailto links
*/
public Set<AnchorURL> getEmaillinks() {
// this is part of the getAnchor-set: only links to email addresses
if (!this.resorted) resortLinks();
return this.emaillinks;
@ -509,7 +512,7 @@ dc_rights
this.videolinks = new LinkedHashMap<AnchorURL, String>();
this.audiolinks = new LinkedHashMap<AnchorURL, String>();
this.applinks = new LinkedHashMap<AnchorURL, String>();
this.emaillinks = new LinkedHashMap<String, String>();
this.emaillinks = new LinkedHashSet<AnchorURL>();
final Map<AnchorURL, ImageEntry> collectedImages = new HashMap<AnchorURL, ImageEntry>(); // this is a set that is collected now and joined later to the imagelinks
for (final Map.Entry<DigestURL, ImageEntry> entry: this.images.entrySet()) {
if (entry.getKey() != null && entry.getKey().getHost() != null && entry.getKey().getHost().equals(thishost)) this.inboundlinks.put(entry.getKey(), "image"); else this.outboundlinks.put(entry.getKey(), "image");
@ -520,7 +523,7 @@ dc_rights
final String name = url.getNameProperty();
// check mailto scheme first (not suppose to get into in/outboundlinks or hyperlinks -> crawler can't process)
if (url.getProtocol().equals("mailto")) {
this.emaillinks.put(u.substring(7), name); // TODO: check why key as string instead of Disgest/AnchorURL
this.emaillinks.add(url);
continue;
}