annotate backend/src/main/java/org/dive4elements/river/importer/common/AbstractParser.java @ 8989:2693bfaf503d

Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
author mschaefer
date Mon, 09 Apr 2018 09:07:00 +0200
parents 50416a0df385
children c43d8c1a4455
rev   line source
8971
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
1 /* Copyright (C) 2017 by Bundesanstalt für Gewässerkunde
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
2 * Software engineering by
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
3 * Björnsen Beratende Ingenieure GmbH
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
4 * Dr. Schumacher Ingenieurbüro für Wasser und Umwelt
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
5 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
6 * This file is Free Software under the GNU AGPL (>=v3)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
7 * and comes with ABSOLUTELY NO WARRANTY! Check out the
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
8 * documentation coming with Dive4Elements River for details.
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
9 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
10
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
11 package org.dive4elements.river.importer.common;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
12
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
13 import java.io.File;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
14 import java.io.FileInputStream;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
15 import java.io.FilenameFilter;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
16 import java.io.IOException;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
17 import java.io.InputStreamReader;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
18 import java.io.LineNumberReader;
8989
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
19 import java.math.BigDecimal;
8971
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
20 import java.text.NumberFormat;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
21 import java.text.ParseException;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
22 import java.util.ArrayList;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
23 import java.util.List;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
24 import java.util.Locale;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
25 import java.util.TreeSet;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
26 import java.util.regex.Matcher;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
27 import java.util.regex.Pattern;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
28
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
29 import org.apache.log4j.Logger;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
30 import org.dive4elements.river.backend.utils.EpsilonComparator;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
31 import org.dive4elements.river.importer.ImportRiver;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
32
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
33 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
34 * Abstract base class for a parser of one FLYS csv data file.<br />
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
35 * The {@link parse} method creates a SERIES object for the meta data
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
36 * and a list of KMLINE objects for the km value lines read from the file.<br />
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
37 * The {@link store} method gets or creates the corresponding database objects
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
38 * by the hibernate binding classes DB_SERIES and DB_KMTUPLE,
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
39 * and updates or inserts them in the database.
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
40 * DB_SERIES has a one-to-many relationship with DB_KMTUPLE.<br />
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
41 * <br />
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
42 * The structure of the file is as follows:<br />
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
43 * <ul>
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
44 * <li>one or more comment lines (#) with the meta info of the data series</li>
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
45 * <li>the comment line with the column titles of values table, starting with the km column</li>
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
46 * <li>the rows of the values table, each one on its own line</li>
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
47 * </ul>
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
48 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
49 * @author Matthias Schäfer
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
50 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
51 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
52 public abstract class AbstractParser<DB_SERIES, DB_KMTUPLE, KMLINE extends AbstractKmLineImport<DB_SERIES, DB_KMTUPLE>, HEADER extends AbstractSeriesImport<DB_SERIES, DB_KMTUPLE, KMLINE>> implements ImportParser {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
53
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
54 /***** FIELDS *****/
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
55
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
56 public static final String ENCODING = "ISO-8859-1";
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
57
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
58 protected static final Locale DEFAULT_LOCALE = Locale.GERMAN;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
59
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
60 public static final String START_META_CHAR = "#";
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
61
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
62 protected static final String SEPARATOR_CHAR = ";";
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
63
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
64 protected static final Pattern META_RIVERNAME = Pattern.compile("^#\\s*((Gew.sser)|(Gewaesser)):\\s*(\\S[^;]*).*", Pattern.CASE_INSENSITIVE);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
65
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
66 protected static final Pattern META_KMRANGE_INFO = Pattern.compile("^#\\s*Strecke:\\s*(\\S[^;]*).*", Pattern.CASE_INSENSITIVE);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
67
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
68 protected static final Pattern META_COMMENTS = Pattern.compile("^#\\s*weitere Bemerkungen:\\s*(\\S[^;]*).*", Pattern.CASE_INSENSITIVE);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
69
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
70 private static final Pattern META_COLUMNTITLES = Pattern.compile("^#*\\s*Fluss.km\\s*;.+", Pattern.CASE_INSENSITIVE);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
71
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
72 private static final Pattern META_SUBGROUP = Pattern.compile("^##.*", Pattern.CASE_INSENSITIVE);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
73
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
74 private static NumberFormat numberFormat = NumberFormat.getInstance(Locale.ROOT);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
75
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
76 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
77 * Path of the file or directory to import from
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
78 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
79 protected final File importPath;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
80
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
81 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
82 * Part of {@link importPath} without the river root dir
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
83 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
84 protected final File rootRelativePath;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
85
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
86 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
87 * River for which the import runs
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
88 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
89 protected final ImportRiver river;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
90
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
91 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
92 * Reader during parse
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
93 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
94 protected LineNumberReader in;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
95
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
96 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
97 * Last line read from in
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
98 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
99 protected String currentLine;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
100
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
101 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
102 * State of the header lines parse loop
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
103 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
104 protected ParsingState headerParsingState;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
105
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
106 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
107 * Series header of the stations table, with the imported meta info.
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
108 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
109 protected HEADER seriesHeader;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
110
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
111 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
112 * List of meta info Pattern matched during {@link handleMetaLine}
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
113 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
114 protected final List<Pattern> metaPatternsMatched;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
115
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
116 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
117 * Column titles of the stations table, starting with the km column.
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
118 * All strings have been trimmed.
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
119 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
120 protected final List<String> columnTitles;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
121
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
122 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
123 * List of the km value tuples imported, no duplicate km
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
124 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
125 protected final List<KMLINE> values;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
126
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
127 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
128 * Ordered list with the imported km to check for duplicates.
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
129 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
130 protected final TreeSet<Double> kmExists;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
131
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
132
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
133 /***** CONSTRUCTORS *****/
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
134
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
135 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
136 * Constructs a parser for an import file
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
137 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
138 public AbstractParser(final File importPath, final File rootRelativePath, final ImportRiver river) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
139 this.importPath = importPath;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
140 this.rootRelativePath = rootRelativePath;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
141 this.river = river;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
142 this.metaPatternsMatched = new ArrayList<>();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
143 this.kmExists = new TreeSet<>(EpsilonComparator.CMP);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
144 this.columnTitles = new ArrayList<>();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
145 this.values = new ArrayList<>();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
146 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
147
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
148
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
149 /***** METHODS *****/
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
150
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
151 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
152 * Lists all files from a directory having a type extension (starting with dot)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
153 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
154 protected static List<File> listFiles(final File importDir, final String extension) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
155 final File[] files = importDir.listFiles(new FilenameFilter() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
156 @Override
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
157 public boolean accept(final File dir, final String name) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
158 return name.toLowerCase().endsWith(extension);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
159 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
160 });
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
161 final List<File> fl = new ArrayList<>();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
162 if (files != null)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
163 for (final File file : files)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
164 fl.add(file);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
165 return fl;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
166 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
167
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
168 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
169 * Parses a file and adds series and values to the parser's collection
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
170 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
171 @Override
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
172 public void parse() throws IOException {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
173 logStartInfo();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
174 this.seriesHeader = createSeriesImport(this.importPath.getName().replaceAll("\\.csv", ""));
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
175 this.metaPatternsMatched.clear();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
176 this.kmExists.clear();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
177 this.headerParsingState = ParsingState.CONTINUE;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
178 try {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
179 try {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
180 this.in = new LineNumberReader(new InputStreamReader(new FileInputStream(this.importPath), ENCODING));
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
181 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
182 catch (final Exception e) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
183 logError("Could not open (" + e.getMessage() + ")");
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
184 this.headerParsingState = ParsingState.STOP;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
185 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
186 this.currentLine = null;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
187 while (this.headerParsingState != ParsingState.STOP) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
188 this.currentLine = this.in.readLine();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
189 if (this.currentLine == null)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
190 break;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
191 this.currentLine = this.currentLine.trim();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
192 if (this.currentLine.isEmpty())
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
193 continue;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
194 if (this.headerParsingState == ParsingState.CONTINUE)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
195 handleMetaLine();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
196 else
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
197 handleDataLine();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
198 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
199 if (this.headerParsingState != ParsingState.STOP)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
200 getLog().info("Number of values found: " + this.seriesHeader.getValueCount());
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
201 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
202 finally {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
203 if (this.in != null) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
204 this.in.close();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
205 this.in = null;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
206 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
207 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
208 if (this.headerParsingState == ParsingState.STOP)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
209 logError("Parsing of the file stopped due to a severe error");
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
210 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
211
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
212 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
213 * Writes the parse start info to the log
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
214 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
215 protected void logStartInfo() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
216 getLog().info("Start parsing:;'" + this.rootRelativePath + "'");
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
217 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
218
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
219 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
220 * Stores the parsed series and values in the database
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
221 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
222 @Override
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
223 public void store() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
224 if (this.headerParsingState != ParsingState.STOP) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
225 this.seriesHeader.store(this.river.getPeer());
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
226 final String counts = String.format("parse=%d, insert=%d, update/ignore=%d", this.seriesHeader.getValueCount(),
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
227 this.seriesHeader.getValueStoreCount(StoreMode.INSERT), this.seriesHeader.getValueStoreCount(StoreMode.UPDATE));
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
228 if (this.seriesHeader.getValueCount() > this.seriesHeader.getValueStoreCount(StoreMode.INSERT))
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
229 logWarning("Number of value inserts less than number parsed: " + counts);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
230 else
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
231 getLog().info("Number of values records: " + counts);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
232 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
233 else
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
234 logWarning("Severe parsing errors, not storing series '" + this.seriesHeader.getFilename() + "'");
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
235 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
236
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
237 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
238 * Strips separator chars from a meta info text, and trims leading and trailing whitespace
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
239 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
240 public static String parseMetaInfo(final String text) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
241 return text.replace(SEPARATOR_CHAR, "").trim();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
242 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
243
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
244 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
245 * Parses a number string with dot or comma as decimal char, and returning null in case of an error
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
246 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
247 public static Number parseDoubleWithNull(final String text) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
248 try {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
249 return parseDouble(text);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
250 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
251 catch (final Exception e) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
252 return null;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
253 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
254 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
255
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
256 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
257 * Parses a number string with dot or comma as decimal char
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
258 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
259 * @throws ParseException
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
260 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
261 public static Number parseDouble(final String text) throws ParseException {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
262 return numberFormat.parse(text.replace(',', '.'));
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
263 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
264
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
265 /**
8989
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
266 * Parses a number string as a BigDecimal, replacing a comma with a dot first
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
267 */
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
268 public static BigDecimal parseDecimal(final String text) throws NumberFormatException {
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
269 return new BigDecimal(text.replace(',', '.'));
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
270 }
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
271
2693bfaf503d Fixed several BigDecimal(double) creations by BigDecimal(String) parsing to avoid unnecessary decimal digits
mschaefer
parents: 8971
diff changeset
272 /**
8971
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
273 * Gets the class's logger
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
274 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
275 protected abstract Logger getLog();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
276
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
277 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
278 * Logs an error message, appending the relative file path
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
279 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
280 protected void logError(final String message) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
281 getLog().error(message + ";" + this.rootRelativePath);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
282 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
283
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
284 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
285 * Logs a warning message, appending the relative file path
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
286 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
287 protected void logWarning(final String message) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
288 getLog().warn(message + ";" + this.rootRelativePath);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
289 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
290
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
291 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
292 * Creates a new series import object
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
293 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
294 protected abstract HEADER createSeriesImport(final String filename);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
295
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
296 protected void handleMetaLine() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
297 if (META_SUBGROUP.matcher(this.currentLine).matches())
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
298 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
299 else if (handleMetaRivername())
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
300 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
301 else if (handleMetaKmrange_info())
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
302 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
303 else if (handleMetaComment())
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
304 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
305 else if (handleMetaOther())
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
306 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
307 else if (handleMetaColumnTitles()) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
308 if (this.headerParsingState != ParsingState.STOP)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
309 this.headerParsingState = ParsingState.DONE;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
310 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
311 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
312 else {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
313 if (this.currentLine.startsWith(START_META_CHAR)) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
314 if (this.headerParsingState != ParsingState.IGNORE)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
315 logWarning("Not matching any known meta type in line " + this.in.getLineNumber() + ", ignored");
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
316 else
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
317 this.headerParsingState = ParsingState.CONTINUE;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
318 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
319 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
320 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
321
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
322 private boolean handleMetaRivername() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
323 if (META_RIVERNAME.matcher(this.currentLine).matches()) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
324 this.metaPatternsMatched.add(META_RIVERNAME);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
325 return true;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
326 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
327 else
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
328 return false;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
329 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
330
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
331 private boolean handleMetaKmrange_info() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
332 final Matcher m = META_KMRANGE_INFO.matcher(this.currentLine);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
333 if (m.matches()) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
334 this.metaPatternsMatched.add(META_KMRANGE_INFO);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
335 this.seriesHeader.setKmrange_info(parseMetaInfo(m.group(1)));
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
336 return true;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
337 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
338 return false;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
339 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
340
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
341 private boolean handleMetaComment() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
342 final Matcher m = META_COMMENTS.matcher(this.currentLine);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
343 if (m.matches()) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
344 this.metaPatternsMatched.add(META_COMMENTS);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
345 this.seriesHeader.setComment(parseMetaInfo(m.group(1)));
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
346 return true;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
347 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
348 return false;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
349 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
350
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
351 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
352 * Parses currentLine for non-default meta info
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
353 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
354 * @return Whether the line has been handled
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
355 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
356 protected boolean handleMetaOther() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
357 return false;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
358 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
359
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
360 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
361 * Parses a header line for the km table column header line
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
362 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
363 * @return Whether the line has been handled and we are ready for reading the km values lines
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
364 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
365 protected boolean handleMetaColumnTitles() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
366 if (META_COLUMNTITLES.matcher(this.currentLine).matches()) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
367 this.metaPatternsMatched.add(META_COLUMNTITLES);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
368 this.columnTitles.clear();
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
369 final String[] titles = this.currentLine.split(SEPARATOR_CHAR, 0);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
370 for (int i = 0; i <= titles.length - 1; i++)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
371 this.columnTitles.add(titles[i].trim());
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
372 return true;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
373 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
374 return false;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
375 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
376
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
377 private void handleDataLine() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
378 final String[] values = this.currentLine.split(SEPARATOR_CHAR, 0);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
379 // Skip import line without data or only km
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
380 if (values.length < 2)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
381 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
382 Double km;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
383 try {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
384 km = Double.valueOf(parseDouble(values[0]).doubleValue());
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
385 if (kmMustBeUnique()) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
386 if (this.kmExists.contains(km)) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
387 logWarning("Ignoring duplicate station '" + values[0] + "' in line " + this.in.getLineNumber());
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
388 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
389 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
390 this.kmExists.add(km);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
391 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
392 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
393 catch (final Exception e) {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
394 logError("Not parseable km in line " + this.in.getLineNumber() + ": " + e.getMessage());
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
395 return;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
396 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
397 final KMLINE value = createKmLineImport(km, values);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
398 if (value != null)
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
399 this.seriesHeader.addValue(value);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
400 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
401
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
402 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
403 * Whether {@link handleDataLine} shall check for and reject km duplicates
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
404 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
405 protected boolean kmMustBeUnique() {
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
406 return true;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
407 }
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
408
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
409 /**
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
410 * Creates a value import item with the km and other fields of the current line;
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
411 * the km has been validated
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
412 *
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
413 * @return value item, or null if parse error
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
414 */
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
415 protected abstract KMLINE createKmLineImport(final Double km, final String[] values);
50416a0df385 Importer for the Schifffahrt (S-INFO) and Oekologie (U-INFO) files
mschaefer
parents:
diff changeset
416 }

http://dive4elements.wald.intevation.org