This is an automated email from the ASF dual-hosted git repository.
lewismc pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/nutch.git
The following commit(s) were added to refs/heads/master by this push:
new f1610723f NUTCH-3216 Fix remaining codespell findings flagged by Yetus
on master (#978)
f1610723f is described below
commit f1610723f016902628427e8361b8e0dd9128d794
Author: Lewis John McGibbney <[email protected]>
AuthorDate: Tue Oct 13 17:34:29 2026 -0700
NUTCH-3216 Fix remaining codespell findings flagged by Yetus on master
(#978)
---
.codespellrc | 8 +++++---
src/test/org/apache/nutch/metadata/TestMetadata.java | 6 +++---
src/test/org/apache/nutch/metadata/TestSpellCheckedMetadata.java | 6 +++---
src/test/org/apache/nutch/parse/parse-plugin-test.xml | 2 +-
src/test/org/apache/nutch/util/TestGZIPUtils.java | 4 ++--
src/test/org/apache/nutch/util/TestURLUtil.java | 6 ++++--
src/testresources/fetch-test-site/pageb.html | 2 +-
7 files changed, 19 insertions(+), 15 deletions(-)
diff --git a/.codespellrc b/.codespellrc
index 9c76c2d71..59200d840 100644
--- a/.codespellrc
+++ b/.codespellrc
@@ -4,9 +4,11 @@
# used to exercise SpellCheckedMetadata error-tolerant Location header
matching.
# uery: intentional query-string token in urlnormalizer-protocol
TestProtocolURLNormalizer.
# fo: intentional trie matcher test prefix in
PrefixStringMatcher/SuffixStringMatcher demo mains.
-ignore-words-list = AfterAll,BeforeAll,Loction,uery,fo
+# atLeast: Mockito verifier (org.mockito.Mockito.atLeast), not the phrase "at
least".
+ignore-words-list = AfterAll,BeforeAll,Loction,uery,fo,atLeast
# CHANGES.md is a historical changelog not maintained by this branch.
# Language-identifier corpora, binary license trees, and naivebayes training
# sample are also listed in .yetus/excludes.txt for full-tree Yetus runs; skip
-# them here so local codespell matches CI.
-skip =
CHANGES.md,src/plugin/language-identifier/src/test,LICENSE-binary,NOTICE-binary,licenses-binary,conf/naivebayes-train.txt.template,src/plugin/language-identifier/src/java/org/apache/nutch/analysis/lang/langmappings.properties,**/naivebayes-model,**/sample/**
+# them here so local codespell matches CI. Path globs must match the ./ prefix
+# that codespell produces when Yetus runs "codespell .".
+skip =
CHANGES.md,*src/plugin/language-identifier/src/test,LICENSE-binary,NOTICE-binary,licenses-binary,*conf/naivebayes-train.txt.template,*src/plugin/language-identifier/src/java/org/apache/nutch/analysis/lang/langmappings.properties,**/naivebayes-model,**/sample/**
diff --git a/src/test/org/apache/nutch/metadata/TestMetadata.java
b/src/test/org/apache/nutch/metadata/TestMetadata.java
index f73fd6cb9..4c314916d 100644
--- a/src/test/org/apache/nutch/metadata/TestMetadata.java
+++ b/src/test/org/apache/nutch/metadata/TestMetadata.java
@@ -267,16 +267,16 @@ public class TestMetadata {
}
private Metadata writeRead(Metadata meta) {
- Metadata readed = new Metadata();
+ Metadata readBack = new Metadata();
try {
ByteArrayOutputStream out = new ByteArrayOutputStream();
meta.write(new DataOutputStream(out));
- readed.readFields(new DataInputStream(new ByteArrayInputStream(out
+ readBack.readFields(new DataInputStream(new ByteArrayInputStream(out
.toByteArray())));
} catch (IOException ioe) {
fail(ioe.toString());
}
- return readed;
+ return readBack;
}
}
diff --git a/src/test/org/apache/nutch/metadata/TestSpellCheckedMetadata.java
b/src/test/org/apache/nutch/metadata/TestSpellCheckedMetadata.java
index c36f02139..62469e00e 100644
--- a/src/test/org/apache/nutch/metadata/TestSpellCheckedMetadata.java
+++ b/src/test/org/apache/nutch/metadata/TestSpellCheckedMetadata.java
@@ -355,16 +355,16 @@ public class TestSpellCheckedMetadata {
}
private SpellCheckedMetadata writeRead(SpellCheckedMetadata meta) {
- SpellCheckedMetadata readed = new SpellCheckedMetadata();
+ SpellCheckedMetadata readBack = new SpellCheckedMetadata();
try {
ByteArrayOutputStream out = new ByteArrayOutputStream();
meta.write(new DataOutputStream(out));
- readed.readFields(new DataInputStream(new ByteArrayInputStream(out
+ readBack.readFields(new DataInputStream(new ByteArrayInputStream(out
.toByteArray())));
} catch (IOException ioe) {
fail(ioe.toString());
}
- return readed;
+ return readBack;
}
/**
diff --git a/src/test/org/apache/nutch/parse/parse-plugin-test.xml
b/src/test/org/apache/nutch/parse/parse-plugin-test.xml
index 2eeb48391..1e3d34c38 100644
--- a/src/test/org/apache/nutch/parse/parse-plugin-test.xml
+++ b/src/test/org/apache/nutch/parse/parse-plugin-test.xml
@@ -30,7 +30,7 @@
<!-- test these 4 plugins -->
<mimeType name="text/html">
<!--
- ! Test that if a parser cannot be instanciated,
+ ! Test that if a parser cannot be instantiated,
! it should not block the process and then the next one is used
!-->
<plugin id="parse-plugin-that-not-exist"/>
diff --git a/src/test/org/apache/nutch/util/TestGZIPUtils.java
b/src/test/org/apache/nutch/util/TestGZIPUtils.java
index 129f7deec..592f2899f 100644
--- a/src/test/org/apache/nutch/util/TestGZIPUtils.java
+++ b/src/test/org/apache/nutch/util/TestGZIPUtils.java
@@ -28,10 +28,10 @@ import static org.junit.jupiter.api.Assertions.fail;
/** Unit tests for GZIPUtils methods. */
public class TestGZIPUtils {
- /* a short, highly compressable, string */
+ /* a short, highly compressible, string */
String SHORT_TEST_STRING =
"aaaaaaaaaaaaaaaabbbbbbbbbbbbbbbbbbbbbcccccccccccccccc";
- /* a short, highly compressable, string */
+ /* a short, highly compressible, string */
String LONGER_TEST_STRING = SHORT_TEST_STRING + SHORT_TEST_STRING
+ SHORT_TEST_STRING + SHORT_TEST_STRING + SHORT_TEST_STRING
+ SHORT_TEST_STRING + SHORT_TEST_STRING + SHORT_TEST_STRING
diff --git a/src/test/org/apache/nutch/util/TestURLUtil.java
b/src/test/org/apache/nutch/util/TestURLUtil.java
index 7c7267ce8..e2fbc5c50 100644
--- a/src/test/org/apache/nutch/util/TestURLUtil.java
+++ b/src/test/org/apache/nutch/util/TestURLUtil.java
@@ -139,10 +139,12 @@ public class TestURLUtil {
url = new URL("http://subdomain.example.edu.tr";);
assertEquals("edu.tr", URLUtil.getDomainSuffix(url));
- url = new URL("http://subdomain.example.presse.fr";);
+ url = new URL(
+ "http://subdomain.example.presse.fr";); // codespell:ignore presse
assertEquals("fr", URLUtil.getDomainSuffix(url));
- url = new URL("http://subdomain.example.presse.tr";);
+ url = new URL(
+ "http://subdomain.example.presse.tr";); // codespell:ignore presse
assertEquals("tr", URLUtil.getDomainSuffix(url));
// plc.co.im is listed as a domain suffix
diff --git a/src/testresources/fetch-test-site/pageb.html
b/src/testresources/fetch-test-site/pageb.html
index cf77ff4f7..d9018be55 100644
--- a/src/testresources/fetch-test-site/pageb.html
+++ b/src/testresources/fetch-test-site/pageb.html
@@ -16,7 +16,7 @@
-->
<html>
<head>
- <title>bage b</title>
+ <title>page b</title>
</head>
<body>
This is page b
