Tutorial 6 - More complex noise reduction
This tutoral is a continuation of the Tutorial 5.
This tutorial will demonstrate how to use Archery to extract data from one Excel spreadsheet. The document is not well-formed and very noisy. To demonstrate the usage of this framework, we will load a document with a somewhat complex layout, as seen here:
Setup Archery
Import the packages and setup the main class:
package com.github.romualdrousseau.archery.examples;
import java.util.EnumSet;
import java.util.List;
import com.github.romualdrousseau.archery.Document;
import com.github.romualdrousseau.archery.DocumentFactory;
import com.github.romualdrousseau.archery.parser.LayexTableParser;
public class Tutorial6 implements Runnable {
public Tutorial6() {
}
@Override
public void run() {
// Code will come here
}
public static void main(final String[] args) {
new Tutorial6().run();
}
}
pom.xml
Archery has a very modular design where each functionality can be loaded separatly. We add the "archery-net-classifier" module to enable the tagging capabilities. This module use TensorFlow for Java. The following depedencies are required to run the code of this tutorial:
<!-- Archery Framework -->
<dependency>
<groupId>com.github.romualdrousseau</groupId>
<artifactId>archery</artifactId>
<version>${archery.version}</version>
</dependency>
<dependency>
<groupId>com.github.romualdrousseau</groupId>
<artifactId>archery-layex-parser</artifactId>
<version>${archery.version}</version>
</dependency>
<dependency>
<groupId>com.github.romualdrousseau</groupId>
<artifactId>archery-net-classifier</artifactId>
<version>${archery.version}</version>
</dependency>
<dependency>
<groupId>com.github.romualdrousseau</groupId>
<artifactId>archery-csv</artifactId>
<version>${archery.version}</version>
</dependency>
<dependency>
<groupId>com.github.romualdrousseau</groupId>
<artifactId>archery-excel</artifactId>
<version>${archery.version}</version>
</dependency>
Load base model
To parse a document, Archery needs a model that will contains the parameters required to the parsing. Instead to start from an empty Model (See Tutorial 10), we will start from an existing one and we will adapt it for our document. You can find a list and details of all models here.
The base model, we will use, is "sales-english" that has been trained on 200+ english documents containing distributor data and with a large range of different layouts.
The base model already recognize some entities such as DATE and NUMBER. We will configure a layex to extract the different elements of the documents. You can find more details about layex here.
final var model = Common.loadModelFromGitHub("sales-english");
// Add a layex to the model
final var tableParser = new LayexTableParser(
List.of("(v.$)+"),
List.of("(()(E+$E+$))(()(/^PRODUCTCODE/.+$)*(/PRODUCTCODE/.+$))+()"));
model.registerTableParser(tableParser);
Load the document
We load the document by creating a document instance with the model. The hint "Document.Hint.INTELLI_LAYOUT" tells the document instance that the document has a complex layout. WThe recipe "sheet.setCapillarityThreshold(1.5)" tells the parser engine to extract the features as big as possible. The recipe "sheet.setDataTableParserFactory(\"DataTableGroupSubFooterParserFactory\")" tells to that the footer continas info we are interrested in. Finallym the recipe "sheet.dropRowsWhenFillRatioLessThan(0.2)" tells to cleanup almost empty rows:
final var file = Common.loadData("document with noises.xls", this.getClass());
try (final var doc = DocumentFactory.createInstance(file, "UTF-8")
.setModel(model)
.setHints(EnumSet.of(Document.Hint.INTELLI_LAYOUT))
.setRecipe(
"sheet.setCapillarityThreshold(1.5)",
"sheet.setDataTableParserFactory(\"DataTableGroupSubFooterParserFactory\")",
"sheet.dropRowsWhenFillRatioLessThan(0.2)")) {
...
}
Output the tabular result
Finally, we iterate over the sheets, rows and cells and output the data on the console:
doc.sheets().forEach(s -> Common.addSheetDebugger(s).getTable().ifPresent(t -> {
Common.printHeaders(t.headers());
Common.printRows(t.rows());
}));
2024-03-13 09:51:56 INFO Common:43 - Loaded model: sales-english
2024-03-13 09:51:56 INFO Common:60 - Loaded resource: /data/document with noises.xls
2024-03-13 09:51:58 DEBUG Common:87 - Extracting features ...
2024-03-13 09:51:58 DEBUG Common:91 - Generating Layout Graph ...
2024-03-13 09:51:58 DEBUG Common:95 - Assembling Tabular Output ...
============================== DUMP GRAPH ===============================
MSRVTA.rpt
|- Item no. Description Start stock Qty rec. Start +rec. Qty sold Month sales Sampl/exp. Dam./corr. Closing stock Qty on order Ytd qty Ytd sales PRODUCTCODE #GROUP? DATA(0, 0, 11, 18, 19, 17) (1)
================================== END ==================================
2024-03-13 09:51:59 DEBUG Common:100 - Done.
Item no. Description Start stock Qty rec. Start +rec. Qty sold Month sales Sampl/exp. Dam./ Closing stock Qty on order Ytd qty Ytd sales PRODUCTCODE #GRO
10255 AAA 3 3 URSR0009
10143 BBB 10 10 URSR0014
10203 CCC 7 7 URSR0014
10209 DDD 16 16 URSR0014
10211 EEE 4 4 URSR0014
10248 FFF 40 40 URSR0014
10197 BBB 8 8 URSR0015
10200 CCC 20 20 URSR0015
10202 DDD 8 8 URSR0015
10248 EEE 14 14 URSR0015
10255 FFF 10 10 URSR0015
On this output, we print out the graph of the document built during the parsing and we can see clearly the relation between the elements of the spreadsheet and how there are structured in tabular form.
Conclusion
Congratulations! You have loaded documents using Archery.
For more examples of using Archery, check out the tutorials.