From 8e4390f20fb400a0a39853877c8fbea48b22ee25 Mon Sep 17 00:00:00 2001
From: Long Pham <ltpham3@illinois.edu>
Date: Thu, 24 May 2018 13:40:09 -0500
Subject: [PATCH] add another try catch

---
 .idea/workspace.xml                           | 35 +++++++++----------
 .../HTMLDocumentIngestionManager.java         | 14 ++++++--
 2 files changed, 27 insertions(+), 22 deletions(-)

diff --git a/.idea/workspace.xml b/.idea/workspace.xml
index 2befe4a..ad17103 100644
--- a/.idea/workspace.xml
+++ b/.idea/workspace.xml
@@ -2,10 +2,7 @@
 <project version="4">
   <component name="ChangeListManager">
     <list default="true" id="bbbbc773-7714-4942-be21-383885d0c5dc" name="Default" comment="">
-      <change beforePath="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestReadObject.java" beforeDir="false" afterPath="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestReadObject.java" afterDir="false" />
-      <change beforePath="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestSelenium.java" beforeDir="false" afterPath="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestSelenium.java" afterDir="false" />
       <change beforePath="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java" beforeDir="false" afterPath="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java" afterDir="false" />
-      <change beforePath="$PROJECT_DIR$/test_urls.csv" beforeDir="false" afterPath="$PROJECT_DIR$/test_urls.csv" afterDir="false" />
     </list>
     <ignored path="$PROJECT_DIR$/out/" />
     <ignored path="$PROJECT_DIR$/target/" />
@@ -32,11 +29,11 @@
           </provider>
         </entry>
       </file>
-      <file leaf-file-name="HTMLDocumentIngestionManager.java" pinned="false" current-in-tab="false">
+      <file leaf-file-name="HTMLDocumentIngestionManager.java" pinned="false" current-in-tab="true">
         <entry file="file://$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java">
           <provider selected="true" editor-type-id="text-editor">
-            <state relative-caret-position="156">
-              <caret line="299" column="48" lean-forward="true" selection-start-line="299" selection-start-column="48" selection-end-line="299" selection-end-column="48" />
+            <state relative-caret-position="213">
+              <caret line="350" column="66" selection-start-line="350" selection-start-column="66" selection-end-line="350" selection-end-column="66" />
               <folding>
                 <element signature="imports" expanded="true" />
               </folding>
@@ -103,11 +100,11 @@
           </provider>
         </entry>
       </file>
-      <file leaf-file-name="TestReadObject.java" pinned="false" current-in-tab="true">
+      <file leaf-file-name="TestReadObject.java" pinned="false" current-in-tab="false">
         <entry file="file://$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestReadObject.java">
           <provider selected="true" editor-type-id="text-editor">
-            <state relative-caret-position="240">
-              <caret line="16" column="51" selection-start-line="16" selection-start-column="51" selection-end-line="16" selection-end-column="51" />
+            <state relative-caret-position="135">
+              <caret line="9" lean-forward="true" selection-start-line="9" selection-end-line="9" />
               <folding>
                 <element signature="imports" expanded="true" />
               </folding>
@@ -226,8 +223,8 @@
         <option value="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ner/annotation/AnnotatorFactory.java" />
         <option value="$PROJECT_DIR$/pom.xml" />
         <option value="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestSelenium.java" />
-        <option value="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java" />
         <option value="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestReadObject.java" />
+        <option value="$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java" />
       </list>
     </option>
   </component>
@@ -429,8 +426,8 @@
     <recent_temporary>
       <list>
         <item itemvalue="Application.TestReadObject" />
-        <item itemvalue="Application.TestSelenium (1)" />
         <item itemvalue="Application.TestSelenium" />
+        <item itemvalue="Application.TestSelenium (1)" />
         <item itemvalue="Application.BasicPipelineExample" />
         <item itemvalue="Application.PipelineHelper" />
       </list>
@@ -455,12 +452,12 @@
       <workItem from="1526678891338" duration="5160000" />
       <workItem from="1526698295267" duration="21787000" />
       <workItem from="1526919640549" duration="22246000" />
-      <workItem from="1527016798854" duration="31029000" />
+      <workItem from="1527016798854" duration="32003000" />
     </task>
     <servers />
   </component>
   <component name="TimeTrackingManager">
-    <option name="totallyTimeSpent" value="259044000" />
+    <option name="totallyTimeSpent" value="260018000" />
   </component>
   <component name="ToolWindowManager">
     <frame x="6" y="23" width="1421" height="839" extended-state="0" />
@@ -932,20 +929,20 @@
         </state>
       </provider>
     </entry>
-    <entry file="file://$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java">
+    <entry file="file://$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestReadObject.java">
       <provider selected="true" editor-type-id="text-editor">
-        <state relative-caret-position="156">
-          <caret line="299" column="48" lean-forward="true" selection-start-line="299" selection-start-column="48" selection-end-line="299" selection-end-column="48" />
+        <state relative-caret-position="135">
+          <caret line="9" lean-forward="true" selection-start-line="9" selection-end-line="9" />
           <folding>
             <element signature="imports" expanded="true" />
           </folding>
         </state>
       </provider>
     </entry>
-    <entry file="file://$PROJECT_DIR$/src/main/java/org/forward/entitysearch/experiment/TestReadObject.java">
+    <entry file="file://$PROJECT_DIR$/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java">
       <provider selected="true" editor-type-id="text-editor">
-        <state relative-caret-position="240">
-          <caret line="16" column="51" selection-start-line="16" selection-start-column="51" selection-end-line="16" selection-end-column="51" />
+        <state relative-caret-position="213">
+          <caret line="350" column="66" selection-start-line="350" selection-start-column="66" selection-end-line="350" selection-end-column="66" />
           <folding>
             <element signature="imports" expanded="true" />
           </folding>
diff --git a/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java b/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java
index ff95660..06f7f48 100644
--- a/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java
+++ b/src/main/java/org/forward/entitysearch/ingestion/HTMLDocumentIngestionManager.java
@@ -311,7 +311,15 @@ public static void main(String[] args) {
 
             System.out.println(filename + "\t" +  baseUrl);
 
-            ESAnnotatedHTMLDocument document = getHTMLDocumentForAnnotation(baseUrl, driver);
+            ESAnnotatedHTMLDocument document = null;
+            try {
+                document = getHTMLDocumentForAnnotation(baseUrl, driver);
+            } catch(Exception ex) {
+                System.out.println("There is some exception when parsing the document ");
+                System.err.println("There is some exception when parsing the document in this URL: " + baseUrl);
+                System.err.println(ex.getClass());
+                continue;
+            }
             time = System.currentTimeMillis();
             System.out.println("Finish creating document for annotation " + (time-start)/1000 + " seconds");
             start = time;
@@ -339,8 +347,8 @@ public static void main(String[] args) {
             try {
                 AnnotatorFactory.getInstance().getAnnotationPipeline().annotate(document);
             } catch(StaleElementReferenceException ex) {
-                System.out.println("Some dangling nodes are found in this URL");
-                System.err.println("Some dangling nodes are found in this URL: " + baseUrl);
+                System.out.println("There is an exception when annotating this document");
+                System.err.println("There is an exception when the document in this URL: " + baseUrl);
                 continue;
             }