]> _ Git - cubeextranet.git/commitdiff
(no commit message)
authorvincent@cubedesigners.com <vincent@cubedesigners.com@f5622870-0f3c-0410-866d-9cb505b7a8ef>
Sat, 30 Apr 2011 17:27:35 +0000 (17:27 +0000)
committervincent@cubedesigners.com <vincent@cubedesigners.com@f5622870-0f3c-0410-866d-9cb505b7a8ef>
Sat, 30 Apr 2011 17:27:35 +0000 (17:27 +0000)
fluidbook/tools/fwstk/project_resources/org/apache/pdfbox/resources/LayoutStripper.properties [new file with mode: 0644]

diff --git a/fluidbook/tools/fwstk/project_resources/org/apache/pdfbox/resources/LayoutStripper.properties b/fluidbook/tools/fwstk/project_resources/org/apache/pdfbox/resources/LayoutStripper.properties
new file mode 100644 (file)
index 0000000..360519a
--- /dev/null
@@ -0,0 +1,97 @@
+# Licensed to the Apache Software Foundation (ASF) under one or more\r
+# contributor license agreements.  See the NOTICE file distributed with\r
+# this work for additional information regarding copyright ownership.\r
+# The ASF licenses this file to You under the Apache License, Version 2.0\r
+# (the "License"); you may not use this file except in compliance with\r
+# the License.  You may obtain a copy of the License at\r
+#\r
+#      http://www.apache.org/licenses/LICENSE-2.0\r
+#\r
+# Unless required by applicable law or agreed to in writing, software\r
+# distributed under the License is distributed on an "AS IS" BASIS,\r
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\r
+# See the License for the specific language governing permissions and\r
+# limitations under the License.\r
+\r
+# This table is maps PDF stream operators to concrete OperatorProcessor\r
+# subclasses that are used by the PDFStreamEngine class to interpret the\r
+# PDF document. The classes configured here allow the PDFTextStripper\r
+# subclass of PDFStreamEngine to extract text content of the document.\r
+\r
+BT = org.apache.pdfbox.util.operator.BeginText\r
+cm = org.apache.pdfbox.util.operator.Concatenate\r
+CS=org.apache.pdfbox.util.operator.SetStrokingColorSpace\r
+cs=org.apache.pdfbox.util.operator.SetNonStrokingColorSpace\r
+Do = org.apache.pdfbox.util.operator.Invoke\r
+ET = org.apache.pdfbox.util.operator.EndText\r
+gs = org.apache.pdfbox.util.operator.SetGraphicsStateParameters\r
+G=org.apache.pdfbox.util.operator.SetStrokingGrayColor\r
+g=org.apache.pdfbox.util.operator.SetNonStrokingGrayColor\r
+q  = org.apache.pdfbox.util.operator.GSave\r
+Q  = org.apache.pdfbox.util.operator.GRestore\r
+K=org.apache.pdfbox.util.operator.SetStrokingCMYKColor\r
+k=org.apache.pdfbox.util.operator.SetNonStrokingCMYKColor\r
+RG=org.apache.pdfbox.util.operator.SetStrokingRGBColor\r
+rg=org.apache.pdfbox.util.operator.SetNonStrokingRGBColor\r
+SC=org.apache.pdfbox.util.operator.SetStrokingColor\r
+sc=org.apache.pdfbox.util.operator.SetNonStrokingColor\r
+SCN=org.apache.pdfbox.util.operator.SetStrokingColor\r
+scn=org.apache.pdfbox.util.operator.SetNonStrokingColor\r
+T* = org.apache.pdfbox.util.operator.NextLine\r
+Tc = org.apache.pdfbox.util.operator.SetCharSpacing\r
+Td = org.apache.pdfbox.util.operator.MoveText\r
+TD = org.apache.pdfbox.util.operator.MoveTextSetLeading\r
+Tf = org.apache.pdfbox.util.operator.SetTextFont\r
+Tj = org.apache.pdfbox.util.operator.ShowText\r
+TJ = org.apache.pdfbox.util.operator.ShowTextGlyph\r
+TL = org.apache.pdfbox.util.operator.SetTextLeading\r
+Tm = org.apache.pdfbox.util.operator.SetMatrix\r
+Tr = org.apache.pdfbox.util.operator.SetTextRenderingMode\r
+Ts = org.apache.pdfbox.util.operator.SetTextRise\r
+Tw = org.apache.pdfbox.util.operator.SetWordSpacing\r
+Tz = org.apache.pdfbox.util.operator.SetHorizontalTextScaling\r
+w  = org.apache.pdfbox.util.operator.SetLineWidth\r
+\' = org.apache.pdfbox.util.operator.MoveAndShow\r
+\" = org.apache.pdfbox.util.operator.SetMoveAndShow\r
+\r
+# The following operators are not relevant to text extraction,\r
+# so we can silently ignore them.\r
+\r
+b\r
+B\r
+b*\r
+B*\r
+BDC\r
+BI\r
+BMC\r
+BX\r
+c\r
+d\r
+d0\r
+d1\r
+DP\r
+El\r
+EMC\r
+EX\r
+f\r
+F\r
+f*\r
+h\r
+i\r
+ID\r
+j\r
+J\r
+l\r
+m\r
+M\r
+MP\r
+n\r
+re\r
+ri\r
+s\r
+S\r
+sh\r
+v\r
+W\r
+W*\r
+y\r