From: Niki Roo <niki@nikiroo.be>
Date: Sun, 25 Mar 2018 12:37:48 +0000 (+0200)
Subject: Fix subtitles and too much content in EreNumerique
X-Git-Url: https://git.nikiroo.be/?a=commitdiff_plain;h=b9afb12e17825f363d3679fcac75095fb1e9dc6d;p=gofetch.git

Fix subtitles and too much content in EreNumerique
---

diff --git a/src/be/nikiroo/gofetch/support/BasicSupport.java b/src/be/nikiroo/gofetch/support/BasicSupport.java
index 8fc259a..b15fac7 100644
--- a/src/be/nikiroo/gofetch/support/BasicSupport.java
+++ b/src/be/nikiroo/gofetch/support/BasicSupport.java
@@ -100,6 +100,17 @@ public abstract class BasicSupport {
 		 *         the usual automatic processing if not NULL
 		 */
 		public String manualProcessing(Node node);
+
+		/**
+		 * This {@link Node} is a subtitle and should be treated as such
+		 * (highlighted).
+		 * 
+		 * @param node
+		 *            the node to check
+		 * 
+		 * @return NULL if it is not a subtitle, the subtitle to use if it is
+		 */
+		public String isSubtitle(Node node);
 	}
 
 	/**
@@ -128,6 +139,11 @@ public abstract class BasicSupport {
 		public String manualProcessing(Node node) {
 			return null;
 		}
+
+		@Override
+		public String isSubtitle(Node node) {
+			return null;
+		}
 	}
 
 	static private String preselector;
@@ -334,6 +350,7 @@ public abstract class BasicSupport {
 					String manual = null;
 					boolean ignore = elementProcessor.ignoreNode(node)
 							|| ignoredNodes.contains(node.parentNode());
+					// Manual processing
 					if (!ignore) {
 						manual = elementProcessor.manualProcessing(node);
 						if (manual != null) {
@@ -342,6 +359,16 @@ public abstract class BasicSupport {
 						}
 					}
 
+					// Subtitle check
+					if (!ignore) {
+						String subtitle = elementProcessor.isSubtitle(node);
+						if (subtitle != null) {
+							subtitle = subtitle.trim();
+							currentLine.append("\n[ " + subtitle + " ]\n");
+							ignore = true;
+						}
+					}
+
 					if (ignore) {
 						ignoredNodes.add(node);
 						return;
diff --git a/src/be/nikiroo/gofetch/support/EreNumerique.java b/src/be/nikiroo/gofetch/support/EreNumerique.java
index bef677d..b6a7598 100644
--- a/src/be/nikiroo/gofetch/support/EreNumerique.java
+++ b/src/be/nikiroo/gofetch/support/EreNumerique.java
@@ -100,10 +100,29 @@ public class EreNumerique extends BasicSupport {
 		try {
 			Document doc = DataUtil.load(in, "UTF-8", url.toString());
 			Element article = doc.getElementsByTag("article").first();
+			if (article != null) {
+				article = article.getElementsByAttributeValue("itemprop",
+						"articleBody").first();
+			}
 			if (article != null) {
 				for (String line : toLines(article,
 						new BasicElementProcessor() {
-							// TODO: ignore headlines/pub
+							@Override
+							public boolean ignoreNode(Node node) {
+								return node.attr("class").contains("chapo");
+							}
+
+							@Override
+							public String isSubtitle(Node node) {
+								if (node instanceof Element) {
+									Element element = (Element) node;
+									if (element.tagName().startsWith("h")
+											&& element.tagName().length() == 2) {
+										return element.text();
+									}
+								}
+								return null;
+							}
 						})) {
 					fullContent += line + "\n";
 				}
diff --git a/src/be/nikiroo/gofetch/support/LeMonde.java b/src/be/nikiroo/gofetch/support/LeMonde.java
index 4ec2c30..235f7ee 100644
--- a/src/be/nikiroo/gofetch/support/LeMonde.java
+++ b/src/be/nikiroo/gofetch/support/LeMonde.java
@@ -98,11 +98,11 @@ public class LeMonde extends BasicSupport {
 				}
 
 				@Override
-				public String manualProcessing(Node node) {
+				public String isSubtitle(Node node) {
 					if (node instanceof Element) {
 						Element element = (Element) node;
 						if (element.hasClass("intertitre")) {
-							return "\n[ " + element.text() + " ]\n";
+							return element.text();
 						}
 					}
 					return null;